POWER DESIGNS PRECISION POWER SUPPLIES

[idpromnut]

@linux-works:

I've looked at the remote control circuit, and it seems that there is a MAX_RANGE voltage (so say 60V in the case of my 6050C or D) that is placed across a pot. There is a constant current circuit that feeds this pot and what you end up with is the voltage drop across the pot is the set output voltage of the supply. I will see if I can grab a picture of the circuit and post it. Perhaps in a new thread

[linux-works]

its just a voltage programmed PSU.  so anything that can let you program in a control voltage (not much current is needed, of course) will let you remotely control the PDI psu.

I've been able to put 60v easily into my resistor/relay atten board and it works fine.  it has no active parts in the way (no pre or post buffers) and so its fully passive and can take even 100v if you really want to.  to test 127db of atten, I did have to put 60v thru it just to see the few microvolts coming out the other end.

the only thing that would need changing is the Rs and Rp values to make it linear instead of log (audio).

you would hear relay clacking as you change the control voltage, but its not very loud if its inside a box and those are telecom relays, not big loud power ones, so they don't make too much of a bang-bang sound.  in my circuit, it uses latching relays and so the voltage only happens (across their coils) for a few ms and then the magnetics hold the relay in A or B position until you change it.  that's kind of neat in that the circuit has a 'memory' even when powered down.

[timb]

That's over complicating for this use, I think. Here's my plan:

Grab a 60V linear MOSFET or transistor and simply hook the source and drain up to the RV lines. Then use an OpAmp with the output hooked to the gate and the non-inverting input hooked to the PSU output. Then you can use a DAC or PWM signal on the inverting input to control the whole thing. (You'll also have to either attenuate the PSU output or boost the DAC's output for feeding the main OpAmp.)

I'll draw the circuit up later.


Sent from my Smartphone

[idpromnut]

That's over complicating for this use, I think. Here's my plan:

Grab a 60V linear MOSFET or transistor and simply hook the source and drain up to the RV lines. Then use an OpAmp with the output hooked to the gate and the non-inverting input hooked to the PSU output. Then you can use a DAC or PWM signal on the inverting input to control the whole thing. (You'll also have to either attenuate the PSU output or boost the DAC's output for feeding the main OpAmp.)

I'll draw the circuit up later.


Sent from my Smartphone

Hmm, seems simple enough. I might try that out tonight. However, did you reverse your inverting/non-inverting connections for the Op-Amp? I.e. feed the non-inverting with the smoothed PWM signal and tie the inverting input to the PSU output).  But I like it!

[robrenz]

That's over complicating for this use, I think. Here's my plan:

Grab a 60V linear MOSFET or transistor and simply hook the source and drain up to the RV lines. Then use an OpAmp with the output hooked to the gate and the non-inverting input hooked to the PSU output. Then you can use a DAC or PWM signal on the inverting input to control the whole thing. (You'll also have to either attenuate the PSU output or boost the DAC's output for feeding the main OpAmp.)

I'll draw the circuit up later.


Sent from my Smartphone

Have you not just cut out all of the original stability of the supply and taken that task on in your circuit?

[timb]

We're just using the FET as a variable resistor essentially. So I don't see how, as that's how the remote voltage works on the 20xx series. (You're bypassing the internal voltage knob resistors with an external resistor.) So I'd personally use a fast, low noise, low drift OpAmp.


Sent from my Smartphone

[robrenz]

I am no EE but I think as soon as you use the PS output to drive your op amp you have two control loops.  If you supply a fixed voltage or resistance the control loop is only the original PS.

[idpromnut]

I seem to remember reading in the operating manual (+theory of operation) that any noise introduced onto the remote programming line will be present on the PSU's outputs. The idea was that you would use shielded cables + "shmick" pots to have near or equal noise specs at the output. But I could be very wrong about that. I need to consult The Fine Manual.

[timb]

I seem to remember reading in the operating manual (+theory of operation) that any noise introduced onto the remote programming line will be present on the PSU's outputs. The idea was that you would use shielded cables + "shmick" pots to have near or equal noise specs at the output. But I could be very wrong about that. I need to consult The Fine Manual.

Yup, this is correct. They also recommend using Make-before-Break relays if you're switching in resistors, because when it goes open circuit it causes a loss of regulation which would induce transients into your output.

And nobody wants their PSU to be like Portland.


Sent from my Smartphone

[idpromnut]

Yup, this is correct. They also recommend using Make-before-Break relays if you're switching in resistors, because when it goes open circuit it causes a loss of regulation which would induce transients into your output.

Actually I believe it's "worse" than that... when the control path goes open circuit, the crowbar kicks and you need to power cycle the supply. Not damaging, but it would ruin your day.

[timb]

So, my 2020B came in today… Of course the seller packed it extremely well!



Look at the layers and layers of bubble wrap! Of course, you only need bubble wrap on the top, right?



Unfortunately, despite the expert packing, it seems the fuse holder was broken in transit.



Not enough to glue it back together. I also shocked the shit out of myself after plugging it in when I went to move it, because I forgot about the exposed fuse. I’ll just replace it with a modern holder.

So, upon turning it on, it does go into overload like the seller stated. I measured 40VDC on the output. Upon checking the back, I noticed there were no straps between the Sense and Output terminals, so I added some wires.



Despite this, it still was outputting 40 volts. I pulled out the current limit knob and a puff of magic smoke came out the top! I opened the bottom cover and the internal fuse was blown. It also appears that R18 is deformed, so I’m thinking that’s what was burning (it smells like it).



So now I’m going to review the schematics and do some tracing. Any recommendations for some things I should check for?

Finally, here’s a nice picture of the entire board. Other than R18, everything else seems in good shape.

[idpromnut]

So I threw together the circuit you were describing timb, and yes, it works great!   I tuned it to use a control voltage of 0-6V for the full 60V range (and 0-0.6V for the 6V range).

[idpromnut]

Also, @timb: right above (in the pictures you posted) the burnt out resistor, it looks like there is a fuse holder. There should be a fuse in there (if there wasn't one when you got it).

[timb]

Awesome, good to hear that circuit works! And yeah, there was a fuse there. It blew when the smoke came out.

Oddly enough that resistor measures fine! I guess the coating just melted. (Still might replace both.)

Can anyone with a 2020B measure the diode drop of Q5, Q6 and Q7 for me? (The transistors mounted to the frame of the case.)

I'm getting a drop of 0.08V on Q6 & Q7, but I'm measuring in circuit so not sure if that's the cause or not.


Sent from my Smartphone

[SeanB]

One of them is shorted then.

[timb]

One of them is shorted then.

Damn, was afraid of that. They're unabtanium, too. I can't find them in any transistor cross reference or even a data sheet online. They're MS1700G's.

Hmmm, looking at the manual, the C500 used IR4230's and the 5020 used FS1700E's. Maybe I'll have more luck finding a match for those. They're both slightly higher spec models, so I imagine they'd work. I just need some basic specs to base a modern replacement on. (Outside of voltage and current.)


Sent from my Smartphone

[idpromnut]

@timb: I found this, which might give you a hint to some of the specs of that transistor: http://www.parttarget.com/5961-01-094-7330_5961010947330_A65212.html/-DB6C6E5C-DADE-4CD7-8AA4-A5D858577F9D


Good luck!

[SeanB]

It is a 2N1700, with a Motorola house coding and possibly a gain binning at production. If it is in a case that bolts to the chassis I actually have some............ Or at least one, and probably more in another box somewhere.

[timb]

Yup, it's in a TO-2 package.

If you could check and see if you have two, I will send you some $$$ for shipping (plus whatever you want for the parts) along with my eternal love!

I assume the FS1700 is a Fairchild version of the 2N1700 as well? Curious why they used a different brand for the drive transistor...


Sent from my Smartphone

[SeanB]

One 2N1721 made 0780 necxt to me right now, with its little mount and nut, but no insulating kit though ( I probably lost those somewhere) and I will have to look to find a mate for it. Sad to say I cannot post it at present, unless you are willing to pay the $150 (Yes, got a quote Friday for another reason) fee to send via courier. SAPO is doing the best it can to prove I live in a fourth world country by escalating the 6 week strike to it being total. SA is making Detroit look good as a vacation spot, I might go there instead of going to relax in Syria or Iran, even Afghanistan looks good.

[timb]

Damn dude, that sucks! Didn't know about the strike. I might be able to find a local replacement, if not I can wait!

It's weird, I can find 2n1700's on eBay, but they're all in TO-39 metal cans, not in the TO-3 package I have.

If I were to go with a more modern replacement, what parameter do I care about matching the most (assuming Vcb, Vce and Ic are all => the current unit)? I'm thinking Hfe, right?


Sent from my Smartphone

[timb]

This looks promising: http://www.onsemi.com/pub_link/Collateral/2N5038-D.PDF

Available on Mouser and eBay. What do you think?

Edit: Whoops! Wrong link. Found a suitable replacement anyway, see below.


Sent from my Smartphone

[timb]

Okay, I took out all three power transistors on the side of the case. It’s Q7 that’s shorted. The driver transistor (Q5) other pass transistor and both fine. I’ll most likely still replace them as a pair.

[SeanB]

Put in a 2N3773 if it is a TO3 package, it will be a good replacement, and will work well within it's SOA limits as well.

[timb]

Put in a 2N3773 if it is a TO3 package, it will be a good replacement, and will work well within it's SOA limits as well.

Ah yes, designed for linear operation. Looks pretty good! Found a bunch on eBay, so that’ll work out perfectly.

Okay, so like I said above, I pulled out all three power transistors and found the one with the short. To determine if that was the only problem, I popped the best fuse I had in (1A Slow-Blo) and hooked the driver transistor and the one pass transistor up via alligator clips:



And…



This thing is pretty much bang on, and I haven’t even touched the calibration pots.



So after testing R17 and R18 with makeshift Kelvin probes, they’re both exactly at 0.75Ohms, despite R18 being a bit deformed. However, after testing most of the resistors on the board, I did find R34 (the matched resistor to the LM399 reference) is measuring 3K~, instead of the marked 7.32K. Though, it doesn’t seem to be affecting accuracy? I might still replace it; unfortunately I can’t find any 0.3W replacements on Mouser (can only find 0.125W ones in-stock). So I may have to resort to using an SMD version with a little adapter board I’ve got laying around.