You and Adrian from Adrian’s digital basement both go with actual troubleshooting before going whole hog swapping parts. I think that’s part of why I enjoy both of your channels.
I absolutely loved hanging out with Adrian during VCF East. He was one of those people that you meet and within 10 minutes you immediately know that you're gonna be best friends! He does really great work, and I love his style of digging deep into a broken system and hunting down exactly what's wrong and why.
Doing things the quick and easy way is all well and good if you are doing something as a job and need to be quick. Doing things the slow and hard way teaches you the hows and whys, and is often more rewarding if you are doing something as a hobby. That's my take at least. Edit: Oh, and Adrian's great. Adrian's Digital Basement, Usagi Electric, and Ben Eater are the triumvirate of retro/hobby electronics on youtube if you ask me. There's other greats, like James Sharman, Curious Marc, GreatScott!, Zack Freedman, Slu4, and many, many more, but Adrian, David, and Ben are who I consider the greatest personally for their combination of sheer shining love for what they do, warmth, ability to present information in an interesting, engaging, and understandable manner, and for feeling approachable in a way that's usually hard to feel through a screen. I'd include Daniel Shiffman of The Coding Train among them, but since his wheelhouse is solely and solidly in programming and has nothing to do with hardware or anything retro I will leave the triumvirate as it is, though he does get this honorable mention.
Important note on the CVT (constant voltage transformer) is that it handles sharp increases in current draw nicely due to the "Tank circuit", which can beneficial with variable load systems.
Good video... Thank you for explaining the Ferroresonant transformer linear power supply design and reason for the Ferroresonant design. Basically the Ferroresonant tank circuit is set to resonate at 60Hz. The stored excess energy in the tank circuit from the primary winding does two things, smooths the AC power into the secondary windings and can even dump power into the secondary windings of the primary input power inline drops a few power cycles, sages or spikes. Back in the 1970 when Centurion switched to a ferroresonat power supply design a company called "Sola" sold (still sells) a AC power line conditioner to place between the AC line source and high cost electronic equipment. The basic Sola product is metal box with line cord, ferroresonant transformer, tank capacitor and outlet socket that sold for around $800 in the 1970's. Warrex-Centurion did not need to have the customer buy a Sola conditioner, we built it into our supply.
Reminds me of the power factor correction systems saw in the late 70s One we had in a grocery store was AC-AC so it corrected the line in. The cash registers were digital and real sensitive enough that compressors jumping on and off the AC would knock out the register. Thanks for the added info
I have one of those Sola conditioners. It helped me find the cause of prematurely failing compact fluorescent lights in my house some years ago. Or, technically that conditioned power gave greatly improved (X10) lamp lifetime.
Ken, I want to thank you for your involvement in the Centurion project. Time and time again, you, David, and several others working on this have hit and overcome obstacles with your massive combined brainpower. All of you comprise what is certainly the preeminent authority on Centurion today. That said, you in particular came along at a time when the Centurion was a pile of dusty blue boxes. Without you, the project would certainly be nowhere close to the state it's in now, and perhaps derailed entirely. Your knowledge, resources and experience working on the Centurion architecture arrived decades later, exactly where it was needed. It has been immensely rewarding, learning all of this about a machine that was all but unknown in recent times, and would have fallen into of those black holes in history if not for you. I know that others have contributed a lot but you were the first, and without your particular expertise, further progress would have been problematic at best. Not to understate the significance of David's work and that of the others, you are very likely the one person on Earth whose involvement was essential. I do not really have the words to express my own gratitude, except to thank you again Ken. But don't hang up the probes just yet. There is a lot more to do!
@@Starchface Thank you. Getting the Centurion running and working to save Warrex-Centurion's little place in minicomputer history has been a team project of 10's of people from around the world. The best part is the young engineers may have learned a little about how we did things in the past and why something are as they are day.
Nice work! Just wrapped up a power supply marathon myself. It's pretty satisfying to chase the signals and find that one component that's causing the trouble - then replace it. FWIW, I'm with you on NOT wholesale recapping. Every component replaced is a potential mishap. If it ain't broke, don't break it.
I second that! I may feel different if i were repairing for a customer, but since you can fix it again if you have another cap fail, if it ain't broke, leave it alone!
IMO it depends on the device and the capacitors. If it is difficult to take the device apart and would be annoying if it failed again, I just recap it if some caps are bad. Better than taking it apart 10 times, each time replacing a single cap. Similar if I use the device all the time - I do not want too much downtime here. Also, if the capacitors used are known for going bad (especially if they tend to short, which can blow up other components) then I just recap it. For example, Revox A77 tape deck uses caps that like to short and some of them are rather difficult to get to - recap. Old stuff made in the USSR usually has a lot of bad capacitors, but by now those that were going bad, did go bad, so replace them and others with the same part number. Other part number may have been manufactured better (if none of them are bad by now). There was a time when you could get a tape deck repaired and then another cap would fail in a week. Bad times for young-me who did not know how to replace capacitors at the time, there was pretty much always something wrong with my tape deck. Vacuum tube stuff - replace all paper caps, most of them are leaky and others will become leaky (and testing for leakage means I have to desolder one leg of the cap anyway). I have one tape recorder where I had to replace the paper caps because of leakage, but all electrolytics made in 1968 still work. PC motherboard - if the caps start going bad, I just replace them, because taking apart a server to get to the motherboard is super annoying and the cap can fail at some inconvenient time leaving me with a badly working server for a while until I find time to replace the cap. Same with a PC power supply. If repairing something for someone else and it has a few bad caps - I tend to do a full recap if it's not expensive, since I do not want the device to fail again for that person.
I agree that it is better to recap when it doubt. If one has gone bad, the others are most likely not far behind. I also agree that its good to learn the troubleshooting techniques and it was worth making the schematic but electrolytic capacitors are far more likely to fail than other components so one _should_ treat them differently
I think you might want to sleeve those pins that you put in the connector. I would be terrified that one will work it's way partially out and short, in tern breaking the power supply you just fixed.
Good explanation. I've used CVTs (Ferroresonant Transformers) in offshore installations to regulate 240VAC for sensitive equipment. A nice property is that it also removes noise and spikes on the input due to the secondaries running in saturation and with the tank.
We had ferroresonant transformers in the Navy on the subs I rode. We called them 'sola transformers' because of the brand name. Or 'voltage regulating' as you describe, they fed a steady voltage to some electronics despite fluctuations on the power supplied by ship's generators. Yeah, we were taught that shunt had a critical airgap that was the 'magic' part of how these things work. Pretty old technology, but then when I was in we used magnetic-amplifiers (mag-amps) for a lot of things, so....
Navy loves magnetic magic, I saw servo loop running almost bow to stern and across several decks, where resolver fed a motor-generator used as mag-amp to power giant AC motor with a gearbox whos output shaft had a synchro that closed the servo loop by spinning the resolver's rotor. I stand by my believe that designer decided to use ALL magnetic devices he knew about in the circuit just because they could.
Aircraft still rely on magamps these days, though often these feed almost immediately into a synchro to digital converter in the avionics computers, and then back out to a digital to synchro converter for the those instruments that are still not screens. the standby ones that do not use this are entirely either self powered, or have 115VAC supplies only, often from an inverter fed directly from the emergency 28V bus. The compass in the cockpit is still using fluxgate sensors in the aircraft, though location is hard, you need to find a place with little wiring, and thus they live in odd places, though the main gyro assemblies have to live at the wing root, as the turn moments are all based off there.
Slightly off topic, I had a 1970's Zenith Duramodule Color TV with a Ferro-resonant power supply. That set wasn't bothered by brownouts, heavy appliances, etc. It was awesome and also had the hands down best picture in the House! We always watched the Rose parade on it, as it looked like a studio monitor. They made them well. 👍 It lasted 25 years before I gave it to a friend.... who knows how much longer it ran.
Great diagnosis. Creating the schematic for the power supply was great and now you have a reference if you run into any power supply issues in the future. Also, kudos to the team of folks that worked with you on the problem.
It's so great seeing the fully complete and working Centurion. It's been an incredible journey, from the day you brought the truckload of dusty mystery parts home until today. I am virtually in tears man. You need to be proud of this accomplishment. Not only have you rescued, assembled and brought the Centurion that would have been lost to history back to life, you have done the same with the Centurion A-Team. You all have proven that with enough determination and elbow grease, nothing can stand in your way. I know the journey isn't over yet. There is still the printer to work on. In my mind's eye I can see the Centurion running at full tilt, updating the financial records at Usagi Oil & Gas headquarters. You're about to become an accountant and oil executive. Congratulations David!
Thank you so much! It's been a wild journey for sure, and there were multiple times where I thought we were dead in the water, only for the community to rally together and get me over the seemingly impossible hurdle. It's been so much fun! And, as you said, we're far from actually done! We still have to revive the Vintage Geek Finch drive and pull the contents off of it. Then we're on to the printer, which is going to be a serious undertaking. I also want to try to whisper a bit of life back into the Phoenix. And, I want to build a custom paper tape interface card and reader that will interface with the Litton minicomputer and have the Centurion act like the storage device for it. Lots more to come for sure!
Have enjoyed this entire series. You did a great job explaining how the supply works. I’ve repaired countless ferroresonant power supplies in all my years working on commercial transmitters. You may want to use a “brown out protection” supply for your finch drives 120vac input. It’s basically a ferroresonant 1 to 1 transformer with an extra large tank capacitor bank. Usually 2 to 4 AC capacitors in parallel. It will maintain the output AC voltage when you have a heavy load start, like large compressor’s, HVAC units and the like. They are not used much anymore with electronics these days not being as sensitive to voltage dips or spikes. My last job, I had several large ones stuffed in a closet that I just couldn’t throw out. Literally, because I didn’t have a heavy duty dolly and the people to help me get them down the stairs and in the dumpster.😊
Good job troubleshooting! I see on the map that you're swinging north through my part of the world. Wave to me as you go by exit 25 off I-90! Last vintage power supply I worked on didn't have a ferroresonant transformer, but it did have a magnetic amplifier - yet another device of magnetic wizardry (excite the auxiliary winding, drop the inductance to near zero as the core saturates).. Fun stuff! One of these days, I'm going to track down the background noise of my 1973-vintage church organ. I'm pretty sure it's a marginal cap somewhere in the 6L6 final amps. I'm with you on NOT wholesale recapping. The noise is enough like the background noise that you'd get from the air pump on a real pipe organ that I'm not that bothered by it. The instrument is surely playable.
smallest electrolytics always dry up first especially if they have low voltage across them. have seen dead coupling caps in audio amplifiers of transistor radios. these are less than 10 microfarads. Same with SMPS.
There doesn’t seem to be a consistent failure mechanism for aluminum electrolytic capacitors. Operationally, high AC ripple current seems to lead normal polarized capacitors to an early demise. High temperature environments and low humidity are also problematic. Electrolytic capacitors are hygroscopic and absorb humidity. There are also a lot of fake capacitors from China which lack appropriate stabilants in the polysaccharide based electrolyte. The fake capacitors started showing up in the mid 1990s. Sometimes you just don’t know. I recently revived a nice digital tuner GE clock radio manufactured in the late 1970s. Probably 3/4 of the approximately 30 electrolytic capacitors had high ESRs and only had 10% to 30% of their designated capacitances. For good measure, I replaced every electrolytic capacitor. The GE radio now works quite well.
@@wtmayhew yep 'capacitor plague' , many things were afflicted by them, especially computer motherboards and power supplies, came across many, and still come across popped ones now, another problem is i think miniaturising them too much, i have a 1958 pye transistor radio, completely virgin and works well, most of my old transistor sets are unrecapped and still work well
@@andygozzo72 Exactly. I have a Zenith battery operated transistor radio my grandmother bought in the late 1950s which still works with the original parts. I’m pretty sure the circuit tolerances were pretty forgiving and early electrolytic capacitors had a lot of safety margin built in, which engineers tried to design out of more recent parts not so well. The down side of the old Zenith is that it runs on four AA cells for maybe three hours. Back then, carbon-zinc cells were pretty much the only type available to consumers, but they were very inexpensive. I worked in a data center which had APC/Schneider Symmetra on-line UPSs. The modular inverters had a bank of big blue capacitors the size of soup cans screwed to bus bars. The inverters needed to be swapped out for re-capping about every five years. They could go longer, but you were rolling the dice. I had one inverter go bad in fewer than five years due to capacitor failure. The UPSs were N+1 redundant with three inverters. The unit took the bad inverter off line and soldiered on with the two remaining good inverters.
6:51, yes I'm a nerd cause all I can think of is, "is this how a Flux Capacitor works?" Truly appreciate your work and videos, always excited when you have a new post.
The 9-pin Molex. I went downstairs to my shop and in 30 seconds had two of them in hand. Unfortunately I am in Honduras and it would take a long time and much money to ship. I think youy will find the in stock at Mouser or Digi-Key. But do put some heat shrink sleeving around those bare pins. Great work. I do not remember studying ferroresonant transformers at Michigan Tech 53 years ago but your tutorial seems to be spot on. Great sleuthing on that cap. I would not swap the other caps at this point. You are Captain Dave of the starhip Centurion. Make your own calls 🙂
Where would be the fun in recapping everything? You are absolutely correct and I agree with your rant. Those who favor recapping everything just don't get it. There is a satisfaction in fixing only what's broken, and nothing more.
Ferroresonant transformers are alive and well. You see them used in industrial applications from time to time, as they do some things that are just hard to do otherwise. They are great at helping to regulate the output AC, and also tend to work well at filtering out harmonics. We had to use these to keep out office equipment happy, due to the sheer amount of electrical noise generated by the industrial processes. A giant DC arc furnace tends to generate a crap ton of harmonics.
I usually would recap as a precaution, but I am also a big fan of troubleshooting and finding out how things work and not judt being a "board jockey" as we use to call people that just swapped parts without knowing why. Great job finding the fault on the power supply!
From a pile of "What is this?" to a fully restored and functional minicomputer this journey has been entertaining and informative for the entire ride. Thanks so much to everyone involved in bringing back this piece of history.
Ferroresonant transformers also have a thing in that they distort the waveform on the secondary, making the top a little flat as the secondary side slides into and out of saturation, so the top of the waveform, when the diodes turn on to charge the capacitors after the bridge rectifier, means a longer on time for the diodes, and thus lower peak currents through the diodes. This also translates to lower ripple current in the capacitor, and slightly lower losses in the diodes and capacitor, so they run cooler, and last a lot longer. As well lower start up inrush, as the resonant tank takes all the energy till it runs into saturation, plus the power supply will ignore up to about a dozen mains cycle interruption before the output voltage starts to decay, so that power supplies with this will ride through momentary glitches. Also will mostly ignore brownouts and spikes as well, though, as noted you have a much larger fixed power dissipation in the transformer to deal with. They make great line conditioners, though you also have a big fixed loss. I used a 2kVA one for years, as it make for a really quiet mains input to the audio equipment I was using, and it sat in the one corner humming to itself, and generating around 300W of heat 24/7. But no mains induced noise at all from the power rail side. Yes you are discovering the world of power supply sequencing, it is very important on many large pieces of equipment that power rails come up in a particular order, and this one ensures that first rail up is always the 5V supply, then the 12V supplies come up, and finally the 24V supply. Some equipment will let out smoke if they come up in the wrong order. As well some will also have connected to the power rails a power good pin, seen on PC power supplies, though often simply now a simple RC delay off the supervisory chip power rail, that is there to tell the computer that mains power has been lost, and that the CPU has around 50-100ms of power before the power rails go away, so it is time to do a semi orderly power down. Mostly write the system clock to RTC clock, abort all writes currently in progress, and send shutdown commands to all disk drives to flush caches on them. Hopefully the BIOS actually implemented it, and the drives action it, but hard to tell as you do not write any logs. I will add, as the Centurion is a rare beast, to do as Marc did, and add in power supply crowbars to the power supply, Marc found out the hard way about that. He did design some of his own, and can hopefully send you some of the spare ones he made, and instructions of where to wire them in on the 5V rails of the power supplies. I have had that occur, and found out also the crowbar on that power supply was also faulty (it never is tested, as that is kind of destructive), and the power supply failed short, giving this half million dollar computer on an aircraft the unregulated 14V input for an unknown time. Replaced the entire 5V regulator side, down to the resistors, and then when it was working went looking for the failed TTL IC's. Only one, a SN5403PM3, ironically a hex inverter with 30V rated open collector outputs, which was really easy to find, as it had unsoldered the lid on it as it fried. Just replaced the entire board, as it was very rare to fail, and I had lots of spares. Not the power supply, the price on that was more than the volume in gold. Those also had nasty failure modes, which then involved an ultrasonic bath, and a session with some black paint to repaint the cabinet. 200l of dry cleaning fluid.....
The Centurion power supply has Current Limit and over voltage Crowbar circuits built into the design. I worked at Warrex-Centurion for over 10 years and never recall seeing a system damaged by a failed system power supply fault. We did have a Centurion system damaged when the customer AC outlet had the neutral and hot wire flipped in the outlet box and the other outlet was wired correctly. The system with the drives plugged into the bad outlet and the scope plugged correct outlet... when the scope's ground lead was connected to the computer's chassis.....max smoke was released. It was a bad day for the Centurion Dealer and the customer's electrician that had just wired the new office for the new Centurion system. Also bad for all for not checking all the outlets before plugging in the hardware.
23:14 Hey, another Forgotten Machines T-shirt sighting...thank you!!! Just when you thought these videos couldn't get any more exciting, then Usagi goes and does this... Seriously, so cool to document this level of technical detail on such amazing Forgotten Machines, indeed!
As powergrid wasn't best in 60s USSR (esp. in rural areas) standalone ferroresonant autotransformers were used to power things that were sencitive to voltage fluctuations (fridges, TVs, etc.)
Woow!!! You dragged me back some 40 years when we were selling a.c voltage stabilizers, small inverters and ups systems which were based on ferroresonant transformers as their final stage, 3 American manufacturers pop in my memory: Gould, Topaz and Superior Electric Your explanation on the ferro resonant principal of operation was the best I ever heard!! Modern electronics pushes away all these old days magnetics which were built to last for ever and wasted tons of energy
Brings me back to a Sunday at the plant ...Overtime . Anyway I get a call from a mechanic " the lights just went out in the break room. Yep sure enough go around the corner and open up a two door 36kva xformer singing away. What on earth could cause that. Then I notice the refrigeration room was quiet. This room has several large ammonia compressors ranging from 150hp up to 500hp and fed from a bus. The mechanic had shut them all down to take on some ammonia delivery and to purge air. It quickly came to me in the vault are some large PF capacitors. We were supposed to have automatic ones but someone upstairs decided to take our money thus we got cheep and had manual switches. Run up to the vault turn of the capacitors.
So timely... I recently had an 2007-era power supply stop functioning, and I've been trying to figure out which component is the culprit. So far, I have been focused on the larger caps... but your video tells me to look at the smaller ones. I got the 5V rail back, but I guess 3.3V is still broken (I didn't test 3.3V before threw my arms up in frustration and shelved the project last week. ;-) ). Disclaimer: I know this PS was built during the cap plague, and that all caps probably should be replaced... and that a new PS is a smarter upgrade than fixing this one. This is an exercise in board level diagnosis, not easiest repair.
Nice work, guys, I wasn't even aware of ferroresonant transformers before this video, now at least I know of the term. Sometimes you have to haul out the big analysis toys to see stuff that a simple multimeter doesn't show. It's rewarding to finally get something of this age working again. In addition, you've used it as a learning experience. I fixed a $2,300 (probably original price) computer merely by accidentally touching one leg of a resistor, if I hadn't done that, I would have given up on the whole repair job. Now I remember to check the actual values against the reported values because of that.
Completely agree, regarding cap replacement. If it was built prior to ~2000, cap failure is as rare as any other component. Except for tantalums. And Macs :D
Good to see you not shot gun the cap in the power supply, that good troubleshooting in my book. It easy to replace ever thing and get it going but you learn nothing from that. Thanks for the video.
Sola CVS Line Conditioner is a giant ferroresonant transformer. It'd a be nice period correct piece for your lab. I first saw one in a TV remote truck. I think it was a 7.5kW unit connected to an Onan generator. The Sola has a huge 190-260V input range. A whopping 365lbs according to the spec sheet. They are also excellent at removing fast voltage transitions. Providing super clean AC. Congrats on getting the Centurion finished! I've got to go back and watch the other episodes. So cool seeing that huge hard drive seeking.
That was a really useful description of a ferro resonant transformer, thanks. I’ve often found that everyone describes topics in the same way, so if you don’t “get it” then you are kinda stuck, so seeing a new visual description was great 🙂
FR transformers are super common for charging forklift batteries. They don't use any electronic regulation, just a fixed capacitor. And, these transformers can be found in ballasts for mercury vapor and similar lighting. Also, it would be a really good idea to install some sort of over voltage protection on all of the outputs to save the equipment should an over volt condition exist.
Agreed... The only time where I had (well, a friend did, because it was SMD caps, and with my terrible soldering skills.. Let's not tempt fate, shall we?) to do a full recap was on a Macintosh motherboard/logic board, specifically a LC II board, every single cap was leaking.. So in these instances, ain't no other solutions.. I still habr the power supply that I need to tackle too, but the caps on them don't look too bad, save for one or two that looks questionable (deformed top), so these two I'll probably swap them out for new ones.
Missing from your rant: modern electrolytic capacitors tends to be less reliable than those made in the golden age (1970s-1980s). So the replacements may die before the working originals would.
Microwave ovens also have ferroresonant transformers, by the way, but basically just used as a ballast just so the whole thing doesn't melt down. Magnetron tubes require around 300 to 1,000 milliamps depending on where they're used in, at between 4,000 to 8,600 Volts, just so they can cook food rapidly and it requires specific transformer that can handle the punishment. Inverter microwave ovens on the other hand use very different transformer, in this case flyback transformers with higher current primary winding just so they can pump in a lot of magnetic strength in the ferro-ceramic core for sufficient amperage to be obtained in the secondary winding (ie. 0.5 Amp at 5,200 Volts DC power at the output of voltage multiplier for 6 - 8 Amps at 110 Volts AC) and you can imagine that the whole thing is very complicated.
Probably could do with some over voltage or crow bar circuitry on that old supply though. Wasn't it CuriousMarc's HP9825 power supply that went and caused all that grief for him? Pass transistor could easily fail short and cause big issues. I would agree though, not all capacitors need to be changed straight out. Love your videos Usagi.
One thing I remember is never bring up a Ferroresonant slowly on a variac, you'll blow the fuse or breaker on the variac. You can crank it up fairly quick and be ok, but usually better to just flip the switch at the rated voltage. You explained why 👍 I worked at a plant that produced these types of transformers in the 70's and I just remember watching some of the techs in the magnetic's lab running test on these types of transformers. I love the judicious use of the good old 723 regulator chip's. I still use a 30 amp home brew supply for my Ham radio equipment that is based on that IC. Still a great device for linear type supply's.
I hardly recognised you without the beard, makes you look younger. I'd never even heard of a ferroresonant transformer, I've learnt something today. It's giving a consistent output with a large variation of input voltage, but at the upper end it must waste a lot of energy in heat. Thank you and well done fixing the 24V output.
The reason for replacing all capacitors is usually a combination of saving time and the risk that a faulty capacitor can take out something else when it gives up the ghost. I for example recap amplifies completely if I find a single bad capacitor, or if the capacitors are over 50 years old, just because having an amp go down during a gig can be quite a costly thing for a musician.
watching this brought a small tear to my eye. these old grandads of computers need to be rescued how else are future generations going to know were it all started and the speed of progress ! up to todays quantum computers. we lean from history and just sometime looking back we can find solutions to future problems. thank you so much for rescuing this old boy and bring it back to life.
Very nice troubleshooting, and congrats on the achievement. As a trained EE, I had not come across ferroresonance at least as far as explained or that I have had to debug. I have one piece of equipment that may in fact have one of those transformers, as I was confused when inspecting the hardware (no schematic) as to what was going on when I saw a cap alone connected to a winding.
I've had a lot of LM723/uA723 chips fail so I would have blamed that. And been wrong. These are used in a 1978 video cassette recorder Philips N1700 using the VCR-LP format and frequently cause no-go. Fortunately they don't tend to drive the regulator transistor hard on and so stuff 17V onto a 12V line, that would be bad.
Learning should always be fun. ( well most ) I worked with ferroresonant power back in the 80's Only now I understand how and why they were used. Nice work.
Nice video! My experience in electronics starts when was in elementary school fixing a Commodore 64 with a possessed disk drive . for some reason after play with it several hours the disk drive start to runs without any reason and of course it's stop reading files from any disk. After send it to some repair shops and this issue keeps happening I decide to fix it by myself. This is some that not all the electronics guys know but some devices fails only when they are under load and reach some temperature. In my case was the 5V rail bridge rectifier. One diode from the bridge was oscillating from normal diode to open circuit. Testing the bridge (module) when was cold all the diodes give the same junction value but when I heated up the module one diode from the bridge disappeared. With capacitors first is test the voltage across them with the scope, if the ripple level is correct leave them alone but if not replace immediately. If some cases with lower values than 10uF if they fail the value goes up not down. This fail is very common in low pass filters in PLL circuits in TV or any other communication equipments. 1uF or less avoid to use electrolytics replace them with multilayer caps.
Was at VCF Midwest. No surprise there were power fluctuations. The show was massive in an exhibition hall attached to a hotel that I am sure was never designed for that kind of current draw. Now I am wondering if they may have brought in outside power generation. As for the Centurion, I presume the designers planned that it would be installed in environments that were highly variable, electrically speaking. Back in that day, voltages supplied in the USA varied widely. In my area, suburb of Chicago, historically the voltage was 110, then I think crept up to 115, then 120-ish now. For many decades we just referred to the supply as 110, long before 120/240 became more widespread. Very interesting and informative presentation about transformer theory and operation. Like EE 101 in ten minutes.
I'll be honest, I also would not have foreseen this particular computer becoming fully functional after the first couple of episodes lol. Not due to any lack of skill on your part, but simply because the technical challenges seemed to be insurmountable due to lack of documentation and spare parts. I'm just massively impressed by you and all the work you've done on that machine. Great job all round!
Awesome work! I think you did a really good job on explaining transformers, some points weren't even talked about at my school (I am an EE). Also I think it's a good idea not to change everything as well. If the capacitors are spec'd correctly then they should "re-wear in" and function correctly when the power supply comes on.
Dave: Well Centurion, I'm damned if I can find anything wrong with it. Centurion: Yes, it's puzzling. I don't think I've ever seen anything quite like this before..... 🙂
Bravo for trouble shooting down to the one failed component! My birth father went through DeVry in Chicago during the 1960's. He got to be in the very last class being taught tube theory, then the next year tube instruction was completely gone. Vocationally he was an engineer repairing X-Ray equipment. He knew well how to track down to the exact component which had failed, replace it, get the X-Ray unit up and running again. First after university he worked for GE Medical. His training in tube theory benefited as legacy X-Ray equipment had tube based X-Ray generators. Then solid state was also in the works of being rolled out. 1970's vintage brand new X-Ray generators were the MPX series of generators. Those had proprietary busses and cards in the slot, quite similar to the Centurion's design. He knew how to trouble shoot those down to the failed component. After MPX came the Advantex line of generators. Those had Mini computers in them, thus industry standard. Then singled a shift to exchanging entire boards. He worked on the GE 8800 CT Scanner, which was powered by a Data General Eclipse S 14 0 mainframe computer. That was a neat machine. Washing machine hard drives. Raised floor / chilled computer room. The medical center system he was last employed at, for around 25 to 30 years, as an in-house engineer. They also had in the Neuro department a Technicare 2060 CT Scanner, which ran on a PDP11. I remember one time the hard drive crashed in that, and it had been a Swan brand. They switched primary X-Ray equipment brands over from GE to Picker CT's. I remember the first unit was a PQ 8000. It ran on a real official flavor of System V Unix, which Picker had bought out the developer, so it became their own flavor. LAN networking was very new back then, so I was brought in as a contract consultant to run the ThinNet Ethernet just to make the Picker CT work with its dedicated workstations. Supporting the Picker equipment he utilized a mix of tracking down to either the failed component, or having to oblige and trace down only to a failed board. So he would work anywhere between millivolts and thousands of volts... which is what it takes to make an X-Ray tube make an exposure. You explaining circuit trouble shooting I still find familiar and foreign at the same time. I have a mind for writing and debugging computer software source code, not the hard electronics that power the computer.
I had so much fun listening to your Ferroresonant transformer explanation on top of the fun I already had observing your team effort to make the Centurion usable again that I finally subscribed.
Rule of thumb is that smaller capacitors tend to die first. I recently fixed a Soviet B/W TV where all the 1uF caps went open circuit, and the was a leaky 22uF cap in the beam turn off during retrace circuit causing a horizontal line to show up at higher brightness when the TV was cold. The bigger 2200uF and such were kinda high ESR but they were in series with inductors so it wasn't that big of an issue. I did decide to recap the whole thing in the end though. Soviet electrolysis are notoriously bad, especially the ones made in Armenia.
I'm a liiiiiiiittle bit sad you didn't cut to a clip of ElectroBOOM yelling "FUUUUUUUULL BRIDGE RECTIFIER!" at 12:36 haha. But honestly though, I love your channel and I learned some things today that I didn't know before, so it's still a win. (Editing to add now that I've finished watching the video) This IS epic! I've only just found your channel in the last month or so, so I didn't realise this Centurion project had been going on so long. I'm so glad you were able to reach such a milestone. I guess now I have to go watch some more older videos to see where this started.
Epic completed. Congrats. Nice intro into this power supply w/ special transformer. Moving between countries of different AC voltages, I was given such / similar power supply or which I wanted to 're-wind' the transformer to keep using my AC small appliances. After finding affordable simple transformers, this project's importance faded and the PS is collecting dust ever since. Your excursion triggers me to take a look again at it. Thanks for sharing. ---- In deed, the habit of going thru the pain of finding an issue and learn something is unfortunately under siege by the short term thinking. Goes for the SW as well... why debug and not just write it new and bring in another bunch of errors that may have been sleuthed out in the current implementation? --- Keeping standing on giants shoulders and going a bit under their skin yields the most efficient progeress.
Yep, makes sense. Usually a small cap at too low of a value. (It won't be the only one...) This seems to be especially true for Switching power supplies. Strange that whatever Q4 is controlling wasn't effected.
27mV is not enough to get the transistor to turn on. You need to forward bias the BE junction which starts at at least 400mV with a silicon transistor.
If you want to learn about the inductors/transformers and the LC, RC, RLC or RL circuits and their uses, I used the old ARRL handbooks for that. Since that is used in radio transmitters and recievers, you will find lots of use cases for these types of circuits and related understanding
The "make or break" jumpers are dip shunts. The idea back in the day was that they would be cut in the factory or by the user once, and never touched again, as cutting the shunts incorrectly had a potential to cause damage to the equipment or make it not function, and it scared off unknowing users from cutting them. In addition, they were cheaper at the time.
I did learn something new today! It's quite logical and a pretty good idea with the ferroresonant transformer. Modern switched power supplies are more efficient, but this transformer could swallow a lot of dirt on the power line. If you have one bad capacitor I'd recommend to at least check the other major capacitors so that they are at least within a reasonable range.
Really interesting! I saw only one mistake in your presentation, near 13:40: in application, the output of the VR would always be tied to the inverting input of the error amp, not the non-inverting input. If you tie the output to the non-inverting input, you will simply lock the output to one rail or the other.
My father called those voltage regulators by saturated core stabilizers, but those I saw (that were used as protectors for TVs) used a DC coil instead of a resonant coil. But they distorted the waveform creating harmonics that just wasted energy and were prejudicial to certain circuits.
Dude, congratulations! It is so awesome to see this thing in all it's glory, as someone who has been following this from the beginning, and even got to meet it's mini-me (and yourself of course) at VCF east. Keep up the awesome work!!
For reverse engineering a schematic, I start by making a list of all the components, designators, and pin numbers (Excel really helps in this regard). Then I start by making a list of what pins are on the ground bus, then repeat for the voltage buses. As I document each connection, I remove the corresponding entry on the list of pins I made. I then repeat the process for the remaining connector pins, then go through the list of remaining pins to develop list of internal nodes. The leftovers at the end should all be unconnected pins. That node chart really helps when it is time to draw the schematic, especially when it comes to quirks that turn up in the design..!
Anyone who grew up in the 1980s should know that when it comes to transformers, there's more than meets the eye.
Deceptive tantalums, any autobots worst nightmare
The one that blew my mind the most was Soundwave. He could shrink from a big ol' robot to a hip, cool boombox!
@@untitled_person1941 Pretty sure it's the electrolytic caps that are way more deceptive + scary ... just waitin' to leak/splode everywhere!
Transformers, fighting the evil forces of the decepticaps!
Ferroresonant transformers, roll out!
You and Adrian from Adrian’s digital basement both go with actual troubleshooting before going whole hog swapping parts. I think that’s part of why I enjoy both of your channels.
I absolutely loved hanging out with Adrian during VCF East. He was one of those people that you meet and within 10 minutes you immediately know that you're gonna be best friends!
He does really great work, and I love his style of digging deep into a broken system and hunting down exactly what's wrong and why.
Doing things the quick and easy way is all well and good if you are doing something as a job and need to be quick. Doing things the slow and hard way teaches you the hows and whys, and is often more rewarding if you are doing something as a hobby.
That's my take at least.
Edit: Oh, and Adrian's great. Adrian's Digital Basement, Usagi Electric, and Ben Eater are the triumvirate of retro/hobby electronics on youtube if you ask me. There's other greats, like James Sharman, Curious Marc, GreatScott!, Zack Freedman, Slu4, and many, many more, but Adrian, David, and Ben are who I consider the greatest personally for their combination of sheer shining love for what they do, warmth, ability to present information in an interesting, engaging, and understandable manner, and for feeling approachable in a way that's usually hard to feel through a screen.
I'd include Daniel Shiffman of The Coding Train among them, but since his wheelhouse is solely and solidly in programming and has nothing to do with hardware or anything retro I will leave the triumvirate as it is, though he does get this honorable mention.
From a pile of scrap to a fully working and complete Centurion computer. Absolutely fantastic.
I can't believe how far the machine has come! It wouldn't have even been remotely possible without the help of the fantastic community!
@@UsagiElectric You have all done a wonderful, (what at first seemed an impossible), restoration. You all deserve a medal.
You are working with an ACTUAL flux capacitor.
Important note on the CVT (constant voltage transformer) is that it handles sharp increases in current draw nicely due to the "Tank circuit", which can beneficial with variable load systems.
They also work as very effective line filters and surge suppressors and even work as UPSs for partial cycle supply dips.
Good video... Thank you for explaining the Ferroresonant transformer linear power supply design and reason for the Ferroresonant design. Basically the Ferroresonant tank circuit is set to resonate at 60Hz. The stored excess energy in the tank circuit from the primary winding does two things, smooths the AC power into the secondary windings and can even dump power into the secondary windings of the primary input power inline drops a few power cycles, sages or spikes. Back in the 1970 when Centurion switched to a ferroresonat power supply design a company called "Sola" sold (still sells) a AC power line conditioner to place between the AC line source and high cost electronic equipment. The basic Sola product is metal box with line cord, ferroresonant transformer, tank capacitor and outlet socket that sold for around $800 in the 1970's. Warrex-Centurion did not need to have the customer buy a Sola conditioner, we built it into our supply.
Reminds me of the power factor correction systems saw in the late 70s One we had in a grocery store was AC-AC so it corrected the line in. The cash registers were digital and real sensitive enough that compressors jumping on and off the AC would knock out the register. Thanks for the added info
I have one of those Sola conditioners. It helped me find the cause of prematurely failing compact fluorescent lights in my house some years ago. Or, technically that conditioned power gave greatly improved (X10) lamp lifetime.
Ken, I want to thank you for your involvement in the Centurion project. Time and time again, you, David, and several others working on this have hit and overcome obstacles with your massive combined brainpower. All of you comprise what is certainly the preeminent authority on Centurion today.
That said, you in particular came along at a time when the Centurion was a pile of dusty blue boxes. Without you, the project would certainly be nowhere close to the state it's in now, and perhaps derailed entirely. Your knowledge, resources and experience working on the Centurion architecture arrived decades later, exactly where it was needed.
It has been immensely rewarding, learning all of this about a machine that was all but unknown in recent times, and would have fallen into of those black holes in history if not for you. I know that others have contributed a lot but you were the first, and without your particular expertise, further progress would have been problematic at best.
Not to understate the significance of David's work and that of the others, you are very likely the one person on Earth whose involvement was essential. I do not really have the words to express my own gratitude, except to thank you again Ken. But don't hang up the probes just yet. There is a lot more to do!
@@Starchface Thank you. Getting the Centurion running and working to save Warrex-Centurion's little place in minicomputer history has been a team project of 10's of people from around the world. The best part is the young engineers may have learned a little about how we did things in the past and why something are as they are day.
Congratulations on a completely populated and working Centurion cabinet! This has been so fun to watch progress on
And congrats on the shave.
Thanks so much for sticking with me through the build!
Nice work! Just wrapped up a power supply marathon myself. It's pretty satisfying to chase the signals and find that one component that's causing the trouble - then replace it. FWIW, I'm with you on NOT wholesale recapping. Every component replaced is a potential mishap. If it ain't broke, don't break it.
I second that! I may feel different if i were repairing for a customer, but since you can fix it again if you have another cap fail, if it ain't broke, leave it alone!
IMO it depends on the device and the capacitors. If it is difficult to take the device apart and would be annoying if it failed again, I just recap it if some caps are bad. Better than taking it apart 10 times, each time replacing a single cap. Similar if I use the device all the time - I do not want too much downtime here.
Also, if the capacitors used are known for going bad (especially if they tend to short, which can blow up other components) then I just recap it.
For example, Revox A77 tape deck uses caps that like to short and some of them are rather difficult to get to - recap.
Old stuff made in the USSR usually has a lot of bad capacitors, but by now those that were going bad, did go bad, so replace them and others with the same part number. Other part number may have been manufactured better (if none of them are bad by now). There was a time when you could get a tape deck repaired and then another cap would fail in a week. Bad times for young-me who did not know how to replace capacitors at the time, there was pretty much always something wrong with my tape deck.
Vacuum tube stuff - replace all paper caps, most of them are leaky and others will become leaky (and testing for leakage means I have to desolder one leg of the cap anyway). I have one tape recorder where I had to replace the paper caps because of leakage, but all electrolytics made in 1968 still work.
PC motherboard - if the caps start going bad, I just replace them, because taking apart a server to get to the motherboard is super annoying and the cap can fail at some inconvenient time leaving me with a badly working server for a while until I find time to replace the cap. Same with a PC power supply.
If repairing something for someone else and it has a few bad caps - I tend to do a full recap if it's not expensive, since I do not want the device to fail again for that person.
If there is a row of capacitors of same type and manufacturer and one has failed you can bet the others are marginal or more likely retain
Thanks!
It's always satisfying when the electrons do what they're supposed to do!
I agree that it is better to recap when it doubt. If one has gone bad, the others are most likely not far behind. I also agree that its good to learn the troubleshooting techniques and it was worth making the schematic but electrolytic capacitors are far more likely to fail than other components so one _should_ treat them differently
I think you might want to sleeve those pins that you put in the connector. I would be terrified that one will work it's way partially out and short, in tern breaking the power supply you just fixed.
Good explanation. I've used CVTs (Ferroresonant Transformers) in offshore installations to regulate 240VAC for sensitive equipment. A nice property is that it also removes noise and spikes on the input due to the secondaries running in saturation and with the tank.
Ferroresonant power conditioners are a thing too... used to stabilize really unstable input voltages, very often used for computers lol
If you live out in the boondocks with too weak power lines then they could be useful too in order to compensate for voltage loss in the power lines.
@@ehsnils They used to be very common behind the Iron Curtain and I believe they're still popular in North Korea.
Crazy engineer here, looks like you nailed the description of ferroresonance for the channel.
I like working above the knee.....
It's much better on your back.
We had ferroresonant transformers in the Navy on the subs I rode. We called them 'sola transformers' because of the brand name. Or 'voltage regulating' as you describe, they fed a steady voltage to some electronics despite fluctuations on the power supplied by ship's generators. Yeah, we were taught that shunt had a critical airgap that was the 'magic' part of how these things work. Pretty old technology, but then when I was in we used magnetic-amplifiers (mag-amps) for a lot of things, so....
Navy loves magnetic magic, I saw servo loop running almost bow to stern and across several decks, where resolver fed a motor-generator used as mag-amp to power giant AC motor with a gearbox whos output shaft had a synchro that closed the servo loop by spinning the resolver's rotor.
I stand by my believe that designer decided to use ALL magnetic devices he knew about in the circuit just because they could.
Aircraft still rely on magamps these days, though often these feed almost immediately into a synchro to digital converter in the avionics computers, and then back out to a digital to synchro converter for the those instruments that are still not screens. the standby ones that do not use this are entirely either self powered, or have 115VAC supplies only, often from an inverter fed directly from the emergency 28V bus. The compass in the cockpit is still using fluxgate sensors in the aircraft, though location is hard, you need to find a place with little wiring, and thus they live in odd places, though the main gyro assemblies have to live at the wing root, as the turn moments are all based off there.
I used to know all of the math for those. Magic, they are not. Voodoo is more like it.
@@denisdrozdoff2926What? No Amplydine???? LOL Those were another, "trust me, you change this and that changes and VOILA!!!" lol
Slightly off topic, I had a 1970's Zenith Duramodule Color TV with a Ferro-resonant power supply. That set wasn't bothered by brownouts, heavy appliances, etc. It was awesome and also had the hands down best picture in the House! We always watched the Rose parade on it, as it looked like a studio monitor. They made them well. 👍 It lasted 25 years before I gave it to a friend.... who knows how much longer it ran.
Great diagnosis. Creating the schematic for the power supply was great and now you have a reference if you run into any power supply issues in the future. Also, kudos to the team of folks that worked with you on the problem.
11:10 Why did I expect Mehdi to get shocked right there.
It's so great seeing the fully complete and working Centurion. It's been an incredible journey, from the day you brought the truckload of dusty mystery parts home until today. I am virtually in tears man.
You need to be proud of this accomplishment. Not only have you rescued, assembled and brought the Centurion that would have been lost to history back to life, you have done the same with the Centurion A-Team. You all have proven that with enough determination and elbow grease, nothing can stand in your way.
I know the journey isn't over yet. There is still the printer to work on. In my mind's eye I can see the Centurion running at full tilt, updating the financial records at Usagi Oil & Gas headquarters. You're about to become an accountant and oil executive. Congratulations David!
Thank you so much!
It's been a wild journey for sure, and there were multiple times where I thought we were dead in the water, only for the community to rally together and get me over the seemingly impossible hurdle. It's been so much fun!
And, as you said, we're far from actually done! We still have to revive the Vintage Geek Finch drive and pull the contents off of it. Then we're on to the printer, which is going to be a serious undertaking. I also want to try to whisper a bit of life back into the Phoenix. And, I want to build a custom paper tape interface card and reader that will interface with the Litton minicomputer and have the Centurion act like the storage device for it.
Lots more to come for sure!
Nicely done! I 100% agree with your delightful rant (nice timer!)
Have enjoyed this entire series. You did a great job explaining how the supply works. I’ve repaired countless ferroresonant power supplies in all my years working on commercial transmitters. You may want to use a “brown out protection” supply for your finch drives 120vac input. It’s basically a ferroresonant 1 to 1 transformer with an extra large tank capacitor bank. Usually 2 to 4 AC capacitors in parallel. It will maintain the output AC voltage when you have a heavy load start, like large compressor’s, HVAC units and the like. They are not used much anymore with electronics these days not being as sensitive to voltage dips or spikes. My last job, I had several large ones stuffed in a closet that I just couldn’t throw out. Literally, because I didn’t have a heavy duty dolly and the people to help me get them down the stairs and in the dumpster.😊
Outstanding! A little touch-up on that scratch in the front panel to the right of the floppy drive and it will be perfect!
Good job troubleshooting!
I see on the map that you're swinging north through my part of the world. Wave to me as you go by exit 25 off I-90!
Last vintage power supply I worked on didn't have a ferroresonant transformer, but it did have a magnetic amplifier - yet another device of magnetic wizardry (excite the auxiliary winding, drop the inductance to near zero as the core saturates).. Fun stuff!
One of these days, I'm going to track down the background noise of my 1973-vintage church organ. I'm pretty sure it's a marginal cap somewhere in the 6L6 final amps. I'm with you on NOT wholesale recapping. The noise is enough like the background noise that you'd get from the air pump on a real pipe organ that I'm not that bothered by it. The instrument is surely playable.
Good to see the first pipe organ comment on this channel. There had to be a crossover at some point.
smallest electrolytics always dry up first especially if they have low voltage across them. have seen dead coupling caps in audio amplifiers of transistor radios. these are less than 10 microfarads. Same with SMPS.
not always, any under high stress such as temperature or ripple may give a short life
There doesn’t seem to be a consistent failure mechanism for aluminum electrolytic capacitors. Operationally, high AC ripple current seems to lead normal polarized capacitors to an early demise. High temperature environments and low humidity are also problematic. Electrolytic capacitors are hygroscopic and absorb humidity. There are also a lot of fake capacitors from China which lack appropriate stabilants in the polysaccharide based electrolyte. The fake capacitors started showing up in the mid 1990s. Sometimes you just don’t know. I recently revived a nice digital tuner GE clock radio manufactured in the late 1970s. Probably 3/4 of the approximately 30 electrolytic capacitors had high ESRs and only had 10% to 30% of their designated capacitances. For good measure, I replaced every electrolytic capacitor. The GE radio now works quite well.
@@wtmayhew yep 'capacitor plague' , many things were afflicted by them, especially computer motherboards and power supplies, came across many, and still come across popped ones now, another problem is i think miniaturising them too much, i have a 1958 pye transistor radio, completely virgin and works well, most of my old transistor sets are unrecapped and still work well
@@andygozzo72 Exactly. I have a Zenith battery operated transistor radio my grandmother bought in the late 1950s which still works with the original parts. I’m pretty sure the circuit tolerances were pretty forgiving and early electrolytic capacitors had a lot of safety margin built in, which engineers tried to design out of more recent parts not so well. The down side of the old Zenith is that it runs on four AA cells for maybe three hours. Back then, carbon-zinc cells were pretty much the only type available to consumers, but they were very inexpensive.
I worked in a data center which had APC/Schneider Symmetra on-line UPSs. The modular inverters had a bank of big blue capacitors the size of soup cans screwed to bus bars. The inverters needed to be swapped out for re-capping about every five years. They could go longer, but you were rolling the dice. I had one inverter go bad in fewer than five years due to capacitor failure. The UPSs were N+1 redundant with three inverters. The unit took the bad inverter off line and soldiered on with the two remaining good inverters.
Holy shit, a commercial use of 2n2222, basically the only transistor I'm actually familiar with by name.
I have designed circuits using 2222 transistors that are still shipping
6:51, yes I'm a nerd cause all I can think of is, "is this how a Flux Capacitor works?" Truly appreciate your work and videos, always excited when you have a new post.
it literally is one
The 9-pin Molex. I went downstairs to my shop and in 30 seconds had two of them in hand. Unfortunately I am in Honduras and it would take a long time and much money to ship. I think youy will find the in stock at Mouser or Digi-Key. But do put some heat shrink sleeving around those bare pins. Great work. I do not remember studying ferroresonant transformers at Michigan Tech 53 years ago but your tutorial seems to be spot on. Great sleuthing on that cap. I would not swap the other caps at this point. You are Captain Dave of the starhip Centurion. Make your own calls 🙂
Where would be the fun in recapping everything? You are absolutely correct and I agree with your rant. Those who favor recapping everything just don't get it.
There is a satisfaction in fixing only what's broken, and nothing more.
Ferroresonant transformers are alive and well. You see them used in industrial applications from time to time, as they do some things that are just hard to do otherwise. They are great at helping to regulate the output AC, and also tend to work well at filtering out harmonics. We had to use these to keep out office equipment happy, due to the sheer amount of electrical noise generated by the industrial processes. A giant DC arc furnace tends to generate a crap ton of harmonics.
Man, that's a nice looking minicomputer you've assembled there. Well done!
I usually would recap as a precaution, but I am also a big fan of troubleshooting and finding out how things work and not judt being a "board jockey" as we use to call people that just swapped parts without knowing why.
Great job finding the fault on the power supply!
Also called shotgun troubleshooting.
Congrats on the PS repair! And adding the schematic to history!
Thanks for keeping history alive!
I'll try to make it out to the museum to say hi!
From a pile of "What is this?" to a fully restored and functional minicomputer this journey has been entertaining and informative for the entire ride. Thanks so much to everyone involved in bringing back this piece of history.
Ferroresonant transformers also have a thing in that they distort the waveform on the secondary, making the top a little flat as the secondary side slides into and out of saturation, so the top of the waveform, when the diodes turn on to charge the capacitors after the bridge rectifier, means a longer on time for the diodes, and thus lower peak currents through the diodes. This also translates to lower ripple current in the capacitor, and slightly lower losses in the diodes and capacitor, so they run cooler, and last a lot longer. As well lower start up inrush, as the resonant tank takes all the energy till it runs into saturation, plus the power supply will ignore up to about a dozen mains cycle interruption before the output voltage starts to decay, so that power supplies with this will ride through momentary glitches.
Also will mostly ignore brownouts and spikes as well, though, as noted you have a much larger fixed power dissipation in the transformer to deal with. They make great line conditioners, though you also have a big fixed loss. I used a 2kVA one for years, as it make for a really quiet mains input to the audio equipment I was using, and it sat in the one corner humming to itself, and generating around 300W of heat 24/7. But no mains induced noise at all from the power rail side.
Yes you are discovering the world of power supply sequencing, it is very important on many large pieces of equipment that power rails come up in a particular order, and this one ensures that first rail up is always the 5V supply, then the 12V supplies come up, and finally the 24V supply. Some equipment will let out smoke if they come up in the wrong order. As well some will also have connected to the power rails a power good pin, seen on PC power supplies, though often simply now a simple RC delay off the supervisory chip power rail, that is there to tell the computer that mains power has been lost, and that the CPU has around 50-100ms of power before the power rails go away, so it is time to do a semi orderly power down. Mostly write the system clock to RTC clock, abort all writes currently in progress, and send shutdown commands to all disk drives to flush caches on them. Hopefully the BIOS actually implemented it, and the drives action it, but hard to tell as you do not write any logs.
I will add, as the Centurion is a rare beast, to do as Marc did, and add in power supply crowbars to the power supply, Marc found out the hard way about that. He did design some of his own, and can hopefully send you some of the spare ones he made, and instructions of where to wire them in on the 5V rails of the power supplies. I have had that occur, and found out also the crowbar on that power supply was also faulty (it never is tested, as that is kind of destructive), and the power supply failed short, giving this half million dollar computer on an aircraft the unregulated 14V input for an unknown time. Replaced the entire 5V regulator side, down to the resistors, and then when it was working went looking for the failed TTL IC's. Only one, a SN5403PM3, ironically a hex inverter with 30V rated open collector outputs, which was really easy to find, as it had unsoldered the lid on it as it fried. Just replaced the entire board, as it was very rare to fail, and I had lots of spares. Not the power supply, the price on that was more than the volume in gold. Those also had nasty failure modes, which then involved an ultrasonic bath, and a session with some black paint to repaint the cabinet. 200l of dry cleaning fluid.....
The Centurion power supply has Current Limit and over voltage Crowbar circuits built into the design. I worked at Warrex-Centurion for over 10 years and never recall seeing a system damaged by a failed system power supply fault. We did have a Centurion system damaged when the customer AC outlet had the neutral and hot wire flipped in the outlet box and the other outlet was wired correctly. The system with the drives plugged into the bad outlet and the scope plugged correct outlet... when the scope's ground lead was connected to the computer's chassis.....max smoke was released. It was a bad day for the Centurion Dealer and the customer's electrician that had just wired the new office for the new Centurion system. Also bad for all for not checking all the outlets before plugging in the hardware.
23:14 Hey, another Forgotten Machines T-shirt sighting...thank you!!! Just when you thought these videos couldn't get any more exciting, then Usagi goes and does this... Seriously, so cool to document this level of technical detail on such amazing Forgotten Machines, indeed!
As powergrid wasn't best in 60s USSR (esp. in rural areas) standalone ferroresonant autotransformers were used to power things that were sencitive to voltage fluctuations (fridges, TVs, etc.)
Woow!!! You dragged me back some 40 years when we were selling a.c voltage stabilizers, small inverters and ups systems which were based on ferroresonant transformers as their final stage, 3 American manufacturers pop in my memory: Gould, Topaz and Superior Electric
Your explanation on the ferro resonant principal of operation was the best I ever heard!!
Modern electronics pushes away all these old days magnetics which were built to last for ever and wasted tons of energy
Nice work on the power supply and congratulations on getting everything working in one cabinet! Good luck on the trip
Brings me back to a Sunday at the plant ...Overtime . Anyway I get a call from a mechanic " the lights just went out in the break room. Yep sure enough go around the corner and open up a two door 36kva xformer singing away. What on earth could cause that. Then I notice the refrigeration room was quiet. This room has several large ammonia compressors ranging from 150hp up to 500hp and fed from a bus. The mechanic had shut them all down to take on some ammonia delivery and to purge air. It quickly came to me in the vault are some large PF capacitors. We were supposed to have automatic ones but someone upstairs decided to take our money thus we got cheep and had manual switches. Run up to the vault turn of the capacitors.
Congratulations on getting the Centurion fully functional. Best of luck with the printer.
So timely... I recently had an 2007-era power supply stop functioning, and I've been trying to figure out which component is the culprit. So far, I have been focused on the larger caps... but your video tells me to look at the smaller ones. I got the 5V rail back, but I guess 3.3V is still broken (I didn't test 3.3V before threw my arms up in frustration and shelved the project last week. ;-) ).
Disclaimer: I know this PS was built during the cap plague, and that all caps probably should be replaced... and that a new PS is a smarter upgrade than fixing this one. This is an exercise in board level diagnosis, not easiest repair.
Nice work, guys, I wasn't even aware of ferroresonant transformers before this video, now at least I know of the term. Sometimes you have to haul out the big analysis toys to see stuff that a simple multimeter doesn't show. It's rewarding to finally get something of this age working again. In addition, you've used it as a learning experience. I fixed a $2,300 (probably original price) computer merely by accidentally touching one leg of a resistor, if I hadn't done that, I would have given up on the whole repair job. Now I remember to check the actual values against the reported values because of that.
Wow, what a journey! You are your own harshest critic, but yet it drives you to success. Congrats on getting everything up and going.
Completely agree, regarding cap replacement. If it was built prior to ~2000, cap failure is as rare as any other component. Except for tantalums. And Macs :D
Good to see you not shot gun the cap in the power supply, that good troubleshooting in my book. It easy to replace ever thing and get it going but you learn nothing from that. Thanks for the video.
The journey to date has been an object lesson in dedication, perseverance, hard work and a sheer unwillingness to give up. It's a joy to watch!
BTW, similar circuits are found in older DC to AC inverters.
The tank circuit is setup to resonate at 50 to 60hz.
As a long time lurker here, it's been a blast watching this come together. Congrats on a superb restoration.
Sola CVS Line Conditioner is a giant ferroresonant transformer. It'd a be nice period correct piece for your lab. I first saw one in a TV remote truck. I think it was a 7.5kW unit connected to an Onan generator. The Sola has a huge 190-260V input range. A whopping 365lbs according to the spec sheet.
They are also excellent at removing fast voltage transitions. Providing super clean AC.
Congrats on getting the Centurion finished! I've got to go back and watch the other episodes. So cool seeing that huge hard drive seeking.
That was a really useful description of a ferro resonant transformer, thanks. I’ve often found that everyone describes topics in the same way, so if you don’t “get it” then you are kinda stuck, so seeing a new visual description was great 🙂
Once again you leave me with a massive grin on my face, awesome work everyone!
FR transformers are super common for charging forklift batteries. They don't use any electronic regulation, just a fixed capacitor. And, these transformers can be found in ballasts for mercury vapor and similar lighting. Also, it would be a really good idea to install some sort of over voltage protection on all of the outputs to save the equipment should an over volt condition exist.
Agreed... The only time where I had (well, a friend did, because it was SMD caps, and with my terrible soldering skills.. Let's not tempt fate, shall we?) to do a full recap was on a Macintosh motherboard/logic board, specifically a LC II board, every single cap was leaking.. So in these instances, ain't no other solutions..
I still habr the power supply that I need to tackle too, but the caps on them don't look too bad, save for one or two that looks questionable (deformed top), so these two I'll probably swap them out for new ones.
Missing from your rant: modern electrolytic capacitors tends to be less reliable than those made in the golden age (1970s-1980s). So the replacements may die before the working originals would.
If you like a ferroresonant transformer to play with look for SOLA brand constant voltage transformer. Common in industry to protect PLCs.
A great deep dive into the power-supply design. Thanks for this!
Awesome!!! I've enjoyed the journey with you as you explain stuff so well. Thank-you!
Epic just amazing. I feel so happy for you. So Good to see a computer that old working so well.
Microwave ovens also have ferroresonant transformers, by the way, but basically just used as a ballast just so the whole thing doesn't melt down. Magnetron tubes require around 300 to 1,000 milliamps depending on where they're used in, at between 4,000 to 8,600 Volts, just so they can cook food rapidly and it requires specific transformer that can handle the punishment.
Inverter microwave ovens on the other hand use very different transformer, in this case flyback transformers with higher current primary winding just so they can pump in a lot of magnetic strength in the ferro-ceramic core for sufficient amperage to be obtained in the secondary winding (ie. 0.5 Amp at 5,200 Volts DC power at the output of voltage multiplier for 6 - 8 Amps at 110 Volts AC) and you can imagine that the whole thing is very complicated.
Probably could do with some over voltage or crow bar circuitry on that old supply though. Wasn't it CuriousMarc's HP9825 power supply that went and caused all that grief for him? Pass transistor could easily fail short and cause big issues. I would agree though, not all capacitors need to be changed straight out. Love your videos Usagi.
One thing I remember is never bring up a Ferroresonant slowly on a variac, you'll blow the fuse or breaker on the variac. You can crank it up fairly quick and be ok, but usually better to just flip the switch at the rated voltage. You explained why 👍 I worked at a plant that produced these types of transformers in the 70's and I just remember watching some of the techs in the magnetic's lab running test on these types of transformers. I love the judicious use of the good old 723 regulator chip's. I still use a 30 amp home brew supply for my Ham radio equipment that is based on that IC. Still a great device for linear type supply's.
You (+ the community) are amazing for seeing this through. Well done. I should have signed up on patreon long ago, now done!
I hardly recognised you without the beard, makes you look younger. I'd never even heard of a ferroresonant transformer, I've learnt something today. It's giving a consistent output with a large variation of input voltage, but at the upper end it must waste a lot of energy in heat. Thank you and well done fixing the 24V output.
The reason for replacing all capacitors is usually a combination of saving time and the risk that a faulty capacitor can take out something else when it gives up the ghost. I for example recap amplifies completely if I find a single bad capacitor, or if the capacitors are over 50 years old, just because having an amp go down during a gig can be quite a costly thing for a musician.
watching this brought a small tear to my eye. these old grandads of computers need to be rescued how else are future generations going to know were it all started and the speed of progress ! up to todays quantum computers. we lean from history and just sometime looking back we can find solutions to future problems. thank you so much for rescuing this old boy and bring it back to life.
Very nice troubleshooting, and congrats on the achievement. As a trained EE, I had not come across ferroresonance at least as far as explained or that I have had to debug. I have one piece of equipment that may in fact have one of those transformers, as I was confused when inspecting the hardware (no schematic) as to what was going on when I saw a cap alone connected to a winding.
I've had a lot of LM723/uA723 chips fail so I would have blamed that. And been wrong. These are used in a 1978 video cassette recorder Philips N1700 using the VCR-LP format and frequently cause no-go. Fortunately they don't tend to drive the regulator transistor hard on and so stuff 17V onto a 12V line, that would be bad.
Learning should always be fun. ( well most ) I worked with ferroresonant power back in the 80's Only now I understand how and why they were used. Nice work.
Nice video! My experience in electronics starts when was in elementary school fixing a Commodore 64 with a possessed disk drive . for some reason after play with it several hours the disk drive start to runs without any reason and of course it's stop reading files from any disk. After send it to some repair shops and this issue keeps happening I decide to fix it by myself.
This is some that not all the electronics guys know but some devices fails only when they are under load and reach some temperature. In my case was the 5V rail bridge rectifier. One diode from the bridge was oscillating from normal diode to open circuit.
Testing the bridge (module) when was cold all the diodes give the same junction value but when I heated up the module one diode from the bridge disappeared.
With capacitors first is test the voltage across them with the scope, if the ripple level is correct leave them alone but if not replace immediately.
If some cases with lower values than 10uF if they fail the value goes up not down. This fail is very common in low pass filters in PLL circuits in TV or any other communication equipments. 1uF or less avoid to use electrolytics replace them with multilayer caps.
What an amazing journey! From seemingly impossible to truly complete, you hit this out of the park!
Was at VCF Midwest. No surprise there were power fluctuations. The show was massive in an exhibition hall attached to a hotel that I am sure was never designed for that kind of current draw. Now I am wondering if they may have brought in outside power generation.
As for the Centurion, I presume the designers planned that it would be installed in environments that were highly variable, electrically speaking. Back in that day, voltages supplied in the USA varied widely. In my area, suburb of Chicago, historically the voltage was 110, then I think crept up to 115, then 120-ish now. For many decades we just referred to the supply as 110, long before 120/240 became more widespread.
Very interesting and informative presentation about transformer theory and operation. Like EE 101 in ten minutes.
Nice job! One quibble with your graphic at 13:34 - the output voltage needs to feedback to the inverting input. It's correct on the next graphic.
I dont know why I like your intro so much. Must be that No one else has something like it.
I'll be honest, I also would not have foreseen this particular computer becoming fully functional after the first couple of episodes lol. Not due to any lack of skill on your part, but simply because the technical challenges seemed to be insurmountable due to lack of documentation and spare parts.
I'm just massively impressed by you and all the work you've done on that machine. Great job all round!
Awesome work! I think you did a really good job on explaining transformers, some points weren't even talked about at my school (I am an EE). Also I think it's a good idea not to change everything as well. If the capacitors are spec'd correctly then they should "re-wear in" and function correctly when the power supply comes on.
Dave: Well Centurion, I'm damned if I can find anything wrong with it.
Centurion: Yes, it's puzzling. I don't think I've ever seen anything quite like this before..... 🙂
Fantastic job getting all the parts of the Centurion restored!
As a bachelor in mechatronics and EE nerd, I'd say that your explanation of transformers and the ferroresonant transformer is quite good.
yay! two years has it really been that long! superb progress good to see the joy it brings you too to get this stuff working
Congrats on getting the Centurion tower completely working!
Congratulations. I'm so happy to see it fully loaded and running.
Bravo for trouble shooting down to the one failed component! My birth father went through DeVry in Chicago during the 1960's. He got to be in the very last class being taught tube theory, then the next year tube instruction was completely gone. Vocationally he was an engineer repairing X-Ray equipment. He knew well how to track down to the exact component which had failed, replace it, get the X-Ray unit up and running again. First after university he worked for GE Medical. His training in tube theory benefited as legacy X-Ray equipment had tube based X-Ray generators. Then solid state was also in the works of being rolled out. 1970's vintage brand new X-Ray generators were the MPX series of generators. Those had proprietary busses and cards in the slot, quite similar to the Centurion's design. He knew how to trouble shoot those down to the failed component. After MPX came the Advantex line of generators. Those had Mini computers in them, thus industry standard. Then singled a shift to exchanging entire boards. He worked on the GE 8800 CT Scanner, which was powered by a Data General Eclipse S 14 0 mainframe computer. That was a neat machine. Washing machine hard drives. Raised floor / chilled computer room. The medical center system he was last employed at, for around 25 to 30 years, as an in-house engineer. They also had in the Neuro department a Technicare 2060 CT Scanner, which ran on a PDP11. I remember one time the hard drive crashed in that, and it had been a Swan brand. They switched primary X-Ray equipment brands over from GE to Picker CT's. I remember the first unit was a PQ 8000. It ran on a real official flavor of System V Unix, which Picker had bought out the developer, so it became their own flavor. LAN networking was very new back then, so I was brought in as a contract consultant to run the ThinNet Ethernet just to make the Picker CT work with its dedicated workstations. Supporting the Picker equipment he utilized a mix of tracking down to either the failed component, or having to oblige and trace down only to a failed board. So he would work anywhere between millivolts and thousands of volts... which is what it takes to make an X-Ray tube make an exposure. You explaining circuit trouble shooting I still find familiar and foreign at the same time. I have a mind for writing and debugging computer software source code, not the hard electronics that power the computer.
I had so much fun listening to your Ferroresonant transformer explanation on top of the fun I already had observing your team effort to make the Centurion usable again that I finally subscribed.
Woo, congratulations!
If you ever have to decommission something your bunny pals can help with those tasty tasty wires.
Thank you for answering a question that has puzzled me for quite awhile .
Congratulations 🎉 what a great accomplishment!!!
Rule of thumb is that smaller capacitors tend to die first. I recently fixed a Soviet B/W TV where all the 1uF caps went open circuit, and the was a leaky 22uF cap in the beam turn off during retrace circuit causing a horizontal line to show up at higher brightness when the TV was cold. The bigger 2200uF and such were kinda high ESR but they were in series with inductors so it wasn't that big of an issue. I did decide to recap the whole thing in the end though. Soviet electrolysis are notoriously bad, especially the ones made in Armenia.
I'm a liiiiiiiittle bit sad you didn't cut to a clip of ElectroBOOM yelling "FUUUUUUUULL BRIDGE RECTIFIER!" at 12:36 haha.
But honestly though, I love your channel and I learned some things today that I didn't know before, so it's still a win.
(Editing to add now that I've finished watching the video) This IS epic! I've only just found your channel in the last month or so, so I didn't realise this Centurion project had been going on so long. I'm so glad you were able to reach such a milestone. I guess now I have to go watch some more older videos to see where this started.
Epic completed. Congrats. Nice intro into this power supply w/ special transformer. Moving between countries of different AC voltages, I was given such / similar power supply or which I wanted to 're-wind' the transformer to keep using my AC small appliances. After finding affordable simple transformers, this project's importance faded and the PS is collecting dust ever since. Your excursion triggers me to take a look again at it. Thanks for sharing. ---- In deed, the habit of going thru the pain of finding an issue and learn something is unfortunately under siege by the short term thinking. Goes for the SW as well... why debug and not just write it new and bring in another bunch of errors that may have been sleuthed out in the current implementation? --- Keeping standing on giants shoulders and going a bit under their skin yields the most efficient progeress.
This is a great expository video illustrating how to do trouble shooting correctly. Thanks for sharing.
Yep, makes sense. Usually a small cap at too low of a value. (It won't be the only one...) This seems to be especially true for Switching power supplies. Strange that whatever Q4 is controlling wasn't effected.
27mV is not enough to get the transistor to turn on. You need to forward bias the BE junction which starts at at least 400mV with a silicon transistor.
If you want to learn about the inductors/transformers and the LC, RC, RLC or RL circuits and their uses, I used the old ARRL handbooks for that. Since that is used in radio transmitters and recievers, you will find lots of use cases for these types of circuits and related understanding
The "make or break" jumpers are dip shunts. The idea back in the day was that they would be cut in the factory or by the user once, and never touched again, as cutting the shunts incorrectly had a potential to cause damage to the equipment or make it not function, and it scared off unknowing users from cutting them. In addition, they were cheaper at the time.
Always nice when a project gets completed
Congrats to such a complete system
I did learn something new today! It's quite logical and a pretty good idea with the ferroresonant transformer.
Modern switched power supplies are more efficient, but this transformer could swallow a lot of dirt on the power line.
If you have one bad capacitor I'd recommend to at least check the other major capacitors so that they are at least within a reasonable range.
Really interesting! I saw only one mistake in your presentation, near 13:40: in application, the output of the VR would always be tied to the inverting input of the error amp, not the non-inverting input. If you tie the output to the non-inverting input, you will simply lock the output to one rail or the other.
My father called those voltage regulators by saturated core stabilizers, but those I saw (that were used as protectors for TVs) used a DC coil instead of a resonant coil. But they distorted the waveform creating harmonics that just wasted energy and were prejudicial to certain circuits.
It was such a learning experience seeing you troubleshoot this!
Dude, congratulations! It is so awesome to see this thing in all it's glory, as someone who has been following this from the beginning, and even got to meet it's mini-me (and yourself of course) at VCF east. Keep up the awesome work!!
For reverse engineering a schematic, I start by making a list of all the components, designators, and pin numbers (Excel really helps in this regard). Then I start by making a list of what pins are on the ground bus, then repeat for the voltage buses. As I document each connection, I remove the corresponding entry on the list of pins I made. I then repeat the process for the remaining connector pins, then go through the list of remaining pins to develop list of internal nodes. The leftovers at the end should all be unconnected pins. That node chart really helps when it is time to draw the schematic, especially when it comes to quirks that turn up in the design..!