Test Equipment Anonymous (TEA) group therapy thread

[med6753]

I pulled out the 7904 power supply...it's ticking. Found one shorted cap on the rectifier board, removed it, but it's still ticking (though the front panel lamps are lighting up briefly with each pulse now). The 7904 power supply service document leads me to believe that the inverter controller IC is toast based on the evidence of how it is behaving with the shorted cap removed. I think my next move will be to at a minimum replace all the tantalum caps and try to find the IC (Tek custom, of course).

I'd continue on the basis that there is still something shorted; my troubleshooting here is usually to connect a meter with grabbits to each rail in turn, and measure in MIN-MAX mode while it powers up. Look for one that still doesn't try to come up when it turns on or with the tick cycles. Sometimes the difference is as small as one rail only comes up to 1-2 volts on MIN-MAX while the others come up to 50% or more of rated voltage.

If all of them seem to come up to roughly the same percentage of rated voltage before going into shutdown, that's when I usually start looking in the controller or pre-reg sections.

If you've done this troubleshooting and are still at this point, disregard and assume it's just for those playing along at home.

mnem
 

Good idea, though actually a VOM is a better choice as my Fluke 87 seems to be unable to latch very well in min max mode (maybe the rep rate is too fast?). I pulled out the trusty Simpson 260 and it appears they are all rising to about the same (rather small) % of rated voltage before shutting down.

I pulled out the 7904 power supply...it's ticking. Found one shorted cap on the rectifier board, removed it, but it's still ticking (though the front panel lamps are lighting up briefly with each pulse now). The 7904 power supply service document leads me to believe that the inverter controller IC is toast based on the evidence of how it is behaving with the shorted cap removed. I think my next move will be to at a minimum replace all the tantalum caps and try to find the IC (Tek custom, of course).

Shit simple to test with an external PSU and another scope....just do it with the 7904 all powered down.
Does the IC have a VCC spec ? Drive it at VCC and scope its output.

Will probably be the next thing I try. Need to go have a squint at the schematic....

I found a datasheet for the IC on tekwiki, but it would seem that it is indeed, dead. No outputs at all, and sometimes it randomly started acting like it was shorting, drawing a lot of current and getting very hot.

I fell asleep and missed all the excitement. I never had to venture into the Inverter board side of things with my 7904 so you are in uncharted territory.  All my troubles were on the Rectifier/Regulator boards.

I'll bet the shorted tant was 47uF...right? 

Oh yeah...it's cute and even blue.  I think the IC might actually be working though...the inverter stop output signal is, in fact, present and runs exactly in time with the ticking. I'm not sure what else it could be at this point. Maybe more shorted caps buried inside?

Check ALL the bulk supplies on the rectifier board for short to ground. I'll bet you got another shorted tant.

[mnementh]

https://i.imgur.com/WVXlXga.mp4

 

mnem

That will buff out. 

Actually, they picked right off with a fingernail and left no damage. I'm just glad they were between the work and my eyeballs, which I'm pretty sure would not have fared so well. 

That's why it's sortof my own personal "WCPGW...?" totem.

mnem
 

[med6753]

https://i.imgur.com/WVXlXga.mp4

 

mnem

That will buff out. 

Actually, they picked right off with a fingernail and left no damage. I'm just glad they were between the work and my eyeballs, which I'm pretty sure would not have fared so well. 

That's why it's sortof my own personal "WCPGW...?" totem.

mnem
 

I wasn't referring to the glasses. 

[0culus]

I pulled out the 7904 power supply...it's ticking. Found one shorted cap on the rectifier board, removed it, but it's still ticking (though the front panel lamps are lighting up briefly with each pulse now). The 7904 power supply service document leads me to believe that the inverter controller IC is toast based on the evidence of how it is behaving with the shorted cap removed. I think my next move will be to at a minimum replace all the tantalum caps and try to find the IC (Tek custom, of course).

I'd continue on the basis that there is still something shorted; my troubleshooting here is usually to connect a meter with grabbits to each rail in turn, and measure in MIN-MAX mode while it powers up. Look for one that still doesn't try to come up when it turns on or with the tick cycles. Sometimes the difference is as small as one rail only comes up to 1-2 volts on MIN-MAX while the others come up to 50% or more of rated voltage.

If all of them seem to come up to roughly the same percentage of rated voltage before going into shutdown, that's when I usually start looking in the controller or pre-reg sections.

If you've done this troubleshooting and are still at this point, disregard and assume it's just for those playing along at home.

mnem
 

Good idea, though actually a VOM is a better choice as my Fluke 87 seems to be unable to latch very well in min max mode (maybe the rep rate is too fast?). I pulled out the trusty Simpson 260 and it appears they are all rising to about the same (rather small) % of rated voltage before shutting down.

I pulled out the 7904 power supply...it's ticking. Found one shorted cap on the rectifier board, removed it, but it's still ticking (though the front panel lamps are lighting up briefly with each pulse now). The 7904 power supply service document leads me to believe that the inverter controller IC is toast based on the evidence of how it is behaving with the shorted cap removed. I think my next move will be to at a minimum replace all the tantalum caps and try to find the IC (Tek custom, of course).

Shit simple to test with an external PSU and another scope....just do it with the 7904 all powered down.
Does the IC have a VCC spec ? Drive it at VCC and scope its output.

Will probably be the next thing I try. Need to go have a squint at the schematic....

I found a datasheet for the IC on tekwiki, but it would seem that it is indeed, dead. No outputs at all, and sometimes it randomly started acting like it was shorting, drawing a lot of current and getting very hot.

I fell asleep and missed all the excitement. I never had to venture into the Inverter board side of things with my 7904 so you are in uncharted territory.  All my troubles were on the Rectifier/Regulator boards.

I'll bet the shorted tant was 47uF...right? 

Oh yeah...it's cute and even blue.  I think the IC might actually be working though...the inverter stop output signal is, in fact, present and runs exactly in time with the ticking. I'm not sure what else it could be at this point. Maybe more shorted caps buried inside?

Check ALL the bulk supplies on the rectifier board for short to ground. I'll bet you got another shorted tant.

None are shorted. 

[med6753]

Looking wonderful Med, did you have any problems with excessive noise being introduced at all with the longer leads?

No, those leads are not grounds, they are the B+ lines. No issues with noise.

Hmm, I would have thought that noise would have been introduced on the B+ lines and noise and ripple etc are then filtered through the caps to ground and the chassis was effectively a screen, a Faraday cage of sorts keeping externally introduced noise from reaching the sensitive circuit within the chassis, hence why some tubes / valves have metal screening caps which are connected to chassis ground?

No, you want to keep the return path short and direct. All 3 of those caps are mounted on a 5 lug terminal strip where the center lug is at chassis potential (ground) via the mounting screw. Since these are positive supplies the negative lead of all 3 caps is connected to the center strip which forms a direct and short return path.

[med6753]

None are shorted. 

OK, just for giggles. After I had my 7904 for several months the PSU started going into tick mode again intermittently and then suddenly solid tick mode. I discovered that if I disconnected the HV lead to the CRT the PSU came up. Long story short.....the CRT had developed a HV leakage path to ground. I had to replace the CRT.

Try disconnecting the HV lead and see if it powers up. But be careful. There's 12kV-14kV in that area more than willing to bite your careless arse. 

[0culus]

None are shorted. 

OK, just for giggles. After I had my 7904 for several months the PSU started going into tick mode again intermittently and then suddenly solid tick mode. I discovered that if I disconnected the HV lead to the CRT the PSU came up. Long story short.....the CRT had developed a HV leakage path to ground. I had to replace the CRT.

Try disconnecting the HV lead and see if it powers up. But be careful. There's 12kV-14kV in that area more than willing to bite your careless arse. 

Might be worth a try...tomorrow.  I already got zapped earlier by the anode lead from the crt side. Wasn't much worse than a bad static shock, but it sure woke me up.  I studied it and came up with a better way of undoing that connector that ensures that won't happen again.

[0culus]

Ehh, went ahead and tried it. No change. 

[mnementh]

https://i.imgur.com/WVXlXga.mp4

 

mnem

That will buff out. 

Actually, they picked right off with a fingernail and left no damage. I'm just glad they were between the work and my eyeballs, which I'm pretty sure would not have fared so well. 

That's why it's sortof my own personal "WCPGW...?" totem.

mnem
 

I wasn't referring to the glasses. 

I know, right...? Just like a train wreck; you know what's going to happen, but you just can't look away... 

mnem
*toddles off to ded*

[Neomys Sapiens]

Oops I did it again ...

On new year's eve I decided to implement a foldback current limit for my HV supply. While it worked in principle, it had a tendency to turn the supply into a power oscillator, which was not exactly what I had in mind.
After waking (at 5.43 am) on new year's day, I gave up on the foldback - my analog experience is simply not enough to solve the issue. Instead I searched my box for better heat sinks and decided to use 4 pass transistors in parallel. Not because of the current, but because the highest R_th is from case to heat sink. That should cut this crucial value by half (I had also tried mica pad + silicone grease, but that resulted in a 1 K hotter transistor than the sil pad). With that and a reduced max current of 250 mA I supposed a shorted output could be handled.
After hours (mechanical work takes ages with me - there was drilling and thread-cutting involved), I started to test the new arrangement. With a 60 W incandescent lamp as load everything was fine, the transistors got warm.
Then I turned the current limit to zero and shorted the output. That was when the fireworks started (a few hours too late for celebrating the new year). I saw a flash and a heard a sharp bang!
Again, a gate resistor had blown, also the current equalizer at the source, and, of course, the H6N100 had lost a bit of its plastic case. This left me so devastated that I switched everything off and went for a walk (where the shoe tormented the nail of my big toe, getting me a bloody sock). During the walk I tried to think about what might have caused that violent dead of another MOSFET (and why the same shoe that was perfectly fine before created so much pain today!). I just couldn't think of an explanation.
Restless, I returned to my lab later in the evening. I replaced the transistor and the resistors, checked that the others were still OK and tried again with the lamp. No go. There wasn't any current at all. Measuring some voltages, the reason was obvious: the fuse in my 34401 must have blown! I also learned what a dumb idiot I was: When I slung in the 34401 for measuring the current, I connected it to ground instead of the current sense resistor, thus disabling the current limitation altogether. Suddenly, everything fell into place.
I'm such an idiot!
At least my changes saved the rest of the circuit and constrained the carnage.

Happy new year everybody!

What was your intent when you set 1k as gate resistors at the power MOSFETS? Makes it just harder for the control loops to clear the >2nF gate capacitance.
Also, do you have an independent transformer for the control voltages? If so, I would introduce some mechanism (relay/SSR in the primary, preregulator inhibit, whatever) that shuts the main circuit off untill all control voltages have ramped up. I use this by default even on power supplies with much less energy.
But I have not yet arrived at a root cause for your first mishap.

[Brumby]

Road Trip to the middle of nowhere. Might be worth dropping the seller a message if anyone is keen eBay auction: #154269650090 $1 USD pickup only.

Good effort, Bean.

Just need to tweak the geography element.

[beanflying]

Road Trip to the middle of nowhere. Might be worth dropping the seller a message if anyone is keen eBay auction: #154269650090 $1 USD pickup only.

Good effort, Bean.

Just need to tweak the geography element.

From this end of the planet a bit of an epic and for some reason I feel safer here with our National Farce than that of the source country  Kansas is about 1500 Miles from everywhere in the USA 

[med6753]

I've been to Kansas. Despite Dorothy in the "Wizard of Oz" saying there's no place like home there's nothing that gives me the desire to ever go back. 

[Ero-Shan]

Oops I did it again ...

On new year's eve I decided to implement a foldback current limit for my HV supply. While it worked in principle, it had a tendency to turn the supply into a power oscillator, which was not exactly what I had in mind.
After waking (at 5.43 am) on new year's day, I gave up on the foldback - my analog experience is simply not enough to solve the issue. Instead I searched my box for better heat sinks and decided to use 4 pass transistors in parallel. Not because of the current, but because the highest R_th is from case to heat sink. That should cut this crucial value by half (I had also tried mica pad + silicone grease, but that resulted in a 1 K hotter transistor than the sil pad). With that and a reduced max current of 250 mA I supposed a shorted output could be handled.
After hours (mechanical work takes ages with me - there was drilling and thread-cutting involved), I started to test the new arrangement. With a 60 W incandescent lamp as load everything was fine, the transistors got warm.
Then I turned the current limit to zero and shorted the output. That was when the fireworks started (a few hours too late for celebrating the new year). I saw a flash and a heard a sharp bang!
Again, a gate resistor had blown, also the current equalizer at the source, and, of course, the H6N100 had lost a bit of its plastic case. This left me so devastated that I switched everything off and went for a walk (where the shoe tormented the nail of my big toe, getting me a bloody sock). During the walk I tried to think about what might have caused that violent dead of another MOSFET (and why the same shoe that was perfectly fine before created so much pain today!). I just couldn't think of an explanation.
Restless, I returned to my lab later in the evening. I replaced the transistor and the resistors, checked that the others were still OK and tried again with the lamp. No go. There wasn't any current at all. Measuring some voltages, the reason was obvious: the fuse in my 34401 must have blown! I also learned what a dumb idiot I was: When I slung in the 34401 for measuring the current, I connected it to ground instead of the current sense resistor, thus disabling the current limitation altogether. Suddenly, everything fell into place.
I'm such an idiot!
At least my changes saved the rest of the circuit and constrained the carnage.

Happy new year everybody!

What was your intent when you set 1k as gate resistors at the power MOSFETS? Makes it just harder for the control loops to clear the >2nF gate capacitance.
Also, do you have an independent transformer for the control voltages? If so, I would introduce some mechanism (relay/SSR in the primary, preregulator inhibit, whatever) that shuts the main circuit off untill all control voltages have ramped up. I use this by default even on power supplies with much less energy.
But I have not yet arrived at a root cause for your first mishap.

Blimey! Those gate resistors are and were always 100 Ω, the schematic is wrong.
I made some tests where I switched the control voltages independently from the main voltage. I could see no problems with that (the depletion mode MOSFET driver works as I hoped for).
As to the root cause for the first "mishap", my best bet is that I simply overpowered the pass transistor. My temperature measurement was totally inadequate.

I've just rebuilt the rectifier/filtering/bleeder assembly today, but something's fishy - to limit inrush current, I have 2 4.7 Ω resistors (4 W) before the rectifier, and they heat up beyond 70°C with 150 mA at the output! And I did not find other unexpected heat sources ...

[bd139]

Yep. Seems to be a design idiom that stupid large gate resistors are usually chucked in empirically for stability due to the large RC time constant of the average gate with one in it. A few ohms is about right though, at least on RF amps and switching which is what I've used them for (2.2->22 ohms).

Is it sitting in the linear region too long when it hits the foldback? That's usually what violently exceeds the SOA. I have a scar on my hand from an accident doing that 

[GreyWoolfe]

I've been to Kansas. Despite Dorothy in the "Wizard of Oz" saying there's no place like home there's nothing that gives me the desire to ever go back. 

Once spent 3 weeks working in Nebraska for the company.  I feel the same way.

[mnementh]

Road Trip to the middle of nowhere. Might be worth dropping the seller a message if anyone is keen eBay auction: #154269650090 $1 USD pickup only.

Good effort, Bean.   Just need to tweak the geography element.

From this end of the planet a bit of an epic and for some reason I feel safer here with our National Farce than that of the source country  Kansas is about 1500 Miles from everywhere in the USA 

I actually bothered to map it... 20 hours one way.      But.... a little too close to the heart of Dumbfuckistan for my tastebuds.

mnem
Tejas is a black hole.

[McBryce]

A Tektronix rarity: https://www.ebay.de/itm/Computer-Joystick-Vintage-Tektronix/284136994190

Anyone here ever used the system this connected to?

McBryce.

[bd139]

A Tektronix rarity: https://www.ebay.de/itm/Computer-Joystick-Vintage-Tektronix/284136994190

Anyone here ever used the system this connected to?

McBryce.

Very cool and also quite shonky. Probably the 4010 display units

Edit: apparently for use with the 4002A.

[Cerebus]

Yep. Seems to be a design idiom that stupid large gate resistors are usually chucked in empirically for stability due to the large RC time constant of the average gate with one in it. A few ohms is about right though, at least on RF amps and switching which is what I've used them for (2.2->22 ohms).

Is it sitting in the linear region too long when it hits the foldback? That's usually what violently exceeds the SOA. I have a scar on my hand from an accident doing that 

Yeah, that's quite typical of "stability by fiddling" rather than stability by design. If the output transistors of an amplifier are faster than the voltage amplifier driving them you're going to have stability problems - fixing it by slowing down the output transistors by completely slugging them is not the way to fix that problem.    I'm being a bit unkind as sometimes that scenario is one of the hardest to get right because you're fighting slew rate limitations but all you can see is there don't appear to be any poles hiding in corners waiting for the right moment to pop out and make a mockery of your stability calculations. Reducing the slew rate of the final output stage will save you, but at the cost of both slew and bandwidth which isn't exactly ideal and you'll still only get marginal stability for your pains.

[Cerebus]

A Tektronix rarity: https://www.ebay.de/itm/Computer-Joystick-Vintage-Tektronix/284136994190

Anyone here ever used the system this connected to?

McBryce.

Probably. Back in the day when I started out the only graphics terminals available to us were Tektronix vector graphics terminals that used a storage tube. Never actually used one with a joystick but I have used them with light pens. They were weird to use, unlike typical "glass teletypes" they didn't scroll in the normal fashion, they'd write a whole screen top to bottom and when you needed to move on the the next line they'd erase the whole screen and start again.

[mnementh]

A Tektronix rarity: https://www.ebay.de/itm/Computer-Joystick-Vintage-Tektronix/284136994190

Anyone here ever used the system this connected to?

McBryce.

Very cool and also quite shonky. Probably the 4010 display units

Edit: apparently for use with the 4002A.

How so...?

I mean, yeah... I prefer a full ball-bearing gimbal on my RC gear, but for a PC pointing device it sure beats a TrackPoint all to Hell. Quality Bourns pots, nice heavy CNC billet AL base... and it looks like it has a very nice range of motion. Getting that working on my PC would be a piece of cake; just need to sacrifice a old XBOX 360 controller to the project. Could even repurpose some of the original pot-holes on the bottom as trigger/function buttons.

mnem
 

[Specmaster]

Time to tighten the hatches down folks I reckon that as of midnight tonight or tomorrow, we will be in another lockdown which will be enforceable to remain at indoors unless you are a key worker, going out to get medication or food shopping, with police forces doing the enforcing.

SWMBO works in the local Crown Court, where life goes on as usual, juries are still sitting, court cases still going on, and they have already had something like 8 or 9 positive cases among staff there. How many people work in these regional courts?

Surely it's time to consider moving Crown Court sittings onto electronic platforms like Zoom, with 4 or 5 courts in operation per day, thats 48 or 60 members of the public being introduced into an already fairly cramped environment for several hours a day plus all the required officials. Similarly, with schools, universities etc is it any wonder that the graph for new cases is now flattening out, but only if you rotate the graph clockwise by 90 degrees.
Forget the so-called economy, put people first and foremost, the economy will take care of itself, as long as the people survive, the economy will pick up again.

Edit: The PM, Boris Johnson is to address the nation tonight at 8.00pm.

[PA0PBZ]

Wow, that really looks bad for the UK. We are on the way down for the moment but if that continues I have no idea.
Good luck for all your guys on the island! 

[mnementh]

Looks like the usual holiday stir-crazy, only amplified a hundredfold due to the COVID clusterfuck.



People can only handle living in terror for so long before they wig out... not exactly a surprise, especially with our so-called leaders at the front of the pack, either doing nothing or wigging out themselves. 

mnem