Exploding Zeners. Can't figure out why

[Blueman2]

As somewhat of a learning experience, I decided to build a PSU using the kgbhultraminipsV4 board (KGBH).  All seemed to go well in assembly, and I tested every adjacent solder point for proper continuity, and ensured no cold solder joints.  Used proper heat sinks on the silicon devices wires while soldering to avoid overheating them.  Cleaned the board with 99.99 alcohol.  I set the board up with 2x100v and 1xs150V zeners to output 350V.  I plugged it into a 300V x2 transformer (smartly using a very long extension cord in the garage) it I got a HUGE pop.  I was using 5W zeners, and when they go, they really go!  It was the Positive side of the PSU.  One zener totally evaporated (100V), and all the others (100V, 150V and even the 24V) all went to shorted condition upon testing them.  I replaced them all, along with the 8N80C for good measure.  Rechecked everything even more carefully.  No shorts, no cold joints, all zeners in proper direction.  This time, it lasted about 2 seconds and made an even larger POP than last time.  Loud enough for my wife to hear inside the house.  I decided to test it while still plugged in, and found the negative side was fine.  Outputting a very nice -345V.  On the positive side, the bias was fine, and I was able to tune it to +580 with good stability.  But the positive rail was outputting +420V.  So pretty much what the rectifier should be doing without the zeners.    

I am somewhat at a loss at this point.  I have included pictures I took before the first burn up.  Do you see anything wrong enough to cause the zeners to commit suicide?  I must be missing something pretty obvious for this big of a problem?   

Edited January 27, 2016 by Blueman2

[sorenb]

Wonder how you did connect the 2x 300V ... the two 335VAC seems pretty 'unused' ...?

[mwl168]

This may not explain the zener problem, but what shoulder washers are you using for the 10M90S? There are some discussions over at the KGSSHV thread on using washers that prevent shorts between the screw and the tab of 10M90S for high voltage applications. Basically, the "tube" of the washer needs to be long enough to extend into the insulation pad below to prevent accidental shorts.

Edited January 26, 2016 by mwl168

[joehpj]

 

Same as I met. I eventually rebuilt another PSU with new board. The bias rail and negative rail are really close. And may cause short in high humidity weather. Maybe it's the problem. The real cause still remains unknown. I had KGST powered up every night for about 4hrs for over three months. No further problem found.

 

 

[xianghao]

using the nylon screws with 10m90s。

if you use the iron one，BOOM will come up。

 and also to check lm7815 lm7915 in right place.

[Pars]

I would also ohm out the zeners to ground on the positive side (every junction). Any short on these would cause problems I would think?

[Blueman2]

42 minutes ago, xianghao said:

using the nylon screws with 10m90s。

if you use the iron one，BOOM will come up。

 and also to check lm7815 lm7915 in right place.

I will do that.  I checked  carefully, and used the correct sleeved insulator so that no contact was possible and the sleeve should have prevented arcing.  But I will do this next.  Just afraid this will not be enough!

 

1 hour ago, mwl168 said:

This may not explain the zener problem, but what shoulder washers are you using for the 10M90S? There are some discussions over at the KGSSHV thread on using washers that prevent shorts between the screw and the tab of 10M90S for high voltage applications. Basically, the "tube" of the washer needs to be long enough to extend into the insulation pad below to prevent accidental shorts.

Used Aavid kit,  Aavid 4880SG.  seemed to be the right sleeve.  

1 hour ago, sorenb said:

Wonder how you did connect the 2x 300V ... the two 335VAC seems pretty 'unused' ...?

Picture was before I added the connectors.  Last picture I had before the big bang.  

[mypasswordis]

Don't forget to replace that 10m90s as well. An easier option to finding the right shoulder washer is to use an insulating screw, such as the aforementioned nylon.

Joehpj has a different PCB version; our negative rail isn't next to the bias.

[kevin gilmore]

you are buying the 10m90 from mouser right?

 

[Blueman2]

1 hour ago, kevin gilmore said:

you are buying the 10m90 from mouser right?

 

Yes.  Though I did have 2 from a friend who had extras. Not sure his source, and not sure which ones I put in!  

I ran resistance measurements across everything and the positive side seems to match the negative pretty accurately. No shorts, no cold joints, nothing appears amiss.  

I did NOT change the 10m90, though.  I will do that now as well.  I have put nylon screws in all the sinks as a precaution as well. 

Even after I replace the 10m90, I am really nervous about powering this thing up again!!!

[mwl168]

11 minutes ago, Blueman2 said:

Even after I replace the 10m90, I am really nervous about powering this thing up again!!!

A variac will be your friend in this case. I bought one and have found it to be a valuable tool to have especially for building and testing these high voltage circuits.

Edited January 26, 2016 by mwl168

[Blueman2]

On 1/26/2016 at 1:50 PM, mwl168 said:

A variac will be your friend in this case. I bought one and have found it to be a valuable tool to have especially for building and testing these high voltage circuits.

Hmmm.  No variac, but would there be any benefit for me to hook up 110VAC to just the positive side, rather than 300VAC from the transformer?  If nothing else, maybe less of an bang and less-vaporized zeners? If the circuit is working OK, it would put out some DC voltage still I assume, like maybe 150VDC? (1.4x110VAC)?

Or, I have a 24VAC transformer as well I could try with.  

 

Edited October 28, 2017 by Blueman2

[johnwmclean]

or a low rated fuse/s over the secondaries...

Edited January 26, 2016 by johnwmclean

[Blueman2]

Well, I just ran a test with a 24VAC transformer.  On the negative rail that was good, it reads -34VDC and holds it pretty steady.  When I connect the 24VAC to the positive rail, I get 26V or so, and it slows drops as I measure it.  If I remove the test leads and wait a few seconds, I will return to 26V or so but then slowly drop again.  I tested the Bias output at the test point, and it gave 46V, and again dropped slowly as I kept the test leads connected.  Is that normal?  I was guessing that since the positive side is also driving the Bias current, it might behave differently at very low voltages. 

I like johnmclean's idea of putting fuses on the secondaries.  Duh, should have done that in the first place.  Any idea how low amperage of a fuse I can get by with for no load on this PSU? Maybe 1 amp fuse on the positive, and only test the positive side for now?  

Oh, and any harm in powering up one side and not the other??  I assume not, but wanted to double check.  

 

EDIT: I switched to using a cheap, but apparently high impedance, DVMM that showed higher readings and did not 'draw' so much current causing the reading to drop.  Positive side now holds at about 30V, and negative side at about 35V, about same as before when on 24VAC transformer.  

Then, I took another step forward.  I drove my 300V x2 toroidal transformer by a 24VAC transformer.  Hooked it up to both side of the PSU.  The 'good' negative side now shows 95VDC and holds steady.  The positive side showed about 60VDC. Bias test point was about 120VDC.  This has me a bit concerned that the positive side is so much lower, but it could be because of the additional power drain needed to run the bias, maybe??

EDIT 2: Well, I found out why the positive side was 'dropping'.  I has my PCB sitting on a stack of paper.  I did not think paper would be a problem.  It was.  It turns out the 8N80C was touching the paper and for some reason stoppled providing current to that rail.  I put some plastic studs on the board to keep it away from anything, and it is now rock solid.  Same voltage being held on both rails.  At least, so far....

Edited January 27, 2016 by Blueman2

[johnwmclean]

What is the unloaded voltages of the transformer secondaries, did you check?

[Blueman2]

19 minutes ago, johnwmclean said:

What is the unloaded voltages of the transformer secondaries, did you check?

320V on both sides.  +/- 2-3 volts over time.   Boy, the more I think about it, I was stupid for not putting a fuse on this in the first place.  Might have saved a lot of de-soldering/re-soldering and throwing away 6 zeners and 2 other devices.  Dumb.  Dumb.  Dumb. 

 

 

Edited January 27, 2016 by Blueman2

[johnwmclean]

Yep, always fuse the primary that goes without saying.

[Craig Sawyers]

Fuses are fine for safety, but I usually find that the silicon blows before the fuse does.

[JimL]

1 hour ago, Craig Sawyers said:

Fuses are fine for safety, but I usually find that the silicon blows before the fuse does.

This is why transistors are also called 3 legged fuses

[Blueman2]

Well, progress.  I connected the PSU with the completely rebuilt positive side to the 300V x2 transformer, and no smoke / no POP!  Getting 345V on negative rail.  But only getting 310V on positive rail.  Able to trim bias to 580.  PSU is very stable at those readings.  My guess is that I have a bad zener in the positive rail?  I have 2x100 and 1x150.  They all test fine as far as infinite ohms in one direction and 300-400 ohms in the other.  Any way to tell which is bad?  I tried connecting a lead to one side of the ladder and testing at each step, but the results (on both rails) were not what I expected.  Almost no voltage on first 2, then quite a bit on final one, and a higher amount (more than the sum) across all 3.  Maybe I am doing it wrong?

BTW, the 1 amp fuses used on the secondary windings held just fine in testing.  But the 1 amp and then 2 amp fuses on used on primary (110V going to the transformer) both blew pretty quickly (when I was plugging the transformer in after several times).  A 5 amp on the primary is working great.   

Edited January 27, 2016 by Blueman2

[Dusty Chalk]

Goddammit, why isn't there a video in this thread?  Every time it gets updated, I open it, hoping for a video...

[gepardcv]

I use a DY294 to check zeners (and other semiconductor parts the tester supports) before soldering them in.

For fuses, I use a 3.15A slow-blow on my KGST and KGSSHV builds (for North America 115V). Could probably go down to under 3A, but I figure the onrush current is likely to blow the fuses annoyingly often. Some people on the KGSSHV thread have recommended thermistors and lower-current fuses, but things seem to be fine with my builds as is.

[mypasswordis]

Everything including the voltage across each zener in the string should be what you expect on the negative rail since you're getting -345VDC output, then you can figure out what is wrong on the positive rail. Be careful and also measure the voltage across the 430k resistor.

Does anyone's wall voltage measure less than 120V these days? I only have a couple test points, which ranged from 122-126V. It would make choosing transformers and other parts easier.

[Blueman2]

27 minutes ago, mypasswordis said:

Everything including the voltage across each zener in the string should be what you expect on the negative rail since you're getting -345VDC output, then you can figure out what is wrong on the positive rail. Be careful and also measure the voltage across the 430k resistor.

Does anyone's wall voltage measure less than 120V these days? I only have a couple test points, which ranged from 122-126V. It would make choosing transformers and other parts easier.

Maybe I am not reading voltages across the zeners correctly then.  For example, on both rails (one that works, one that is 10% low), the voltage reading across the 'bottom' zener is about 0.9VDC on BOTH rails!  I was expecting 100V since they are both 100V zeners.  This is what seemed 'not right' to me.  But I am new to this. 

What voltage should I be looking for across the 430K resistor and what does that indicate? 

Home voltage here in SFO area is 122 to 125 range over the last year.  I have a home energy monitoring system that tracks usage and records voltages over time.  In the past year, it has not gone outside that range.  

[mypasswordis]

Replace your zener strings with fresh zeners, they should not be measuring near shorts. 100V zener should in fact measure ~100V. You should measure ~320VAC*1.414 so around 450VDC across the 430k bleeder resistor.