Sorensen / Xantrex XT 15-4 Power Supply Repair / Calibration

[kelchm]

I recently picked up two Sorensen XT 15-4 power supplies with the GPIB option on eBay relatively cheaply (~$100 shipped). One of them worked well out of the box while the other reads voltage quite inaccurately. The manual (attached) briefly discusses the theory of operation as well as calibration. I'm trying to work from this to determine what is wrong with the misbehaving supply.

The symptoms I've noticed so far:
- Displayed voltage is off anywhere from 0.5V to 2V compared to actual
- Voltage bar graph seems to be stuck all the way filled

So far I've verified that R45 is set correctly such that 10.01V is measured across C5. Since I have another fully functioning unit, I've also tried swapping the entire front panel (A1) onto the malfunctioning unit. Interestingly, after performing this swap I see similarly inaccurate voltage readings reading displayed as well as the voltage bar graph stuck at 100%.

I think the problem at this point must reside on the power circuit (A2), but I'm not sure where to begin digging.

[macboy]

...
The symptoms I've noticed so far:
- Displayed voltage is off anywhere from 0.5V to 2V compared to actual
- Voltage bar graph seems to be stuck all the way filled
...

It is probably oscillating like crazy. Do you have a scope? The output should be steady DC with very little noise or ripple. Your DMM set to AC may or may not be able to read an AC voltage (it may not if it has poor enough bandwidth).

You may find the attachment below very useful    I scanned it from a paper copy of the manual.

Check ESR and capacitance of all electrolytic caps. You can bet some are dried out, even the big filter caps for the output.

[duak]

The above schematic didn't display properly for me.  Here's another link:
 https://www.eevblog.com/forum/projects/dual-output-transformer-with-relay/?action=dlattach;attach=152407

The voltage monitor circuit looks quite simple: a voltage divider, connector driving a bargraph indicator and an Intersil 7107 DVM circuit.  As Macboy indicates I'd check for oscillations or other AC noise on the output.  I'd also check the power supply voltages to the indicator and DVM circuits for the correct level.

BTW, these supplies were designed and built just down the road from me and I have a dual output unit on my bench.  Mine is 30 years old and I wouldn't be surprised if electrolytic caps had failed.  I believe the only problem this unit had is the power switch failing.

Best o' luck,

[kelchm]

Thanks for the advice!

I'm still learning how to use my oscilloscope, but here's a screenshot of what I'm seeing currently. I should note that this is with my probe set to 10x, so the peak to peak is actually 10x worse than than what's being displayed under the measurements, right?

[cdev]

I have a working triple unit (1-6v 2x20 v) and would be happy to help...

Output is rock solid on a working unit. Almost no noise.

[macboy]

The above schematic didn't display properly for me.  ...

I did scan at a very high resolution so it's a big image, over 10k pixels wide. Perfect for printing on B/W laser, not great for viewing on screen. It does display fine for me in Chrome, FWIW.  I scaled it down for a more reasonable pixel count (now 6 MP instead of 60 MP):

[macboy]

Thanks for the advice!

I'm still learning how to use my oscilloscope, but here's a screenshot of what I'm seeing currently. I should note that this is with my probe set to 10x, so the peak to peak is actually 10x worse than than what's being displayed under the measurements, right?

That's with the output set to zero or so? Yeah, doesn't look good.
Your scope+probe probably don't automatically switch when you switch the 1x/10x switch so you'll need to do the math in your head or set the scale factor for that channel in a menu (if possible). So, yes, the voltages are 10x bigger than implied by what you see.

I have one of these (a 30 V, 2 A model) in similar condition, but mine works up to ~ 1/2 rated voltage, then begins to oscillate. So the bar graph looks fine from 0-15 V, then turns into a blurred full-scale mess above that.  My plan (if I ever get around to it) is to re-cap first, then hunt down remaining problems, if any. I already replaced one big filter cap after finding it had nearly no capacitance and high ESR.

If you don't already have one, get one of those cheap component testers as discussed in this thread. They will measure capacitance and ESR for capacitors, and will identify, test, and measure almost any two or three pin device (resistor, capacitor, inductor, diode, zener, NPN, PNP, JFET, MOSFET, Triac, SCR, etc. etc.).

[duak]

I've seen a waveform like that in a Kikusui power supply.  Turned out to be the electrolytic cap across the output terminals.  I would first try a 100 uF or so el-cap across the output to see what happens.

MacBoy, thx kindly for a revised file.  I'm using Linux with its document viewer & I think I was probably getting display artifacts due to the high res.  As it turned out the file I uploaded wouldn't print properly so yours will come in handy.

Cheers,

[cdev]

If I am the only one of you with a working unit, I would be happy to help you get yours working, what I mean is, if you are at a point where you need to know what does normal operation of X look like on a voltmeter, etc. I can do it.

(I also can sample current if its on a wire that I can fit through my clamp on meter)

I have a triple unit one 7 volt-6 amp unit and two 20 volt 3 amp units. Its nice, it has the big bright LEDs and the linear display.  Its especially nice to be able to limit current to very low values accurately. (it is accurate, even now after 30+ years, my unit was manufactured in the 1980s)

So they are worth getting working. My only bitch about mine is that it has a hum.

Not in the output, the transformers are just old and dried out and they hum.

[kelchm]

Glad this topic has gotten lot's of attention since I posted it. Thanks for all the input guys!

Since my last post I also took a look at my 'working' supply on the oscilloscope and found that it also has a nasty oscillation in the output. I guess that explains the giant red "QA REJECTED" stickers that were on both of these when I received them. 

I did decide to go ahead and tear into the worse of the two today. I've removed all the caps on the A2 board and tested them with my AY-AT transistor tester. Here's a quick list of the worst offenders from my notes:

Location	Mfg / Value	Measured Capacitance	Measured ESR
C4	Nichicon 10µF 50V	7390nF	52.29 ohm
C15	Nichicon 10µF 50V	7710nF	43.40 ohm
C16	Nichicon 10µF 50V	7977nF	7.99 ohm
C16A	Nichicon 33µF 25V	9255nF	12.59 ohm
C17	Nichicon 10µF 50V	7705nF	34.22 ohm
C18	Nichicon 10µF 50V	7286nF	19.05 ohm
C27	Nichicon 10µF 50V	8356nF	3.26 ohm

I thinks basically every Nichicon capacitor in this supply was bad or going bad expect for one 2200µF. With a date of manufacture in 1999, could this be a remnant of the 'bad caps' era? Unfortunately there are also some 10µF Nichicon caps on the A1 board that I will also need to pull and replace. My second unit appears to have only a small handful on the Nichicon caps -- anyone want to put money on those few being bad as well?

I'm going to be putting in an order all of the required replacements tomorrow. Have not yet decided I will be replacing everything or just the caps that tested bad.

[bitseeker]

I don't see leaking electrolyte, so it's probably not due to the "capacitor plague," just age, operating environment, etc. Hope all is well after the recap.