PWM Woes....

So for some reason, my power supply for my HeCd laser keeps eating PWM cards... I'm thinking that the problem lies in the main 'driver' IC, which is the long IC in the picture below. Everything else on the card is pretty basic... resistors, caps, a couple little transistors. The IC is the only thing I could think of that would completely stop all function. The smaller IC is a little op-amp.

Anyway, one card failed about two months ago, I swapped it with a spare that I had and everything was fine. Then just today another one died. I swapped in the last spare that I had and it's happy again, but I would like to rebuild these two 'bad' cards that I have, just to have them around and functional in case I need them.

The design looks very simple, each card (there are two in the supply, for the two separate HV units) is tied to a pair of power transistors which then power the HV module.

Does this IC seem like the most likely component to fail?

Any of you electronics gurus have any tips? The ICs are still being made (scratch that... no longer made but I found a couple for $2ea) and they are socketed so changing them will be very easy.

There are no symptoms, it will work flawlessly one day and the next day one side of the PSU just wont work. Sometimes after that it will decide to randomly work, but that's rare.

Here's the board:

http://i269.photobucket.com/albums/j...s/101_0052.jpg

I know there are at least a few people here who might have an idea... dont be shy *retardo*retardo*retardo

Steve R, Dr. Lava, I'm looking your way!! *retardo

hey there, if it's 'eating' these cards and it didn't used to, I'd tent to suspect something connected to the card, not the card itself, is the cause of failure. Bad caps or surge supressors allowing voltage transients to the pwm card or something. But for the cards that have failed, you're probably on the right track, ICs tend to be the most sensitive. That is one old school looking card you have there.. hard to beleive it was made in '81 ! :)

Quote:

---

Originally Posted by drlava

hey there, if it's 'eating' these cards and it didn't used to, I'd tent to suspect something connected to the card, not the card itself, is the cause of failure. Bad caps or surge supressors allowing voltage transients to the pwm card or something. But for the cards that have failed, you're probably on the right track, ICs tend to be the most sensitive. That is one old school looking card you have there.. hard to beleive it was made in '81 ! :)

---

-----------------

I'd tend to agree with my esteemed Colleague,

It took less then 3 seconds to realize the words "Gate Driver" when looking at that card. HeCad supplies are nasty and like to kill themselves, 22 Kv or starting, 1500-3200 V running @ 60 mA, and seem to have a tendency to get power "reflected" from the load like it was RF and bad SWR.

It is a protective device somewhere failing. You have two inductors there that do gate drive or base drive for the switching devices, which I'd guess are a pair of big PNPs, that doesn't quite look modern enough to be FETS.

Since the first thing close to the chip that can create reflected or undamped power is those two inductors, which get some really meaty fast pulses running through them, I'd look at what drives them, those around the two transistors, next to the coils. Ie bad cap, marginal clamping zener or TVS diode. O'd then look at the next components in the power stream and the protective stuff around them. Without a schema I can only speculate, but it is informed speculation.


For a example of a similar circuit, Take a look at the gate drive circuit in the Omni 150 schematic in the FAQ, your dealing with what looks like two of those stages, with pulse transformers being those two ferrite pot cores on the board. See if the pot cores are inductors or transformers. Based on the date of the technology, I'd guess the later. And these days I tell everyone to look at B+ (or B-) on the drive card circuitry as well and see if its clean. If you use a scope, and it had best be a cheap scope, use a 10-1 probe and "Lift" the scope from ground with a isolation transformer, and keep the scope limited to the drive card. You never know who used a "Hot Chassis" scope blowing design "feature" in that era, most switchers were. Those two parts are socketed for a reason, and it is not because someone at Liconix owned stock in Augat, who made the sockets.

BTW, cheap pwm chips had a lot of variability then, does it seem quiet and happy when it runs?

Those hecad supplies are nasty in terms of potential lethality, BE CAREFUL!

Steve

Well in the first 5 minutes or so of run time one can hear a high pitched switching that goes away. That was with the old card... Since I threw in the replacement it seems quiet from the start. I ordered two of the ICs so hopefully if it goes tits up again that will patch me up. I think board that brings power to the pwm card LOOKS ok... But that doesn't mean anything. I'll snap some pictures of it. The weird thing is that it hasn't been the same side to die. It was one side then the opposite side. I have an entire psu of spare parts, should I need them.

I can't say if this supply has had problems with the cards before, as I haven't really had it long enough. I know that both of the original pwm cards bit the dust and I took the two out of my parts supply to get it working again.

well the replacement also failed to work today... Just quit working.

I started swapping IC's from 'good' to 'bad' and eliminated them as the cause. Since the problem is the same on all three cards and seems to be intermittant on at least two of them, I had a feeling the IC's probably weren't to blame.

I also elimated the big transistors, and the HV modules as cause of the problem by various tests.

At this point I was just frustrated... I pulled out the DMM and literally began measuring every component's resistance to see if I could find something broken. I finished the entire board, comparing all compoents on the 'good' to one of the 'bad' boards.

Everything was the same. No problems found.

Then I flipped the board over... there is NOTHING on the other side except a tiny cap, a tiny resistor, and a tiny zener diode between two legs on a little transistor-looking component.

As I understand zener diodes, resistance should rise while measuring them, as when they see voltage the resistance increases (or the other way around...) correct?? This is how the diode on the 'good' board was reading.

Then I had an epiphany... all three of the diodes on the bad boards read as almost a short... .002 megaohm either way, no rising or falling of resistance. The 'good' diode would start at like .1 megaohm and climb up to at least .6 Megaohm before I got tired of measuring. So At this point I took the board which was the 'most' intermittant and put the leads on the diode, same thing .005megaohm and no change. Well after about 4 minutes resistance actually began to increase and from there on it appeared to be working. I had an idea at this point... I plugged the card in and it worked.

So now the laser is working... lasing away, and it seems I need to order a few zener diodes to fix these boards.

I dont know much about PWM... does this sound like a feasable problem? Out of however many components are on the board... probably around 100, this was the only component that I could get a reading different from the 'good' board.

EDIT: Part number is a 1n4743A *retardo

1n4743 is a 13V planar zener. Odds are its a clamp if as you say, its across a transistor. Just as predicted by Dr Lava.

easy test, find a 15V power supply or use the 30V power supply from your galvos, hook a 1/2 watt 470 ohm or similar resistor in series with the zener and wire it across the power supply. Your meter , when across the diode, should read about .6 to .8 Volts one direction (forward biased) and 13V +/- .1 or so (ZENER ) the other. At 30V with a 470 ohm R, thats 60 mA through the device, a little high, but still a good test. Rated I Surge is 340 mA and rated current is 64 mA, A 640 ohm would be OK as well. At 15V its 31 mA, well within the ratings.

Whats the number on la bete noir avec sa deux pieds?

Happy ST George's Day,

Steve

Whats the number on the black animal with two feet? Lol.... The number on the transistor-looking component which the zener diode is soldered to is..... lemme grab a lens so I can read it...

Looks like a Motorola 2N6028 *retardo

If that's what you even meant :P

Quote:

---

Originally Posted by GooeyGus

Whats the number on the black animal with two feet? Lol.... The number on the transistor-looking component which the zener diode is soldered to is..... lemme grab a lens so I can read it...

Looks like a Motorola 2N6028 *retardo

If that's what you even meant :P

---

Mais Oui. and its "black bug"...

Thats a unijunction transistor. Are you sure they are only using 2 legs?

Steve.

Well.. they are using all three legs of the transistor, the zener diode only directly contacts two but I think a resistor joins it to the third. Yup.. that's how it works. Zener goes from one pin to the middle, and then a resistor goes from the middle to the third. Here's a picture:

http://i269.photobucket.com/albums/j...s/100_0057.jpg

You can see the laser in the background on this one hehe... shining on a broken HGM5

http://i269.photobucket.com/albums/j...s/100_0059.jpg

*notice both cards lit up and everything working in this pic.... I'm sure it wont be like this tomorrow *retardo

http://i269.photobucket.com/albums/j...s/100_0060.jpg

seems like the zener diode on the 'intermittant' board is just very slow to respond.. I got home and it didn't work, I pulled the board out and charged up the diode myself and put it back and it worked *retardo

Looking forward to getting my new diodes in the mail... hopefully it will take care of this problem.

Although I do have a backup plan... I picked up another 4240PS. I guess you can just never have too many :p

two cards , one for each side. Ok.
Thats not bad.

Steve

Typically when you see the 'resistance' increasing on your multimeter as you probe components, it is because you are charging a capacitor that is parallel somewhere in the circuit. There is a rather large tantalum right in that area. Tantalum caps are very sensitive to reverse voltage and over-voltage because of their extremely thin dielectric. What could have happened is the dielectric has a micro 'pierce' that increases the caps leakage. You might try desoldering the cap and testing it, and if its bad replacing that cap with a higher voltage rated cap and see if the 'intermittent' problem goes away.

Yup that cap is tied to the Zener. Maybe that is the issue. I'll order some tantalums. It's weird though that if I charge it with my dmm it works, but if it sees the stronger voltage of the psu it doesn't. Shouldn't a Zener gain resisitance as voltage is applied as well or am I wrong?

Quote:

---

Originally Posted by GooeyGus

Yup that cap is tied to the Zener. Maybe that is the issue. I'll order some tantalums. It's weird though that if I charge it with my dmm it works, but if it sees the stronger voltage of the psu it doesn't. Shouldn't a Zener gain resisitance as voltage is applied as well or am I wrong?

---

No, not really. they do very voltage with respect to current and temperature, but they dont really shoot up in resistance on a ohm meter.

Its the cap breaking down, which dumps pulses into the zener, which causes your chain of failure. Right about the time your PSU was made, there was a scandal about bad tants flooding the market, Then again in the 90s, and again around 2002. i avoid tantalums like the plauge, but sometimes you need them, and often Uncle Samual and Uncle Ivan insist on them.

Steve

Yahtzee...

I swapped the cap on one of the boards that was totally dead, it clicked right on *retardo

Thank you both very much...

So, in this application, I'm guessing It's necessary to use the tantalum cap right there? On some of the newer boards I have, all the tants have been replaced with electrolytics EXCEPT that one, so I'm guessing it needs to stay tantalum.

I tore my house apart trying to find a 25V 33u tantalum cap... I realized that the spare boards I had, had a few of the same caps in different areas so I just borrowed one. I guess I'll have to order some now.

Thanks again... my hair was getting a bit thin because of this thing :D

Consider yourself very lucky! Your photos and testing feedback were good enough to allow remote informed conjecture/diagnosis. That sort of thing is pretty rare!
:)

If the new boards still use tantalum there, then I'd assume it's for a good reason. Tants are typically more expensive than their electrolytic counterparts. You could probably get away with a ceramic/electolytic parallel pair, or just keep a few spare tantalums on hand.

I just cant believe that all three caps failed within a couple months of each other...

Maybe it really was effected by that 'bad capacitor' batch, or maybe they were just getting old. The funny part is on that board where the cap wasn't totally dead but it still wasn't 'good', every time I turned the key off that side of the power supply would fire for a split second. I think the higher voltage was enough so that it wouldn't work, due to your explanation of a 'micro-hole' inside the cap. BUT if I used my low voltage on my DMM to charge up the cap first and THEN plug it in, it would work... crazy electronics... :rolleyes::D

Tantalums will gradually fail if left unused or overvoltaged. My guess is that those ones were just old.

The little circuit you have there with the zener and the UJT seems to be a relaxation oscillator of some kind. My guess is that it's responsible for starting the whole shebang going. And by shebang I mean switchmode.

Good news... I got my replacement 25V 33uf tantalum capacitors today. I changed every tantalum capacitor on the boards with the corresponding values (I had a bunch of other tantalums already, but none of the bigger 25v 33uf). Every card fires right up now!! Even the totally dead ones clicked right on. I cant thank you guys enough :D

Wow... more troubles today. I dont know if these caps are bad or what but i'm having issues again. Everything worked flawlessly last night. Ran it for a few hours. I went to start it today and the cards would only give weak little pulses... clicking at maybe 2Hz while trying to start. Usually it starts instantly, but if it has to click a few times, it's MUCH faster. On the order of 50Hz or so until it starts, so you really only hear it click once or twice. It's almost like these new caps cant dump in pulses to that little starter circuit fast enough... I swapped the caps back to the 'old' ones on two of the boards and slapped them in and they worked and its running now... but it's really weird that everything worked fine last night and today it wouldn't even fire up.

One of the boards was acting like it did when it had the 'almost dead' capacitor in it. I would start it and it wouldn't do anything, but then when I turned the key off and the voltage dropped it would click on for a split second. Very weird. I'll bring these caps to work with me tomorrow to test the capacitance.

Old Cap on the right, new cap on the left. Am I missing something? "336" means 33uf, right? OR are the old caps in the same notation? Meaning '33' would actually be 33pF. That cap seems a bit too big to be only 33pF!!! It must be uF...

Opinions?

http://i269.photobucket.com/albums/j...s/100_0098.jpg

Hum 103 is .01 uf. 104 is a .1 uf, 105 would be 1 uf. so 106 must be 10 uf,
so 336 SHOULD be 33 uf. I'm sure heroic will check my math in a few hours.


Steve

Quote:

---

Originally Posted by mixedgas

Hum 103 is .01 uf. 104 is a .1 uf, 105 would be 1 uf. so 106 must be 10 uf,
so 336 SHOULD be 33 uf. I'm sure heroic will check my math in a few hours.

---

336 is indeed 33 microfarads.

I know you're not stupid and all... but did you respect their polarity? Tantalum caps are polarized.

-J.

Quote:

---

Originally Posted by heroic

336 is indeed 33 microfarads.

I know you're not stupid and all... but did you respect their polarity? Tantalum caps are polarized.

-J.

---

Yup I made sure the side with the line on it, on the yellow cap, went to the positive (+ is actually marked on the board on every capacitor... makes it kinda nice!). Also, one lead was longer than the other, and I've always known that to be the anode lead, so I had triple verification on polarity.

So now I'm just stumped... these caps came from an ebay seller from Taiwan so who knows.. maybe they are just not good caps. The 25 has a 'K' after it... don't think that means anything though, probably the brand.

So on the red capacitor (the old one) since the 33 doesn't have another digit after it, am I just to assume that it's 33uf? I dont think they used the three-digit notation that long ago.. but if they did it would be a 33pf cap, which just doesn't sound right to me.

The PSU has absolutely ZERO issues once the PWM circuits get going... it's just the actual starting of the card that is miffed up. It has to be something to do with these new caps since it worked once I put the old ones back in.

While I am loath to suggest a change, a 33 uf non polarized electrolytic or poly cap might work just fine there, just pick the right voltage, hopefully twice what it sees in the circuit.

Steve

I am far from an expert in this subject but one thing reminded me of the past.

Those boards are old and so are the connections, is it possible that all those caps are failing due to poor contacts from the board to the socket?

I have this feeling you fixed the symptoms but might not have cured the problem.

if you have further failures in the futer, before swapping the boards again, have a good look at those contacts.

Best Regards

Andy

It could be a poor connection at the socket, although I thought of that before and gave the cards a light 'wiggle' with no effects. I'll give the leads that go into the socket a light tinning for better contact.

I got my (third) new power supply today anyway, and it seems to work perfectly so I'll have plenty of time to mess with this other one *DZ

Its a couple years newer than the other two, so I'm curious to check out the PWM cards to see if they changed those caps to electrolytics or anything other than tants.

I've only ran it for today, and the other supplies seemed to work for the first day and then have problems, so I guess we'll see tomorrow! :p

So either I just have horrible luck, or there is something else wrong here. The third power supply worked flawlessly yesterday. I went to start it this morning just to see if everything was still as it should be and only one side of the psu lit, just like the other two. I really don't think there is anything at all in the head itself that could cause this... It's really just a tube and a very simple control card for the He and Cd heaters...



So frustrating....

I'm so sorry, but I actually laughed when I read that :(

maybe it's a bad connection/wire somewhere or something in the head. Surely THREE PSUs can't all have the same fault !

the boards in the new PSU are a new design. I replaced the same capacitor (different location, but the one that performs the same function) and what do you know... it works again... man these PSU's are a barrel of fun. :rolleyes:

So it worked great for a night and now the new cap is dead. Either I got a abad batch or something else is going on here.

Is it possible that the cap is seeing a higher voltage than it can handle due to the tube being a bit high pressure? I'm going to swap out the 25v 33uf for a 50v 33uf cap and see if that helps. Fingers crossed!

Interesting new development.

I measured voltage across the capacitor after the laser has already started. I made sure the polarity was correct and the voltage reads about -3VDC. I'm wondering if the negative voltage over a few hours of run time is what is killing these capacitors...

It also seems like the starting of the circuit depends on the speed at which the capacitor can 'charge up'. I have one of the capacitors here that tests to 33uf on my multimeter, yet it charges VERY slowly. It wont start the board when I just turn the key on, but if I click the key back and fourth a few times (charging the cap) then it lights. Also, if I charge the cap by hand and then install the card it also starts. weird stuff here.

Thoughts?

Quote:

---

Originally Posted by GooeyGus

Interesting new development.

I measured voltage across the capacitor after the laser has already started. I made sure the polarity was correct and the voltage reads about -3VDC. I'm wondering if the negative voltage over a few hours of run time is what is killing these capacitors...

Thoughts?

---

Yes.

Replace with polyester, but more worryingly, why is it doing this? you may have a leaky diode somewhere or something.

Quote:

---

Originally Posted by heroic

Yes.

Replace with polyester, but more worryingly, why is it doing this? you may have a leaky diode somewhere or something.

---

I dont think it's a fault in the PSU... I think it may be related to the head. I think it could be that the head is a bit high pressure, causing higher than normal tube voltage, which is causing this. I just doesn't make sense for three totally separate PSU's to have the exact same failure. The only constant in this situation is the head.

Oh, also I wanted to add that both caps on both PWM cards read -3VDC. They operate completely independant of each other

EDIT: Also, will a polyester cap have the same performance as tantalum? I assume they are not polarized which is why you recommend them, correct?

EDIT AGAIN: Finding a 33uf polyester cap @ 25v or more is proving to be quite difficult...

Quote:

---

Originally Posted by GooeyGus

I dont think it's a fault in the PSU... I think it may be related to the head. I think it could be that the head is a bit high pressure, causing higher than normal tube voltage, which is causing this. I just doesn't make sense for three totally separate PSU's to have the exact same failure. The only constant in this situation is the head.

Oh, also I wanted to add that both caps on both PWM cards read -3VDC. They operate completely independant of each other

EDIT: Also, will a polyester cap have the same performance as tantalum? I assume they are not polarized which is why you recommend them, correct?

EDIT AGAIN: Finding a 33uf polyester cap @ 25v or more is proving to be quite difficult...

---

Digikey 478-2589-ND

thanks! But WOW... almost 13 bucks for one capacitor... that's nuts!! I guess I'll buy a couple and swap them in here and see what happens. And the thing the huge... 1" lead spacing. Hopefully it will fit with the cover on the PSU!!

Is there any other 'type' of capacitor that would be ok for this application? I really cant see why it needs to be tantalum, especially since it doesn't seem to do much of anything once the circuit is started.

Any thoughts on this guy?

http://www.mouser.com/Search/Product...0VwGLdEQ%3d%3d

Quote:

---

Originally Posted by GooeyGus

Any thoughts on this guy?

http://www.mouser.com/Search/Product...0VwGLdEQ%3d%3d

---

At 26 cents, might be worth a try I'm about to place a digi order, why dont I save you some woe and put it on my order and then just mail them to you.

Or better yet, parallel a bunch of the mylar/ poly ones here :

http://www.goldmine-elec-products.co....asp?dept=1027

at 50 cents each its not that bad a ouch
Steve

Quote:

---

Originally Posted by GooeyGus

Any thoughts on this guy?

http://www.mouser.com/Search/Product...0VwGLdEQ%3d%3d

---

The polypropylene ones will definitely be stronger and have a lower selfinductance, but that may not be a problem in your application; I'd try the cheap non-pol electrolytic and see if it works :-)

I just found that there are two 1n4935 diodes on each board near the big inductors.

I measured only 4ohms resistance across these diodes, both ways, on both boards. These are 200v 1A diodes. I measured resistance using my DMM... I'm guessing this is an issue?

Quote:

---

Originally Posted by GooeyGus

I just found that there are two 1n4935 diodes on each board near the big inductors.

I measured only 4ohms resistance across these diodes, both ways, on both boards. These are 200v 1A diodes. I measured resistance using my DMM... I'm guessing this is an issue?

---

That depends what's in parallel with them really- can you remove them from the PCB and test them again? They may be essential to the proper operation of the switcher- there'll be a flywheel diode and a rectifier diode in most buck or boost topologies. If the flywheel diode goes, efficiency drops and the chopper transistor eventually fails, but if the rectifier goes the whole thing goes foom.

Quote:

---

Originally Posted by heroic

That depends what's in parallel with them really- can you remove them from the PCB and test them again? They may be essential to the proper operation of the switcher- there'll be a flywheel diode and a rectifier diode in most buck or boost topologies. If the flywheel diode goes, efficiency drops and the chopper transistor eventually fails, but if the rectifier goes the whole thing goes foom.

---

Yeah you were right... they test fine when removed *retardo