I would really like to see you and Adrian Black (Adrian's Digital Basement) do some sort of PET project together. Perhaps you could rescue an abused unit and then add some LED to it. Who knows what would come of it.
I actually reached out to Adrian about this one! I got in contact with him and then he went radio silent and I haven't been able to get a response out of him since I told him the AmberPET idea, so I just went ahead and did it on my own, but think it would have made a great collab. I even proposed that he could explain the video sync to me as part of the episode since I assume he knows that part better than me. No idea what's up. If someone knows him, ask him why he stopped responding! I can't imagine he's off doing a competing amber pet or something :-)
I'm unsure when you were communicating with him, but he relatively recently had surgery on his hand/wrist. So that might have been the explanation for the radio silence.
After just watching this video, I too was going to recommend Adrian. I was surprised to hear he went silent on you but then seeing the comment he recently had surgery, that explains it.
That was very educational. I struggle with the analog electronics too. I would have never guessed that the 5155 monitor was composite input. That means in theory you could modify a standard IBM 5155 to watch television on the internal amber screen! How neat would that be? And you should have tried out Petscii Robots on it! In fact, I have a new version I'm about to release for the PET which has an on-screen map using the pseudo bit-map mode of the PET.
Thanks! I too was quite surprised to find out it was composite only. I was tempted to see if I could inject my sync and luma directly, thinking the composite wouldn't be crisp, but I guess it's plenty crisp for broadcast standard resolution! I'll make sure to grab a copy of PETSCII robots from your site so that I can demo it in the future!
I had an early monochrome mac, the electronics were fried, so I was going to build a TV inside it, however I waited too long, and the whole idea become too dated - as in most people would not recognise the early mac in a modern sea of IBM clones & laptops, plus with the advent of digital TV it would have just shown static so I dropped it into the bin - a decision I now regret.
The PET video is clearly monochrome (literally “one color”) and the schematics show that the luminance signal (video) is generated using digital logic, so this means every pixel is either off (black) or on (full brightness). Did any of the PETs or similar computers of the era have multilevel monochrome video output? For a black-and-white CRT this would be a “greyscale” display, but there’s probably no reason that you couldn’t do the same with green or amber phosphors. Composite greyscale analog video should be much simpler to generate than color.
Hi David, also wanted to add that the 5155 also has a composite video output out back, which outputs color video, it's very easy to use this with composite/hdmi converters and hdmi capture devices seem to accept it.
[11:05] That capacitor there is part of a pulse former, in combination with the pull-down resistor next to it. The horizontal drive signal coming from the PET is much too wide to be used directly in a composite video signal. The resistor and capacitor are there to shorten the high part of the pulse to about 3.3us (microseconds). Which is why a 0.1uF (100nF) cap wouldn't work in its place; that gives a 150us pulse, which is way too long.
Correct. The capacitor lets the current through for a short time while it's charging, and then blocks the current when fully charged. Note that when the input flips back to zero, the capacitor is discharged while the output was already pulled down to zero by the resistor. This will result in a negative voltage pulse at the input of the NOR-gate. Digital gates should have protection diodes to handle inputs below 0V or above 5V, but generally don't like this. Even worse, the negative pulse current is not really limited by anything, except for the output resistance of whatever produced the original signal.
I have some more questions: why connect both inputs of the gate used as an inverter to the signal and not just one and the other to GND? That would add less load to the signal, which I would think would be better. And the second question: how can the combination of the signal in the last stemp work? The signal for the sync pulses is connected to the output of the lowest NOR gate and from there the signal goes to a another resitor. But because it goes through the part directly connected to the output, shouldn't that cause the signal from the sync pulses to get "overridden"? EDIT: Or does that pull the signal below 0V, when a sync pulse comes and the video signal inverter is not driving the signal high in that moment?
Dave is so wholesome - I love this content, just a retired guy having fun sharing what he wants with the world. I appreciate your contributions Dave, have a great day buddy!
Pleased you liked the Mini PET 40/80D, and sorry you if there was an issue with support, I don't know what happened there. For future reference, there are 2 pads marked J6 next to the composite video jack that I added for exactly the purpose of connecting an internal composite video monitor. There is also a side reset button so you shouldn't need to quickly cycle the power to reset the machine (and also pads to add your own reset switch if you want to).
I am (or was) an analog hardware guy and enjoyed your simple and clever solutions to this problem. I have done some work in the opposite direction, breaking up composite video into the sync and signal. The LM1881 chip makes this easy and hanging a scope on one will really teach you about composite video signals. BTW, I started an MSC in Physics at U of S (Saskatoon campus) when you were 3 weeks old, That was when I learned of small computers, a PDP8 and that lead me to my passion for software and hardware. Still doing it over 50 years later!
I was blessed enough to have a Commodore 64, a 1541 DD, a dot-matrix printer, and a gorgeous amber monitor. I had a geek dad and know exactly how privileged I was to experience this stuff first hand. I might add ... my eldest boy is now a fully qualified mechatronic engineer on his way to a PhD ... the first university educated person in my immediate family. Cheers Dave. $0.02
I will always have a soft spot in my heart for Commodore PETs. We had a few of them in my Computer Math class back in 1981/82. We had several 40xx models and one 80xx model. We also had an Exidy Sorcerer. In total, we had five computers in the class for about 20 kids. I was addicted to this class; before a month in the semester had passed I knew more about programming in BASIC than everyone else including the teacher. I was drawn to programming and it all simply made sense to me. That all lead to the start of a great career. I still love programming today and it all started 40 years ago with my preference in using the PETs in that class.
That PET, with the cassette deck and chicklet keyboard, was the first computer I'd ever touched. That was in high school, in the late 70s. Memories! Now celebrating 38 years in IT.
Dave, that burnt component marked L2 is likely only a small inductor, you could like just replace it with a jumper. You could also throw in a 10K resistor, if you are uncomfortable with the jumper; just to test everything. When working with video signals an oscilloscope is very useful to make sure all the voltages and timing relationships are correct.
This may be one of my favorite PET repair videos! I would have never imagined an IBM 5155 monitor would have worked! I'm currently working on a monitor repair on one of my vintage systems, a NeXTStation, although finding the appropriate replacement CRT for that system, or even one that will fit the existing chassis has proven to be nearly impossible.
Fantastic. I'd seen the amber PET in the background and wondered about it. The connector on the CGA card in the 5155 was originally used for an external RF modulator option on the 5150, and there is a power output pin for it.
Dave: your attention to production detail with the colours and bokeh balls are great! But your balance between what you know and don't know about electronics is perfectly calibrated, and this results in a great balance between modesty and arrogance. I love your insights and explanations. Keep it up!
Dave, Composite mono video has 3 significant voltage levels; "black" level is at ~0.3v, "white" is at ~1v and the sync pulses are "blacker than black" at 0v. While video is being displayed, D1 is clips the mixed sync to 0.6v (the forward voltage of the diode) which is divided by R2,4 & 4 to 0.3v when the video is black, and 1v when it is white. During the sync periods the AC coupling of C2 takes the voltage at D1/R2 negative and as the video is always black, R2 is able to pull the output level down to 0v.
@@DavesGarage Any voltage lower than "black" is considered to be a sync pulse. anything above 0.3v is black to white. Of course on a monitor that only shows black or white, the voltage should only be 0v, 0.3v or 1v. Sometimes the output levels of composite video stages are higher than "standard" and rely on the input impedance of the monitor to drop the voltages to the correct level, so if you measure them with no monitor or the wrong monitor connected they can be higher than expected, although older TVs tend to tolerate this somewhat. Note these voltage levels are approximate in reality which is why televisions and monitors have brightness and contrast controls! There is a very good wikipedia page "Composite video" that has some great diagrams that explain and show this.
The 1-kohm and 390-ohm resistors form a voltage divider which multiplies the voltage output by the luminance (video) NOR gate (acting as a digital inverter) by the ratio given by 390 / (390 + 1000) = 0.28 In other words, if the voltage at the luminance NOR output is Vx, then the voltage at the output (where the 390 and 1k resistors meet) is Vout = (390/1390) x Vx = 0.28 x Vx For a 74LS02 with a 5-Volt supply, the typical logic-high output level is about 3.4 V. Putting this value into the formula (as Vx) gives 0.95 V at the output, which should correspond to the assumed highest voltage level of 1.0 V.
This voltage-divider ratio assumes the the “input impedance” (basically resistance) of the circuit that this one is driving is much higher than 390 ohms. If this is not the case, then the “390” term in the ratio has to be replaced by the “parallel resistance” of a 390-ohm resistor and the driven-circuit’s input impedance (call it Zin): Zpar = 1 / (1/390 + 1/Zin) = 390xZin/(390+Zin) Zpar will always be less than the smaller value of the two resistances. If the input impedance is not much higher than 390 ohms, the voltage-divider ratio will decrease from 0.28, so the output voltage may be too small.
Vetco saved my butt in college robotics. We blew up a motor driver a couple days before we had to leave for competition, and they had one we could make work on hand. Pretty great place!
This brings back memories...back in the 80s I used to earn a bit of extra money adding the wiring to fit the extra RAM to the PET motherboards. The local Commodore dealer was a friend and used to drop one or two a week off for mods. We used wire wrap wire for the jumpers and I got to play with the PETs until they went back to the shop,,,,,
11:11 - Dave I think this cap is an 'AC coupling' cap - it blocks DC and only lets through a short pulse when there is change. You should invest in (if you haven't already) a cheap Siglent scope. Seeing these signals on a fast-refreshing a trigger-locking screen can be an amazing eye opener as to what's going on.
Thank you. FWVLIW: I would suggest the best person to teach something is someone articulate who has just learned themselves. Why? -They actually know rather than thinking they still do, while remembering what ignorance is like. -They are unencumbered by nuance that will only confuse new students.
Ah... the lengths we go to make 1's 0's and 0's 1's for the correct duration. I learned a lot about the NTSC signal back in the day building and adjusting descrambler boxes for those signals you mentioned 🙂
Thank you for the video, I look forward to each one. Ben Eater has a great video series on building a video card from scratch using logic chips which, I think, may help answer some of your video questions.
This is true, but to be fair, Ben's videos are for a digital VGA, which is way different than the analog composite he's doing here, although there's a lot to learn about sync signals and "front porch/back porch" I guess... In any case, I agree, EVERYONE should watch all of Ben's videos too - they're amazing!
Hey Dave, I have an original 4032 PET sitting in a closet with the rare purple screen. (in addition to a 2001 and a SuperPET 9000) These are where I learned to code so I'll never get tired of content on these machines.
Those good ole days when you actually for a schematic with a product. We should go back to those days. That capacitor in combination with the resistor is a high pass filter. It will only pass a portion of the sync signal through. Probably to shorten the sync. And that’s why your large cap skewed the sync, as it was there too long.
C1 is used to make the vertical sync a short pulse. For something that small you could get away with a "gimmick" capacitor. C2 and D1 make a DC restorer. You need a signal that goes from 0v to 5V. Without the DC restorer the signal at that point would be from -2.5V to +2.5V. Resistors R2, R3, and R4 are used to make the 5V peak signal into a 1.5V peak signal. The sync pulses and the video are needed at different voltage levels. The fancy resistor work not only divides each signal into a proper level but also "adds" them together, plus makes the output impedance match the monitor impedance. From the values given it looks like the monitor is expecting about 300 ohms impedance.
Having watched as many videos I have on old televisions I knew that first capacitor would be an issue. I must admit I to probably would have forgotten that power switch on the pet as well. Sadly the last fail I had was not bothering to clean my work space before servicing one of my own computers and dropped a 40 GB onto the concrete floor. The computer with out a doubt refuses to post with it installed so it is safe to say it died. Sorry for the long comment but having just replaced a ten year old keyboard it feels surprisingly good to type on a new Logitech keyboard. It isn't the kind of keyboard I'm 100% happy to have as it doesn't have the mechanical aspect I would want from a good keyboard but for a membrane it feels good enough to enjoy typing on my keyboard again.
Regarding the 2200pF cap & 1.5k resistor, it's for converting the PET's horizontal sync pulse to the right length for composite video compatibility. Get yourself a scope, and it'll all make sense, as a gadget guy you know you want to! Rigol DS1054Z is remains a common entry-level crowd pleaser for example. For those with gear acquisition syndrome, Keysight (the good bit of what was HP) and Tektronix are the brand leaders: I use all three on my bench, but Keysight is my go to scope due to its snappy UI.
Danger Will Robinson! Danger! An O'scope can become an interest, confusing, educational er "fidget toy". On the other paw, all the calculus and mumbo-jumbo aside analog electronics are fairly easy to get a feel for, especially if you tinker around a little. The practical stuff falls back to relatively simple algebra. More advanced stuff falls into basic trigonometric identities and arithmetic. It's only when you start piddling around with much higher frequencies and ham radio sorts of things like fancy antennas that it become heavy lifting math to understand - and arithmetic coupled with faith to use. An O'scope is a WONDERFUL tool for visualizing what is going on. {^_-}
16:30 About the 7812 regulator: If you want some more juice, just solder a second 7812 straight onto the other's pins. The circuit inside the 7812 can handle multiple outputs tied together. With double the current you can't go wrong.
I thought when you did that, one tended to try to power the entire load because of slight differences in the output voltages? I was always taught NOT to do that.
My first home-built computer got an old monochrome TV as monitor. Just that I bypassed the normal electronics for modulating the beam to get 80 decent characters/line instead of somewhere 32 or 40 that was more in line with expected TV transmission bandwidth. A bit extra work since the TV did not have any video input so first task was to figure out where that signal first showed up from processing the antenna signal. But I still needed to inject image data there for the horisontal and vertical sync part. "Decoding" electronics without access to any schematics can be quite fun.
So glad I ran into your video. I also have one of these service part monitors in the same packaging from Computer Reset. Now I know what it is. I was planning to fit it in a dead crt Mac 512ke that I have to make a amber Mac!
The value of C1 and R1 are both critical because together they establish an RC circuit establishing a timing circuit. This basically creates a specific delay between the incoming signal and the signal going into pin 6 on the gate. R1 will start out discharged, and then will charge up at a specific rate through C1. When the voltage level at the junction of C1 and R1 rises to the proper level the input to the gate will be activated. The time necessary for a capacitor to charge up to one time constant, (1T), may be expressed mathematically as RC Time Constant, Tau: τ=R×C This RC time constant simply indicates a charge rate, with R in Ω and C in Farads.
I enjoy the old hardware talk as much as the software. I much prefer learning ASM and the 6502 and the base hardware over a python script any day. Great stuff. (Also a big fan of Ben Eater and his 6502 project)
The 6502 is way too limiting imho. I learned on an IBM 1401 then an IBM-370. Best teacher in the world. The o/s could not be beat and live worked on .or machine than people have read about. When I played with the Intel series (4004, 8080, 8088 and up) I thought what a cute toy. The instruction set is very limiting, yet people seem to enjoy tinkering with it. I felt that IBM should have invested more in schools with mainframes, but they were beyond the finances of most schools. I was very fortunate to have learned it when I did. And as far as im concerned, there is IBM mainframes, then the rest of the toys out there. There is sheer joy on coding on a mainframe and assembly was my main language for over 50 years
Our first PET 2001, although 8K, came with built-in cassette player and the small “chiclet” keyboard. We didn’t get a standard(ish) keyboard until we moved up to the 4032.
18:55 you mentioned having to solder wires on the board to get the composite output because of the RCA connector's location. At 18:10 we see the bottom of the board, follow the traces... the composite video signal appears on the unpopulated 2-pin header J6. A molex KK100 2-pin connector would be dandy for this output.
Get yourself 'The Art of Electronics' by Horowitz and Hill. I would suggest getting the companion lab manual as well and start playing around. LTSpice is a great circuit simulator also. Analog electronics can be great fun.
@16:00 Great call on using a dedicated PSU. I'd have done the same. Power supplies and circuitry fail pretty often. This way, if the PSU failed it's easy to swap out and likely won't damage the main board and other parts (as modern good quality PSUs have over voltage/over current protection and so on).
Hello sir. Interesting timing. I am currently working on composite out for my PET 4032. It took a while to find some schematics. Eventually found 2 of them. First was the one you are using and the other was from a book called The PET Revealed by Nick Hampshire. Too bad that an assembled solution are being sold anymore. Once I pick the best solution, I will be working with a friend to get them out there as a kit or pre assembled unit. Thanks for sharing and love the amber video over the green!
Hi Dave, As always, I enjoy your videos, especially when you switch to a bit of hardware 😉 Anyway, regarding video etc... a book that I had which I found the most useful when I was trying to sort out S-Video output from an FPGA project was called, "Video Demystified - by Keith Jack" I'm sure it's either still available or even as electronic version. It is quite a large read but was the one source I found useful in explaining and getting the interface correct, in my case. Anyway, keep up the great videos you do.
Really envy your access to Pet computers so early in your life. I am about your age and had the first micro when I was 17 (Sinatra?), but it was the Atari 800XL. As to the circuit -> I would personally dig into the monitor's schematics, to find the 3 points the H-Sync, the V-Sync and the Luma are internally generated from the Composite input signal. Then I would connect the output of the (inverted if needed) separate signals from Pet directly there. BTW, any separate switching (high noise) power supply is an overkill - the 3 gates used in 7402 require negligible power - you'd tap it from the linear supply of a Pet or the 5V of a monitor.
Mono monitors don't degauss, that's only a colour tube thing so it's a different fault, if you hear a 'thump' when you switch on it could be excessive inrush current which might be as a result of a shorted diode in one of the bridge rectifiers. I'd definitely not run the monitor and board off that 7812, the Commodore original monitor had its own onboard 7812 which would suggest to me that it would be a good idea to add a regulator from whatever supply connection the original one had.
Just measure the current the 7812 is already using and add it up. If it’s under the limit, than your generally okay. Just as important is how much voltage is feeding the 7812. The voltage drop multiplied times the current for the Wattage. This is going to give you an idea of how much heat is going to be produced. Run it a while and check the heat. If it’s too much then you’ll have to add cooling. Then if it’s still to much, then leave it on separate supplies. But I strongly suspect it’ll be just fine.
The cooling solution needs to be taken into account. 7812s on 1.5A needs a reasonable heatsink and at least some airflow (passive from the rising air due to heat could be okay). Also it might introduce weird noise from the digital board into the CRT or the other way around. And also you would need to check if the CRT supply needs to be isolated in which case you just can't attach it to the internal regulator. I doubt it needs to be but better make sure.
Why not adding another one, from the input of the existing one? That way, no interference between digital and analog boards, and also no overheat of the IC. Also, I believe that 12V is used in digital board just for op-amps and maybe for serial RS232 if exists (it requires +12/-12V).
I can feel your trepidation about doing the composite signal conversion. Analog electronics are voodoo compared to digital electronics. Very nice work getting that up and running. I would have, just as a personal preference, done whatever I needed to to retain the original PET motherboard. But its nice to see that the mini PET is a viable replacement. Thanks for sharing this. Thumbs up! :)
Capacitor looks like it is to block DC. with a digital gate, it will shorten pulses to a fixed time. when you used an 0.1u it probably output a pulse that was too long.
Dave, talk to Electroboom about your circuit. He could talk about this stuff all day, and according to other collabs he's worked, he will and does talk about this stuff all day. :D
A 7812 has internal protection so if the voltage does not drop as it heats up, it is not catastrophically overloaded. It may still run hotter than may be desirable for reliability. A second regulator helps a lot since you are adding a second package with twice the surface area contacting the heat sink, so the die temperature is much lower with 2 regulators on 2 small heatsinks compared to one regulator on a larger heatsink equal to the the two small ones.
Funny how all our rabbit holes seem to crisscross. Whether it is Ben Eater, LGR, 8-bit guy, Adrian's Basement, or your channel we are all raising our inner nerd.
I work as a broadcast engineer for a large television broadcaster in England where we had the superior PAL colour TV system and the correct spelling of color ;-) before everything went all digital. The NTSC system could not correct for colour phasing errors causing hue issues. In fact we nicknamed it Never Twice (the) Same Colour. The reason for the three grounds is each cable screen gets its own dedicated ground pin rather than trying to land three screens on one connector pin. Noise can only be eliminated by using differential circuits where the signal on one wire is 180deg out of phase with the other wire. If noise is introduced on the wires both would be the same phase so on the receive end the two in phase noise signals cancel each other out. On differential circuits a ground is not needed technically but often used to ground the overall screen surrounding the two wires
Haha NTSC was first and was standardized in the earl 1950 to be fully compatible with the millions of b&w television sets that were already in the homes.. You guys were late to the electronic TV broadcast game and didn't re-engineer color TV til the late 60s. The UK also didnt have to contend with the millions of television receivers thay were already in homes. Not to mention you only had 1 television station (BBC1) where we had thousands. In 1964 the UK had 2 stations. Also the BBC suspended operations from 1939-1946 die the the war, but here in the US we still had stations on the air, specifically WABD (the 3rd network) continuously since it first fired up it transmitters in 1938 making it the longest continuously running television network in the world. Even longer than WCBS, WNBC & WABC To change a standard in a country that had millions of TVs and thousands of television broadcast stations at the time would be cost prohibitive. Its always easier to improve on someone elses standard. Its much more difficult inventing things in the first place. Electronic TV was an effort by many people and not one company (or person) can take credit, although Alan Sarnoff of RCA tried to lay claim for the invention of electronic television, by buying out companies that held certain patents. Just like bill gates did. I worked at WABD and I lived the history. Now I volunteer to restore broadcast video tape machines made by Ampex to keep that important history alive
@@rty1955 - All colo(u)r TV standards had to be backwards compatible with existing black and white televisions in customer homes so yes we had the same problem when NTSC Color was introduced. The issue was how do you send colour information over the air to new sets and not impact the reception of existing black and white televisions. This was achieved in both solutions by superimposing a colour subcarrier on to the TV signal at such a high frequency the black and white television does not see it. However with a suitable decoder you could extract the luminance (black & white) and colour difference signals (R-Y) and (B-Y). From that you can then create the red, green and blue signals to the guns on the cathode ray tube. So yes NTSC Color came before 625 line colour which in turn had many variants with PAL improving on NTSC’s handing of differential phase errors but incorrect that we did not have a large user base of existing black & white televisions to support as well.
A 74LS02 takes TTL logic levels which are distinctly DC. I'm guessing that the signal from J1 pin 3 is AC. Components C1 and R1 filter the input to match the voltage levels expected by the nor gate, blocking the DC offset from the signal and providing something more appropriate.
At the time, the CRTs were only meant to display image on the screen, there was no difference between the CRTs used in common B&W TVs and those used in computers monitors. Wasn't possible to swap just the picture tube? Or the monitor board was nonfunctional inn the PET? I swapped a CRT twice and worked well, once a green phosphor with a amber and once with a white one. All the CRTs were the same size and had the same pin-out and size at the socket. Great to see an amber screen computer, I always loved them. About the 7812, if unsure, you can add another one, from the same input as the existing one to power just the monitor. Make sure to add the capacitors required and a heatsink.
Fantastic video and project from start to finish Dave! So far anyway, I expect to see more on that PET in the future. Entertaining and very educational. I'm a digital software guy too (not on your level though), but I also try to learn about analog hardware as well.
Ah, Vetco. That place is such a gem. With Fry's being gone, there's really no other place you can go to pick up a Raspberry Pi or Arduino clone locally.
I would love it if you started a series of tutorials for OS development, explaining the concepts, tools, requirements in terms of knowledge, materials and books. Of course based on the X86 architecture. 🙏🙏🙏🙏
R3/R4 voltage divider is there to lower the 5V from the logic chip to around 1.4V, which you can see listed in the description of the composite video specs of the monitor.
I'm kind of interested in acquiring a PET, however they are rare as hens teeth on the other side of the planet from you. The mini-pet looks like a good solution. I have a selection of composite monitors, including a brand new green CRT, However the case and keyboard are stumbling blocks for me. I did see a Commodore inspired mechanical keyboard on Drop/MassDrop, not sure if they still have them, but that could be a start. Now all I need is some drawings and a sheet metal fabricator to whip me up a case.
The capacitor generates a short PULSE when the input level changes. The sync signal from the board is too "wide" for the composite signal standard. Time constant of that circuit is given by the R C values (product).
Correction at 6:06 - the IBM Portable PC does not have a full monochrome display adapter in it. The IBM Portable PC comes with a color graphics adapter which can provide a composite NTSC video out to allow it to be connected to an external monitor, as well as a TTL output to connect to a IBM Cooler Monitor. The inbuilt screen is monochrome only, and is horrendous to use long term (hence the use of composite out to allow the poor sod who carried it home from work to plug it into a better screen.
11:05 in Russian we usually call this type of capacitors "barrier" or "connection" caps. Basically it prevents DC from passing while lets AC and signals to go through... I might be wrong but meh whatever, guess is a guess after all :)
My guess would be that those caps are working as some sort of filter, and the mismatched value caused it to filter out some of the signals you actually needed.
I've tried to watch this video like 5 times today in the last 2 hours, and I've not been able to get it all in my head due to external factors. Unfortunately, those external factors pay the bills, so, your view count is going up significantly today Dave. heh
*Why so many grounds:* For one, it'll reduce the resistance of the ground connection between circuits, which is always helpful. More than likely, though, the ground-per-signal is to reduce inductance. Especially since the signal and ground are twisted together. With a high-frequency analog signal, you would probably do better with a coaxial cable instead. (A small coax wire is fine, doesn't need to be like satellite TV cable.) *Using a 7812:* Don't try to draw nearly an amp from a 7812. You won't likely be able to keep it cool enough without some heroics - at least depending what the input voltage is, and thus, how much it has to drop. I would take a look at the monitor circuit to see what its B+ voltage is. It may run quite well from whatever your unregulated supply voltage is. Monitors usually don't need a tightly-regulated B+. They derive their voltages from that rail and regulate them internally. *Composite Video:* I think you might be better off approaching this solution the other way around. Instead of supplying composite to the monitor, why not look at where the composite input is divided into sync and signal? Mono video + VSync + HSync is far easier to deal with -- it isn't, but may as well be, a digital signal. At least when "mono" is actually just, literally, black and white, and not grayscale. It may be more convenient to use composite with the retrofit motherboard, but if you have a working original (at least with a drop-in RAM substitute) then why not use that? Otherwise, just emulate the entire thing. :-)
The capacitor (C1) is acting in conjunction with the resistor (R1) as a high-pass filter. My guess is that the line before the capacitor has a dc offset issue that the high-pass filter is correcting for. And maybe also helps with pulse shaping, but I don't know enough about this specific standard to say one way or the other. Typically pulse shaping is not of importance in digital circuits as the very high open loop gain of the follow on digital devices will slam the signal to the rails given even a tiny offset from the mid rail, thus destroying any pulse shaping some one is trying to accomplish (excusing pwm). Now C2, D1, R2, R3, and R4 might be something closer to a pulse former. I'm fairly comfortable saying that R4 is helping with impedance matching. Given that the NAND gate will present as a very high conductance to AC ground, we can see R3 and R4 make a parallel combination to AC ground and present a load of about 280 ohms to the "composite video out" line. The weird component to me is D1 as that will basically clamp the anode side to the forward bias of D1, what purpose this has I'm not sure of just yet. It would be very interesting to see what an oscilloscope shows at the various nodes in this circuit.
I recommend you watch Ben Eater's video about his video card, although it deals with VGA signals, the idea of sync pulses and blanking time is still present due to vga signals being based on crt style monitors.
Although I would agree that Ben’s “world’s worst video card” video (or something like that) was a great introduction to how HSYNC and VSYNC and component video (RGB) signals work together (but as separate signals), the circuit used here combines the logic-level synchronization signals and the one-bit luminance signal (off or on) into a single signal that will be used as the single input to a monochrome monitor. It would be great if Dave had a multi-channel oscilloscope and could show his results (and oops moments) like Ben Eater does.
Horizontal position on old mono monitors is sometimes a tall(ish) inductor, it would need to be adjusted with the correct plastic tool, never use a screwdriver it shatters the inductor slug.
@@DavesGarage always use a plastic adjustment tool. the metal screwdriver will load the induction tuning down and change the resonance value, and well as possibly shatter the ferrite tuning slug... then bad things happen to good people! :)
@@DavesGarage Yep, that is it. Might be Horiz, size though, depends on how the monitor is designed (plus, it has also been 30 years ...). It has a slug of ferrite in it, can be very brittle stuff!
@@eDoc2020 Yes, I think you are correct. I seem to remember having to adjust CRT monitor chassis using a plastic trimmer, some controls on pots, a few on inductors, but it has been 20 years since I did it for real though. I think you are corrent and hpos was normally on a pot.
That's what I was thinking, after having watched a lot of Uncle Doug's amp repair videos. But I wasn't sure that would be the case in such a different circuit and there are other comments here that sound like they think it's something else, to shorten the pulse or something.
@@davidtaylor6124 Yes, the lower de caps capacitance, the higher the transmitted AC frequencies, the shorter the pulse. Actually, it could be ment to make two short pulses of one long one, but we can't tell because the connector is an enigma. But I am not an electronics engineer, I just have had electronics as a hobby some weekend haha ;) Have a great day.
@@davidtaylor6124 Actually, I think you are right; the capacitor is probably ment to convert a long digital pulse into two short ones. It makes no sense to decouple just one port, so I think you are right.
I'm guessing that your existing PET monitor had more than just a simple bad CRT issue? Like power or control circuitry was also bad? Because it would've been trivial to pull that CRT out of the IBM monitor and drop it into the PET's monitor, then adapt the PET's CRT connector to that amber CRT's connector. Heck, the pins might have matched exactly, since there are only so many ways to wire up a monochrome CRT internally.
I would really like to see you and Adrian Black (Adrian's Digital Basement) do some sort of PET project together. Perhaps you could rescue an abused unit and then add some LED to it. Who knows what would come of it.
I actually reached out to Adrian about this one! I got in contact with him and then he went radio silent and I haven't been able to get a response out of him since I told him the AmberPET idea, so I just went ahead and did it on my own, but think it would have made a great collab. I even proposed that he could explain the video sync to me as part of the episode since I assume he knows that part better than me.
No idea what's up. If someone knows him, ask him why he stopped responding! I can't imagine he's off doing a competing amber pet or something :-)
I'm unsure when you were communicating with him, but he relatively recently had surgery on his hand/wrist. So that might have been the explanation for the radio silence.
Excellent! I just typed out a message suggesting the same and then saw this pinned comment. Note to self, look at pinned comments first. 😀
Love to see this too hope it works out.
After just watching this video, I too was going to recommend Adrian. I was surprised to hear he went silent on you but then seeing the comment he recently had surgery, that explains it.
That was very educational. I struggle with the analog electronics too. I would have never guessed that the 5155 monitor was composite input. That means in theory you could modify a standard IBM 5155 to watch television on the internal amber screen! How neat would that be? And you should have tried out Petscii Robots on it! In fact, I have a new version I'm about to release for the PET which has an on-screen map using the pseudo bit-map mode of the PET.
Thanks! I too was quite surprised to find out it was composite only. I was tempted to see if I could inject my sync and luma directly, thinking the composite wouldn't be crisp, but I guess it's plenty crisp for broadcast standard resolution!
I'll make sure to grab a copy of PETSCII robots from your site so that I can demo it in the future!
Nice plug there ;)
I had an early monochrome mac, the electronics were fried, so I was going to build a TV inside it, however I waited too long, and the whole idea become too dated - as in most people would not recognise the early mac in a modern sea of IBM clones & laptops, plus with the advent of digital TV it would have just shown static so I dropped it into the bin - a decision I now regret.
The PET video is clearly monochrome (literally “one color”) and the schematics show that the luminance signal (video) is generated using digital logic, so this means every pixel is either off (black) or on (full brightness). Did any of the PETs or similar computers of the era have multilevel monochrome video output? For a black-and-white CRT this would be a “greyscale” display, but there’s probably no reason that you couldn’t do the same with green or amber phosphors. Composite greyscale analog video should be much simpler to generate than color.
Hi David, also wanted to add that the 5155 also has a composite video output out back, which outputs color video, it's very easy to use this with composite/hdmi converters and hdmi capture devices seem to accept it.
[11:05] That capacitor there is part of a pulse former, in combination with the pull-down resistor next to it.
The horizontal drive signal coming from the PET is much too wide to be used directly in a composite video signal.
The resistor and capacitor are there to shorten the high part of the pulse to about 3.3us (microseconds).
Which is why a 0.1uF (100nF) cap wouldn't work in its place; that gives a 150us pulse, which is way too long.
I wondered about that... I thought maybe that was what was happening there, but wasn't sure, without looking at it on an oscilloscope.
Its' a very basic high-pass filter, passing the high frequency of the leading edge of the pulse but filtering out the longer part of the pulse.
Correct. The capacitor lets the current through for a short time while it's charging, and then blocks the current when fully charged. Note that when the input flips back to zero, the capacitor is discharged while the output was already pulled down to zero by the resistor. This will result in a negative voltage pulse at the input of the NOR-gate. Digital gates should have protection diodes to handle inputs below 0V or above 5V, but generally don't like this. Even worse, the negative pulse current is not really limited by anything, except for the output resistance of whatever produced the original signal.
@@jwstolk Yeah, probably a good idea to use some uncritical series resistance (maybe 470 Ohms so it doesn't drop too much voltage)
I have some more questions: why connect both inputs of the gate used as an inverter to the signal and not just one and the other to GND? That would add less load to the signal, which I would think would be better.
And the second question: how can the combination of the signal in the last stemp work? The signal for the sync pulses is connected to the output of the lowest NOR gate and from there the signal goes to a another resitor. But because it goes through the part directly connected to the output, shouldn't that cause the signal from the sync pulses to get "overridden"?
EDIT: Or does that pull the signal below 0V, when a sync pulse comes and the video signal inverter is not driving the signal high in that moment?
Dave is so wholesome - I love this content, just a retired guy having fun sharing what he wants with the world. I appreciate your contributions Dave, have a great day buddy!
I appreciate that!
Pleased you liked the Mini PET 40/80D, and sorry you if there was an issue with support, I don't know what happened there. For future reference, there are 2 pads marked J6 next to the composite video jack that I added for exactly the purpose of connecting an internal composite video monitor. There is also a side reset button so you shouldn't need to quickly cycle the power to reset the machine (and also pads to add your own reset switch if you want to).
I am (or was) an analog hardware guy and enjoyed your simple and clever solutions to this problem. I have done some work in the opposite direction, breaking up composite video into the sync and signal. The LM1881 chip makes this easy and hanging a scope on one will really teach you about composite video signals.
BTW, I started an MSC in Physics at U of S (Saskatoon campus) when you were 3 weeks old, That was when I learned of small computers, a PDP8 and that lead me to my passion for software and hardware. Still doing it over 50 years later!
I was blessed enough to have a Commodore 64, a 1541 DD, a dot-matrix printer, and a gorgeous amber monitor. I had a geek dad and know exactly how privileged I was to experience this stuff first hand. I might add ... my eldest boy is now a fully qualified mechatronic engineer on his way to a PhD ... the first university educated person in my immediate family. Cheers Dave. $0.02
I will always have a soft spot in my heart for Commodore PETs. We had a few of them in my Computer Math class back in 1981/82. We had several 40xx models and one 80xx model. We also had an Exidy Sorcerer. In total, we had five computers in the class for about 20 kids. I was addicted to this class; before a month in the semester had passed I knew more about programming in BASIC than everyone else including the teacher. I was drawn to programming and it all simply made sense to me. That all lead to the start of a great career. I still love programming today and it all started 40 years ago with my preference in using the PETs in that class.
That PET, with the cassette deck and chicklet keyboard, was the first computer I'd ever touched. That was in high school, in the late 70s. Memories! Now celebrating 38 years in IT.
Dave, that burnt component marked L2 is likely only a small inductor, you could like just replace it with a jumper.
You could also throw in a 10K resistor, if you are uncomfortable with the jumper; just to test everything.
When working with video signals an oscilloscope is very useful to make sure all the voltages and timing relationships are correct.
this
When you mentioned the LGR vid of computer reset I literally went down the biggest rabbit trail of my life, that story is really interesting. Thx!
This may be one of my favorite PET repair videos! I would have never imagined an IBM 5155 monitor would have worked!
I'm currently working on a monitor repair on one of my vintage systems, a NeXTStation, although finding the appropriate replacement CRT for that system, or even one that will fit the existing chassis has proven to be nearly impossible.
The purpose of the 2200pF cap (C1) on the schematic is to block any DC coming in on that line.
Fantastic. I'd seen the amber PET in the background and wondered about it. The connector on the CGA card in the 5155 was originally used for an external RF modulator option on the 5150, and there is a power output pin for it.
Dave: your attention to production detail with the colours and bokeh balls are great! But your balance between what you know and don't know about electronics is perfectly calibrated, and this results in a great balance between modesty and arrogance. I love your insights and explanations. Keep it up!
Much appreciated!
Here's a proper UA-cam channel in the old spirit of things. Just a guy in his shop sharing what he likes for the fun of it. Kudos to you Dave.
Regina pride! Can't believe I randomly stumbled across this channel a long time ago and still watching it to this day. I also live in Regina!
Dave, Composite mono video has 3 significant voltage levels; "black" level is at ~0.3v, "white" is at ~1v and the sync pulses are "blacker than black" at 0v. While video is being displayed, D1 is clips the mixed sync to 0.6v (the forward voltage of the diode) which is divided by R2,4 & 4 to 0.3v when the video is black, and 1v when it is white. During the sync periods the AC coupling of C2 takes the voltage at D1/R2 negative and as the video is always black, R2 is able to pull the output level down to 0v.
Hmm... for some reason I thought black lever could vary from 0.3v (black) up to the 1.5V, no? It's purely tristate?
@@DavesGarage Any voltage lower than "black" is considered to be a sync pulse. anything above 0.3v is black to white. Of course on a monitor that only shows black or white, the voltage should only be 0v, 0.3v or 1v. Sometimes the output levels of composite video stages are higher than "standard" and rely on the input impedance of the monitor to drop the voltages to the correct level, so if you measure them with no monitor or the wrong monitor connected they can be higher than expected, although older TVs tend to tolerate this somewhat. Note these voltage levels are approximate in reality which is why televisions and monitors have brightness and contrast controls! There is a very good wikipedia page "Composite video" that has some great diagrams that explain and show this.
@@DavesGarage Technology Connections has a fantastic series on how analogue TV actually works.
The 1-kohm and 390-ohm resistors form a voltage divider which multiplies the voltage output by the luminance (video) NOR gate (acting as a digital inverter) by the ratio given by
390 / (390 + 1000) = 0.28
In other words, if the voltage at the luminance NOR output is Vx, then the voltage at the output (where the 390 and 1k resistors meet) is
Vout = (390/1390) x Vx = 0.28 x Vx
For a 74LS02 with a 5-Volt supply, the typical logic-high output level is about 3.4 V. Putting this value into the formula (as Vx) gives 0.95 V at the output, which should correspond to the assumed highest voltage level of 1.0 V.
This voltage-divider ratio assumes the the “input impedance” (basically resistance) of the circuit that this one is driving is much higher than 390 ohms. If this is not the case, then the “390” term in the ratio has to be replaced by the “parallel resistance” of a 390-ohm resistor and the driven-circuit’s input impedance (call it Zin):
Zpar = 1 / (1/390 + 1/Zin) = 390xZin/(390+Zin)
Zpar will always be less than the smaller value of the two resistances.
If the input impedance is not much higher than 390 ohms, the voltage-divider ratio will decrease from 0.28, so the output voltage may be too small.
Vetco saved my butt in college robotics. We blew up a motor driver a couple days before we had to leave for competition, and they had one we could make work on hand. Pretty great place!
This brings back memories...back in the 80s I used to earn a bit of extra money adding the wiring to fit the extra RAM to the PET motherboards. The local Commodore dealer was a friend and used to drop one or two a week off for mods. We used wire wrap wire for the jumpers and I got to play with the PETs until they went back to the shop,,,,,
11:11 - Dave I think this cap is an 'AC coupling' cap - it blocks DC and only lets through a short pulse when there is change. You should invest in (if you haven't already) a cheap Siglent scope. Seeing these signals on a fast-refreshing a trigger-locking screen can be an amazing eye opener as to what's going on.
Thank you.
FWVLIW: I would suggest the best person to teach something is someone articulate who has just learned themselves. Why?
-They actually know rather than thinking they still do, while remembering what ignorance is like.
-They are unencumbered by nuance that will only confuse new students.
I find your videos informative and interesting, but I think it's your presentation style that most appeals to me, which I find calming.
I stumbled across this channel a short while ago and couldn't be happier with the contents.
Brings me back to one of my childhood obsessions of wanting to build a MAME arcade cabinet with a proper CRT screen.
Ah... the lengths we go to make 1's 0's and 0's 1's for the correct duration. I learned a lot about the NTSC signal back in the day building and adjusting descrambler boxes for those signals you mentioned 🙂
Thank you for the video, I look forward to each one. Ben Eater has a great video series on building a video card from scratch using logic chips which, I think, may help answer some of your video questions.
This is true, but to be fair, Ben's videos are for a digital VGA, which is way different than the analog composite he's doing here, although there's a lot to learn about sync signals and "front porch/back porch" I guess... In any case, I agree, EVERYONE should watch all of Ben's videos too - they're amazing!
Hey Dave, I have an original 4032 PET sitting in a closet with the rare purple screen. (in addition to a 2001 and a SuperPET 9000) These are where I learned to code so I'll never get tired of content on these machines.
A purple CRT? That sounds awesome. Do you have any pictures of yours or similar ones?
these videos fly way over my head, but i always hope just a little crumb of smartness will rub off on me... so i watch them all the time.
Those good ole days when you actually for a schematic with a product. We should go back to those days.
That capacitor in combination with the resistor is a high pass filter. It will only pass a portion of the sync signal through. Probably to shorten the sync. And that’s why your large cap skewed the sync, as it was there too long.
C1 is used to make the vertical sync a short pulse. For something that small you could get away with a "gimmick" capacitor. C2 and D1 make a DC restorer. You need a signal that goes from 0v to 5V. Without the DC restorer the signal at that point would be from -2.5V to +2.5V. Resistors R2, R3, and R4 are used to make the 5V peak signal into a 1.5V peak signal. The sync pulses and the video are needed at different voltage levels. The fancy resistor work not only divides each signal into a proper level but also "adds" them together, plus makes the output impedance match the monitor impedance. From the values given it looks like the monitor is expecting about 300 ohms impedance.
Kudos for giving Vetco a shout-out. I LOVE that place!
Having watched as many videos I have on old televisions I knew that first capacitor would be an issue. I must admit I to probably would have forgotten that power switch on the pet as well. Sadly the last fail I had was not bothering to clean my work space before servicing one of my own computers and dropped a 40 GB onto the concrete floor. The computer with out a doubt refuses to post with it installed so it is safe to say it died. Sorry for the long comment but having just replaced a ten year old keyboard it feels surprisingly good to type on a new Logitech keyboard. It isn't the kind of keyboard I'm 100% happy to have as it doesn't have the mechanical aspect I would want from a good keyboard but for a membrane it feels good enough to enjoy typing on my keyboard again.
Even if we're a small group, we appreciate this content Dave!
Regarding the 2200pF cap & 1.5k resistor, it's for converting the PET's horizontal sync pulse to the right length for composite video compatibility.
Get yourself a scope, and it'll all make sense, as a gadget guy you know you want to! Rigol DS1054Z is remains a common entry-level crowd pleaser for example. For those with gear acquisition syndrome, Keysight (the good bit of what was HP) and Tektronix are the brand leaders: I use all three on my bench, but Keysight is my go to scope due to its snappy UI.
Danger Will Robinson! Danger! An O'scope can become an interest, confusing, educational er "fidget toy".
On the other paw, all the calculus and mumbo-jumbo aside analog electronics are fairly easy to get a feel for, especially if you tinker around a little. The practical stuff falls back to relatively simple algebra. More advanced stuff falls into basic trigonometric identities and arithmetic. It's only when you start piddling around with much higher frequencies and ham radio sorts of things like fancy antennas that it become heavy lifting math to understand - and arithmetic coupled with faith to use. An O'scope is a WONDERFUL tool for visualizing what is going on.
{^_-}
16:30 About the 7812 regulator: If you want some more juice, just solder a second 7812 straight onto the other's pins. The circuit inside the 7812 can handle multiple outputs tied together. With double the current you can't go wrong.
I thought when you did that, one tended to try to power the entire load because of slight differences in the output voltages? I was always taught NOT to do that.
A fun trip down memory lane... Thanks for putting this out.
Monochrome CRTs don't have a degauss cycle - only colour ones, where the shadowmask can get magnetised
I got that far into the video and was about to post the same comment! I'll watch the rest of the vid, now! :-)
My first home-built computer got an old monochrome TV as monitor. Just that I bypassed the normal electronics for modulating the beam to get 80 decent characters/line instead of somewhere 32 or 40 that was more in line with expected TV transmission bandwidth.
A bit extra work since the TV did not have any video input so first task was to figure out where that signal first showed up from processing the antenna signal. But I still needed to inject image data there for the horisontal and vertical sync part.
"Decoding" electronics without access to any schematics can be quite fun.
So glad I ran into your video. I also have one of these service part monitors in the same packaging from Computer Reset. Now I know what it is. I was planning to fit it in a dead crt Mac 512ke that I have to make a amber Mac!
The value of C1 and R1 are both critical because together they establish an RC circuit establishing a timing circuit. This basically creates a specific delay between the incoming signal and the signal going into pin 6 on the gate. R1 will start out discharged, and then will charge up at a specific rate through C1. When the voltage level at the junction of C1 and R1 rises to the proper level the input to the gate will be activated.
The time necessary for a capacitor to charge up to one time constant, (1T), may be expressed mathematically as RC Time Constant, Tau:
τ=R×C
This RC time constant simply indicates a charge rate, with R in Ω and C in Farads.
I enjoy the old hardware talk as much as the software. I much prefer learning ASM and the 6502 and the base hardware over a python script any day. Great stuff. (Also a big fan of Ben Eater and his 6502 project)
The 6502 is way too limiting imho.
I learned on an IBM 1401 then an IBM-370. Best teacher in the world. The o/s could not be beat and live worked on .or machine than people have read about.
When I played with the Intel series (4004, 8080, 8088 and up) I thought what a cute toy. The instruction set is very limiting, yet people seem to enjoy tinkering with it. I felt that IBM should have invested more in schools with mainframes, but they were beyond the finances of most schools. I was very fortunate to have learned it when I did. And as far as im concerned, there is IBM mainframes, then the rest of the toys out there.
There is sheer joy on coding on a mainframe and assembly was my main language for over 50 years
Our first PET 2001, although 8K, came with built-in cassette player and the small “chiclet” keyboard. We didn’t get a standard(ish) keyboard until we moved up to the 4032.
There are brand new amber screens? That's amazing! I love amber monitors!
Technically it's new old stock. I'd imagine if you know where to look you can still pick one up.
18:55 you mentioned having to solder wires on the board to get the composite output because of the RCA connector's location.
At 18:10 we see the bottom of the board, follow the traces... the composite video signal appears on the unpopulated 2-pin header J6. A molex KK100 2-pin connector would be dandy for this output.
Biggest Legend in computing. Love you Dave!
Get yourself 'The Art of Electronics' by Horowitz and Hill. I would suggest getting the companion lab manual as well and start playing around. LTSpice is a great circuit simulator also. Analog electronics can be great fun.
@16:00 Great call on using a dedicated PSU. I'd have done the same. Power supplies and circuitry fail pretty often. This way, if the PSU failed it's easy to swap out and likely won't damage the main board and other parts (as modern good quality PSUs have over voltage/over current protection and so on).
Hello sir. Interesting timing. I am currently working on composite out for my PET 4032. It took a while to find some schematics. Eventually found 2 of them. First was the one you are using and the other was from a book called The PET Revealed by Nick Hampshire. Too bad that an assembled solution are being sold anymore. Once I pick the best solution, I will be working with a friend to get them out there as a kit or pre assembled unit. Thanks for sharing and love the amber video over the green!
Hi Dave,
As always, I enjoy your videos, especially when you switch to a bit of hardware 😉
Anyway, regarding video etc... a book that I had which I found the most useful when I was trying to sort out S-Video output from an FPGA project was called, "Video Demystified - by Keith Jack"
I'm sure it's either still available or even as electronic version. It is quite a large read but was the one source I found useful in explaining and getting the interface correct, in my case.
Anyway, keep up the great videos you do.
Excellent book (with several revisions, I believe)!
Really envy your access to Pet computers so early in your life. I am about your age and had the first micro when I was 17 (Sinatra?), but it was the Atari 800XL. As to the circuit -> I would personally dig into the monitor's schematics, to find the 3 points the H-Sync, the V-Sync and the Luma are internally generated from the Composite input signal. Then I would connect the output of the (inverted if needed) separate signals from Pet directly there.
BTW, any separate switching (high noise) power supply is an overkill - the 3 gates used in 7402 require negligible power - you'd tap it from the linear supply of a Pet or the 5V of a monitor.
Very cool. I learned to program on a 4K pet, so this is a real blast from the past.
Mono monitors don't degauss, that's only a colour tube thing so it's a different fault, if you hear a 'thump' when you switch on it could be excessive inrush current which might be as a result of a shorted diode in one of the bridge rectifiers.
I'd definitely not run the monitor and board off that 7812, the Commodore original monitor had its own onboard 7812 which would suggest to me that it would be a good idea to add a regulator from whatever supply connection the original one had.
Also, it might not be a bad idea to add a crowbar circuit, in case the regulator fails, as happened with one of CuriousMarc's computers.
Just measure the current the 7812 is already using and add it up. If it’s under the limit, than your generally okay.
Just as important is how much voltage is feeding the 7812. The voltage drop multiplied times the current for the Wattage. This is going to give you an idea of how much heat is going to be produced. Run it a while and check the heat. If it’s too much then you’ll have to add cooling. Then if it’s still to much, then leave it on separate supplies. But I strongly suspect it’ll be just fine.
The cooling solution needs to be taken into account. 7812s on 1.5A needs a reasonable heatsink and at least some airflow (passive from the rising air due to heat could be okay).
Also it might introduce weird noise from the digital board into the CRT or the other way around. And also you would need to check if the CRT supply needs to be isolated in which case you just can't attach it to the internal regulator. I doubt it needs to be but better make sure.
Why not adding another one, from the input of the existing one? That way, no interference between digital and analog boards, and also no overheat of the IC.
Also, I believe that 12V is used in digital board just for op-amps and maybe for serial RS232 if exists (it requires +12/-12V).
0.8A is a lot to ask from a 7812. It _can_ do 1.5A, but keeping it cool while doing so is going to be non-trivial.
Great job Dave, brings me back to 1982. I have an old Odyssey computer / gaming system if you really want to go retro
Can you imagine a collab between Dave Plummer and Dave Jones from the EEVBlog?
One can only dream...
I can feel your trepidation about doing the composite signal conversion. Analog electronics are voodoo compared to digital electronics. Very nice work getting that up and running. I would have, just as a personal preference, done whatever I needed to to retain the original PET motherboard. But its nice to see that the mini PET is a viable replacement.
Thanks for sharing this. Thumbs up! :)
I still have it, and will go back to it when I get the ROMUlator if I can, and will do a little followup on that!
Small nitpick: CVBS means Composite Video Baseband Signal. Baseband in this case means the signal is not carried over a UHF or VHF RF band.
According to Wikipedia both are correct.
@@DavesGarage Cunningham’s law meets shroedingers principle
Capacitor looks like it is to block DC. with a digital gate, it will shorten pulses to a fixed time. when you used an 0.1u it probably output a pulse that was too long.
Dave, talk to Electroboom about your circuit. He could talk about this stuff all day, and according to other collabs he's worked, he will and does talk about this stuff all day. :D
I always thought that stood for Composite Video Baseband Signal.
This kind of content is worth more than gold!
I love the @LGR plug. Clint is awesome!
A 7812 has internal protection so if the voltage does not drop as it heats up, it is not catastrophically overloaded. It may still run hotter than may be desirable for reliability.
A second regulator helps a lot since you are adding a second package with twice the surface area contacting the heat sink, so the die temperature is much lower with 2 regulators on 2 small heatsinks compared to one regulator on a larger heatsink equal to the the two small ones.
Funny how all our rabbit holes seem to crisscross. Whether it is Ben Eater, LGR, 8-bit guy, Adrian's Basement, or your channel we are all raising our inner nerd.
Thanks for another neat video Dave. I love the amber display... I always preferred amber or white CRT phospor to green *hulp*...
I work as a broadcast engineer for a large television broadcaster in England where we had the superior PAL colour TV system and the correct spelling of color ;-) before everything went all digital. The NTSC system could not correct for colour phasing errors causing hue issues. In fact we nicknamed it Never Twice (the) Same Colour. The reason for the three grounds is each cable screen gets its own dedicated ground pin rather than trying to land three screens on one connector pin.
Noise can only be eliminated by using differential circuits where the signal on one wire is 180deg out of phase with the other wire. If noise is introduced on the wires both would be the same phase so on the receive end the two in phase noise signals cancel each other out. On differential circuits a ground is not needed technically but often used to ground the overall screen surrounding the two wires
Hmmhmogmmgmgggg
Haha NTSC was first and was standardized in the earl 1950 to be fully compatible with the millions of b&w television sets that were already in the homes.. You guys were late to the electronic TV broadcast game and didn't re-engineer color TV til the late 60s. The UK also didnt have to contend with the millions of television receivers thay were already in homes. Not to mention you only had 1 television station (BBC1) where we had thousands.
In 1964 the UK had 2 stations.
Also the BBC suspended operations from 1939-1946 die the the war, but here in the US we still had stations on the air, specifically WABD (the 3rd network) continuously since it first fired up it transmitters in 1938 making it the longest continuously running television network in the world. Even longer than WCBS, WNBC & WABC
To change a standard in a country that had millions of TVs and thousands of television broadcast stations at the time would be cost prohibitive.
Its always easier to improve on someone elses standard. Its much more difficult inventing things in the first place.
Electronic TV was an effort by many people and not one company (or person) can take credit, although Alan Sarnoff of RCA tried to lay claim for the invention of electronic television, by buying out companies that held certain patents. Just like bill gates did.
I worked at WABD and I lived the history. Now I volunteer to restore broadcast video tape machines made by Ampex to keep that important history alive
@@rty1955 - All colo(u)r TV standards had to be backwards compatible with existing black and white televisions in customer homes so yes we had the same problem when NTSC Color was introduced.
The issue was how do you send colour information over the air to new sets and not impact the reception of existing black and white televisions. This was achieved in both solutions by superimposing a colour subcarrier on to the TV signal at such a high frequency the black and white television does not see it.
However with a suitable decoder you could extract the luminance (black & white) and colour difference signals (R-Y) and (B-Y). From that you can then create the red, green and blue signals to the guns on the cathode ray tube.
So yes NTSC Color came before 625 line colour which in turn had many variants with PAL improving on NTSC’s handing of differential phase errors but incorrect that we did not have a large user base of existing black & white televisions to support as well.
@@gordonm2821 tell me how many receivers & broadcast stations the UK had when the UK implemented the PAL standard
@@rty1955 - Not sure why you are asking. You said we did not have to worry about existing television when we launch colour TV but we did.
A 74LS02 takes TTL logic levels which are distinctly DC. I'm guessing that the signal from J1 pin 3 is AC. Components C1 and R1 filter the input to match the voltage levels expected by the nor gate, blocking the DC offset from the signal and providing something more appropriate.
At the time, the CRTs were only meant to display image on the screen, there was no difference between the CRTs used in common B&W TVs and those used in computers monitors. Wasn't possible to swap just the picture tube? Or the monitor board was nonfunctional inn the PET? I swapped a CRT twice and worked well, once a green phosphor with a amber and once with a white one. All the CRTs were the same size and had the same pin-out and size at the socket.
Great to see an amber screen computer, I always loved them.
About the 7812, if unsure, you can add another one, from the same input as the existing one to power just the monitor. Make sure to add the capacitors required and a heatsink.
Fantastic video and project from start to finish Dave! So far anyway, I expect to see more on that PET in the future. Entertaining and very educational. I'm a digital software guy too (not on your level though), but I also try to learn about analog hardware as well.
0:35 you did the salute with more than 3 fingers
Thanks for this, analog video is huge blind spot and every little helps !
Ah, Vetco. That place is such a gem. With Fry's being gone, there's really no other place you can go to pick up a Raspberry Pi or Arduino clone locally.
I would love it if you started a series of tutorials for OS development, explaining the concepts, tools, requirements in terms of knowledge, materials and books. Of course based on the X86 architecture. 🙏🙏🙏🙏
R3/R4 voltage divider is there to lower the 5V from the logic chip to around 1.4V, which you can see listed in the description of the composite video specs of the monitor.
I'm kind of interested in acquiring a PET, however they are rare as hens teeth on the other side of the planet from you. The mini-pet looks like a good solution. I have a selection of composite monitors, including a brand new green CRT, However the case and keyboard are stumbling blocks for me.
I did see a Commodore inspired mechanical keyboard on Drop/MassDrop, not sure if they still have them, but that could be a start. Now all I need is some drawings and a sheet metal fabricator to whip me up a case.
"I'm a digital software guy and this is an analog hardware problem ... but I'll make it work anyway." LOL AWESOME!
The capacitor generates a short PULSE when the input level changes. The sync signal from the board is too "wide" for the composite signal standard. Time constant of that circuit is given by the R C values (product).
this was very informative - I like this style of video
Correction at 6:06 - the IBM Portable PC does not have a full monochrome display adapter in it. The IBM Portable PC comes with a color graphics adapter which can provide a composite NTSC video out to allow it to be connected to an external monitor, as well as a TTL output to connect to a IBM Cooler Monitor. The inbuilt screen is monochrome only, and is horrendous to use long term (hence the use of composite out to allow the poor sod who carried it home from work to plug it into a better screen.
11:05 in Russian we usually call this type of capacitors "barrier" or "connection" caps. Basically it prevents DC from passing while lets AC and signals to go through...
I might be wrong but meh whatever, guess is a guess after all :)
Awesome video Dave, loved it!
Any chance of doing something with the amiga would love to here your storys about them
My guess would be that those caps are working as some sort of filter, and the mismatched value caused it to filter out some of the signals you actually needed.
That capacitor is allowing the ac signal through but blocking dc. Probably to avoid any dc ground loops.
beautiful machine and that screen is awesome..
Love this. Old school repair guy.
I've tried to watch this video like 5 times today in the last 2 hours, and I've not been able to get it all in my head due to external factors. Unfortunately, those external factors pay the bills, so, your view count is going up significantly today Dave. heh
Adrian Black from Adrian's Digital Basement could help. I think I have seen him troubleshoot and fix a few Pets.
I think Adrian is in the same area as you as well.
Yup. See the pinned comment. No sure why he's not answering email!
@@DavesGarage I think he is swamped eventually he should get back.
*Why so many grounds:*
For one, it'll reduce the resistance of the ground connection between circuits, which is always helpful. More than likely, though, the ground-per-signal is to reduce inductance. Especially since the signal and ground are twisted together. With a high-frequency analog signal, you would probably do better with a coaxial cable instead. (A small coax wire is fine, doesn't need to be like satellite TV cable.)
*Using a 7812:*
Don't try to draw nearly an amp from a 7812. You won't likely be able to keep it cool enough without some heroics - at least depending what the input voltage is, and thus, how much it has to drop. I would take a look at the monitor circuit to see what its B+ voltage is. It may run quite well from whatever your unregulated supply voltage is. Monitors usually don't need a tightly-regulated B+. They derive their voltages from that rail and regulate them internally.
*Composite Video:*
I think you might be better off approaching this solution the other way around. Instead of supplying composite to the monitor, why not look at where the composite input is divided into sync and signal? Mono video + VSync + HSync is far easier to deal with -- it isn't, but may as well be, a digital signal. At least when "mono" is actually just, literally, black and white, and not grayscale. It may be more convenient to use composite with the retrofit motherboard, but if you have a working original (at least with a drop-in RAM substitute) then why not use that? Otherwise, just emulate the entire thing. :-)
Dave Jones needs to do a review or collab with this Dave
The capacitor (C1) is acting in conjunction with the resistor (R1) as a high-pass filter. My guess is that the line before the capacitor has a dc offset issue that the high-pass filter is correcting for. And maybe also helps with pulse shaping, but I don't know enough about this specific standard to say one way or the other. Typically pulse shaping is not of importance in digital circuits as the very high open loop gain of the follow on digital devices will slam the signal to the rails given even a tiny offset from the mid rail, thus destroying any pulse shaping some one is trying to accomplish (excusing pwm). Now C2, D1, R2, R3, and R4 might be something closer to a pulse former. I'm fairly comfortable saying that R4 is helping with impedance matching. Given that the NAND gate will present as a very high conductance to AC ground, we can see R3 and R4 make a parallel combination to AC ground and present a load of about 280 ohms to the "composite video out" line. The weird component to me is D1 as that will basically clamp the anode side to the forward bias of D1, what purpose this has I'm not sure of just yet. It would be very interesting to see what an oscilloscope shows at the various nodes in this circuit.
I recommend you watch Ben Eater's video about his video card, although it deals with VGA signals, the idea of sync pulses and blanking time is still present due to vga signals being based on crt style monitors.
Although I would agree that Ben’s “world’s worst video card” video (or something like that) was a great introduction to how HSYNC and VSYNC and component video (RGB) signals work together (but as separate signals), the circuit used here combines the logic-level synchronization signals and the one-bit luminance signal (off or on) into a single signal that will be used as the single input to a monochrome monitor.
It would be great if Dave had a multi-channel oscilloscope and could show his results (and oops moments) like Ben Eater does.
great coverage again Dave!
👍
Glad you enjoyed it
Horizontal position on old mono monitors is sometimes a tall(ish) inductor, it would need to be adjusted with the correct plastic tool, never use a screwdriver it shatters the inductor slug.
Is that the plastic tube dohickey wraped in wire coil? Always wondered what that was! In my old Telstar touching it seems to slow the clock down?
@@DavesGarage always use a plastic adjustment tool. the metal screwdriver will load the induction tuning down and change the resonance value, and well as possibly shatter the ferrite tuning slug... then bad things happen to good people! :)
I don't think I've heard of using coils like that for horizontal position. I would suspect horizontal size or frequency instead.
@@DavesGarage Yep, that is it. Might be Horiz, size though, depends on how the monitor is designed (plus, it has also been 30 years ...). It has a slug of ferrite in it, can be very brittle stuff!
@@eDoc2020 Yes, I think you are correct. I seem to remember having to adjust CRT monitor chassis using a plastic trimmer, some controls on pots, a few on inductors, but it has been 20 years since I did it for real though. I think you are corrent and hpos was normally on a pot.
C1 is a decoupling capacitor; it transmits AC while blocking DC. That way you can eliminate any offset DC.
That's what I was thinking, after having watched a lot of Uncle Doug's amp repair videos. But I wasn't sure that would be the case in such a different circuit and there are other comments here that sound like they think it's something else, to shorten the pulse or something.
@@davidtaylor6124 Yes, the lower de caps capacitance, the higher the transmitted AC frequencies, the shorter the pulse. Actually, it could be ment to make two short pulses of one long one, but we can't tell because the connector is an enigma. But I am not an electronics engineer, I just have had electronics as a hobby some weekend haha ;) Have a great day.
@@davidtaylor6124 Actually, I think you are right; the capacitor is probably ment to convert a long digital pulse into two short ones. It makes no sense to decouple just one port, so I think you are right.
Amazing times when computers came with schematics and service manuals.
Wow that Vetco looks nice. We have nothing like that in Lansing.
That was the first computer I had as a child. Still have it too.
I'm guessing that your existing PET monitor had more than just a simple bad CRT issue? Like power or control circuitry was also bad? Because it would've been trivial to pull that CRT out of the IBM monitor and drop it into the PET's monitor, then adapt the PET's CRT connector to that amber CRT's connector. Heck, the pins might have matched exactly, since there are only so many ways to wire up a monochrome CRT internally.