Test Equipment Anonymous (TEA) group therapy thread

[Vince]

Could not help. Had a quick go at it, I at least wanted to measure all the rails before going to bed, to get a taste of what would await me tomorrow evening.

We have :

+5V
-15V
+15V
-50V
+50V
+130V

All rails are present and spot on with low ripple except for the negative 15V rail which reads 0.75V mmmmm... like in, clamped by a diode or something ?

Anyway, so at least I narrowed it down.... it's 90% there, just that one rail that needs investigating...

 

[TERRA Operative]

I've been right where you are many times.... My money is it's on a forgotten connector or something plugged in wrong (from personal experience...).

At least it shouldn't be too hard to trace out the -15V rail, the PSU looks relatively simple. I would start by checking you are getting -18V into the board, then just work towards the -15V output, there isn't much there to go wrong at all.

The Tekwiki even has a really nice schematic too!
https://w140.com/tekwiki/images/b/b4/Tek_7603_power_supply_A10_and_A11_schematic_%28redrawn%29.pdf

Also, looks like there's a bunch of mods there too, if you feel like some extra fun on a rainy day....
https://w140.com/tekwiki/wiki/7603

[srb1954]

TEK 7603 PSU RECAP

22H30, I am just finished recapping the PSU after 3 solid hours of painful labour. I hate the construction of '70s scopes with a passion my limited vocabulary would fail to describe.

The 7603 PSU looks nice and easy to work on compared to working on the switching PSUs in the larger models like the 7854 and 7104. 

Before commencing a repair on the PSU in one of those models you will have expand your vocabulary of swear words very considerably. 

[oz2cpu]

high input impedance ?? what is that ??

Keithley Instruments 610C Solid State Electrometer fet volt amp ohm meter

this is really still an impressive voltmeter, i tried to put a 1G ohm resistor with my measured voltage, and was not able to detect any voltage drop..

https://youtu.be/05zAVuIkNwY

[Vince]

TEK 7603 PSU RECAP

FIXED !

No connectors were to blame. I had made a mistake when soldering the replacement caps for that -15V rail.

Old can had of course 4 negative pins, and the circuit board relied on that, it used the can as a jumper to complete the circuit. So basically the cap was not even connected to the connector that sends that -15V to the regulator board.
Of course I knew this could be an issue when I started recapping this PSU, but I got caught anyway. I did pay attention to that, concerning another can, because it was screaming at me : its 4 pins/pads were not connected to anything, none of them ! So of course that caught my attention and I looked underneath the board wit a torch light to study the traces and figure out what I need to do exactly to get it to work. Ended up adding two jumper wires.

Anyway, so I soldered a jumper wire to complete the circuit of that faulty rail, and the scope now works, yeah... however I am still NOT happy !

.. because it has not solved my issues ! The display is still as crap as it was before recapping the PSU ! 
Fat fuzzy trace and readout, readout that "wobbles" like a jack in the wind, and very noisy signal trace. The flat line It looks like it were a pipe with little balls rippling inside it, so to speak.

Well I guess recapping this old scope was a good thing in the long term anyway, and I can now rule out the PSU, so I am making progress towards a diagnosis I guess.

So if the PSU is not to blame, I guess it's the CRT circuitry itself that's to blame ? I am out of my comfort zone here.
Maybe there are some crapacitors in there as well, that cause the noise / fuzziness and wobbling ?
Need to look at the schematic to see if there are electrolytic caps in there.


A piece od good news though, regarding the time base plugin. I have two of them, same model, and I could not get the dual time base to work. So I RTFM and experimented a bit, I got the hang of it now. It's not defective, it's just that I had too high an expectation : I assumed a dual time base could display both the main and second time base at the same time.... but this scope can't. You can only display one or the other, but never both on the screen at the same time.

However there is still something I can't figure out : how to tell the scope to trigger on a rising or falling edge ? Only button I see related to that, is located on the time base plug-in, but it has no effect : the scope triggers invariably on the falling edge, no matter the position of that button.
Looking more closely, looks like it might be normal : that trigger edge button looks like it's meant to acts on the plugin's external trigger input... not the signals that you apply to the vertical plugins.

So, how the f*** do you select rising or falling edge ?!

I can believe that such a fancy state of the art expensive scope, that it was at the time..... can't even let yo select the trigger edge.... 

Oh, and the graticule illumination is piss poor as well. I guess the incandescent bulbs in there are very tired to say the least.

Oh, and the CRT itself looks very tired... with the time base at full speed you can barely distinguish the trace when using a single channel. Press the x10 sweep magnification, or just enable 2 or 3 or 4 channels, and the trace disappears completely... meaning the scope is useless. No point offering specs and features if you can't make use of them in practice 

So either the CRT is very tired, or this scope was never that good to begin with ? Trouble is, that's my first 7000 series scope, so I can't compare... I can only judge it compared to other old analog scopes.

Anyway, that's why I just love my 2467B so much. It's a technical marvel. Even with all 4 traces on the screen, sweep speed to the max, x10 mag added, the CRT still has plenty of brightness under the hood to make you blind in broad day light. One needs to see it to believe it. I saw it, so now I believe it.  Long live science and engineering, a winning duo 

So, the 7603 is basicallt recapped and working again, so I could button it up for now, and work on the remaining issues god knows when. Later.
Will have a quick go at the CRT section tomorrow, in case I can find some low hanging fruits in the schematics...
To begin with, I could put some contact cleaner in the brightness and focus pot, they are really begging for it. Hoping they are not of the sealed type, as I do'nt feel like drilling a hole into theiy body to get the cleaner in there. With my luck I would damage the pots and make things worse... 

[wkb]

New(to me) HP34401A arrived today. Love it! TEA virus bit me again..

[tautech]

Vince
You should have a edge selector or button somewhere.

For clean EHT you need a sound primary side first so any caps around the switcher circuit and then the ceramics in DC restorer on the secondary side. After that any high value (M ) resistors should be treated as suspect and all checked.
Then you must set that all up correctly and have EHT meet spec. < very first target !

Trace intensity is also in this region BUT for a dual delayed timebase there are special requirements for setting this up correctly....RTFM again and again and again.
Typically in the delayed mode trace intensity should be just able to be blanked out at Min Intensity and this is your primary adjustment for the internal intensity trimmer with the front panel Intensity control set as prescribed.
Did I mention RTFM ! 

[Vince]

Thanks for the info, though it looks like I might still be trying to fix this thing in 20 years from now. Problem is that there are, I am sure, no test points and specs in the manual to tell you what the voltage and ripple should be to be in spec with regards to trace focus/sharpness. So I could measure anything and everything but would not be able to interpret it one way or another.

I can't of course replace every component in sight, too expensive and too much work, and not exactly satisfying from an intellectual POV...

I have no plans to have that scope on the bench in bits to diagnose this problem... the PSU Recap was done precisely so I could put it back together in the first place 

Will try and see what I can do within a reason. No wild dismantling, and no staying on the bench for days, keeping me from doing other things.


So as I said the most obvious suspects are electrolytic caps if I can find any in the CRT department, if just to rule them out.


Trigger slope : RTFM and found it : it's not a push button, but rather a knob. It's the inner / small tip of the trigger level control pot, in the top left corner of the front panel.
Just tried it, works just fine, yeah....... 

As I was playing with that, a problem showed up, which I have already experienced but not sure if I mentionned it or not : from cold, sometimes the entire display jumps up and down, quite a lot, like 1 or 2 or 3 small divisions, erratically and randomly : can do it for a few seconds or a split second, then stay stable/good for a minute or two, then move a bit... then it progressively gets better as the scope warms up.
So it could be again a bad electrolytic cap reforming at power up, just maybe. Worth a shot. Once that's fixed, maybe it will improve or modify my other symptoms and might make my diagnosis easier.

The entire display moves : both the readout  and the 4 signal traces, in perfect unison, so that would tend to indicate an issue in the deflection plate amplifiers, the vertical amp more specifically, since the display moves strictly, only, along the vertical axis.

It definitely needs fixing to make the scope usable, and it sounds like it should not be too difficult to diagnose (the cold start and random aspects of it don't help though...), so I will put that in the low hanging fruits category and give it a shot tomorrow...for now it's 23h40 so bed time..... 

[tautech]

Too tired to think clearly Vince, fuck off to bed now and dream of freeze spray. 

[TERRA Operative]

TEK 7603 PSU RECAP

FIXED !

No connectors were to blame. I had made a mistake when soldering the replacement caps for that -15V rail.

Old can had of course 4 negative pins, and the circuit board relied on that, it used the can as a jumper to complete the circuit. So basically the cap was not even connected to the connector that sends that -15V to the regulator board.
Of course I knew this could be an issue when I started recapping this PSU, but I got caught anyway. I did pay attention to that, concerning another can, because it was screaming at me : its 4 pins/pads were not connected to anything, none of them ! So of course that caught my attention and I looked underneath the board wit a torch light to study the traces and figure out what I need to do exactly to get it to work. Ended up adding two jumper wires.

Anyway, so I soldered a jumper wire to complete the circuit of that faulty rail, and the scope now works, yeah... however I am still NOT happy !

Aaah, that old chestnut... HP seem to like doing that apparently. I haven't found it being done in Tek gear yet, but I'm sure there's one or two out there.
I just replaced the same type of caps in my TM506 mainframe and remembered to check if I needed jumpers. Luckily I didn't.

At least you'll remember from now on...

.. because it has not solved my issues ! The display is still as crap as it was before recapping the PSU ! 
Fat fuzzy trace and readout, readout that "wobbles" like a jack in the wind, and very noisy signal trace. The flat line It looks like it were a pipe with little balls rippling inside it, so to speak.

Well I guess recapping this old scope was a good thing in the long term anyway, and I can now rule out the PSU, so I am making progress towards a diagnosis I guess.

So if the PSU is not to blame, I guess it's the CRT circuitry itself that's to blame ? I am out of my comfort zone here.
Maybe there are some crapacitors in there as well, that cause the noise / fuzziness and wobbling ?
Need to look at the schematic to see if there are electrolytic caps in there.

See if this mod is installed and try it if not.. Apparently it reduces the fuzzyness which is a GHz range oscillation. It might help a little.
Shouldn't need much more than a bit of copper or brass strap.

https://w140.com/tekwiki/images/0/0e/7603mod36005.pdf

Also replace more caps! And check for drifted high value carbon composite resistors in HV circuits and replace as needed (I replace almost all carbon composite resistors with metal film by default these days, besides those in HF signal path where some care is required).

[duckduck]

First post here... I seem to have developed a bit of a problem after watching Curious Marc repair a bunch of old hp gear. Coupled with a place up by work that does local pickup on dodgy old test equipment I have fixed up several patients so far. So far mostly just power supply issues. Tonight's rescue was an old HP 8642b. Turned out the only immediate issue were the blown +/- 15V fuses. I will do a more testing over the weekend, but at least the frequency seems to be fairly close:
(Attachment Link) " alt="" class="bbc_img" />
Since I haven't finished my 10Mhz GPDO yet, I am using the A/B function on my 5316a and the 1pps out of my gps receiver for now.

More to come, espicially my "tower of Power" (4 6253A's , all just had cosmetic issues.)

Welcome to the support group. If you are already repairing HP rackmount TE then you are going to fit right in around here.

[ch_scr]

New(to me) HP34401A arrived today. Love it! TEA virus bit me again..

Check your firmware version and then here to enable the hidden functions! Additional Math functions, PT100 temperature measurement, save/recall state, 10mA range etc. They figured it out now, how to enable them over SCPI 

[hvontres]

First post here... I seem to have developed a bit of a problem after watching Curious Marc repair a bunch of old hp gear. Coupled with a place up by work that does local pickup on dodgy old test equipment I have fixed up several patients so far. So far mostly just power supply issues. Tonight's rescue was an old HP 8642b. Turned out the only immediate issue were the blown +/- 15V fuses. I will do a more testing over the weekend, but at least the frequency seems to be fairly close:
(Attachment Link) " alt="" class="bbc_img" />
Since I haven't finished my 10Mhz GPDO yet, I am using the A/B function on my 5316a and the 1pps out of my gps receiver for now.

More to come, espicially my "tower of Power" (4 6253A's , all just had cosmetic issues.)

Welcome to the support group. If you are already repairing HP rackmount TE then you are going to fit right in around here.

Thanks It seems that the magical RF guts seem to be fine. I played with the amplitude and some of the modulation settings and it seems to work well enough for a home gamer... Issues left to fix:

• Genral de-gunking inside and out
  
• Keys need cleaing
• one of the backlight bulbs is out, I am thinking of putting in some low profile amber LEDs leftover from a Keyboard build
• check the filter caps on the LCR meter at work, replace as needed

My other recent projects:

• TDS744a - replaces Dallas devices, fixed bad standby supply, partial recalibration. Still needs: Mod to 784a spec, full cal, possibly oscilator replacement, screen replacement (external VGA works OK for now)
• HP 3314a - had a bad transior on the -15V rail that caused the overvoltage protection to trip and blow the main fuse. On visual inspection, bad part was missing a leg  . Replaced and also put in new filter caps. Needs adjustment but works OK otherwise
• HP 5316A - had some crusty switches, but a little deoxit cleaned that right up. Was going to install the option 004 OXCO option, but the two fleabay oscilators are junk (one is about 15Hz low, the other looses signal as it warms up)
• HP 4294a - folowed me home from work, source is bad. Isolated 24V supply seems to be dead, but this is new enough to not have the classic HP service manual with schematics, so in the back of the pile right now.
• HP 34401a - decided I should have one for doing calibration adjustment work on my gear. Had been borrowing one from work. VFD is a bit weak, but otherwise seems to be in good knick. Still looking for a reasonable calibration house in my area to get it in good order.

will post some pictures once my bench looks a little less like Bob Pease's....

[TERRA Operative]

TDS744a - replaces Dallas devices, fixed bad standby supply, partial recalibration. Still needs: Mod to 784a spec, full cal, possibly oscilator replacement, screen replacement (external VGA works OK for now)

What firmware you running?
Might be worth updating if it is an earlier version. If your version isn't listed here, I wouldn't mind a copy to upload too.
(Generally you can upgrade within a major firmware revision number, going up a major revision number requires the corresponding hardware)

https://w140.com/tekwiki/wiki/TDS744

[wkb]

New(to me) HP34401A arrived today. Love it! TEA virus bit me again..

Check your firmware version and then here to enable the hidden functions! Additional Math functions, PT100 temperature measurement, save/recall state, 10mA range etc. They figured it out now, how to enable them over SCPI 

It runs 10-05-02. And the extra functions have been enabled already 💪🏻🔧


cheers,
Wilko

[hvontres]

TDS744a - replaces Dallas devices, fixed bad standby supply, partial recalibration. Still needs: Mod to 784a spec, full cal, possibly oscilator replacement, screen replacement (external VGA works OK for now)

What firmware you running?
Might be worth updating if it is an earlier version. If your version isn't listed here, I wouldn't mind a copy to upload too.
(Generally you can upgrade within a major firmware revision number, going up a major revision number requires the corresponding hardware)

https://w140.com/tekwiki/wiki/TDS744

It was already at 4.21e when I got it. I think the previous owner upgraded the FW but then never did the calibration. Apperently, between the 1.x and 4.x versions they changed the storage format for the cal data and it was throwing some errors on the self test. Once I did a partial cal, the errors went away. I got stuck on the 50 Ohm gain cal becuase my homemade voltage source can't drive 20ma into a 50ohm load (yet....) Also for the HF cal, I didn't have a signal generator that could push 16dBm at ~500 Mhz (the R&S I borrowed from work can only output 13dBm). I was going to use a cheap RF amp and a power meter to boost the output, but now that I have a working HP 8642B, I should have plenty of power to get to ~4V p-p into 50 Ohm. Once I manage to do the Calibration as a 744a, I will look at doing the 784 upgrade. I also need to figure out how to set up my Debian/freedos machine in the garage to do backups of the NvRam and cal data just in case.

[hvontres]

New(to me) HP34401A arrived today. Love it! TEA virus bit me again..

Check your firmware version and then here to enable the hidden functions! Additional Math functions, PT100 temperature measurement, save/recall state, 10mA range etc. They figured it out now, how to enable them over SCPI 

It runs 10-05-02. And the extra functions have been enabled already 💪🏻🔧


cheers,
Wilko

Nice Mine is a 06-04-01, so I don't thin I get all the goodness. Need to check after I do the VFD replacement.

[TERRA Operative]

TDS744a - replaces Dallas devices, fixed bad standby supply, partial recalibration. Still needs: Mod to 784a spec, full cal, possibly oscilator replacement, screen replacement (external VGA works OK for now)

What firmware you running?
Might be worth updating if it is an earlier version. If your version isn't listed here, I wouldn't mind a copy to upload too.
(Generally you can upgrade within a major firmware revision number, going up a major revision number requires the corresponding hardware)

https://w140.com/tekwiki/wiki/TDS744

It was already at 4.21e when I got it. I think the previous owner upgraded the FW but then never did the calibration. Apperently, between the 1.x and 4.x versions they changed the storage format for the cal data and it was throwing some errors on the self test. Once I did a partial cal, the errors went away. I got stuck on the 50 Ohm gain cal becuase my homemade voltage source can't drive 20ma into a 50ohm load (yet....) Also for the HF cal, I didn't have a signal generator that could push 16dBm at ~500 Mhz (the R&S I borrowed from work can only output 13dBm). I was going to use a cheap RF amp and a power meter to boost the output, but now that I have a working HP 8642B, I should have plenty of power to get to ~4V p-p into 50 Ohm. Once I manage to do the Calibration as a 744a, I will look at doing the 784 upgrade. I also need to figure out how to set up my Debian/freedos machine in the garage to do backups of the NvRam and cal data just in case.

There is a set of utilities floating around that lets you use the floppy drive to copy out (and in) NVRAM and Cal data.
Can't extract the firmware this way though as it's too big, but you got the latest that I know of anyway so you're set there.

https://www.eevblog.com/forum/testgear/tektronix-tds500600700-nvram-floppy-dump-tool/

[Ice-Tea]

TDS3054B. Power supply trying hard but not quite making it.

Unplugging the power brick (standard affair, so that's good in case...). Stable 15V output. Ah. Probably not the power supply, then. Still, using external power supply to power the motherboard so it's easier to observe.

Unplugged what I could from the motherboard: still pulls whatever the PSU would deliver.

Bringing in the FLIR. Oh, well that's one toasty FET, eh? It might still be so hot because of a short further down the line but the FLIR doesn't show much else. I'm feeling lucky. Digging the FET out and it's short. Doesn't mean the rest of the DC/DC can't be cooked (taking out the FET) but... Looking better.



Fast forward a few weeks: installed new FET.

Powers up, there we go!



Done yet? Well, no. The display is not properly set, the intensity control knob is knackered, one flex connector is missing the securing mechanism. BUt: so far so good!

[DC1MC]

OK everybody, not one but TWO  shameless shilling:

1) Due to some interest in the last AR488 GPIB buffered interface I have somehow three persons asking for it  , so I had to order another set of PCB to make everybody happy, that means that 8 are still available either as blanco PCB, or assembled and tested and for the reason that JLPCB works during the Chinese New Year holiday the order will reach me early next week  , so if you still don't have a GPIB adapter for your TEA and not flush with money to get Prologix or NI, now is the moment to pounce if you're in DE, or around.

2) The more important for me, I'm planning to do an improved Arduino Nano compatible board, plus something that is now not coalesced yet completely, I have created a thread here:

https://www.eevblog.com/forum/projects/improved-performance-5v-arduino-nanomega-compatible/

Kindly please read and give me your opinion or even contribute a bit.


 Cheers,
 DC1MC

[Vince]

TEK 7603 restoration

See if this mod is installed and try it if not.. Apparently it reduces the fuzzyness which is a GHz range oscillation. It might help a little.
Shouldn't need much more than a bit of copper or brass strap.

https://w140.com/tekwiki/images/0/0e/7603mod36005.pdf

Thanks !

So Tek was aware of the problem... yet they put that scope on the market anyway 
The guys in the marketing dept must have told management : " We have a narrow window to launch these new 7000 series of scope , it's NOW and not in 3 months ".
... "You can always improve it later, but it needs to go public NOW ".

Oh dear..

That board is easy to access from the side of the chassis, once you remove a large access door, so I had a go at it.

So I added a strap, well two, because the one shown in the Tek illustration is super wide, don't have that.
Best I could come up with in a hurry was to find a use for my 35 year old 15 meter roll of 3mm wide tinned solder wick !
Too coarse to work on your average electronic board, and too much of a thermal mass for my 50W iron to heat it up quickly.

So I put two pieces of braid in  //, each insulated with heatshrink tubing. Blue tubing of course, because.. it's a Tek ! 
I don't know, I felt in the mood of using some colour..

Given how old that braid is, needless to say solder doesn't stick to it any more... but if you heat it up long enough and use enough of ym long expired 30cc syringe of flux... you can bring it back from the dead...

So the result ? Well, scope still works and no smoke, so that's a good start. I would need to male a side by side comparison to be objective.. Which is not possible of course.
So I have to go with my memory here, of what it was like before I put the straps in place.  I would say it's better. Hardly "Tektronix sharp", but better enough that I think it exists and is not a placebo effect. So I will take that... thanks Terra for the link.

I also noticed that the display at power up does not move and up erratically any more... well it might come back later who knows, but for now I don't see an ounce of it any longer, so I will take that as well !   We will see in the longer run.
Maybe the various connectors just needed to be reseated / cycled..

Also replace more caps!

I found 4 electrolytic caps on that board. No alu ones, but 4 dipped tantalums, the old variety with colour bands and dots on them. All the same value it looks like, though I don't know what value that is... need to google how to decode the markings. So I could replace those I guess.

And check for drifted high value carbon composite resistors in HV circuits and replace as needed (I replace almost all carbon composite resistors with metal film by default these days, besides those in HF signal path where some care is required).

Will look at the HV section later on, but I fear it might not be feasible as access looks to be... impossible. I mean I can't even SEE where the HV stuff is located  in that chassis !    That's definitely not promising to me...


23H45.... 

[tautech]

I found 4 electrolytic caps on that board. No alu ones, but 4 dipped tantalums, the old variety with colour bands and dots on them. All the same value it looks like, though I don't know what value that is... need to google how to decode the markings. So I could replace those I guess.

Don't you have the service manual with the BOM ?

Brown, Black, Green should be 10uF....measure one.

[Vince]

No it should be 1uF not 10 ! 

Schematic confirms it, all 1uF to decouple the 4 rails that power that board : +/- 50V and +/-15V.

[tautech]

No it should be 1uF not 10 ! 

Schematic confirms it, all 1uF to decouple the 4 rails that power that board : +/- 50V and +/-15V.

Just be sure to derate them all with 50V tants. Quick rule of thumb for tants is to at least double the rail rating for the cap itself.
You might remember some older Tek stuff used 15V tants on 12V rails which was a really common failure point and our old buddy bd139 made lots of coin buying particular old Teks with this common fault.....like a 5 min fix.

[TERRA Operative]

TEK 7603 restoration

See if this mod is installed and try it if not.. Apparently it reduces the fuzzyness which is a GHz range oscillation. It might help a little.
Shouldn't need much more than a bit of copper or brass strap.

https://w140.com/tekwiki/images/0/0e/7603mod36005.pdf

So I put two pieces of braid in  //, each insulated with heatshrink tubing. Blue tubing of course, because.. it's a Tek ! 
I don't know, I felt in the mood of using some colour..

So the result ? Well, scope still works and no smoke, so that's a good start. I would need to male a side by side comparison to be objective.. Which is not possible of course.
So I have to go with my memory here, of what it was like before I put the straps in place.  I would say it's better. Hardly "Tektronix sharp", but better enough that I think it exists and is not a placebo effect. So I will take that... thanks Terra for the link.

As I posted the info on the ground strap I thought of the solderwick braid trick, but I guessed you'd figure it out.

Looks like great minds do think alike after all.....


On my bench is still the TM506. I replaced all the pass transistors with TIP35C and TIP36C as a couple of the originals were blown, so who knows how much abuse the others took, and I got a bushel of TIP3xC transistors for a good price, so in they all went.
I had to turn them around and put them on little flyleads as the pinout was reversed, but that made the next step easier.
I have found that the NPN transistors often like to oscillate when testing them with my mainframe tester.
I don't know if this is a result of something about the mainframe tester or inherant to the mainframe design itself, but the PNP transistors don't do it and the design is exactly as per the original Tek design and the +Ve and -Ve sides of the test module are mirror images of each other, practically down to the PCB layout.


Noisy noise is too noisy... ENHANCE!


That doesn't look right! ENHANCE!


Woop, there it is! 4.7MHz!


This is what it's supposed to look like, from the PNP transistors.

I measured the oscillation at 4.7MHz, so I played with some ferrites and managed to kill the oscillation completely, so after a trip to akihabara to get more, the mainframe is mostly happy.
I ended up with two ferrite beads on the base of each NPN transistor (most transistors only needed one but a few were still marginal, so I put two on them all for good measure) and one on each PNP, just to be sure they won't oscillate in other load conditions.

Now, I said the mainframe is mostly happy...
The high-power compartment is being a bit stubborn still..
The PNP transistor os rock solid, but the NPN is still being naughty.
I added a few ferrites and it solved the 4-ish MHz oscillation, but now there's a 9-something kHz oscillation to kill too.
The maximum ripple is stated at 600mV, but the ripple is pushing it out to 650mV..


Well, at least it's better, but still not great.


ENHANCE! again and we see 9.4kHz.

So, next is to dampen that 9kHz ripple, I'm sure the 40cm long wires from the backplane PCB to the pass transistors located on the other side of the mainframe (good job on that one Tek engineers...) isn't helping things.
The usual 10-100 ohm series resistor on the base does nothing, 0.1uF cap between base and emitter makes it worse, and more ferrites has no effect as the ripple frequency is so low...

Looks like we fixed the oscillation due to the parasitics, but now we have a loop instability problem. Might need some RC network lovin' applied in a special place.

Gotta hunt for the solution this weekend..