Agilent 66309D Power Supply Repair (schematic)

[Jester]

I picked up an Agilent 66309D power supply on ebay in non working condition.

First impression was that channel 1 was working, and channel 2 was dead (based on DVM readings).

I traced the channel 2 fault to a fuse and after replacing the fuse channel 2 appears to work fine, meters correctly and produces full output current at 12V.

Back to channel 1: the voltage setting agrees with external DVM measurement, however the output waveform looks awful when loaded (very clean when unloaded)

I downloaded the so called service manual, however it does not include detailed schematics just block diagrams and some abbreviated portions of the schematic that lacks the details required to really dig into the problem. Are detailed schematics available?

This is what the channel 1 output looks like when set to 10V and loaded to about 1.25A


Update 1.............................
CH 1 no longer outputs the setpoint voltage, it now outputs about -3.2Vdc, so whatever was bad is now slightly worse. (It did this without any intervention from me, other than having the unit powered up and on)

I opened things up a bit and found something unrelated but interesting, see broken 1r resistor in bottom right corner of image. Appears someone was messing with this as the broken resistor is solder tacked to a pad and bridging across an adjacent track (good thing for the solder mask), not sure what the previous guy was doing, the odd thing this resistor looks to be associated with the now good channel 2. I'm going to remove it to make sure it does not short the trace and attempt to troubleshoot Ch1 without the schematic.

I did a bit of digging on the web for a proper schematic and so far nothing, looking at the service manual for the popular 6612C power supply they appear to use the same "A1" power board you would think someone must have posted a schematic for this circuit as there a gazillion of them out there?

Any troubleshooting hints are welcome


Update 2........
Found that F309 had blown (CH1), replaced it and now CH1 is in regulation again (back to lousy waveform), will try to limit the on time while I troubleshoot

Update3.....
I zoomed in on the spike and it looks like this:


This is the partial schematic


With no load the output looks perfect, with even a small load the spiked output occurs. Hmmmm.....

[EPTech]

Hi there,

Concerning the broken resistor, looks like the previous owner tried to disable the crowbar. maybe it was on constantly. However, the fuse F309 comes after the crowbar. And is the output is not loaded and there are only high impedance measure circuits on the output, the fuse should not blow. Unless the measurement circuits are not so high impedance anymore, possibly shorted out. That would also explain why the output is not tracking anymore. So I would check whether there is any low impedance path to either rail on the input circuits (opmaps). Also check the protections devices. My guess is somebody put a way to high external voltage on the supply, causing the voltage feed-back circuit to break down. It may also be the series regulating transistors but that would be too obvious. 

Good luck.

[Jester]

Without a full schematic, I'm not sure what the broken resistor is used for, it appears to be connected to Ch2 which appears to be working fine now.

New observations:
 - with a 1k load and less than 4V out the output is clean, if I increase the output voltage, the spike begins as soon as the output current is greater than about 5mA, if I reduce the load resistance the voltage at which the spiking occurs is also reduced accordingly, so the problem definitely appears to be current related.

- the U309 rails go way beyond their normal +/-15V while the spike occurs, not sure if this as a result of the spike or causing it?


I might try disabling the down programmer, followed by lifting U309-1 and applying an external setpoint voltage to the gate of Q303.

[Kleinstein]

The broken resistor could be the low range shunt - it's bypassed in higher current mode, so one might not notice is.

The fuse might blow with an external reverse power source, or the crowbar engaging while charging a battery.

The spikes look like some kind of residue from some oscillation. Disabling the down programmer might be a good test. Though the onset with only 5 mA somewhat suggests that there is no current flowing from the down programmer anyway. In some cases even a non working down programmer could cause oscillations - tough usually only with large capacitive loads and after load steps. There might be a pot to trim a standing current.

[Jester]

Update 4.... getting closer

The down programmer has a 5A fuse in series (not shown in the schematic) so it was easy to disable it. With the down programmer disconnected the burst (spike) looks a bit different so I suspect that the overvoltage aspect of the spike was triggering the down programmer so with that out of the way I then lifted u309-1 and connected a lab source to the gate of Q303.

The good news is that with a clean signal driving Q303 the output is spike free (with load) and I can adjust the output voltage up and down as desired by tweaking the lab supply so It appears the main output regulator Q308 works.

The voltage metering is perfect, however the current metering although in the ballpark is not quite right (when the setpoint (Vg) is external)

I also monitored U309-1 while I was testing, no spikes.

Load = 1kOhm, Vpgm=3.0V(setpoint via the UI)
=======================================
Vout     Vg     U309A-1     Imeas       Idisp
5.0       7.3     -9.1             5ma          5ma
8.0     10.4     -6.1             8ma        11ma
10      12.7     -3.8           10ma        17ma
12      14.9     -1.7           12ma        24ma
14      17.2     +0.5          14ma        32ma


Load = 47Ohm, Vpgm=3.0V(setpoint via the UI)
=======================================
Vout     Vg     U309A-1     Imeas       Idisp
5.0       8.9                        86ma       106ma
8.0     12.4     -4.8           180ma       215ma
10      14.4     -2.9           221ma       281ma
12      16.7     -0.7           268ma       322ma
14      18.7    +1.1           311ma       390ma

I then reconnected U309-1 to the gate to compare the internal setpoint to what I had to use from the external supply.

Load = 47Ohm
=======================================
Vsp     Vout     Vg      Imeas       Idisp
3.0      3.2       4.3      71ma        71ma
5.0      5.1       6.3    114ma       113ma
8.0      8.1       9.4    180ma       179ma

I'm not sure why the external setpoint value is so much higher?

Another observation, the power supply for the U309A actually shifts up as the output voltage increases for example:
Vout = 3V  V(-) = -10 and V(+) =19
Vout = 5V  V(-) =   -8 and V(+) =20
Vout = 8V  V(-) =   -6 and V(+) =24
I assume to keep the rails high enough to drive Q303, while not exceeding the 35V PS limit.

The op amp power supply rail also has the disturbance on it, perhaps this is the cause of the problem?   Thoughts?



There is nothing detailed in the service manual  for the opamp supply just the note +/-15V gated bias. I need that schematic

Perhaps I will try providing the rails to U309A with a lab supply and see if that works.

This appears to be the gated opamp supply (less the details):
The SecCom (SGND) point tracks the output voltage, at first I thought this might be by design, but now think this in unlikely. Perhaps there is a shorted device between the +output and SecCom. It appears than when loaded, the negative rail approaches the required setpoint voltage at U309A-1 and this is what causes the burst, if the negative rail was lower I'm pretty sure everything would work fine. Need to figure out why SecCom is tracking the output voltage

[Jester]

I ended up probing some of the relevant circuits and then adding the component designators to the block diagram (see below). I'm pretty sure the problem in in the Positive Gross Current Limit circuit. Something is pulling the D317 anode very close to ground and it's not D317 or D321 as the cathode of both are Positive, so that leaves the voltage clamp circuit shown with the red arrow below. Unfortunately I can't find it on the board. I connected an Ohm meter to D317 Anode and probed every diode and op amp around and none of them are connected?

Too bad Agilent was not a bit more liberal with the designators on the block diagram.

If anyone has ever had to work on the current limit circuit for a 6611-6613C, or a 66309D perhaps they can tell me where this circuit is hiding or the designator[/b

[Kleinstein]

One point would be looking for the 330 Hz clock that somehow triggers the spikes. For me possibly candidates might be the MUX part (e.g. broken LM311 and mux253) or the DAC part for the set points. Some of the supplies use a single DAC and than separate S&H for the set voltages - a malfunction there could cause spikes.

The voltages at U328 look suspicious the comparator should be high in this case.

[y2khris]

Jester, any luck with getting it working correctly?

[Jester]

I forgot to come back to this thread. Yes I fixed it.

This supply has a setting for output compensation, and does not like to run in "H" mode with no load capacitance. So using either "L" mode or having some capacitance at the load results in a nice clean output.

[pelule]

Congratulation.
I have registered this thread, if once my 66309D gets defect.