Help with power supply repair

[Enginuitor]

The other day I stopped by Halted Specialties Company in Santa Rosa (an electronics surplus place) and managed to talk them down to $25 for a large 0-40VDC @ 0-10A lab supply. It's by "Power Ten Inc." and the model number is 3101-4010. It's a big rackmount chassis with a pretty basic design. Coarse/fine adjustment knobs for both voltage and current, bananna plug outputs, analog meters.
Internally, it's basically a 3-part design. First the input stage, where the incoming line voltage passes through a switch and fuse, then through a rectifier bridge, then is smoothed out to about 65 volts DC by a large capacitor. This plugs into a port on the board. The output port of the board connects to an identical capacitor, then to the screw terminals on the back, then to the front bananna jacks.
Three additional taps on the transformer connect via a header. This appears to be rectified, regulated and filtered into positive and negative low-voltage lines, which presumably power the logic (among other things).
There is a circuit with a 12 zener diode and 1.5k resistor which both get very hot. However, I am overlooking it at this point because the temperatures seem to be within their operating ranges. However, there is evidence of extreme overheating in the past (the board below the resistor is charred) so I have replaced both of these components with higher-wattage values just in case.
When powered on, the unit makes a pleasing hum, and the voltage needle drifts slightly to the left, below zero (hmm, wierd...). Output voltage is about -0.5VDC. Turning the voltage setting down results in a slightly greater negative voltage, while turning it up results in a smaller negative voltage (closer to 0).

More info to come. I'm basically going to keep posting here as I discover and test things, and would appreciate suggestions on things to check or possible solutions.

View attachment 35620

 

[Enginuitor]

In addition to the power output terminals, there is a screw terminal block on the back with the following terminals from left to right: A1, A2, A3, A4, -S, -, +, +S, A5, A6, A7. -S is jumpered to -, and +S is jumpered to +. Originally there were also a few more jumpers near A1-A4, but being the genius I am I just went and took 'em all off without even noting their positions.
All the traces going to these terminals are very small, so these aren't power hookups. However, some brief trace-following showed the + and - terminals to be connected to the power output.

I suspect these terminals are for external control and/or monitoring. Anybody got any ideas?

 

[Enginuitor]

By the way, the "chip in charge" here is an SG3524. The only other IC on the board is a photodarlington optocoupler, which appears to activate the shutdown relay. I'm going to probe around the '3524 with my oscilloscope and see if I can find any signs of malfunction.

 

[Enginuitor]

After some meter/scope work, I have determined that the oscillator network for the '3524 is operational, as pin 3 (osc/sync) is generating 3.5V pulses at about 30kHz as specified by the datasheet; however there is no signal from the outputs! I am now going to start backtracing from the regulator control lines and see what I find...

 

[Enginuitor]

An update on the overheating components. It appears it is due to a poor design rather than some kind of failure.
The resisitor and zener diode are essentially being used as a crude voltage regulator. The diode drops 12 volts in reverse, so by running power across it (I.O.W. putting it straight across the positive supply and ground) you get a nice 12-volt potential between the anode and cathode. However, as we know, using a semiconductor to complete a short circuit across an extremely powerful voltage supply tends to result in a spectacular loss of magic smoke and resulting failure of said component. That's what the resistor is for; it is placed in series with the diode, limiting the amount of current so that bystanders will not have to have tiny diode bits surgically removed from their faces. This is why the resistor gets so hot; it has to drop over 50 volts between the cathode and the 65V positive supply. It is dissipating around 2 watts, which, over time, can make the component quite hot to the touch.

 

[Enginuitor]

Found an operation & maintenance manual! Luck just doesn't get any better than this...
http://www.elgar.com/pdfs/manuals/PowerTen/3101A%20Series%20Models%204006%20and%204010%20Operation%20and%20Maintenance%20Manual.pdf

Remember that strange screw-terminal header I mentioned? As I gathered based on the internal connections, it is used to set up the unit for remote or local control and sensing. Actually, you can do a lot of neat things like operating two units in parallel or series, in a master/slave configuration for added power (both are controlled by the "master" unit's knobs).

I will set the jumpers up properly for local control, then report back...

 

[Enginuitor]

It miraculously worked!...
For a short period of time.

Everything was working fine, so to test the current output I put a small wire across the output. I started easing the current up, and it did fine up to about 3 amps, then suddenly the needles dropped to zero. I think I may have heard a small click, but I'm not sure. This is a 10-amp power supply -- What happened??


Now it's in a truly wierd state. When I turn it on, the voltage slowly rises, charging the output capacitor, until it hits the OVP cutoff point (set by a small potentiometer; I have it at about 46 volts) at which point the relay shuts down the output stage. This happens regardless of the voltage settings, though a higher setting seems to make it happen more quickly. Obviously the system is no longer regulating properly.

By the way, there is a full schematic in the pdf I linked to.
Any suggestions?

 

[Enginuitor]

I was halfway through the calibration steps when a wispy trail of magic smoke lead me to R43 (see schematic). Perhaps something has gone wrong with U2 (MC1723CG) or the surrounding circuitry. I would appreciate anyone's input.

I have posted the full schematic so people no longer have to download the PDF:

View attachment 35626

 

[ante1]

Greg,

So, you are in two topics with this one? Presuming the supply worked for a short time before the

 

[Enginuitor]

Presuming the supply worked for a short time before the

 

[ante1]

Greg,

I can

 

[Enginuitor]

I can

 

[ante1]

Greg,

It

 

[Enginuitor]

Thanks again Ante!
After I get all my real-life duties taken care of today, I'll hunker down and start taking measurements.
Updates to come...

 

[Enginuitor]

You have to start from the rectifier and proceed down the line, U9 input and then U9 output and also what happens there if the controls are moved.

Interesting. The U9 input is at about 65 volts (it's connected directly to the input) while the U9 output behaves exactly as the final voltage output; it ramps slowly up to about 45 volts before the OVP kicks in. Next, I will disconnect C5 on the input side so I can see exactly what's coming out of U9.

 

[Enginuitor]

Took some more measurements on U9. Its output with the cap disconnected reads on the meter at about 19VDC with voltage knobs turned all the way down, and 23VDC when turned all the way up. In addition, when switching the meter to AC, it reads from about 1.1 volt to 0.8 volts respectively, at approx. 15.5kHz. The current knobs appear to have no effect. Unfortunately, I don't think my oscilloscope can handle very high voltages on DC coupling, so I can't see exactly what is going on here. Any ideas?

 

[Enginuitor]

Q1 base (12V)? Q1 emitter R53 (~10 V?). And Q9 base (?) and Q9 emitter (?) and to they follow the controls?

Q1 base is about 12.6 volts. Q1 emitter is about 12.2 volts. Q9 base is about 3 volts with C5 disconnected from input side, and does the ramping thing (0 to 45 volts) when C5 is fully connected. Q9 emitter is of course the final voltage output, and also does "the ramping thing". Very curious. I am ordering replacements for Q9 and U9. Anything else I might want to replace?

 

[Enginuitor]

Uh oh...
Just got a quote from MicroSemi, the only known maker of the PIC647:

PIC647 23pcs min @ $45.46 each

o__0

Methinks it may be time to research an acceptable alternative to this IC!
Here is the manufacturer's datasheet:
http://www.microsemi.com/datasheets/WT7-12.PDF

Below is a schematic of the PIC647. It is essentially a small transistor driving a bigger one. If I were to basically create a makeshift substitute for this part, can anyone reccomend values for the components?
View attachment 35654

 

[ante1]

Greg,

I just found the PDF file you submitted, I just had the schematic up untill now. This confirms some of my thoughts but it also confuses me some. In 4-1 it is stated that the pre-regulation circuit operates at approx. 35kHz and you say 15.5kHz so here is something wrong. Since it works as a PWM regulation the frequency should not be that much off pace.

Ante :