TSP #65 - Teardown & Repair of a Matsusada RE-600-1.6 600V 1.6A Power Supply
Вставка
- Опубліковано 15 жов 2015
- In this episode Shahriar attempts the repair of a Matsusada RE-600-1.6V 1.6A power supply. This low profile power supply is intended for rack-mount application and has been likely used in a vapor deposition system before it has failed. The scent from the unit hints at capacitor failures due to thermal stress over a long period of time. The teardown of the unit reveals the low-profile construction as well as the ventilation air flow. The faulty capacitors are replaced with higher temperature rated units. Unfortunately further faults are detected and it is revealed that the main switching transistors have also failed. All part number of all ICs and control modules are either sanded off, or permanently potted. Since no schematic is available, the repair has to be temporarily put on hold.
The Signal Path
www.TheSignalPath.com - Наука та технологія
Just started watching, but I have never before heard such words: *"My girlfriend wanted to help me repair this broken power supply"*
+pillsnotbills . Did you get your girlfriend from Ebay, and do you have the link, because I want one?
+pillsnotbills Haha. :)
When you measured the short on the gate I was like "There goes the driver". Nice video!
I'm very much enjoying the frequent videos from you lately! They are always a pleasure to watch.
There are two ways to get inside the module. (1) Find the chemical that melts the black epoxy. (2) Lots of heat, so that the epoxy becomes soft enough to dig out. I have done this and it works. (3) dig out the epoxy with the Dremel and a cutting bit. I have had success this way too.
When powering up something like this after a repair, it's a good idea to put a resistor in place of the fuse to limit current if things go wrong.
Amazed to see such amateurish methods used in a Japanese PSU, but who the hell has heard of Matsusada nowadays... PSUs are always subject to failure, so making one unrepairable is unforgivable. There is no reason for the drive stuff to be a module except to chisel money out of customers for spare parts.
+mikeselectricstuff He should send the module to you for X-raying and bandsaw-extreme-teardowning :D
+Róbert Valdimarsson
That would be incredibly awesome
i use resistor + fuse combination.
+mikeselectricstuff when the fuse is on the primary i usually use the poor-mans-limiter incandescent light bulb. At least the brightness gives a indication of current and limit at the same time. I would prefer a proper limiting PSU thought :(
+mikeselectricstuff Hey Mike, yes, as always being a little lazy costs extra time! I am going to invest in a good Variac also. Too bad they made it impossible to repair, it would have been nice to have this thing running for future experiments.
I thoroughly enjoyed this video.
That module can be unpotted. I use to work in the automobile aftermarket industry and we used to unpot modules all the time. We had to unpot the modules in order to reverse engineer modules so that we could create our own modules based on the original design. For those in shock out there.... No... It wasn't a copy. We analyzed the OEM circuits and came up with our own designs which were often times better than the OEM. It's quite legal.
The unpotting was similar to how you uncapped an IC in one of your earlier videos.
Unfortunately, I don't remember the chemical used.
If you were to contact the aftermarket auto electronics industry, you may be able to find the name of the chemical.
Well if nothing else we all know what brand not to buy! great video as always
Email me and I' send you the PCB layout and the details to of components layout. Simple. Three diodes, a few caps and two pre driver transistors (one an NPN driving a PNP Darlington that drives a larger driver 2SD880 that drives the N-Mosfets Simple to etch and populate the board, that is driven by a simple PWM using a TL494CN. The Darlington is 2SA1020 driven by a 2SC1815 that drives a 2SD880 that switches the N channel FETs. The diodes are HER208F (FR107) SR260 (FR207 1KV) A pack of Zeners 5W
1N5711 1N5819 1N5399 33uH a smattering of 100nF decoupling caps and couple of 10nF to keep of any stray RF.
Good to go.
As in some of Mike's videos, a Dentist, or Veterinarian might be willing to X-ray the part for you. Depending on the potting compound, you may be able to chemically soften and remove it (I think Mike has had pretty decent success with dichloromethane paint stripper). Ben Krasnow or Jeri Ellsworth might have decapped those transistors to see if there were any markings on the dies (which might be useful if you're successful in de-potting the module). While I suspect this is probably more involved that you'd like to get, you might also be willing to do it based on principle as well.
Hats off to you, Shahriar. You're analytically very strong. So you hate this power supply? Well guess what, I hate EVERY switching power supply ;)
It looks like mosfet driver / amplifier circuit. Maybe you can find input control logic signal and put generic mosfet driver like ADP3633 or for your parameters.
Great video, :)
Dig out a TL494 and an IR2110 or 50 Gate driver to replace the PWM gen block. You used a clone component in your magnetic levitation circuit.
Oh, throw a few KSC3265 and KSA 1298's as push pull drivers and your up and running.
This HV topology is used in several KW Crown PA amps, is bloody reliable and highly configurable...
If you are sufficiently motivated you can unpot those epoxy encapsulated units if you leave them sitting for a few weeks in dichloromethane. Not a pleasant process and they won't be reusable after but you could get the circuit details.
I know how it feels to blow money on parts, and get a bad outcome. But you did a wonderful job. THANKS.
So much respect for your videos and the things you do for the community brother. My only suggestion would be to upgrade to a higher resolution camera to improve the quality of your videos
Bummer they hid the parts like that from you and the community at large. At the very least even if it can't be fixed, you've got a pretty decent collection of HV parts in that thing that could be re-purposed for other experiments. Impressive work as always even getting that far.
I would have liked to see more on the process of measuring the transistors and selecting replacement parts.
A repair tip: Use a variac or a resistor to limit current so you don't blow the input section until you are sure you have fixed any dead shorts.....Would have saved you from blowing those resistors. :)
You can use heatgun for removing the epoxy of the module i've tried before it works pretty well. i like your videos they are very professional
Thank you
Another great video.
Nice video with a nice style of presentation. Too bad about the secretive module at the end that prevented successful repair. Never mind, keep up good work.
I am old school, I pretty much always use my arrangement of switchable light bulbs for current limiting and a variac to bring up the voltage on mains powered equipment under repair to make sure nothing explodes.
Anyway, yeah, those potted modules are pretty awful, seems ridiculous, I am sure anyone wanting to copy their design could just as easy design their own.
In that case should be the gate drive trasformers. Check it close to see if two transformers are sticking out of potting. You cold replicate that device quite easy. It should have two pins for powering the input buffers (like TC4422) something around 12V , also with the scope check to see if you have the PWM signals on some pins. Check also the two resistors that are sticking out of potting, as those could be in series with the GDT.....
In situations like these, I often ask myself: What would Mike do? And Xray comes to mind :) And I guess you can consider yourself lucky that gate driver decided to die while keeping gates in low state...
At least you have a whole bunch of nice parts that you can reassemble into some power supply of your own ;) Btw. for the case of identifying the proper mosfets, these cheap ebay transistor tester thingies allow you to rule out a few candidates of transistors by giving you rough measurements of some datasheet values.
Big electrolytic caps (well, all caps, but especially caps like those big Nichicons, or bigger) can regain up between 10-15% in as little as 30 minutes through dielectric absorption (thin film, poly(ester/propolyene)) are all around 1%, tantalums are around 3%). Always use the bleed resistor that's in series with the cap to drain and confirm with your DMM before performing any work on anything like old TV's or if you're building that awesome Vox AC30 replica
That was an interesting tear down/repair. Any way to de-pot the h-bridge drive even destructively (some kind of solvent or acid) and reverse engineer a replacement?
Do the replacement caps have similar ESR?
That date code and RU logo looks like discrete Avago gatedriver. Depending on what it's expecting on the input, maybe it's not that hard to create a replacement. Did you scope the input side? maybe trick the feedback and see if the signal going to the gate driver module changed?
Could it be that the output caps were blown not by overheating but something bad that came IN through the OUTput terminals ? I don't see much protection circuitry on the output.
Enjoyed diagnosing the H-bridge.
Will the manufacturer sell you a module? Or better: give you one for the exposure showing off the details of their quality product?
GF left just as things got interesting!
Not sure if you will see my comment considering the age of "Teardown & Repair of a Matsusada RE-600-1.6 600V" video... I am trying to bring back to life very similar Matsusada RE35-34-09 supply. It has the same architecture with the exception of output stage, which is low voltage/high current. These supplies need a jumper on DB25 connector between pins 10 and 21, without this jumper the output is disabled.
Is it possible to use plastic safe paint stripper to "unpot" the potted module? It hardened by evaporation of the potting materials vehicle, maybe you can soften it carefully and remove the potting by exposing it to paint stripper fumes. Its really the fumes in the paint stripper that does the work. There are a few manufactures but I use Chameleon Hobby products. I would definitely write down all the visible part numbers and photograph the module and possibly even do a schematic from what you can see before trying this as those part numbers might disappear when doing this. I see you just replaced the larger caps, which test instrument do you recommend for testing ESR on larger caps like 2200+ microfarad caps? Also which inductance meter manufacture and model would you recommend for figuring out a transformers turns?
Shahriar send this balalaika to me, I will make it easy! Actually in one evening with the help of an iron, such a scheme can be done!
What a shame. Obvious question: Is this gate driver module available as a spare part from the manufacturer?
Are we 'sure' that the potted module has failed? Yes, it seems likely to me also. This repair has already gone sideways for several reasons. But is it possible to surmise at what levels and switching rates the old/vs new MOSFETS must see to fire up? We have two "good" old MOSFETS...
It seems like the potted module has 2 sets of half bridge output, so I speculate that each 4 pin output is bootstrap voltage, gate 1, gate 2 and gnd. For the input side, you have (presumably) 2 gnd, 1 vcc and 2 signals, or 1 gnd, 1 vcc, 1 en and 2 signals. Go figure it out and build one gate driver module to drop in. If you can figure out which pin is which, and measure the voltages and frequency as well as physical dimensions, I will be happy to build one for you for free as soon as I got some free time.
When I get something like an IC or module which I suspect but can't find any info on it then I start searching on the board for bad parts. Bad capacitors, resistors, diodes, transistors, etc. I would not give up on it yet especially when you already invested money into it. There is nothing more I hate than investing money and time in a money-pit but that seems to happen quite often to me so I'm used to it... Anyway, nice video. Cheers :)
Mr. Carlson is very knowledgeable concerning electronics maybe he can help you with the schematics. You can find his video's here in you tube.
...I've seen that soooo often in older japanese radios where potted hybrids were used everywhere, making schematic drawing nearly impossible...
But i am surprised to see the MOSFET numbers scratched, which is normally not in Japanese tradition, especially when it is known that Japan has its own semiconductor foundries (i was expecting 2SJ / 2SK devices here, having N/NPN devices in black epoxy and P/PNP devices in green epoxy).
Too bad. I understand you Shariar. It is frustrating and only gives a bad name to Matsusada.
When you are measuring the diode junction in those MOSFETs - that is the voltage that opens the gate, right?
No, that is the reverse diode voltage of the source/drain to bulk.
Sorry the repair didn't make it (yet!). Hopefully someone will cough up either a replacement module or at least some info on the internal workings. I always appreciate when you own up to mistakes along the way, some folks videos are surely edited to delete those "slap my forehead" errors! BTW, did you get the opportunity to update your DSA1030A-TG firmware?
I'm surprised of the huge size of the circuitry, I mean a 600W psu for fex a computer is like 8 times smaller smaller. Yes those doesn't have 0 - 600V regulated capability but that alone shouldn't be the reason for the huge size.
Heat or chemical paint stripper could help to see whats under the epoxy in that custom module, most epoxies become soft at around 100C, but, since the module is quite thick, hot air gun will not work on it, it is better to "soak" it with the heat in an oven and then try to chip away the glue.
The other method (not that great for thick pieces) is to use aircraft paint stripper, or powder coating (which is essentialy an epoxy) remover, these come in aerosol type containers, chemicals are quite nasty, use gloves and preferably do it outside.
I'd use the heat method to get large chunks out, and then clean it up with the paint stripper, note that the paint stripper may attack the plastic container as well, or any other plastic encased parts inside there, or destroy component markings which are painted on the parts.
Chloroform will also dissolve epoxy (it becomes jelly soft like, can be brushed off with something like toothbrush or soft metal brush), but the part needs to soaked and worked on while it's still wet with ChF, it evaporates very quickly and once its gone, epoxy becomes hard again.
+jz1199 I'm guessing your paint stripper had dichloromethane (aka methylene chloride) in it. Wonderful stuff for dissolving epoxy. Too good. It will dissolve component packages eventually too (I found that out the hard way). Also there is some evidence for it being carcinogenic.
+Verm1n yes, DCM alone is not exactly an over the counter house hold chemical :)
+Verm1n
It also welds quite a few of the common plastics, especially acrylic, polycarbonate and ABS. Best thing to repair fresh cracks in cases. It's really annoying that it becomes harder and harder to get for private people here in the EU (because "safety").
+superdau back in 80ties as a kid I used to use chloroform to put together model airplanes, funny thing about those model sets back in the USSR times, they very rarely showed up in shops, and immediately they were bought out - one would think that they got to the shops as soon as factory made them, and production capacity was the limiting factor - but then the glue, that came with the set, was always completely dried out, which would suggest prolonged storage times...
The Japanese at Matsusada seem to be very afraid of the Chinese copying their products, so they put a lot of effort into obfuscating it.
+Raymund Hofmann Except the Japanese at Matsusada seem to be getting their fake Nichicon capacitors from those same Chinese!
Those aren't fake Nichicon capacitors...
@@SteelBlueVision They do not know how to count. In general, such technical solutions suggest that the manufacturer’s technical level is low!
have you checked that little 8-pin dil package next to that white opto-coupler? It is probably the oscillator and drive IC for the mos-fets. Just next to the prool trafo. You can always try to chemically dissolve those epoxy packages. But I don't think there's a problem in there. If so there would be discoloration and crackles, or such. Clearly the safeguard is working fine. So probably the oscillator isn't giving pulses, to do the switching. The circuits are starting up but the main power supply is not. So the drive IC must be damaged or broken. That is that little 8-pin Dil package IC next to that transformer. Namasté.
Ha, I just bought two of these exact supplies 40 Volt 80 Amp with an identical issue with no output. The only IC with the numbers filed off is that little 8 pinner. I haven't gone into any detailed checking yet other then visually which is all good except the first transistor warped the plastic sheet under it. The second one I haven't even opened up yet. Seeing your comment is a sign and I'll let you know when I get time to work on them if that IC is the problem, maybe the second supply still has the numbers intact.
What's causing your problem is the pwm gen isn't working or the gate driver pair. If the gets are shorted gate to drain you can bet your bottom dollar the gate drivers are snotted..
Each and every day I fix SMPS in multi KW and HV applications.. No fun
1M023A050N are good devices, but there are better devises for 500V working. I use 50A 1000V N channel FETs in my PSUs
and find that changing the fast recovery diodes for high power schottkey helps the FETs to stay safe. Just saying.
Are there any date codes on it? The boards look pretty ancient
+cmdstraker The potted gate driver module appears to contain components with date code 0513 or something similar at 29:54 so 10 years old. What makes you say they look ancient?
+KX36 I wouldn't have expected that many through hole components in low profile gear like this. That's all
This is why I like and buy old equipment. You get stuff that you can actually maintain, and the manuals provide full schematics. This modern desire to obfuscate everything is very irksome. Nice video anyway, and hang on to any girlfriend that wants to help you repair a power supply.
Design own control circuitry around smps ic
That's why I don't like close-source hardware. After you bought the units they'll keep information from you. It's a marketing thing not a engineering thing.. Old story.
Waiting processing to 720...hurry up youtube, i need to see this in good quality
If the modules not available IRF and others make simple to use gate driver chips, so I recon you can still get it going. Is that module next to the gate driver module the PWM generator/switching regulator? Please always have the camera running on the first power up :-) I guess you will probably use the light bulb trick or something else for protection next time.
The electrical guys say that hiding parts so they can't be copied doesn't work, yet in this case it did.
Very odd that he didn't do his usual double checks, like for other shorts on the PCB and voltages, and just trusted only eyeballing everything. Seen other elec YT channels do the same thing and blow stuff up.
very nice video always a pleasure to watch @mikeselectricstuff when can we expect a cool video from you agan
Did that green (polyester?) capacitor really fail? Looks more like glue used to hold it down rather than actually spilling it's guts..
+Tim Feleppa Yes, it had actually cracked and melted.
Leaving the module in acetone overnight should do it? I know dichloromethane can do it, but it will also dissolve the IC packages and capacitor labels:/
+TheAmmoniacal acetone will have 0 effect on the cured epoxy
jz1199 Have you tried?
+TheAmmoniacal Acetone has no impact unfortunately.
+TheAmmoniacal not extensively, I have tried it on powdercoat (epoxy), and it would not do anything noticeable after couple hours, and that is a very thin coat compared to the amount of epoxy in that flooded case
Okay then, methoxybenzene next!
Will Matsusada sell you a module? Or is it an ancient supply?
+Laurence Stant No, they don't sell anything but the whole supply.
+The Signal Path Blog Wow, that's customer service for you (not!). This ain't no HP/Agilent/Keysight, that's for sure...
MELT THAT SHIT DOWN frankly i wouldn't rest until i fixed it, you can certainly do it!!
We wanna the big review, be quick!
Can we have some spoiler about it??
Anyway is very very bad that they use custom module, I can understand use custom module in high freq RF stuff but in power supply why the hell use custum part?
Another example is in all TDS500/600/700 after the A revision. In B revision and later revision in the PSU they use a custom transistor that include the pwm modulation in it. No sense!
With all the equipment you have, how come you are relying on eyeballing the capacitors rather than using a ESR meter. Would be interesting to see how bad the caps when tested really are.
+0521Bryan I did. They were shorts.
You said you keep this one, but at some point you have to get rid of some of your equipment. What's your usual way of "disposing" of the devices? Put it on ebay "as is"? Desolder the new components you just installed and scrap the rest? Is it worthwhile for you to desolder other components from the board?
+superdau Well, it depends on what it is. For something like this, I might just de-solder come parts and drop the rest at an electronic recycling trash.
Know it's an older video and the unit is most likely gone now, but sometimes with stuff that's potted like that, you can work it out of spots with hot air...lots of hot air. It is quite stinky and takes...well...time but I have been able to get that stuff off by doing that and see things that were covered up. I'd like to see some sort of referendum put into place for "right to repair" to stop or ban this practice of intentional obfuscation across the industry. Doing that to FETs cannot be good for them having a vibrating abrasive tool hitting them like that either. We have right to repair for vehicles, why not for electronic devices? It seems overdue.
X-RAY that module, so you can have an idea of the circuit, then figure out what can be.
Too many design compromises to fit into 1U chassis.
Also haha man that sucks about the potting material. If it's silicone based (it seemed elastic enough from few frames I saw you handle it) people have had success with simply slowly hacking away at it after applying some silicone sealant removal available at any hardware store. If it's hard potting material (two part epoxy type) don't use methylethylketone/xylene/acetone no matter how tempted, it will destroy the components with the potting material. It's crude but a steady hand and a ball mill on a dremel to slowly file away at it will let you get most of the material off while retaining enough control (and tactile feedback) to figure out if/when you moved from epoxy to component (try to aim for a part on the board without components/traces on it, so you just register with FC4 instead). kevtris.org/Projects/votraxpss/unpot.html This is cool haha
Yeah... I can sort of feel the smell from here and... WAIT!!! I think I left something cooking in the oven! :D
What would be more amazing is to train the cat to smell and find the leaky capacitors ;)
Its a shame good company's trying to protect there design in such way, and make repair almost impossible
Hope you get an answer what that module is, and reverse engineer it.
It is not a сatastrophe. You can buy dimethylsulfoxide in ebay and solve this epoxy, or just heat it up and you can remove it mechanically.
Is your cat grounded?
(4) Ask the parent company to send you a new model.
my estimate: the goal "return of ebay investment by repairing this piece of failing circuit" will soon disappear into a hunt for the golden jigsaw puzzle piece, just in order to solve the lurking mistery lurking somewhere on those few inches of analogue switching design.
It's ridiculous that they would go to the trouble of sanding the mosfets when they are screwed to the inside of the heatsink. What are they hiding, state secrets? Maybe you should put the heatsink under the microscope and check for microdots.
Ah well, good video anyway. I guess Dan will be along soon to complain about how it's a waste of time. Yo Dan! Great video, eh?
+pillsnotbills After what has been found out in the discussion here about the probably faked capacitors, i suspect something going on on those mosfets too.
I seems that somebody really had a crack on the BOM during production. I wonder if they did this on the initial production batch or just as a cost reduction plan between production cycles.
Nevertheless: as this unit looks like to bei at least of 15y+ age, i feel amazed such thing having taken place already in those days.
or IGBT's
Good time to invest in an x-ray machine.
I guess its a chicken or the egg situation now. Did the gate driver go up, turn on both transistors in the halfbridge and roast that, or did the transistors go short, and roast the gate drive.
Either way, its damn shameful to grind the part number on a mains voltage switching mosfet. Its not really a hassle since it shouldnt be super hard to spec a replacement (as demonstrated) but even more insulting than an IC somehow. If it were output transistors on a class AB amp, maybe.
That being said, there isnt that many pins on that module, given that it is gonna have G and S connections for each FET, power in, and feedback. Figure out where the feedback is coming from, get some gate drivers, isolation if needed and see if you can stuff a tl494 into duty!
+CKOD Also, check to see how the potting compound holds up to heat. Most epoxy based compounds arent *that* heat resistant. For me at work, 5 minute epoxy is 'temporary', jb kwik is 'slightly less temporary', and jb weld is 'kinda permanent'
Great girlfriend, 2 show that much interest. Namasté
What a shame. This is why I never buy from anyone I suspect is in the business of stealing IP. If we buy products that are developed with stolen intellectual property then there is a market for it. If there is a market for it, companies are forced to put protective measures into their products. This increases the cost of the product and as we have seen from this excellent repair video, makes repair exceedingly difficult. If the price is too good, you should ask yourself if everything is on the level!
give me it.
this girlfriend of yours sounds too good to be true ;P
By the way I see Matsusada has 4 overseas offices in the US, one of which is in New York. For some reason, I gather that you're in that area. www.matsusada.com/company/offices.html
Also, the positive coefficient thermistors should be looked at as they also have a limited lifespan. The none use of a surge resister is poor design. I suggest the ordering company asked for the units to be made as cheap as humanly possible, like the buyer was an American company and the buyer was some wet behind the ears kid with ZERO experience in the field. One always should leave room for a good diversity factor of at minimum 30% with 50% being preferable. I wonder if the kid's grandfather was one of the designers of the Titanic?
So now its a lot of girlfriend and cats. Does the girlfriend nudge the cat into the picture. Do you have the cat just to show in the video. Is the "girlfriend" just to push the cat into the video near the malfunctioning open high voltage power supply.