Power Designs TP325 - Fix + Restoration !

[Sparky]

Thought I would create a thread to document and share restoration of my Power Designs TP325!

I picked up this "for parts" unit to tweak a TP340A I have...but it's in better shape than I expected soooo... time for PETA! 

Exterior:
I will polish the meter windows with plastic polish for hobby cars.  I cleaned some sticker/marker residue from the faceplate with IPA...but it left the aluminium a bit cloudy in areas, and I'm still able to see some discoloration from where stickers have been.  How to clean the aluminium faceplate properly?

Only 3 physical issues:

1. Missing voltage control knobs.
REPLACEMENT: I got a few of these.  They could be a little taller, but they will do.

2. Broken locking toggle switch for Source A / Source B tracking mode ON/OFF.
REPLACEMENT: bitseeker indicated possible replacement is Gemini/Alco DPDT Locking Toggle Switch (125V, 5A), part #A201KZQ.  This part a little hard to find in single quantity, but I found comparable item that should work: NKK Switches M2022LL1W01 from Mouser.

3. Split red banana jack.
REPLACEMENT: Sourced identically looking red/black pair from eBay.

Photos:
Pictures of top and side PCB are attached.  No obvious failed parts.  Looks like the calibration has not been touched since manufacture.

Operationally:
I have found a problem with Source A:  The corresponding pot does not control the voltage output.  Instead, when the unit is turned on the voltage slowly increases starting from zero.  After a few minutes it has gone from 0V to 21V...it would probably go higher but I turned the unit off. 

I have not checked ripple on the outputs yet.  Will follow-up with more details in the next post.

I got the service manual from KO4BB's archive.   

Does anyone happen to have higher resolution images of the schematics?  Will be great to have for reference and then start the diagnosis  

Edit: Schematics from service manual is sufficient when printed in "poster" format and align the pages.

[Sparky]

I took a few photos showing voltage increase at Source A.  Adjusting the pot had no effect on the voltage at all.

I also attached a photo of the voltage control pots.  For Source A and B they are 5k, 2W, 10 turn from "Helipot" made in El Salvador.  Perhaps they are using this brand before Bourns?

Source C has a much different pot.  Initially I thought the Source C pot had a mechanical problem with the wiper because it required more torque to rotate.  Now it's known to be clearly different type it makes sense and seems operationally fine.

[bitseeker]

I usually start from the rear (i.e., the transformer end) of a power supply and work forward, checking voltages and ripple. Since you have no control over the voltage, apart from a bad pot, my first reaction is usually the pass transistor(s). The voltage doesn't shoot up, like a shorted transistor, so it might be something else or the transistor isn't completely fried. It could also be a combination of things (e.g., electrically leaky filter cap that slowly charges and a shorted transistor). Can it sustain a small load (e.g., 100mA)?

Also, you might want to temporarily swap out the broken locking toggle switch for a normal one so that you don't have the switch in some funky state causing weird behavior.

Although I haven't seen a PD with Helipot multi-turns, they're good stuff, too. So, it's plausible that it could be original. PD did have more variations in their BOM during the later years of the company's history. Even the brushed faceplate wasn't oriented in the same direction all the time (my TP340A being one such example).

Edit: Actually, looking more carefully at the photo, there's a high probability that the switch is the problem. If you turn down the voltage knob on supply B, does A follow it? When in tracking mode, A does not operate. B is the master.

[Sparky]

I usually start from the rear (i.e., the transformer end) of a power supply and work forward, checking voltages and ripple. Since you have no control over the voltage, apart from a bad pot, my first reaction is usually the pass transistor(s). The voltage doesn't shoot up, like a shorted transistor, so it might be something else or the transistor isn't completely fried. It could also be a combination of things (e.g., electrically leaky filter cap that slowly charges and a shorted transistor). Can it sustain a small load (e.g., 100mA)?

Also, you might want to temporarily swap out the broken locking toggle switch for a normal one so that you don't have the switch in some funky state causing weird behavior.

Although I haven't seen a PD with Helipot multi-turns, they're good stuff, too. So, it's plausible that it could be original. PD did have more variations in their BOM during the later years of the company's history. Even the brushed faceplate wasn't oriented in the same direction all the time (my TP340A being one such example).

Edit: Actually, looking more carefully at the photo, there's a high probability that the switch is the problem. If you turn down the voltage knob on supply B, does A follow it? When in tracking mode, A does not operate. B is the master.

Thanks for the guidance bitseeker!  I printed the schematics and PCB component layout on multiple sheets so it's easy read and get started. 

Attached is rotated schematics/component layout pages in PDF for anyone following along   The top PCB is entirely for Source C, whereas Source A and B are on the side PCB.

Definitely I have to replace the toggle switch ASAP...who knows how the contacts are being connected internally...  Internally the switch seems in one position as no matter what way I move the little bit of lever that remains there's no "latching" into any other position and no change in the behavior of Source A/B voltage meters.

Load Test
All channels can support a 0.5ohm (50W) load up to max current limit.
Source A: Amp meter showed increased from 0A up to 1A and then the FAULT light came on.  The current increased automatically as the voltage is not being controlled by the pot.
Source B: Voltage/Current as expected, with FAULT lighting when I got to 2V for the 0.5ohm resistor. 
Source C: Voltage/Current as expected. FAULT light coming on at 3A but it's just the current limit setting.  I tweaked that and went to 5A.

I've measured about 120mV ripple at the output terminals of Source B, C.  My unit has PCB revision B from 1975 (according to schematics) so the electrolytic caps are about 43 years old.  Perhaps I can proceed to replace the output filter capacitors and switch, and then see where we are at?

The good thing is Source B seems to be functioning as expected so that can be a good reference for Source A.

[vindoline]

Exterior:
I will polish the meter windows with plastic polish for hobby cars.  I cleaned some sticker/marker residue from the faceplate with IPA...but it left the aluminium a bit cloudy in areas, and I'm still able to see some discoloration from where stickers have been.  How to clean the aluminium faceplate properly?
.

Many of us here have had excellent results with melamine sponges (also called Magic sponges). Dip it in isopropyl alcohol and rub firmly with the grain of the metal. It does work wonders!

[vindoline]

2. Broken locking toggle switch for Source A / Source B tracking mode ON/OFF.
REPLACEMENT: bitseeker indicated possible replacement is Gemini/Alco DPDT Locking Toggle Switch (125V, 5A), part #A201KZQ.  This part a little hard to find in single quantity, but I found comparable item that should work: NKK Switches M2022LL1W01 from Mouser.

Operationally:
I have found a problem with Source A:  The corresponding pot does not control the voltage output.  Instead, when the unit is turned on the voltage slowly increases starting from zero.  After a few minutes it has gone from 0V to 21V...it would probably go higher but I turned the unit off

.


As bitseeker has suggested, you really need to address the broken switch first. The feedback loop that sets the voltage goes through that switch. In one position both A and B have their own feedback network. In the other position B shares A's network. That's how the tracking works.

[bitseeker]

Thanks for the guidance bitseeker!  I printed the schematics and PCB component layout on multiple sheets so it's easy read and get started.  Attached is rotated schematics/component layout pages in PDF for anyone following along  

Cool. Thanks. That's handy.

The top PCB is entirely for Source C, whereas Source A and B are on the side PCB.

Yep, very similar to the other TP supplies.

Definitely I have to replace the toggle switch ASAP...who knows how the contacts are being connected internally...  Internally the switch seems in one position as no matter what way I move the little bit of lever that remains there's no "latching" into any other position and no change in the behavior of Source A/B voltage meters.

Yeah, just a regular DPDT toggle will be fine for the interim. Since the handle broke off the original one, there's nothing to lock it into position.

Load Test
All channels can support a 0.5ohm (50W) load up to max current limit.
Source A: Amp meter showed increased from 0A up to 1A and then the FAULT light came on.  The current increased automatically as the voltage is not being controlled by the pot.

Source B: Voltage/Current as expected, with FAULT lighting when I got to 2V for the 0.5ohm resistor. 

Yep, good. ~1A is the max for A and B. A's unregulated state is TBD after you get the temporary switch in.

Source C: Voltage/Current as expected. FAULT light coming on at 3A but it's just the current limit setting.  I tweaked that and went to 5A.

Good.

I've measured about 120mV ripple at the output terminals of Source B, C.  My unit has PCB revision B from 1975 (according to schematics) so the electrolytic caps are about 43 years old.  Perhaps I can proceed to replace the output filter capacitors and switch, and then see where we are at?

I wouldn't just yet. This is why I usually work from the back of the suppy, forward. Seeing ripple on the front end doesn't necessarily mean that the output caps are bad. They could be overwhelmed by heavy ripple from the back end.

The good thing is Source B seems to be functioning as expected so that can be a good reference for Source A.

Yep, you're off to a good start. Get that switch replaced, double-check voltage control of A with the tracking switch on and off, then check for problems starting from the output of the rectifier, CR202, and the ripple on the main caps, C203 & C204. If all looks OK, continue forward.

[Sparky]

Many of us here have had excellent results with melamine sponges (also called Magic sponges). Dip it in isopropyl alcohol and rub firmly with the grain of the metal. It does work wonders!

Thanks for the tip!  I have tried it but without success.  My faceplate does not have the brushed/grain finish --- it is flat without texture.  The IPA evaporates and leaves a cloudy residue.  I will have to keep trying something else.

As bitseeker has suggested, you really need to address the broken switch first. The feedback loop that sets the voltage goes through that switch. In one position both A and B have their own feedback network. In the other position B shares A's network. That's how the tracking works.

Yeah, just a regular DPDT toggle will be fine for the interim. Since the handle broke off the original one, there's nothing to lock it into position.

The switch is coming   I didn't have a DPDT handy so ordered the right one.

Cool. Thanks. That's handy.

You guys are welcome! 

I wouldn't just yet. This is why I usually work from the back of the suppy, forward. Seeing ripple on the front end doesn't necessarily mean that the output caps are bad. They could be overwhelmed by heavy ripple from the back end.

Yep, you're off to a good start. Get that switch replaced, double-check voltage control of A with the tracking switch on and off, then check for problems starting from the output of the rectifier, CR202, and the ripple on the main caps, C203 & C204. If all looks OK, continue forward.

Sounds like a good plan.     I will update here once I get the switch in.  Going to be a few days until I get it...

Thanks guys!

[bitseeker]

Many of us here have had excellent results with melamine sponges (also called Magic sponges). Dip it in isopropyl alcohol and rub firmly with the grain of the metal. It does work wonders!

Thanks for the tip!  I have tried it but without success.  My faceplate does not have the brushed/grain finish --- it is flat without texture.  The IPA evaporates and leaves a cloudy residue.  I will have to keep trying something else.

Ah, yes. I have some precision supplies that have a satin finish. Also, it seems that some panels had a lacquer or other clear coat over the aluminum. That could be what's going cloudy. One of mine has some yellowing of that finish coating (you can see it in contrast to the clear spots that were under stickers).

I will update here once I get the switch in.  Going to be a few days until I get it...

No worries. We're not on a schedule.

[mtdoc]

I've measured about 120mV ripple at the output terminals of Source B, C.

Take care in how you measure the output ripple and noise. If you haven’t seen it already, watch Dave’s video on the subject.  Even doing it as Dave demonstrates will tend to overestimate ripple. I recently discovered this on my PD and HP precision supplies. I was getting  about 2 mV of ripple using the best methods Dave demonstrates.  I knew they should do better than that and sure enough, once I took the time to do a proper setup with a Tek AM502 differential amp, the numbers dropped significantly - down to the few hundered microvolt range.  But even if you don’t have a differential amp or probe, following Daves recommendations will get you close.

[Messtechniker]

I very much like those front panel screwdriver operated pots.
Is the the mounting piece for these still in current production?
Where to buy?

[bitseeker]

I'm not sure if they're still available. The usual round panel nuts I see are either knurled or slotted. These tapered ones, with the two holes or knurled edge for gripping the nut, I've only seen on PD supplies. On mine, panel nuts for the switches have knurling and the ones for pots have the little holes. Both types are tapered on top. They look really nice.

[Messtechniker]

Thanks for your hints. By hunting around a bit, I found these:
Mentor 729. Spindle Bushing (Ø) 6 mm
Available from Conrad, for example.
Not quite as nice, but will probably do.

[bitseeker]

I don't understand how that spindle bushing would work with a potentiometer. If you have success with it, do post some photos.

[Sparky]

I'm back, with successful updates!

Ah, yes. I have some precision supplies that have a satin finish. Also, it seems that some panels had a lacquer or other clear coat over the aluminum. That could be what's going cloudy. One of mine has some yellowing of that finish coating (you can see it in contrast to the clear spots that were under stickers).

Yes, that seems exactly what it is.  There was a few small stickers on the front panel and after removing them there is a yellowish tint where they used to be.  I will try Windex or ammonia next...will really like to stop the cloudy surface.

I will update here once I get the switch in.  Going to be a few days until I get it...

No worries. We're not on a schedule.

@bitseeker, vindoline: You guys were exactly right with the locking toggle!  I replaced it (see comparison photo below of replacement: NKK Switches M2022LL1W01 from Mouser Elec.) and fantastically all the problems are solved!  Both Source A and B now work independently, and also tracking mode works correctly!  I have done a quick check of the output ripple with digitizing multimeter and it seems to be significantly improved, but I need to follow mtdoc's advice to review the procedure of Dave's video.

It's great that the switch has resolved all the problems!  But a little disappointing that the circuit sleuthing has come to an end so quickly!  Anyway, I am very happy to have the unit functioning so well again.


Cosmetically still two items can be improved:
1. Front panel: need to remove the cloudy appearance.
2. Split red banana jack. Unfortunately the replacement I sourced looked identical but is physically much larger than the original.  I still need to find a replacement.

[bitseeker]

That's great news, Sparky. The NKK switch looks like a perfect substitute. Looking at my TP340A, TP343B and 2020B, they're all locking types, but the handles are different. The NKK M2022LL1W01 is just like the one on my TP340A (provides some grip), whereas the handles on my TP343B and 2020B are smooth.

Anyway, as I mentioned, PD had variations in the parts they used, even for the same model. So, you've got a fair bit of flexibility.

That said, if you can't find an exact replacement for that broken binding post, you could replace them all so that they're at least consistent with one another. Across all the PD supplies I have are three different binding post styles.

[Sparky]

That's great news, Sparky. The NKK switch looks like a perfect substitute. Looking at my TP340A, TP343B and 2020B, they're all locking types, but the handles are different. The NKK M2022LL1W01 is just like the one on my TP340A (provides some grip), whereas the handles on my TP343B and 2020B are smooth.

Anyway, as I mentioned, PD had variations in the parts they used, even for the same model. So, you've got a fair bit of flexibility.

Yes, I was comparing the logging toggles (NKK and Gemini/Alco) to the one on my TP340A and it had the little bit of increased diameter in the level/handle, too.  I noted the NKK one seemed to match it so that's the one I went for.  The flexibility of sourcing parts is definitely a good thing in case one manufacturer stops producing the part.

That said, if you can't find an exact replacement for that broken binding post, you could replace them all so that they're at least consistent with one another. Across all the PD supplies I have are three different binding post styles.

I might have to, alternatively I was thinking there must be another "for parts" TPxxx out there with perfect binding posts 

[bitseeker]

alternatively I was thinking there must be another "for parts" TPxxx out there with perfect binding posts 

There certainly are. And so begins the slippery slope that is TEA.

[Sparky]

Take care in how you measure the output ripple and noise. If you haven’t seen it already, watch Dave’s video on the subject.  Even doing it as Dave demonstrates will tend to overestimate ripple. I recently discovered this on my PD and HP precision supplies. I was getting  about 2 mV of ripple using the best methods Dave demonstrates.  I knew they should do better than that and sure enough, once I took the time to do a proper setup with a Tek AM502 differential amp, the numbers dropped significantly - down to the few hundered microvolt range.  But even if you don’t have a differential amp or probe, following Daves recommendations will get you close.

Thanks mtdoc for this timely reference and pointer to excellent info on proper measurement technique.  I had not watched this video from Dave.

The measurements I reported earlier in this thread were with no-load connected and using digitizing mode on a bench-top DMM.  Now that I fixed the toggle switch I watched the video and repeated the measurements, this time with oscilloscope. 

I tried first with Rigol DS2000 with 10:1 probes (don't have 1:1 probes...) and doing "Ch1 - Ch2".  Even without the probe GND clips and AC coupling on the scope I had to use 2V/div to see the Ch 1 and Ch 2 waveforms on the screen.  Then I did the MATH function for Ch1 - Ch2 but because of the low resolution on the original channel data the subtraction was very low resolution and blocky.  I would guess something like 20mV peak-peak noise...it was so problematic I did not continue further.

I then switched to a Keysight 3000X series which fortunately I have a differential probe (N2818A 200MHz bandwidth).  It is a fixed 10:1 probe, too.  The differential probe helped a huge amount to eliminate the common mode signal and I can set 5mV/div for vertical scale.  Using the measurement functions I got 6-7mV peak-peak when using a long time base (few milliseconds) and when "zoomed in" (nanosecond time base) the peak-peak measurement fluctuated 1-2mV.  The AC Vrms was always < 800uV.

Unfortunately I could not set the input channel on the Keysight to AC coupling when using the differential probe.  Hmm...it seems not to allow it.  An alternative would be using a low-voltage low-bandwidth differential probe with 1:1 attenuation....I went looking for such a probe and cannot find one for Keysight 3000X series.