POWER DESIGNS PRECISION POWER SUPPLIES

[ManateeMafia]

Is the linear control loop the same as a proportional control loop like the one used in the Fluke 732A? The thermal mass of the 732A seems comparatively large but they can maintain it at +/- 2 deg C.

[motocoder]

Yes, it is pretty much a PID with just the P. But without PWM, it means you are dissapating the excess heater power in some transistor.

[David Hess]

Is the linear control loop the same as a proportional control loop like the one used in the Fluke 732A? The thermal mass of the 732A seems comparatively large but they can maintain it at +/- 2 deg C.

I was not thinking of the Fluke 732A specifically but that is the type of implementation I had in mind.

[timb]

timb -

I went over this before, but maybe you missed it. I already made a thermostat that you can set the hysteresis to whatever value you want (currently set to around 0.6C). However, that won't get you temperature regulation to within 1C. The problem is, when the thermostat shuts off, a big load of heat has already been transferred from the heater coil to the metal of the tube. This heat continues to soak in, and the heat within the tube continues to rise. I don't recall the exact numbers, but it was very significant - way more than the 1C.

Right, but the bi-metal thermostats originally used had 10c of hysteresis, so an IC with 1c (or hell, even 3c which is a common option) would be better than what was on there, right?

[motocoder]

timb -

I went over this before, but maybe you missed it. I already made a thermostat that you can set the hysteresis to whatever value you want (currently set to around 0.6C). However, that won't get you temperature regulation to within 1C. The problem is, when the thermostat shuts off, a big load of heat has already been transferred from the heater coil to the metal of the tube. This heat continues to soak in, and the heat within the tube continues to rise. I don't recall the exact numbers, but it was very significant - way more than the 1C.

Right, but the bi-metal thermostats originally used had 10c of hysteresis, so an IC with 1c (or hell, even 3c which is a common option) would be better than what was on there, right?

Oh, yes definitely. I think I was seeing about 5C of drift with mine. And as you established earlier, I am running it at too high a power,so it should be somewhat better than that running at a more reasonable 6W or so.

[motocoder]

timb -

Here's some simulation results with the LTSpice model created for the original heater thermostat with SCR circuit. An image of the schematic and the LTSpice model are also attached.

The top trace is the AC mains voltage so you can see the relative position of the waveforms in the lower window. The green waveform is the voltage controlling the simulated thermostat cycle. I'm using a voltage controlled switch to model that, so -1V corresponds to an OFF thermostat, and +1V corresponds to thermostat ON. I've deliberately picked timing on that so we can see what happens when it turns on and off in the middle of a cycle. The pink waveform is the current through the heater.

You can see that it largely limits turn on to a zero crossing (there's a little spike because the circuit isn't perfect). The turn-off is restricted to a zero crossing by the behavior of the SCR.

And BTW, if you think the schematic looks sloppy, it's because I was trying to duplicate it exactly as it was drawn on the PD2005A schematic.

[MarkPalmer]

I put the new caps in my 5020 and it all looks to work at least as well as PD intended it to, I’m very pleased with it.  It’s fairly stable right from the get-go.  After about an hour under moderate load I calibrated it using a 3456A.  The dial settings are all considerably more accurate than the .1% +/- .5 mV at any set point, and after running for 10 hours under a 500 mA load overnight it didn’t drift more than the .001% tolerance.  So I’m satisfied that it’s within specs and I can trust it for reliable use. 

I ran in to the same problem with the front panel Timb did.  After removing all the old calibration stickers, there was no way on earth to remove the markings left behind.  I think PD used some type of varnish on the aluminum that gets a yellowish tint to it over time, and it came off with the stickers.  My answer to this will be…. if you could guess.....  a new set of calibration stickers from work to cover the old ugly spots- creative LOL.  I’ll post a pix later after I get a chance to paint the outside cabinet for laughs. 

-Mark-

[robrenz]

I have never had a Power Designs front panel that would not clean up completely when using a magic sponge (melamine foam) and IPA.  They really are magic.

Check out the before and after pics of the top rail on my 3410A restoration.  That was not repainted, just a magic sponge.

[timb]

I have never had a Power Designs front panel that would not clean up completely when using a magic sponge (melamine foam) and IPA.  They really are magic.

I scrubbed the thing to death with a magic eraser (melamine) and 97% IPA. I also tried a mix of 30% peroxide, baking soda, tetraacetylethylenediamine and sodium percarbonate, with a little Dawn liquid dish soap. Still nothing.

I might try to use the peroxide, TAED and PCS I used above, but in a gel form and leave it in the sun for a few hours (UV exposure). [AKA retr0br1ght]


Sent from my Smartphone

[MarkPalmer]

All my other PD power supply aluminum fronts cleaned up well, just not this one.  Any amount more of cleaning just polishes the aluminum where the old stickers were to a shine.  All the old, "lab cigarette tar," came off the rest of the front, but there is definitely some type of varnish or coating there that I don't want to rub through.  I figure I'll just play it safe rather than risk ruining it.

I have plenty of different cal stickers to use.  They are all decent laminated ones too.  It is sort of legitimately calibrated 

-Mark-

[timb]

Mark, I'll try a few more chemical agents and see what I can come up with. You're right though, there is a varnish on there. I can see a brownish/dirty looking area several places that won't come off. I think this is the varnish or coating that has aged from UV over the years.

Worst case I destroy my panel and have a new one made. I'm milling this one out anyway for a new panel meter.


Sent from my Smartphone

[MarkPalmer]

I found a few stickers to use in the meantime- "Periodic Calibration Not Required," and, "For Reference Only."  That way I keep it honest    It's sort of a shame because the panel cleaned up so nice other than where those old stickers were.

I'll let it be a lesson to myself if I ever get another one of these in the future...  leave old patina in place.

The other misfortune is PD didn't etch this front panel, the legends are just printed on.  Otherwise I would buff it out down to bare aluminum.  The panels on my 3650S PD supplies are the only ones I have ran across from PD that have the legends etched in.     

-Mark-

[timb]

timb -

Here's some simulation results with the LTSpice model created for the original heater thermostat with SCR circuit. An image of the schematic and the LTSpice model are also attached.

The top trace is the AC mains voltage so you can see the relative position of the waveforms in the lower window. The green waveform is the voltage controlling the simulated thermostat cycle. I'm using a voltage controlled switch to model that, so -1V corresponds to an OFF thermostat, and +1V corresponds to thermostat ON. I've deliberately picked timing on that so we can see what happens when it turns on and off in the middle of a cycle. The pink waveform is the current through the heater.

You can see that it largely limits turn on to a zero crossing (there's a little spike because the circuit isn't perfect). The turn-off is restricted to a zero crossing by the behavior of the SCR.

And BTW, if you think the schematic looks sloppy, it's because I was trying to duplicate it exactly as it was drawn on the PD2005A schematic.

Interesting... I just added some scope meter measurements to my iCircuit simulation and got the same thing. I bet they did this to prevent interference from the thermostat switch. I have another PD supply with a working oven based on the 2005 design (100V model) and I can see a large transient on the regulated output when the heater turns on and off.

The switch may also spark, because a pair of nearby speakers make a popping noise as well.

Switching at zero crossing would prevent this.

I guess that NPN in the circuit is there to bias the SCR low when the thermostat is off?

Sent from my Smartphone

[motocoder]

Interesting... I just added some scope meter measurements to my iCircuit simulation and got the same thing. I bet they did this to prevent interference from the thermostat switch. I have another PD supply with a working oven based on the 2005 design (100V model) and I can see a large transient on the regulated output when the heater turns on and off.

The switch may also spark, because a pair of nearby speakers make a popping noise as well.

Switching at zero crossing would prevent this.

Yes, this is the usual reason for switching at the zero crossing level; it reduces generated EMI.

I guess that NPN in the circuit is there to bias the SCR low when the thermostat is off?

It pulls the gate of the SCR low when the half-wave rectified mains voltage exceeds 21 volts. This gives enough of a pulse to turn on the SCR, but only near the start of the voltage ramp-up.

[timb]

I see. Makes more sense now. I couldn't see that happening in my simulation and haven't gotten to a PC to load up the LTSpice one.


Sent from my Smartphone

[Luap2]

I have been a reader of this forum for a year of so, lots of info here and a lots of it over my head. This thread got me to buy a 6050a off Ebay.
Like the few pieces HP equipment I have the in sides are a pleasure to look at. On to the 6050a I works ok. However  I have measured 47v across the 40v cap. I have the info that was posted in this thread ( thank you to those that did it )but I need more. I would like some service info to run this down  I do not mind buying it, but would like to do a PDF down load and not pay for the manual to be sent to Alaska. I have looked and all I find is a bound manual  to be shipped. Any one know where on the web I can do a PDF down Load, thanks in advance

  Luap2   

[motocoder]

I have been a reader of this forum for a year of so, lots of info here and a lots of it over my head. This thread got me to buy a 6050a off Ebay.
Like the few pieces HP equipment I have the in sides are a pleasure to look at. On to the 6050a I works ok. However  I have measured 47v across the 40v cap. I have the info that was posted in this thread ( thank you to those that did it )but I need more. I would like some service info to run this down  I do not mind buying it, but would like to do a PDF down load and not pay for the manual to be sent to Alaska. I have looked and all I find is a bound manual  to be shipped. Any one know where on the web I can do a PDF down Load, thanks in advance

  Luap2   

Not sure about 6050A. I see a link to a manual for a TW6050D here:

http://www.ko4bb.com/manuals/index.php?dir=09)_Misc_Test_Equipment/Power_Designs

[timb]

So I got the 2020B all wired back up last night.

I did something a bit different with the 10-20V selection switch. PD had it so that by default the output wire (blue) came directly out of the 10k resistor on the back of the switch board. When you had the switch set to 10V, there were two wires that connected to each end of the resistor and it would short it.

The problem with this approach is that the 10V range would pick up an extra 0.5Ohm from the switch.

So when I rewired it, I hooked the blue output lead to the center of the switch and connected the 10V contact to the top of the 10k resistor (where the output of the switch bank cones in) and the 20V switch contact to the output of the 10k resistor. This way both ranges have the same added resistance.







I got the calibration spot on from 1V to 20V!.. Well, until I hooked up the indicator lamps. As soon as I did that, I started getting weird drift from 20V down. If I calibrate for 20.0000V, I'll be at 1.996xxxV at the bottom.

I think the HVDC going through the same switch is introducing something into the range selection output.


Sent from my Smartphone

[timb]

Tonight I realized I had a small issue with my 2005 heater rebuild. It seems that the thermostat module is what physically presses against the bottom of the ref board socket to hold it against the ring. With the OEM thermostat now gone, I had to find another solution. Enter my friend: JB

A little JB Weld and the socket is now permanently apart of the mounting ring. Huzzah!





Since the socket sticks out a bit from the top of the ring, I couldn't just straight up clamp it. I needed a tube that only touched the edges of the ring... Now I remember why I save pill bottles! (Aside from being handy for parts storage, they're great for stuff like this.)




Sent from my Smartphone

[timb]

So, I got another PD supply in! This one isn’t a precision unit and in fact appears to have been made in the early 90’s, surely one of the last units they made. Here’s the odd thing though, some of the parts date back to the mid-1970’s! I’ve noticed this is some of their precision supplies as well; since they had custom binned parts, I think they kept quite a stock on hand.

They also appear to be using Illinois Capacitor [ic] brand caps, which I had honestly never heard of until I googled them, but turns out they have been around since the 1930’s and hold quite a few important patents on high temperature electrolytes. That said, I’m not certain of their quality.

Overall the construction seems fairly good; I don’t see any glaring safety issues. They’re still even using turret boards! The one thing that does bother me is the fact that the main boards are only secured halfway up. There’s a bracket that attaches to the back of the unit, which in turn extends half the length of the main PCBs on both sides, which they’re screwed into. The remaining length of PCB is just hanging there; I can touch it and make it flex, which seems not optimal.

After inspecting the insides and doing some basic tests, I decided to do a full load test. Out of the box, without tweaking any adjustment pots, I set each source to 5.00A and hooked the outputs directly to two of my 5.5 digit multimeters (which have been calibrated against a transfer standard recently) set to the 10A range. You can see the results for yourself:



The Keithley 197A is connected to Source B; the Tek DMM4020 is hooked to Source A. This is after running for 10 minutes or so. I let it run for an hour (with banana shunts, not through my meters) and nothing got over temperature. I also set the max current adjust pots up to 6.5A (the unit is rated for 5A@0-6V or 1A@0-60V max) and ran it for another 15 minutes; I only saw a 10c rise in the series pass transistors, but it stayed in regulation, so I’d say they’ve got appropriate derating on all the parts.

The unit is physically split into two identical power supplies, completely isolated from each other. Source A is the top board, Source B is the bottom board. The transformer is pretty big with eight secondaries (four for each source), six of which are center tapped.

Anyway, enough talk, here’s some pictures!

[motocoder]

That's pretty cool. I didn't know they had supplies like that. I think you qualify as an official Power Designs addict at this point. How many of these things do you have now?

Fortunately, space constraints on my lab bench limit me from acquiring any more

[timb]

That's pretty cool. I didn't know they had supplies like that. I think you qualify as an official Power Designs addict at this point. How many of these things do you have now?

Fortunately, space constraints on my lab bench limit me from acquiring any more

Five…with a sixth on the way!

Honestly, I do plan on reselling most of these. I’m buying broken ones off eBay to fix up and give new homes to, rather than seeing them end up in a scrap yard.

I’m keeping the TW6050D I just posted, plus the 2020B I’ve been working my ass off to fix.

I’d also like to build a sort of oven restoration kit to sell on eBay, so people can modernize the oven and keep them going. But we’ll see how my own attempts go on this 2005.

Life’s been pretty stressful lately due to a family member with some medical issues and my addict cousin staying with me while he gets clean, so working on these at night gives me some “zen” time.

Anyway, I did get the 2020B case repainted, and it came out pretty great I think!





I also snapped a picture of three other cases together (with the original paint) to show the slight color variations:



It’s interesting to note that the 2005A is about 0.5” longer than the 2005; so they obviously used an entirely different frame. My 1025P was produced in ’67 judging by the caps, while my 2005 has a ’65 date on the oven; meanwhile my 2005A has ’73 stamped on the upper lefthand corner of the from panel’s backside. According to the official 2005 schematics, it would have entered production in ’65; the 2005A schematics are dated ’69 so that would most likely mean early 1970 for the first production run. I can’t find schematics for the 1025P, but it’s constructed like the 2005 (all point to point wiring) so I’d say 1967 is a good bet for initial production. Why am I saying all this? Well, if you look at the 2005’s top cover, you’ll notice it only has a couple of vent slots, meanwhile the 1025P has the hole pattern they used for all the other precision supplies. However, the 1025P and 2005 both have the circular indents, which were intended for stacking the supplies onto of one another; the 2005A omits this. Why?

Well the 2005 and 1025P both included circular rubber feet that simply screwed into the bottom covers, the 2005A on the other hand included molded plastic feet with a metal tilting bail; this setup appears to have been used on all of the following precision supplies (including my 2020B).

Speaking of the bottoms of these supplies, one final curious point: The 2005 and 2005A both used non-painted metal bottom panels, however, the 1025P (which chronological came in-between the two aforementioned supplies) had a painted metal bottom panel, with the same normal hole pattern as the top. This was also the way all future supplies would be constructed, except the 2005A.

I know none of this matters, but for some reason Power Designs really fascinates me. I wish I could find someone who worked at the company for a long time to interview. Companies that produced lab equipment in the 1960’s were constantly popping up and going out of business. There’s a few great articles about multimeters that talk about it. The history of calculators from about 1950 to 1990 is very interesting as well. The ’70s saw an explosion in the production of calculators; it went from huge mechanical desktop adding machines, to solid state, to pocket calculators to personal computers. Okay, I’m rambling now, sorry.

[MarkPalmer]

I enjoyed the photos of the TW6050.  Funny you mention the history of calculators, as that's one of my biggest electronic fascinations- vintage USA made calculators and computers.  I have quite a few early Bowmar, Commodore, Novus, and HP calculators and enjoy fussing around with them.  Other than HP's most vintage calculators can be found dirt cheap.  MY HPIB instrument system works through an HP 85 computer, just as it would have been used in the 1980's and I can actually program it, still remembering BASIC from high school. 

Here are the power supplies I use on my workbenches.  TW5005, 3650S (X2), 5020 (new to me), and 1556 HV, a LKB 2500V/250mA modified electrophoresis supply is on top of it.  I have a few small HP supplies on the shelf.  I just respect Power designs as an American company that (until bought out) didn’t seem to ever compromise their quality principles.  These bench supplies are extremely reliable and I’m sure mine will still function for many generations after I am gone.  There is something much more satisfying in using these supplies along with my other 70’s and 80’s vintage lab grade equipment in comparison to this white plastic with Playskool rubber bumper crap that is made today, much of which only runs a few years until meddled with issues that can’t be resolved in a realistic manner.  My PD supplies I trust and I don’t worry about- when I turn them on to use them, they always work as they should.  And hey, they are cheap enough on E-bay where even too many of them doesn’t set you back much.   







-Mark-

 

[timb]

Wow, that’s awesome! I really want one of those HV units.

So, I got the 2020B all buttoned up. I ended up putting an LT1001 in there, which basically has no drift compared to the OP-02 *and* it has a 4mV offset. So this is basically the highest Op-Amp upgrade you can get. The next down would be the OP-177.







I ended up replacing all the electrolytics, every transistor, both Op-Amps and all the power resistors; taking every pot physically apart, cleaning, re-lubricating (Silicone or White Lithium Grease) and applying DeOXIT Gold or Shield to the contacts. Even the "sealed" Helipots used for zero and cal got the treatment (you can use super glue to reapply the bottom cap afterwards).

The only thing left to do is build a replacement meter. It's either going to be some sort of an LED bargraph or perhaps an Intersil 3.5 Digit LED driving DMM-in-a-chip. (Maxim make newer versions that are Delta Sigma, requiring few external components and can do 4.5 digits.)

Of course I'd use two of those adorable HP four digit 7-segment bubble displays to display voltage and current at the same time. Simply use the mounting holes from the old panel meter with short standoffs to hold the board and get a piece of red acrylic cut to fit into the existing cutout with a bit of glue.


Anyway, the new color looks amazing.

[saturation]

What a great job you did, timb 

I was surprised by how out of shape your unit was and how you've brought it back to life, and looking forward to seeing your upgrades.

I'd be curious to see whether the panel meters will inject a tiny bit of noise into the output if you build it into the chassis. 

I'm like MarkPalmer on maintaining the original look and technology as much as possible, and when I do need better metering of the power output, just drop my eload in series or use separate DMMs. 

I've kept original labels from the primary owner whenever possible [ not the seller].  The top one was from Boeing, and one of the others is either Martin Marietta, or Grumman but fell off, and one from a cal lab that has stickers on the screws not to open it  .  Aside from cleaning and some Dexoit, the units are working as they came OOB from eBay.