-
Agilent E3631A strange output behavior
Posted by
alex.forencich
on 26 Nov, 2014 01:46
-
I just picked up an Agilent E3631A with some issues that I am trying to sort out. The output seems to be behaving rather strangely - I cannot get the output to turn off. The front end seems to be ignoring the 'turn off' signal and continues to output the 'limit' voltage setting. If I connect an ammeter across the output terminals, I get the current limit, even though the output is supposed to be off. The output voltage also seems to rise slowly, by about a mv every second or so.
free_electron mentioned
here, that these supplies have an odd way of turning off the output. I have been looking at the service manual, and the only things that jump out to me as possibilities are programming either the voltage or the current to zero, or this mysterious 'A' signal that is used by the voltage control side of both of the 25 volt supplies and subtracted from the 6v reference voltage with a discrete differential amplifier. I presume this may be the control signal, but it does not seem to be getting driven by the DAC. I count only 6 discrete levels out of the DAC that correspond to the 3 current and 3 voltage settings, but there does not seem to be a 'step' for this 'A' signal.
Does anyone know how the E3631A controls the output state?
-
#1 Reply
Posted by
idpromnut
on 26 Nov, 2014 17:47
-
Is the display reflecting that fact that you turn the output on/off? Perhaps the issue is in the button/front end as you say.
-
-
The display updates to reflect the selected output status (OFF annunciator and OUTPUT OFF display)
-
-
I presume this may be the control signal, but it does not seem to be getting driven by the DAC. I count only 6 discrete levels out of the DAC that correspond to the 3 current and 3 voltage settings, but there does not seem to be a 'step' for this 'A' signal.
I didn't look at what this 'A' signal does, but it does actually seem to come from the DAC. The DAC (well, the 74HC4051 analog switch that it connects to) has 8 outputs, the 6 you mentioned, the 'A' signal and one NC.
-
-
Yes. Like I said, it seems like the DAC should drive it, but it only seems to 'visit' 6 outputs. I'll post a scope trace when I get a chance. The level of the signal at the mux does not change when the output state is changed. In fact, none of the levels out of the DAC change with the output state. Hence why I need a comparison - either there is something very wrong here, or I am looking at the wrong thing.
-
-
After doing some poking around with the a multimeter, it seems that the A signal is drifting at around 1 mv per sec. It seems the leakage current of the op amp is charging the sample+hold R125/C72 and causing the drift. Next step is to stick a logic analyzer on U13 and see which outputs are being selected. Could be a bad U13 if it is actually hitting that channel. If the DAC is skipping the channel, then it seems like there might be a software issue, as the mux is driven by the CPU through U11.
-
-
change the analog mux and the tl074 opamps. i've seen those fried multiple times...
-
-
Traces from DAC output pin:
First image is power on defaults: all voltages zero, all currents max. Does not change with output enable.
Second image is with 6v output turned up.
Third image is with all of the outputs turned up.
As you can see, no DAC cycles are sitting at zero when none of the current and voltage commands are zero. So it seems like it must be skipping over the A channel. I have no idea what could be causing this.
-
-
change the analog mux and the tl074 opamps. i've seen those fried multiple times...
That's certainly an option if the chips are bad. However, right now it looks like they are working properly. The op amps seem to be working correctly, anyway. If the analog mux is actually commanded to that output, then it might be bad. But it seems like the DAC never generates a level for that output, which is completely baffling.
-
-
i need to look at the schematic. one thing i know for sure : the dac scans a fixed pattern. it does not one time scan 5 and another time scan 6... the pattern is fixed
which channel does not turn off ?
-
-
None of the power supply output channels turn off. I can draw full current from all 3 outputs when the supply says 'OUTPUT OFF'.
-
-
Yeah; the entire DAC output is being skipped. LA trace is attached. The MUX only visits outputs 0 through 5 - it never hits output 6, which corresponds to the 'A' output. That is really, really weird. Seems like this must be a firmware issue, then.
The firmware ROM in my unit is labelled E3631-60032 U14 REV 1.4. Top board is labelled E3631-60002-001113. Bottom board is labelled E3631-6004 C-3627-Y3. Unit is HP branded, serial KR64305092. It also seems to have taken a fall at some point, the case is bent.
-
-
no not a firmware issue. the firmware works or does not work.
you say you can draw current... but does it give voltage ?
-
-
Current is fine. Voltage is more or less right - the 6v output rises slowly, about 1 mv per second. This seems to be caused by the input leakage current of U10 slowly charging C72, which is never 'cleared' by the DAC.
-
-
ok so scanning is right as the sequence is:
+25 voltage setpoint
+ 25 current setpoint
- 25 voltage setpoint
- 25 current setpoint
+6 voltage setpoint
+6 current setpoint.
there is nothing else.
that 'A' signal is the analog ground taken after the current sense stage. the node R66-R67 ( 0.2 ohm 10W) is the 'hard point'.
all measurements must be made in reference to the common ground terminal of the symmetric section.
so your scan system is fine.
-
-
It is not the analog ground. There is a ground symbol marked 'A' and then there is another signal which is also marked A, but it is sourced by mux pin X6 through sample+hold R125/C72 and buffered by U10B. See page 129 of the service manual. The MUX never visits X6, so this output is never updated.
The A signal is not used directly in the 6V supply; it is subtracted from the 6V voltage command with U10C. However, the A signal is also connected to U26D and U21A in the 25 volt power supplies.
Did this get changed at some point on the E3631A? I'm wondering if I have a power supply with a top board of the wrong revision.
-
-
ok , we are talking about differen things apparently. i cant find c72
we have established the DAc is scanning correctly.
the dac goes t the 4051 (U13) and then there are 6 filters :
r26/c15
r25/c14
r24/c13
r27/c10
r22/c11
r23/c12
check u10:
voltage on pin 14 should be same as pin 12
1 should be like 3
check u9
voltage on pin 8 should be like 10
14 like 12
1 like 3
7 like 5
all these measurements need to be done with 'A' as reference
if this is all fine the regulator control works properly.
now, find the two lm393's in the 25 volt section ( the big board with the cpu on it )
U28 and U23 is who we are after.
check pin 3 of U28
check pin 2 of U23A
those two should be the same signal.
toggle the suppl output on off. that signal should change
-
-
you have the wrong service manual !
-
-
ok , we are talking about differen things apparently. i cant find c72
we have established the DAc is scanning correctly.
the dac goes t the 4051 (U13) and then there are 6 filters :
r26/c15
r25/c14
r24/c13
r27/c10
r22/c11
r23/c12
check u10:
voltage on pin 14 should be same as pin 12
1 should be like 3
check u9
voltage on pin 8 should be like 10
14 like 12
1 like 3
7 like 5
all these measurements need to be done with 'A' as reference
if this is all fine the regulator control works properly.
now, find the two lm393's in the 25 volt section ( the big board with the cpu on it )
U28 and U23 is who we are after.
check pin 3 of U28
check pin 2 of U23A
those two should be the same signal.
toggle the suppl output on off. that signal should change
Pin 3 of U28 and pin 2 of U23 are listed as NC in my schematic. Are we looking at the same schematic? I am look at this one:
http://exodus.poly.edu/~kurt/manuals/manuals/HP%20Agilent/HP%20E3631A%20Service.pdf .
-
-
that is for revision 2 of the hardware. i think you have a rev-1 ( based on your serial number )
rev 1 is only paper copy...
February 1996 Edition 2 - Printed in Korea
-
-
Aha! I think we found the problem. I believe I have a rev 1 supply with a rev 2 top board.
On my top board CR27, R121, R102, R104, R105 near U28 and R70, R71 near U23 are unpopulated. Also, the firmware ROM is rev 1.4. And the board silkscreen marking has an F-1 sticker on top of it.
However, on the bottom board, U5 is populated, unlike in the teardown unit linked earlier. It seems on the new revision, the shutdown signal was changed.
-
-
correct.
in the original design U28 and U23 on the top board are LM393D's.
pin 3 of U28 has the shutdown signal for the +25 section
pin 2 of U23 has shutdown for negative section.
your firmware revision cannot handle the new shutdown mechanism using the A signal it uses the SD signal instead.
in the orignal design U28 and U23 would pull down the bases of Q8 and Q2 thus shutting down the regulator power stage.
in the new revision this is gone. . they sht down the regulators by controlling signal 'a' which acts as reference for the opamps U26D and U21A.
bringing that signal above the programmed voltage sends the opamp outputs to -15 volt which has the same effect.
these regulation loops work as follows : look at Q8 ( +25 volt section )
Q8 emitter goes to the base of Q7 ( the pass transistor).
all Q8 can do is make the base more 'positive' ( this turning Q7 off.
the current source made with an lm317 ( U27) pushes a constant current into the base of Q7 attempting to turn it off
Q8 does is wick away that current. thus turning Q7 on.
Q8 can only be turned on by current coming through the diodes CR22 and CR23
Q8 itself is kept turned off by default by R85.
so if the opamps stop steering voltage : Q8 slams shut , and Q7 slams shut. so the output voltage is zero.
that is what they do with the 'A signal. if that is 0 volts. the opamps work as regulators. send 'A' above the setpoint and the opamps become comparators sending their output hard low. the diodes CR22 block that so the entire powerstage slams down.
anyway . here is your problem : whoever messed with your machine screwed it up badly as they mixed board revisons and firmware revisions.
looking at the scan pattern of the DAC the cpu only services 6 channels. which are the original 6 of the original hardware.
so the big question is : do you really have a rev 2 top board using all 8 TL084 ( the original boards had two opamps not used )
this fix was implemented because the 6 volt section of the supply could not correctly shut down in the orignal design. when turned of it would go to -0.5 volts and could actually deliver like 1 amp.
the reason was that SD signal does not go to the 6 volt section. the only way to turn that one of was steer the opamps to ground. but, due tot he way that thing works the regulation loop actually shot through to a slight negative supply coming from the current sink in the block
so in rev 2 they fixed that as follows :
they now use one dac channel to control the turn off. ( that is signal 'a' )
the +6v CV ref is now under control of the extra opamp U10C ( U9A used to be +6V CV)
so if 'A' goes to +5 V U10C works as comparator and sets its output low.
so now they have a true shutdown.
You gotta remember this is agilent.... there are revisions and backdating and you really need ALL manuals. some are paper only and were never released as pdf.
so, to fix your thing we need to find out if you really have the circuitry around U10C or not.
my gut says you don't have it since you have a rev 1.4 firmware.
if you do have the circuit then the bozo that tried fixing the machine installed the wrong firmware and this will never work ...
So, of the opamps do not steer
-
-
Looks like I may need to turn my rev 2 board into a rev 1 board. Could you possibly post scans of the DAC and 25 volt power supply sections of the rev 1 schematic, as well as the shutdown signal between the CPU and J3? It seems like the required footprints for many of the parts are present on the board, just unpopulated.
-
-
it may be easier getting the right firmware ( 1.6 as opposed to 1.4)
i will check my machines if i have a 1.6 although i doubt it.
-
-
The board is definitely a rev 2. U10C is definitely working as a differential amp, and U10B pin 7 slowly drifts negative, causing the 6 volt supply to drift. My guess is that A output is set equal to the 6v voltage command to turn the output off, no? Perhaps fixing this is as simple as adding a CMOS switch to drive U10B pin 5 to U9A pin 1 when the SD pin from the CPU shuts the unit off, along with a resistor to pull U10B pin 5 to AGND to stop the drifting output.
However, if v1.6 will solve this, then that is probably the most ideal solution.