power supply return high voltage (40V DC) above earth ground? (shorting out)

[coppercone2]

My sorensen dual supply 60V unit has a problem, in the past I fixed it by replacing all the parts.

Today I tried to hook up the earth ground and I got a big spark then of course the pass transistor blew and I hope thats all that blew, its pinned at 83V

Its hard to probe anything in this thing, but I am measuring the difference between the black return and chassis to be 43V. The other channel is 0V. I needed +-15V earth grounded for a particular circuit and when I hooked a banana plug to connect earth ground to black of course there was a big surge.. however when the channels are isolated hooking up the negative of the other channel to earth ground works as intended.

I never had this kind of issue before. Is it something with the transformer? Why is there a giant potential between the earth ground and the floating ground?

After it sparked it actually was working, but when I raised the voltage to check, when I got to like 60V it jumped to 80 and I guess fully failed (the transistor looks like old bullshit though, it might have been partially damaged to begin with based on how corroded and worn it looks.. should have got another one on ebay, maybe it would survive).. but anyway it looks like the 43V difference in earth and neutral is not related to the surge, and I think if I replace this transistor it will continue to be there, making earth ground unusable.

What causes this in isolated supplies? i never gave it much thought because I always ran fully floating.

I tested a few more supplies and their all close to 0V. Something comes to mind about bleeder resistors or transformer grounds or something but I don't know for sure. never managed to wrap my head too much around connecting precision electronics to my garden but go figure

[coppercone2]

I think I used this schematic before on reply #8, https://www.eevblog.com/forum/projects/dual-output-transformer-with-relay/

not sure wtf

something with the transistor and heat sink maybe? bad alignment with one of the transistor legs touching the heat sink thats grounded and making something in the circuit chassis ground  potential,  ? you need to stick em through a bore hole into a socket maybe it was touching, i replaced the old mica with kapton insulation though.

[coppercone2]

when I put new transistors in I this time measured each pin to chassis and it read open, so I tried it again and I read ~0V between earth ground and ground.

So it must have been a short some where. I don't see how it happened, the bores are properly aligned, the pins were seated.. Maybe I will ultrasonic clean the old transistor pads and look for tears, but it was brand new carefully handled and cleaned kapton with arctic silver. Perhaps arctic silver MX is conductive at higher voltages and not suitable for 80V operation, but the other side did not have a problem with the same thing. I am thinking maybe there was a metal filing or something that pierced the kapton insulator, or something that was in the thermal paste (conductive contamination).. they were pretty generous with the tolerances on the heat sink. I will look with a strong light between the heat sink and PCB to look for any thing else, but I am kind of certain the only way a high current like that can flow is if there was something wrong with the insulation on the main transistors because it would have blew apart a poor connection with the spark that it made.

I could have sworn I checked the body to chassis before, so either it made contact with the onset of paste creep, or I don't know how the fuck the transistor pins could have done it. 

I suppose the screw could have been not aligned but damn its hard to figure that

anyway if you do chassis ground on power supply, its wise to ground it before power on to make sure it current limits rather then sparks. I got used to not having things plugged into a power supply because of overshoot paranoia but I think it might be a good way to prevent explosions ....... I had one PSU do that kinda stuff where it was supposedly safe but as usual, it was damaged after touching a solenoid.

[Kleinstein]

A short at the heat sink is indeed a possible explanation. Thermal paste is not a reliable insulator, as it can become very thin locally. The versions with metallic filler are also rather poor insulators, if at all and may even fill cracks or small gaps in other insulators. So failure at higher voltage is possible.

A nasty source of insulation failure after long time are tin whiskers, that may grow from tined surfaces and lead free solder. These are thin and may evaporate after the fault.

Another possible failure mode is a strong ESD event to trigger a spark over with a relatively short distance somewhere.

It may be a good idea to check the insulation with a slightly higher voltage (e.g. maybe 300 V range), even though it is not mains and in most cases only a low voltage would be present.

[coppercone2]

it might be dremel mini stainless steel wire brush bristles, if you used them at high speed to polish annoying metal you will find them all over the place eventually. i still run into them from 5 years ago. it would be the perfect candidate. I already threw the dead transistors and kapton into a really dirty trash can but I did clean the one that did not fail really good and put new film on it, and same with the one from the parts bin.

It was my first repair in a series of maybe 12? subsequent equipment repairs/restorations since march and it was performed under the worst conditions, when I finished working on that power supply in a trashy lab I ended up cleaning everything and setting up another work bench for assembly/repair work that was cleaned much better and put away maybe 600 jim williamed parts and dust bunnies and did some vacuuming and wiping (the combination of the repair space and lab was like putting a clean room in a foundry)

or it could be a solder bit from mass desoldering with a manual tool, I got my solder xtractor vacuum pump after this first repair job when i became overrun by broken crap, i might have shot solder at the ceiling out of frustration dealing with pumping clogged holes for the 20th time)

[coppercone2]

so how would I test to 300V?

I do have a 250V and 1200V power supply, I don't have a megger. When I was around a megger it was used in a paranoid way (only once because it was said to cause damage).

Would I clip it to the bolt in the transistor and the heat sink and see what current passes? (through voltage divider)

Do you think its safe to use inside of a assembled equipment chassis that is somewhat extensively restored with good capacitors and such?

or do you mean to setup the same experiment on a block of metal that just a test setup to test it on? Testing it inside of the PSU is a little scary. Some kind of current limiting?

Making the heat sink for those oval transistors is a little hard too, I suppose I can cobble one together, but I am almost sure since the other section, and another equal 1 stage sorensen power supply is working fine with the same worksmanship by me, it must be some kind of contamination, because I also did my HP power supplies with the same grease and methods, and my amps, and none of them measure anything to ground, but 80V is the highest of them all.



I was thinking about a megger of some kind to ascertain transformer quality in old equipment but it is a little expensive. Maybe it would help me find other unreliable motors (I have been thinking about my ceiling fans, kitchen extractor fan).. if it is useful around the home maybe

I suppose I should do something because MX thermal grease is getting into everything I Have now.. fairly cheap, good thermal conductor and spreads well and keeps well

[coppercone2]

Since I have more thermal grease and a power supply, do you think I can do some kind of other test on it?

I have capilary tubes.

I was thinking to fill a capillary tube cut to some length with grease, put copper wire electrodes in it, and hook it up to a HV power supply and test it that way. My berylium copper wire is pretty stiff, so I can make 2 long strait bits that I can put into the grease tube and vary their separation in the material while measuring current. Or maybe just use sewing needles

do you think that might be a good experiment? its easier to setup then a replica heat sink or probing inside of the equipment

well at least as a grease qual, I do know you are still supposed to run an earth ground test after assembly, I suppose I should invest in it, maybe you live longer with that test

[coppercone2]

god damn fuck these anthrax power supplies i am tempted to go full vacuum tube on all my power supplies because their serious trouble

I replaced the transistor and it was not going down under 20V then I replaced the little transistors and it stopped working and the 317 was getting ridiclously hot, so i need to replace both power transistors and the little transistors again, i am tempted to make a wooden fixture with holes so I can replace ever single active part on the board at once because thats what it comes down to with this shit. maybe I will socket the rest of the active parts yeah , great reliability. they already socketed the op amps i figure why not socket the plastic transistor packages too

i dont know if we need transistor circuits for linear power supplies, its almost not worth the trouble

i figure if i keep blowing it with bread boards at this rate it will pay off to buy a curve tracer after like 5 more repairs, you can't fix this shit without whack a mole

i think darlingtons suck

[bd139]

Just get a linear supply that isn’t shit and you’ll be fine.

[coppercone2]

anthrax/sorensen and lambda are both tempremental bastards, i like the form factor but there is something to be said about a stack of reliable harrisons

2x the board area and # of power transistors for the same current, noise, etc says something good. they made those dual units compact and sucky.

[coppercone2]

proud new owner of a hp dual supply that looks well built, the sorensen might get 1 shot at fixing it and the lambda is probobly going to get shot because i dont want to trouble shoot 4 heavy brackets attahced by soldered wires in 3d chess you need mr spock for that

[langwadt]

A short at the heat sink is indeed a possible explanation. Thermal paste is not a reliable insulator,

thermal paste does not work as an insulator, you need sil pads, mica pads with paste, or similar

[coppercone2]

i said i put a new fresh cleaned kapton pad under it, it had to be something unusual, i do this alot and this never happens, just thermal paste is stupid and unreliable

[coppercone2]

power supply go bye bye, it crossed me for the last time. there is something stupid about the design making it overly sensitive, anyway I binned it and replaced with HP, good riddens.

highly do not recommend you buy.. pads started coming off after god knows how many transistor replacements and I decided its a thorn in my side and a waste of my time, a totally disproportionate amount of time was wasted on keeping it in service

Hp is
1) lower price
2) same parameters
3) non crappy display
4) 10x more robust housing
5) serviceable without a hardwired transformer to the PCB making rework a nightmare, not to mention with magnet wire, you would need to build an entire stabilization assembly and glue it to the transformer to make crimp connectors viable after splicing on real wires. if you put a transformer on the PCB put the transformer on the PCB, don't put the transformer on a separate plane with fragile wiring not attached to the pcb. you wanna justify this give me 10nv noise not the same noise as stuff 20 years younger. i don't see any improvements over old supplies. they are kind enough to put varnish soaked non flexible insulation over the wires that ensure they break
6) better potentiometers, crowbar, etc.. actually feels moderately durable instead of the usual pcb in front panel with trimmer, they actually give you a nice brass screw
7) does not break from the electronic equivalent of a sneeze
no MTA garbage connectors
9) has nice screw terminals for remote operation, making it rapidly usable with a wiring kit rather then making you fuck around making dsub shit
10) you can pull the cover off without playing games with a clam shell metal design. screeches like a pterodactyl being eaten alive when you wanna take the cover off. the front panel is held on by a PCB that the metal housing literary wraps around, its totally floating.

someone tried to make a good thing cheap and it came out like a scam, I feel like I am doing heart surgery on a turkey working on these shoe boxes 

It was probobly gonna go to 80V output again because it suffered from a 100mA inrush current into a $500 PCB anyway.