Author Topic: Need help with understanding part of the circuit in Agilent PSU to repair it  (Read 2413 times)

0 Members and 1 Guest are viewing this topic.

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Hi, recently got broken Agilent E3647A power supply, voltage readings and setting weren't working, turns out a lot of precision resistors(green ones) were open, double the value, or 5%+ off, replaced 10 of them and it works fine(needs calibration tho).
But then I discovered that on CH1, OVP function trips too early, when output is set over 36V where OVP is set to default 66V, on CH2 it works fine.
Started probing out, and found that resistors are fine, op-amps are fine but output on U31(analog multiplexer) for OVP function for CH1 shows about half the voltage.
To be precise, when OVP is set to default 66V on both channels, then U31: pin4=113mV(working CH2) pin15=76mV(faulty CH1).
So I supect this multiplexer to be faulty, but I'm not sure as I don't really understand how they are using one DAC to control multiple functions, are they switching dac+multiplexer really quickly?
Also I wouldn't suspect the DAC to be faulty because then all voltage/amperage settings wouldn't work properly as well, right?
Is there a way to somehow measure this multiplexer?

Part of the schematic in attachments, manual and all schemtics here: http://ridl.cfd.rit.edu/products/manuals/Agilent/power%20supplies/E3646AE2.pdf
 

Offline orbanp

  • Regular Contributor
  • *
  • Posts: 160
Hi Arek,

An analogue mux/demux is like a rotary switch.
An input is connected to one of the outputs.
So the voltage at the output must be the same as on the input.
Look up the chip's datasheet!

Inputs and outputs are interchangeable, the switch is bidirectional, like a real mechanical contact.

The common input/output is  "Z", pin 3.
The selectable inputs/outputs, the other end of the switches, are "Y0" to "Y7".
The switch selection is done through "S0" to "S2", a binary number, a logic signal.
The enable pin "E" is active low, it has to be at zero for one of the switches to be closed.

For the mux/demux correctly operating you need to select an input by the "SO" - "S2", the enable "E" must be low, and in that case the voltage on the common input/output must be the same as on the selected input/output.

Hope this is clear!

Peter
« Last Edit: April 21, 2018, 05:02:53 pm by orbanp »
 
The following users thanked this post: Arek_R

Offline LateLesley

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: scotland
I've just had a wee nosey at the schematic, and datasheets, and to be honest, I don't think you have a faulty component, if the op-amps and passives tested ok. I did note in the manual there is a calibration procedure, which includes setting the OVP. Have you run through the calibration procedure? (page 193 of the manual)

I would try that first, before condemning parts. If it fails, then i'd be trying replacing U25, then U31.

The circuit is basically using U31, to switch the DAC output between the different op amps. U29 is being used it store data in a shift register, and then output the data to the U31 address lines to address each op amp, and also output the DATA to the DAC input. This probably does run a lot of data constantly from a microprocessor. In fact it all does connect back to the ASIC (U16), which will be instructed from the microcontriller (U17).
 
The following users thanked this post: Arek_R

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Have you checked the actual overvoltage sensors, either U104 and U138? Specifically, I'd suggest working backwards from the datasheet for the MC3423D, starting with the internal 2.6V reference and checking the comparator at the sense 2 pin. I had a E3646A with overvoltage sensing problems that ended up being a broken comparator in the MC3423D.

I'd also echo @LateLesley's suggestion about running the internal calibration routines first. For voltage calibration, you don't need anything fancy, just a regular multimeter. Current calibration requires a specific resistor value, ideally with high precision, but you can skip that for now.
« Last Edit: April 21, 2018, 08:43:23 pm by ddcc »
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Ok, fine, calibration.
Power up + holding viev(calibrate) button, long beep, displays "cal mode"
Next step, press view button again, and it displays "secured" instead of entering first parameter to be calibrated.
Is it somehow locked or what?
« Last Edit: April 21, 2018, 09:17:21 pm by Arek_R »
 

Offline Samogon

  • Frequent Contributor
  • **
  • Posts: 456
  • Country: us
Sure it is secured. Hit secure Use 3647 as a code rotary chooses digit and righ left chooses position at the end hit secure button again
 
The following users thanked this post: Arek_R

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Thanks man!

Ok, tried calibration, first LO value, done, second MI value, displays CH1 TRIPPED(or something like that) and can't do anything...
And when I'm in normal operation mode, it trips at about 1,4V(yes it changed out of nowhere, last time it was about 36V >.<), but it doesn't display "ch1 tripped" it just loads PSU(MC3423D activated SCR and creates a short on output) and then maybe later it displays message about CH1 being tripped.
This makes me thing that maybe ddcc is right and MC3423D is faulty?
ddcc how to check this internal reference of MC3423D?

EDIT: Managet to calibrate everything apart from voltage for CH1 and OVP for CH1(fails).
« Last Edit: April 22, 2018, 12:42:15 am by Arek_R »
 

Offline Samogon

  • Frequent Contributor
  • **
  • Posts: 456
  • Country: us
Each channel has its own crowbar IC and TIN, check them i will post later part of the scheme
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Each channel has its own crowbar IC and TIN, check them i will post later part of the scheme
Maybe some details please?
These are on maind board so I will have to take everything apart...

Another thing I discovered:
CH1 OVP disabled, turns output to 60V, everything fine, 5 minutes later it turns 0.8V 500mA CC, no OVP TRIP message...
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Oops, not sure why I didn't get any post notifications.

My understanding of the OVP mechanism is that one of the microcontrollers on the control board uses a DAC to generate the reference voltages OVP_REF1 and OVP_REF2.2, taking into account the OVP calibration, user-selectable OVP threshold, etc. This signal passes through a connector onto the bottom board, and combines with the output from a sense amplifier to bias the input into the OVP supervisor MC3423D.

You'll need to double check the service manual for the E3647A, I don't remember which MC3423D (either U104 or U138) corresponds to channel 1 or 2. Pull up the datasheet for the MC3423D, and I believe the block diagram shows that the inputs to the two internal comparators are exposed on sense 1 and 2. You can't directly measure the 2.6V internal reference, but if the input to either comparator exceeds 2.6V and it hasn't tripped (or the reverse), then you know there's a problem with the MC3423D. I have a E3646A repair write-up that discusses this part: https://www.dcddcc.com/blog/2017-12-02-agilent-e3646a-power-supply-repair.html .
« Last Edit: April 26, 2018, 12:27:50 am by ddcc »
 
The following users thanked this post: Arek_R

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Ok, made some measurements on MC3423.
Negative probe of my DMM connected to positive banana terminal(ground).
I started with functional channel 2.
Output of PSU set to 0V = pin #2 of MC3423 reads around 0.5V
Output of PSU at maximim 62V = pin#2 of MC3423 reads around 2.4V
OVP disabled = pin #2 of MC3423 reads around -0.6V
For me it seems alrighty, so U104 is fine.

Now faulty CH1.
This one was tricky, voltages jumping all around, sometimes stable or not, so don't use values below for any calculations.
Output of PSU set to 0V = pin #2 of MC3423 reads around 4.9V, sometimes it's tripped at this moment sometimes it's not
Output of PSU at 2V = pin#2 of MC3423 reads around 5.2V, PSU running at CC, no trip message.
OVP disabled = pin #2 of MC3423 reads around 2.4V, and when I increase PSUs output voltage to 8V this pin #2 reaches 2.6V and triggers OVP(with message), also, before I was able to reach 40V instead of 8V.

So extremly unstable.

And so ddcc you say that it's because of faulty MC3423 and not circuits delivering this sense signal?

PS I was able to catch CH2 getting tripped two times as well, but it happens very rarely...
Nothing worse than faulty gear with behaviour that changes randomly :/
« Last Edit: April 25, 2018, 08:50:24 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
I'm suggesting that you start troubleshooting with the MC3423, and work backwards from there to figure out what's broken.

From the datasheet, the MC3423 is supposed to trip if the input on sense 1 (pin #2) exceeds 2.6V. You should verify this on channel 2 by setting the OVP, incrementing the output voltage, and ensuring that it does trip at 2.6V. What was the value of the OVP threshold with the 0V and 62V output voltage? Did it trip when it was supposed to? Likewise, with the OVP disabled, did you check that it does not trip with 0V to 62V output voltage?

Channel 1 is trickier because the input into the MC3423 seems off. When the input was 5.2V, and it did not trip, was OVP enabled? It is strange that it trips even with the OVP disabled, but the MC3423 doesn't seem to have an enable pin, so I don't know if the software will suppress spurious trips. If the MC3423 is working fine, then you'll want to work backwards to check the op amp U136 and the resistor R370, R371 for the sense amplifier.
« Last Edit: April 26, 2018, 12:32:35 am by ddcc »
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
I verified it, see my previous post, basically on ch1 it triggers when it feels like.
OVP thresholds were ofc set to default 66V.
OVP was disabled only on third measurement.
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
It sounds like it sometimes doesn't trip when it should, but does it ever trip when it shouldn't? Note that the MC3423 supports a programmable minimum overvoltage duration before tripping, so this may explain the inconsistency. You could try replacing the MC3423 if you want, but the part is EOL, so it's a bit hard to get ahold of. I ended up ordering what appeared to be new old stock from a Chinese parts broker, Utsource.

But I'd suggest looking farther back through the circuit at the op amp and sense resistors, because the input voltages don't seem right. I believe the power supply service manual shows the expected voltage range from the sense amplifier before it is biased with the reference voltage, so I'd start there. I'd also expect the reference voltages from the microcontroller for both channels to be fairly similar, aside from channel-specific calibration.
 
The following users thanked this post: Arek_R

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Ok so as you said I went back to earlier part of the circuit with DAC and op-amps.
Something changed since first post, now both values from DAC for OVP1/OVP2 are same so it'll be easier to check voltages, and here's what I measured:
Faulty OVP1/U25:
-input pin #10 = 0.0775V
-output pin #8 = -4 to -5.6(floating back and forth)

Working OVP2/U32:
-input pin #10 = 0.0774V
-output pin #8 = 5.951V

Measurements done with active output, active OVP at 66V and everything set to 0V/0A

I had replaced R332 before, it was showing 10k1
R333 was measuring fine, but now I removed it and measured out of circuit, completly fine, but because these precision resistors fail I just placed new one, of course nothing changed apart from decalibrated DAC(0.4V on outputs when set to 0V).

So one channel of quad op-amp U25(TL074CD) is faulty?

PS part of the schematic with these components is in the first post.
« Last Edit: April 27, 2018, 09:29:03 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
I think you went back two steps, instead of just one. Did you check the sense amplifiers and resistors that feed into the MC3423 first? Specifically R370, R371, U136, etc? For pins 1, 2, 3 of U136, lookup the formula for a scaling summing amplifier and check that the output voltage matches. Note that this amplifier is biased to ~6.95V by R401/R402, so you probably need to add this to the calculated output voltage from the formula. For pins 8, 9, 10 of U136, you can probably skip the formula for the differential op amp, and just check that the output is in the range printed on the schematic.

The output of U25 does seem off, but I'd suggest checking the scaling summing amplifier first.
« Last Edit: April 27, 2018, 10:05:17 pm by ddcc »
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Range printed on the schematic?
Not sure what you're talking about.
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Yeah, look at the net connected to pin 8 of U136. There's a "0 ~ -5.625V" label on it.
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Ok I will make these measurements.

But why you tell me to check R370, I can't see it on my schematic, but I found this physically marking on processing daughter board, it's unpopulated...
« Last Edit: April 27, 2018, 10:43:19 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
That's weird, the schematic shows R370 and R371 populated for channel 1, and R300 and R301 populated for channel 2. Are they all missing? These are on the right side of pg. 209 and 210 in the PDF. They are part of the output voltage feedback path into the MC3423s for each channel that you should check. Note that these are on the bottom board, not the control board.
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Couldn't find R371 on schematic or in PSU.
Couldn't find R370 as well but I found it in my PSU on smaller PCB, where we're supposed to talk about main power PCB where U136 and MC3423 is located...

I measured pin #8 on U136 and it reads masically 0V(few mV) where for reference on CH2 I read around 2.5V
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Did you see my previous post with the page numbers?

Is this with 0V output voltage on both channels? It's strange that they differ so much; you might want to also check that both channels remain in the correct voltage range across the full output range.

What about the summing amplifier?
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Yes I saw your previous post about page numbers, I can't find these resistors, where R370 is not used in this device nor presented on the schematic at all...
Yes outputs were at 0V.
Will measure this in a moment.

EDIT:
Ok measured U136(outputs set to 0V):
Pin# | voltage
1 5.8V(floating a bit)
2 2.025
3 2.030V

8 0V
9 0V
10 0V
« Last Edit: April 27, 2018, 11:18:47 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Yes I saw your previous post about page numbers, I can't find these resistors, where R370 is not used in this device nor presented on the schematic at all...

Did you locate the relevant portion of the schematic? I've highlighted the resistors in red, and the other two opamps in blue.

 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Of course I located U136, I just gave you the measurements, are these good?
Oh I see, on the very right...
But still R370 isn't populated.
Also I don't understand what it's doing on schematic for main board when marking R370 is on daughter board(see attachments)
« Last Edit: April 27, 2018, 11:41:53 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Also I don't understand what it's doing on schematic for main board when marking R370 is on daughter board(see attachments)

No, you're looking at the wrong board. Although the component references can overlap between the main board and control board, there's a different one on the main board. You should refer to the component locator on PDF pg. 208, which shows them on the far right. It seems that the two pairs of resistors are on the bottom of the main board, probably close to the output terminals.
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Oh right, sorry :D
Ok so I made more measurements.
To avoid possiblity of me screwing up calculations for these summing ampifiers or misunderstanding something, I just referenced to working CH2, which shows 0.585V at pin #14 of U100, and also it reacts accordingly to changing OVP preset, reaches about 2.56V when OVP set to 1V, seems right.
Where faulty CH1 shows -7.41V at pin #14 of U136, and it doesn't change when I change OVP preset.
I ignored sense amplifiers because it is used when you use remote sense, but anyway I checked at least R370 and R371 and of course they're fine.

So now I'm waiting for your comment on these readings, meanwhile I will also check resistors around the suspect.

EDIT: Checked resitors, haven't found anything...
« Last Edit: April 30, 2018, 10:30:53 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Are you sure you're measuring the right pin? Pin 14 of U100 (channel 2) corresponds to pin 1 of U136 (channel 1), not pin 14.

The remote sense functionality is always being used, because for local operation you put the jumper wires on the back of the unit from S+ to V+ and S- to V-. That's how it ensures that the local output voltage is correct.

At the moment, it sounds like at least one of your opamps are bad, although it's not quite clear which since they're all connected in circuit. To save some time, you could probably just order a couple of TL074CDs, and try replacing U25 and U136, one at a time.

Edit: Have you checked the +/- 15V rails for U136 and U25? They're unlikely to be bad since the other channel works, but might as well double check.
« Last Edit: April 30, 2018, 11:50:45 pm by ddcc »
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Yeah, dot marks first pin, this is 14pin package so on opposite side.

I don't use any jumper wires, is that bad? xD

Ordered 3x of them right now,, mf by ti.

Only on U25 and it's ok.
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Hmm, you're right, there's a local bypass somewhere so the jumpers aren't necessary when remote sense is disabled. I don't know where that is on the schematic though, so your problem may not be where I thought it was. You could check the node labeled 3V connected to CR145, but I don't fully understand this portion of the schematic. It seems that the optoisolated CV_REF1 signal is fed into pin 5 of U136, and the output on pin 7 ends up being sampled by the microcontroller at CV1, but I don't see a connection to OUT1 (unless it's through the V1 net?). It's also a bit confusing because channels 1 and 2 don't quite mirror one another; there's an optoisolator on OVP_REF2.1 to OVP_REF2.2, but not for OVP_REF1.

I'm saying that you seem to be measuring the wrong pin for channel 1. On channel, measuring pin 14 of U100 is correct, because it's the output of the summing amplifier. But for channel 1, the output of the summing amplifier is on pin 1 of U136. So the -7.41V that you see on pin 14 of U136 is normal, because it's a voltage follower connected to a fixed voltage divider. So you should repeat your measurements on pin 1 of U136, and see if the voltage changes when you modify the OVP and the output voltage.

 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
DAC/ADC circuitry is on potential of CH1, so it's connected directly to it, where signals/references for CH2 are fed into analog optoisolators.

EDIT: You can see this vid, dude explains a little bit what's going on, also I had to replace same resistor as him(+ 9 other resistors)
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
AAAnd it's fixed.
Sorry for not replying but I had a lot on my head and then holidays.
Today I decided to replace suspects, started with U25(this quad opamp) and it was it, PSU is now fully functional  ;D
Now it only needs calibration using proper DMM, calibrated with my normie UT139 for now.
https://drive.google.com/file/d/1ttRAYV4fs23mXUCKImqGOAUTbtdvzzd3/preview
« Last Edit: May 09, 2018, 05:48:07 pm by Arek_R »
 

Offline ddcc

  • Regular Contributor
  • *
  • Posts: 102
  • Country: us
Nice, congratulations! You may want to consider looking into the DMMCheck Plus, a new production run was restarted recently.
 

Offline Arek_R

  • Regular Contributor
  • *
  • Posts: 64
  • Country: gb
  • Polish guy living in the UK
Looks nice!
I actually have broken 34401A on way to me, will fix it, send it for proper calibration and then calibrate this PSU(and other stuff).
Also 34401A is actually recommended in manual of this PSU :D
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf