PD 2020B question

[liquibyte]

Can anyone that owns or has worked on a Power Designs 2020B enlighten me on the pinout of the A65555 voltage reference?  Trying to find information on this thing is impossible it seems.  I'm just curious as to what the names/functions of pins 1-4 would be (in, out, ground etc.) as I'm trying to understand the schematic without having actually owned one of these.  What I'm trying to do is model the various parts of this in LTSpice but obviously I can't get my hands on a datasheet or even anything past the service manual and what information is posted on this forum.  There's good information here but I haven't been able to find much info on this part of the circuit.  I don't intend on trying to model this part because I can put a voltage between pins 1 and 2 I'm thinking but the service manual doesn't say what pins 3 and 4 should be showing and I'm clueless about the function of Q4 as it relates to R42 and the reference itself.

[RJFreeman]

Try using LM399 to model the A65555 - very similar or equivalent I believe...

[liquibyte]

Try using LM399 to model the A65555 - very similar or equivalent I believe...

Not much luck finding that model at either TI or Linear.  I tried a cursory web search but no luck there either.  Making my own from the datasheet (399, not 65555) is out of the question because I'd be way out of my depth there.  Plus, the reference voltage on the schematic says 6.6-7.3V which I'm assuming would be based on the choice of R34 (6.87K-7.32K) which is why I was wondering about pins 3 and 4.  Technically I could drop in any sort of reference there but I still don't understand the relationship between Q1, Q4 and R42 as it relates to pin 3 on U3.  Is this a feedback pin that somehow helps with zeroing U2's output?

[tautech]

Try adding the device package identifier to the IC # when searching.
Eg. A65555 dip16 ic

[liquibyte]

Try adding the device package identifier to the IC # when searching.
Eg. A65555 dip16 ic

Trust me, I've tried everything I could think of for several days to find a datasheet or even a bit of info on this.  I don't think there ever was one unless it was internal to the company itself and most of the other info I've found comes from here but no one seems to elaborate too much.  From the images I've seen here, I think it's an assembly of reference plus heater, i.e. big assed can.  Maybe I should just leave that part out and drop in a quick shunt reference to see how things look at the output.  I'm still working on getting things put together and was just trying to understand this part of the circuit to see if it could be modelled because I thought it was weird that pin 2 connects to the positive output jack and that got me questioning the part connections altogether.

[RJFreeman]

Try using LM399 to model the A65555 - very similar or equivalent I believe...

Making my own from the datasheet (399, not 65555) is out of the question because I'd be way out of my depth there.  Plus, the reference voltage on the schematic says 6.6-7.3V which I'm assuming would be based on the choice of R34 (6.87K-7.32K) which is why I was wondering about pins 3 and 4. 

No need to change anything, the 65555-1, 65555-2 or 65555-3 are merely difference tolerance LM399 tested, and then labelled accordingly.
use the mid range value (7.09K) as R34 for an LM399 in the middle of the tolerance range.
In fact many PD2020B use an LM399 (at least by now) and if you look at this video:
http://youtu.be/NkR_faWq1Jw
of one being pulled apart, (at 6:32) U3 is an LM399H while R34 is 6K8.

[liquibyte]

Try using LM399 to model the A65555 - very similar or equivalent I believe...

Making my own from the datasheet (399, not 65555) is out of the question because I'd be way out of my depth there.  Plus, the reference voltage on the schematic says 6.6-7.3V which I'm assuming would be based on the choice of R34 (6.87K-7.32K) which is why I was wondering about pins 3 and 4. 

No need to change anything, the 65555-1, 65555-2 or 65555-3 are merely difference tolerance LM399 tested, and then labelled accordingly.
use the mid range value (7.09K) as R34 for an LM399 in the middle of the tolerance range.
In fact many PD2020B use an LM399 (at least by now) and if you look at this video:
http://youtu.be/NkR_faWq1Jw
of one being pulled apart, (at 6:32) U3 is an LM399H while R34 is 6K8.

Thanks for that info, I now have what I was after and can keep going.  I got a kick out of the PS being referred to a glorified LM317.  I guess for instructional purposes that explains things well enough to start with.  Now if I could just find datasheets on the op amps and transistors I'd be happy.  I know it's probably not going to happen but my quest isn't to find the old parts anyway, it's to model this using newer components to see if I can bring it into this century as a learning excersize.  After the 0-30V fiasco, I thought I might start with a proven design and work backwards to the basics and go from there.  Since everyone raves about the accuracy and stability of these and after having looked at many service manuals, I thought this looked simple enough to not be too daunting to tackle.  I was originally after 0-20V @ 2A anyway.

[RJFreeman]

The manual I have lists 3 different possible Op-amps and Transistors can be so variable you probably don't need to be too fussy.
I missed the 0-30V fiasco but do know that if you are building a voltage regulator/power supply from basic building blocks rather than power supply ASICs you can have a bit of fun  finding a balance between stability and accuracy.....
 

[liquibyte]

The manual I have lists 3 different possible Op-amps and Transistors can be so variable you probably don't need to be too fussy.

Would it be this one perhaps?  I've found quite a few discrepancies in that between the parts list and the schematic.  There are parts in the list that aren't used in the schematic and there are parts in the schematic that aren't on the list.  Granted, I didn't read through the entire thing but things like that really get my goat.

Anyway, the service manual I linked and that I'm working from lists U1 as LH358H or UA798HC and U2 as OP-05CP.  I'm assuming, though I haven't gotten that far yet really, that these are high power versions of the old standards with the real tricks being pulled off via the precision components and the reference which we've now identified as an LM399, at least in the later models.  I wasn't too worried about the transistors because, as we know, the manufacturing process is much, much better today than it was.  Still, I'd have liked to have a gander at the datasheets for the transistors and op amps just because.

I missed the 0-30V fiasco but do know that if you are building a voltage regulator/power supply from basic building blocks rather than power supply ASICs you can have a bit of fun  finding a balance between stability and accuracy.....
 

I ended up bodging in a transient protection mosfet and a normally closed relay to bleed the caps out on the 30V supply.  I haven't built it yet but it simulates nicely so I think it'll work.  Other than the transients, the rest of the supply works nicely for what it is.

As for the rest, that's the fun of learning how all this works.  You can do math all day long and simulations for weeks and what you end up doing in the end is still a bit like magic to the uninitiated.  I would have never thought about slotted circuit boards to keep references stable before I started coming here.  You just don't think about mechanical stresses with electronics unless you've dealt with that before.

[AMR Labs]

I know this is an old thread, but I was wondering how the suggested adaptation of the LM399 as a voltage reference for your 2020B power supply went. Did it work ok? and is it still working fine now or did you have to deal with any issues in terms of stability and/or accuracy?

Just purchased a C500 (0-100V 0-500mA) model power supply from the same manufacturer that is a very similar design to the 2020B and 5020 models (it even shares the same service/user manual), so it also uses that same A65555 voltage reference. I just  did a Duckduckgo search on that part number and this thread came up as the only relevant link.

Seller described it as tested for "power on only" but since the price seemed right and shipping of this somewhat heavy power supply can be one of the dominating factors of the purchase but in this case it was also quite reasonable, so went ahead and pulled the trigger on it. It has a few scratches and minor dings as expected after all these years, but otherwise seems to look ok. Had an ebay search setup for a long time ago and with a decent price limit, so finally this one just came up.

I wonder if I may end up having to face a similar problem with that same voltage reference, which seems to be the only "mystery" unobtanium part in these supplies and no datasheets available. Should get it hopefully in a couple of weeks.

Thanks.

+Added some pictures from the original auction.

[donlisms]

I think if you just assume it should be an LM399, with a different part number on it, you'll be fine.  Worry about something else.  :-)

[AMR Labs]

My C500 power supply arrived, and aside from missing all 4 jumpers on the back terminal strip that are required for normal operation without remote sensing or programming, and a few dirty contacts on the voltage step selectors that seemed to be missing a couple of steps in one of the decades, everything seems to be working fine. Also very little dust inside and otherwise everything looks quite clean and totally unmolested.

I put some deoxit on the selector contacts, and now all voltage steps are working again. The current limit pot is a bit jumpy so it could use some fader lube in there, but it seems to be sealed without any obvious openings where I could get a drop or two inside it. Or perhaps it will clean itself with a bit of exercise.

Has anybody ever opened one of these current limit pots in these power supplies? Its a single turn 1K wirewound control with a pull switch that allows to set the current limit without having to short the output terminals by switching in a 0.4 ohm 3W resistor across the main output.

Zero and full voltage calibration was off by about 50-60 millivolts, so I am now adjusting it according to the manual. So far its meeting specs without any issue aside from the zero calibration drifting around a bit by 5-10uV, so I am adjusting it with the cover on and been letting it stabilize, then readjust again. Seems like with a bit of back and forth adjustments it seems to be gravitating towards even a bit better spec than  +/-50uV output specified when all voltage selectors are set to zero.

Still have to test for ripple under load to check if any of the caps may need to be replaced, for which I will be using the HP6060B electronic load. So far none of the caps show any visible obvious signs of deterioration. Its only been 30 something years since this power supply left the factory.

[AMR Labs]

A quick follow up.

Voltage output gets very stable once the voltage has been selected and after a while the long term drift seems to be minimal as seen on my HP3478A. After setting it up as a test to 13.80V the output was 13.8101V, and after about 40 minutes it was 13.8040. I then turned off the PS to go out (but left the DMM switched on), and turned it back on 5 hours later, and the output voltage was 13.8047, , and 90 minutes later was 13.8042, another 90 minutes then 13.8041, and then after 11 hours of staying on overnight in the morning it was 13.80040. So long term stability seems quite nice.

I am now running the same test as above while the PS is set to output 2.50V, and this will also give me an additional 5th decimal resolution. I haven't pulled out yet the HP3457A that is in storage, and that would give me 6 decimals. Maybe during next week.

My main issue so far is the zero output adjustment. I can set it fairly easily (it a 10T pot adjustment) within 15-20uV or even better (spec says within 50uV) of zero output, but if I then set any output voltage with the controls, and then go back to zero, the "zero" output voltage will have raised to 1.5 to 2mV or thereabouts and will not change much from there even if given time to settle. And if at that point I try to readjust the offset back to zero I will even run out of range on the potentiometer. But if Instead I just turn off the power supply for about a minute and then turn it back on, the offset voltage will once again have fallen to uV levels that I can once again set properly. If I set the output to zero, and adjust it properly, then just turn off the power supply, and then turn it back on, it will correctly come back to the level I had set the zero adjustment.

So I wonder if the OP-05CP Opamp used in this design might have gotten unstable over all these years (and may just sat unused for a good number of those) and is no longer able to maintain stable offset. Datasheet of the OP-05 says its obsolete, recommended substitute is OP-07. But according to its datasheet, the 07 has actually worst drift specifications: drift/temp: 1.3uV/deg-C max, drift/time: 1.5uV/month, against the 05 specs which has drift/temp: 0.5uV/deg-C, and 0.2uV/month.

Seems like it would be better to try with a OP-05CP old stock device first, and if not then the current 07CP. I realize these are all minimal differences and I may never take real advantage of these, but just in the spirit of restoring equipment to its rightful original specs I would like to try and fix this drift issue. On the other hand, perhaps its not the IC that is causing the drift, but something else in the circuit. Still have to take a close look at the board, but I have to say that everything looks absolutely perfect.

Opinions and suggestions welcomed. I have attached the schematic of the power supply, and both datasheets of the 05 and 07 op-amps.