EEVblog #649 - Power Designs 2005 PSU Teardown

[wn1fju]

Just some random thoughts that may or may not already be known.  In my PD2005, the calibration procedure calls for adjusting two pots - a 0V pot and a 20V pot.  However, I was unable to get the 2005 properly adjusted, particularly in the 10 to 20V range.  And that is because the voltage selection switches really only generate intermediate voltages between 0 and 10V; they add an additional 10V (1 ma through a precision 10K resistor) to bring it up to the 10 to 20V range.  If that 10K resistor is off, calibrating it solely at 20V is guaranteed to screw either the 0-10V or 10-20V range, take your pick.  What they really should have done is add a trimmer to the 10K resistor and let you first calibrate it at 10V and then use the trimmer to translate that calibrated range into 10-20V.  And so that is what I did.  Now I am comfortably achieving a maximum voltage error of 0.05% (against the 0.1% spec) with most voltages only off by about 0.02%.  Also, my unit would never achieve the ripple spec.   I hooked up a scope and saw a 1.2 MHz sinewave of several hundred mV.  I was able to eliminate that oscillation by tacking on a small capacitor across Q4, B-C.   Finally, the service manual does have some errata which calls for a few additional caps to be added. 

[HP-ILnerd]

Keep the vintage tech coming, Dave!  It's nice to see all the point-to-point without so much as a single BGA anywhere.

Any chance you could acquire an HP200a?  Bill and Dave's first product would make a splendid episode. 

[corrado33]

That old beauty absolutely deserves a proper referb and cal.  Another vote for making it so, 

Agreed, I'd love to see another video on this lovely bit of old tech! (Plus those types of videos are awesome, albeit most likely frustrating for Dave.)

[williefleete]

That output was out by at least 2.5% at most! if it can't be bought back by tweaking on the pots, either resistors drifted, transistors went leaky or the Zener drifted, if it's the latter (two) and most of the precision (and perhaps not-so precision) resistors check out, he may have to crack open the heater can.

C'mon troubleshoot and repair video

[riconette]

have you seen any whiskers, or otherwise funny/corroded solderpoints?

i wonder - from the video the internals look *so* clean.

[Dr. Frank]

Nice Video, and interesting to see, that in 1964, silicon transistors were very common already.

In the end, the topology of the circuitry is very similar to what Fluke in 1965 (!) designed in the early 332A calibrator,  i.e. an ovenized zener, a stable, inverting OpAmp, a trimmable range resistor, and 4 1/2 decades of wire wound resistors for the sample string. (The 1/2 digit consists of a 10k ww resistor, R45).
The Fluke 332 / 335 instruments are also designed "the other way round", if you study the  "Theory of Operation" in later manuals.

Therefore - no need to turn the schematic of the 2005 PSU upside down, not even in Australia  ...

And that transistor pair Q5 is NOT a current mirror, in this sense, it's simply the front end, i.e. the differential input stage of the OpAmp.
Instead of using chopper stabilization, this differential pair is also temperature stabilized to achieve a certain degree of stability. Clever idea, .. I think this was later realized in an integrated OpAmp from National, or Fairchild, maybe µA 714, or earlier.

Certainly, this Zener diode CR13 has drifted over these 50 years, i.e. its voltage has increased. Maybe also the ww range resistor R11 = 5200 Ohm has decreased..

But there's absolutely no need to crack the oven open for calibration.

R80 and R83 are intended to do the coarse calibration of the range.
Simply measure the Zener diode voltage, and chose these resistors, that the value in kOhm of the complete range string R80 + (R11 || R83) + 1/2 * (R12 || R79) (roughly 5.3kOhm) equals the voltage value (roughly 5.3V).

Then, the linearity of the sample string (10k + 10 x 1k / 100 / 10 / 1 + vernier 1 Ohm) should be checked, i.e. whether it still achieves 0.1% linearity..

Maybe, that the gain (beta) of several of the transistors (outside the oven) also decreased greatly, as this is a typical effect of such old components.

This, and also dried out capacitors, may lead to an unstable output.

But anyhow, I also would like to see a characterization of this historic (?)  instrument, after this easy calibration!

Frank

[geo_leeman]

Great video! I love old PSUs.  I did have a similar issue with another supply recently of not being able to tweak it into cal.  I removed the jumpers on the back, gently polished the contacts with some fine scotch-brite pad and reassembled.  That was all it took.  Some corrosion was messing with the sense and it was quickly brought dead onto spec. 

[robrenz]

I have done multi day and week long burn ins of these units at full load without any capacitor issues so I would not just replace caps because of age.

After a 30 hour burn in at max amps and voltage then calibrated using my Fluke 8846A before I sold this one on the forum.  incredible performance   (All switches were given a cleaning and 2 step deoxit treatment.)

0V = 0.2 µV 
20V = 20.0001V   
10V = 9.9998V
5V = 5.00000V
1V = 1.000095V
0.1V = .0999999V

It was the later model "A" version with the edge meters and pull to set current controls.

[don.r]

As for the unit being out of spec, are you still as "In Spec" as you were at 18? Most likely not. Couple things come to mind. The electrolytic caps are due for a change, and those carbon resistors are going to drift. You could do the lazy mans job (And in this case its the way I would probably go) and bodge in a resistor to bring the divider string into cal, replace the filter capacitors, and call it a day.

Yes, will probably do at least that. But would be nice to hunt down the drifting culprit.

On my 2020B it was one of the carbon resistors in the divider that went south. I replaced it with a pair of hand picked 1%'ers.

[artag]

Good to see this. I hadn't realised how well-loved they are : I've had a 5020 for several years, bought for not much at a hamfest. It works fine, though I've never checked the cal.

[K6TR]

Nice score Dave. Regarding your unit being out of spec I don't have anything new leads for you from what already has been pointed out but will state for emphasis to focus on the Carbon Composition Resistors. I think you would find it very revealing if you removed one and checked it's impedance.  Yes that's right I used the word impedance. Carbon Comps are notorious for not only drifting out of spec but adding reactance as they age. That's part of the reason they stopped making them and went over to metal oxide film resistors.

As to the capacitors they were built far more robust with a longer intended life time but still can fail. They often fail not outright but become noisy.

I sure would welcome a follow-up video to see what you found wrong with it.

[Rutger]

Wonder what the ESR is on those old caps?  I would like to see a calibration/repair on this unit and a review of the individual components, like are they still in spec?

[Agent24]

Add another vote for checking the resistors. I have some bags of quite old resistors and while sorting and testing them I found that quite a few of the values have increased in value even just sitting there unused.

[yym]

The problem is obviously the ovenized reference oscillator (at 8:33) 

[IuriC]

The schematic image link you provided is too tiny. Here is the big one image:

http://www.rako.com/Articles/images/Power_Designs_Model_2005_Schematic.gif

[bob_60]

Again a nice video from you Dave.

When I think back, you have made quite a few videos about different power supplies, and I think I have probably seen them all.

With all those different power supplies, what about making some videos where you look at how the regulating circuits - in detail - is actually designed in the different supplies. In your u-power supply series, you choose an IC as the main regulator, but what designs does the power-supply manufactures choose. And what has changed design-wise - if anything - in the last maybe 20 years.

Personally I would also find it interesting to see how good regulating circuit designs differ from cheap designs like for example the Korad KE300 you have tested - and I'm not thinking of the quality of the connectors, components, the chassis etc. - just the design of the regulating circuit itself.

/Bo

[aroby]

I loved the look of this, so I bought one from eBay.  Got it for $50 because it didn't work.  I just received it - it needs a clean and there are some scratches on the front.  And it only doesn't work because there is no fuse in the fuse holder at the back.  I shorted it out, powered it up and it seems to work fine. The dials were set to .0316v and that's what my meter measured!

The unit seems to be older than Dave's - only one fuse holder on the back and the serial is low down vs high up.  Once I get it cleaned up I'll see what other differences there are.

Anthony

[Macbeth]

Love it. Nice piece of gear that.
I remember back in 1985 at University learning Electronic Engineering I had to do a section on loom wiring and lacing up just like that power supply. Fantastic, IWOOT!

Haha, I remember learning how to lace looms as part of my City & Guilds back in the late 80s/early 90s. It seemed so anachronistic then, but had to be done as a practical  Get the stitching more than a few mm out and lose points!

[SeanB]

If you work on aircraft you learn how to do wire knitting on looms, especially if you have to either replace a wire or add a new one as part of a modification, or if the wire breaks inside the loom. Had plenty of practise changing plugs and sockets as well, where you have up to 200 wires, all white, all the same size and all with the hot stamped lettering that was applied in production ( each wire is individually labelled every 10cm with a designation of the loom, section and wire number with a hot foil stamper) worn off by chafing in the loom. Simplest was to get the new connector, align correctly (round multipole connector with 4 possible ways to mount it, and with 16 variants as to the inner alignment and locating slots as well) and do each wire one by one, use the extract tool to release the connector from the shell, cut the connector off, strip, solder or crimp as required then insert in the new one. Generally inside a panel and with large amounts of other stuff in the way, and often knee deep in hydraulic oil and fuel. for more fun with a hot engine as well for company, and you lying on top of the cowling working inside an access panel with 40 screws holding it. Cable ties not allowed, they work loose or cut into the looms. Still have some rolls of the waxed twine and the needles, handy to do leather work with.