Author Topic: Advantest R6581 8.5 digit DMM mini teardown/repair  (Read 201421 times)

0 Members and 2 Guests are viewing this topic.

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #50 on: January 11, 2018, 03:18:58 pm »
Hi, Mr. Mickle.T ,
I got the meter in hand. It seems look good than before. No A/D check5 error found till now , and still have the inaccuracy issue ,but not so bad. I need to self calibrate the unit. Could you give me soem advice on how to perform the calibration?

Thanks and Best Regards,
szszjdb
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #51 on: January 11, 2018, 04:27:09 pm »
MENU -> CALIBRATION -> INTERNAL -> ALL -> OK
Wait 4 min 20 sec and all is ready.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #52 on: January 11, 2018, 04:56:15 pm »
Hi, Mr. Mickle.T ,

I found the calibration in the menu and performed the internal calibration , but the inaccuracy issues are still there and test data as I sent before.
source , read, range
1,   1.19438   ,1mv
10,   10.20681   ,10mv
100,   100.4422   ,100mv
1000,   1002.125   ,1v
10000,   10018.8844   ,10v

0.001,   0.000991121   ,1ua
0.01,   0.010042577   ,10ua
0.1,   0.10022894   ,100ua
1,   1.002252   ,1ma
10,   10.02333   ,10ma
100,   100.21481   ,100ma

When short the DCV port with the short jumper, the votage display is 0.19806 mv zero offset, very strange. There must have some hardware failure inside the circuit.

Then I get into the diagnostic mode and find the data as below.
X1Z1   -0.0352   v
X1Z2   0.0000359   v
X10Z   1.88976   mv
X100Z   2.08456   mv
7V   7.085432   v
-10vref   -9.8421145   v
-1vref   -982.06829   mv
-0.1vref   -96.29756   mv
temp    30.74   c

10ma   -9.659214   
1ma   -965.6122   
100ua   -96.5745   
10ua   -9.655528   
1000na   -968.3756   
100na   -97.2028   
10na   -8.504   

When to perform A/D check, I press enter but no action taken at all and I press HOME to quit.

Could you kindly give me some advice on how to check in the following.

Best Regards,
szszjdb
« Last Edit: January 11, 2018, 05:37:53 pm by szszjdb »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #53 on: January 11, 2018, 05:48:25 pm »

Further check when open the case, as the picture above.
As mentioned by seller that the meter has been repaired by somebody before.  I can find the U105 is replace by a 1993's LT1220 and the nearby R155 3.3ohm have been soldered.  The U013(hc138) and U014(lm393) also have been soldered. The R103(10K metel foil res) has also been soldered.  The k500 must been replaced as the seller told. The PCB is clean enough.

It is a great challenge for me to fix it. 

Best Regards,
szszjdb
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #54 on: January 11, 2018, 06:21:43 pm »
This really looks like a sick ADC. A broken (highly nonlinear) ADC can also cause the scale being off  - so I would not worry so much about the absolute scale. If you have a suitable equipment, it would help a little to check the linearity with a few more points (e.g. every 0.5 V in the 10 V range) in one range (e.g. 10 V). With some luck this could give a clue what is bad with the ADC / amplifier. There is still a small chance the defect is not in the ADC itself, but the amplifier or mux.

For the time being there is no need to look at the other ranges.
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #55 on: January 11, 2018, 06:30:41 pm »
I think the second step must be External Zero Calibration.

R155 (22 Ohm), D112, D113 and C114 - makes a 13 V positive power supply for U105 opamp. U105 is a output amplifier of Inp. Amp. "precision" channel. It in use while integration time [200 uS...Nx10PLC].
« Last Edit: January 11, 2018, 06:45:00 pm by Mickle T. »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #56 on: January 12, 2018, 12:15:34 am »
Hi,
Yes ,after external zero calibration , the offset error slighty decreased . Test data as below.

before:
short input port zero offset:  0.19806 mv , too large offset error.

source , read, range
1,   1.19438   ,1mv
10,   10.20681   ,10mv
100,   100.4422   ,100mv
1000,   1002.125   ,1v
10000,   10018.8844   ,10v

after:
short input port zero offset:  0.00962 mv , much better, but still large. In my 34401, it's 0.001mv

1,   1.00514,   1mv
10,   10.02054,   10mv
100,   100.14873, 100mv
1000,   1001.74301,1v
10000,   10017.832, 10v

The reading is not so stable in the last 3 digit and I am using the default 10 plc.
And I just performed the external zero calibration, others like DCV, OHM calibration havn't been done. I have performed the internal calibration before, so not doing it this time.

How to do in the next step?

Best Regards,
szszjdb
« Last Edit: January 12, 2018, 06:10:13 am by szszjdb »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #57 on: January 12, 2018, 01:45:27 am »
Hi, Mr.Kleinstein,
Time is too rush to get to work. I will measure more point when get back home. I got R6144, TR6143 and HP34401a in hand and the 34401a is reliable for the further test following.

Thanks a lot!
Best Regards,
szszjdb
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #58 on: January 12, 2018, 05:09:32 pm »
Further external calibration were made with the input 10v for DCV, but can't fix the fullscale error.  It seems the calibration have no effect to the fullscale gain.
below are the 10V linearity test compared to the 34401a for 10v range, open the A ZERO function and plc=10
34401a     6581              error(%)
1.00021   1.002196        0.198558303
1.9997   2.00372          0.201030155
2.99921   3.005257    0.20161976
3.9988   4.006853          0.201385416
4.99942   5.009495     0.201523377
5.99906   6.011147          0.201481565
6.9988   7.012906          0.201548837
7.99856   8.014679         0.201523774
8.99837   9.016499         0.201469822
9.9982   10.018345     0.201486268

below are the 10mv linearity test compared to the 34401a for 10mv range, open the A ZERO function and plc=10
34401a     6581              error(%)
1.0119   0.99621   -1.550548473
2.0115   1.99846   -0.648272434
3.0112   3.00078   -0.346041445
4.0101   4.00188   -0.204982419
5.0113   5.00422   -0.141280706
6.0101   6.00517   -0.082028585
7.0106   7.00771   -0.04122329
8.0107   8.0105   -0.002496661
9.0104   9.01134   0.010432389
10.0108   10.01334   0.025372598

I also check the power supply for U105 , v+=13.6 , V-=-15v , it's ok.

Then I test the U014 (LM393) and find the voltage below:
pin   LM393(v)
1   -17.6
2   2.478
3   0.1
4   -17.6
5   4.95
6   2.478
7   SQUARE wave ,2.4hz, as ADIN waveform
8   4.95

Also test the U105 (LT1220) (v)
1    13.6
2     0
3     0
4    -14.3
5    156MV
6    SQUARE wave ,2.4hz, as ADIN waveform
7     13.7
8     13.7

Both the pin7 of the U014 (LM393)  and the  pin6 of U105 (LT1220) have the same waveform like the ADIN test point. It's a SQUARE wave ,2.4hz, and the amplitude equal to the input DCV.

I also tested the reference pcb output , and it's  read 7.07097v in hp34401a and 7.08520xx in diagnostic mode ,which has small difference.

Is there any thing strange ?  What's the next step?

Best Regards,
szszjdb
« Last Edit: January 12, 2018, 07:01:01 pm by szszjdb »
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #59 on: January 12, 2018, 07:41:37 pm »
These R6144, TR6143 and HP34401a are too bad setup for 6581 testing :(



1. Before any comparing all of the devices must be preheated. 6581 not less than 2 hours!
2. HP34401a readings must be aquired via RS232/GPIB interface terminal with as much digits, as possible.
3. Shorts the 6581 inputs terminals. Wait 10 minutes, run External Front Zero Cal.
4. Connect 10V reference source both to 6581 and 34401A terminals. 34401A must be in >10GOhms mode 100 PLC. Wait 10 minutes, run External DCV Cal and input ref. value from 34401A.
5. Run Internal Cal (All ).
Now you can check the 10V range with reference to 34401A. Please note that 6581 have much better linearity perfomance in negative polarity (and all DCV/DCI/OHms selfcal are carried out with negative artifact sources).

Quote
I also check the power supply for U105 , v+=13.6 , V-=-15v , it's ok.
It isn't Ok, but not critical.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #60 on: January 13, 2018, 12:19:04 am »
Hi, Mr. Mickle.T ,
Thanks a lot!
Yes I know the  R6144, TR6143 and HP34401a are not suit for precision calibration of 6581, I just expect the short term stability match the comparing requirement. After fixing, I will send it to external lab for the final calibration.
I can't understand the data BEST FIT column in your picture, so as to the INL column. The error is reach 20mv when 10V input and 2mv when 1v input  comparing the 2 meter. From my positon , the 34401 is more reliable because I got 2 pcs 34401 and the error between them is at 10uv range.

I will following your suggestion to redo the calibration tonight.

For the moment , I feel very strange that the external DCV 10V calibration take no effect to the measured result of 6581. If in 34401 case, the reading is corrected right now.

Please look at the very first data when I get the meter:
6144      6581
1000,    1002.125   ,1v
10000,   10018.8844   ,10v
The update data after internal and external calibration:
34401a     6581
1.00021   1.002196
9.9982   10.018345
It's very clear that the reading of 6581 nearly no change before and after external calibration and the error is still in the 20mv range for 10V input.

What I do in the external calibration are as below:
1. power up for 2 hours both for r6144,34401 and 6581
2. output 10V from 6144, apply it to 34401(10Mohm default) and 6581 the same time
3. wait for 1 min and get into 6581's external calibration menu and choice the DCV
4. enter the reading from 34401 to menu and comfirm to calibration, and after few seconds ,done
5. the 6581 reading is still the same as before
Are there any step wrong?
I just connected 10V source to HI and LO port and left other port floating include the GRARD ,HI_sense ,LO_sense and DCI port. When calibrating the external zero point, I just short the HI,LO,HI_sense and LO_sense, left GRARD and DCI port floating.

Unlike the fullscale external calibration, the zero calibration take effect right now and reading form 0.19806 mv reducing to 0.002 mv range.

I was wondering if the external 10v calibration signal haven't been apply to  signal chain or beyond the upper and lower limits for 10V calibration and the calibration has actually been ignored . Am I right?  If do , it can be the reason why the previous guy can not fix the reading error issue.

Best Regards,
szszjdb
« Last Edit: January 13, 2018, 06:37:24 am by szszjdb »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #61 on: January 13, 2018, 03:36:39 am »
For further information, the reason I test the  U014 (LM393) and the U105 (LT1220) is because the 2 parts has been soldered or replaced by the former serviceman. If look at U014 (LM393 mark as 393) the V- reach -17.6V and V+ reach +5V.  Is there an -18v power rail in the meter?

Best Regards,
szszjdb
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #62 on: January 13, 2018, 07:21:14 am »
U014 comparators are part of U001 FETs gate drivers. U001 MUX divided in to the two sections: a first includes two Ohms, three DCI and one LM35 inputs, a second - DCV, +7.2V, LO, RMS, HV_DIV and ACAL -0.1/1/10V. U014-1 controls the first section, U014-7 the second.
-18V on the comparators output disables the corresponding section.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #63 on: January 13, 2018, 08:32:22 am »
Hi, Mr. Mickle.T ,
Many Thanks !
That is to say the U014 and  U105 is working fine.
So how about my last post about the external DCV calibration?

I was wondering if the external 10v calibration signal haven't been apply to  signal chain or beyond the upper and lower limits for 10V calibration and the calibration has actually been ignored . Am I right?  If do , it can be the reason why the previous guy can not fix the reading error issue.

How can I check for them?

Best Regards,
szszjdb
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #64 on: January 13, 2018, 08:54:07 am »
The test for linearity does not look that bad. Quite linear (the residual difference could be due to noise / not wel stabilized temperature). However there is a significant offset. This might point to a problem with the MUX part, so that for some reason the internal zero adjustment does not work right. If the internal zero adjustment (that is reading an internal zero with the ADC) does not work correct, there would be quite bit of drift (not just the normal drift, but likely extra drift from the not working zero) and thus the external zero cal would not be stable over time.

If the internal zero is off this might also cause an error for the DCV calibration, though ideally it should not have so much effect. But I would not yet worry so much about the scale. If the zero does not work correct, one can not expect the DCV cal to work correct either. If taking out the offset the scale is not that bad.

It might be worth to check if there is significant offset drift when measuring a short with AZ active and maybe for comparison without AZ. I would expect quite some drift due to the defect. The non AZ mode might not be effected.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #65 on: January 13, 2018, 09:38:51 am »
Hi, Mr.Kleinstein,
Thanks a lot!
Yes there have both zero point and fullscale point error in my 6581.

Before external calibration, the short input port zero offset:  0.19806 mv , too large offset error.
After external calibration ,short input port zero offset:  0.00962 mv(yestorday 0.002xxmv) , much better, but still large. In my 34401, it's 0.001mv

The main issues is the  fullscale point error, which can not be cerrect by external DCV calibration. Please refer to the data below:
below are the 10V linearity test compared to the 34401a for 10v range, open the A ZERO function and plc=10
34401a     6581              error(%)
1.00021   1.002196        0.198558303
1.9997   2.00372          0.201030155
2.99921   3.005257    0.20161976
3.9988   4.006853          0.201385416
4.99942   5.009495     0.201523377
5.99906   6.011147          0.201481565
6.9988   7.012906          0.201548837
7.99856   8.014679         0.201523774
8.99837   9.016499         0.201469822
9.9982   10.018345     0.201486268
When 1v, the error is 2mv, and when 10v, it is 20mv,  10 time the 1v error. If it is just the offset error, it should be 2mv both in 1V and 10v input. Both meter are on fixed 10v range. That should be the A/D 's gain error. The key is the error can not be corrected by external DCV calibration.

I was wondering if the external 10v calibration signal haven't been apply to  signal chain or beyond the upper and lower limits for 10V calibration and the calibration has actually been ignored . Am I right?

Best Regards,
szszjdb
« Last Edit: January 13, 2018, 09:56:03 am by szszjdb »
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #66 on: January 13, 2018, 11:21:27 am »
Ext DCV Calibration uses only three MUX inputs:
U001-10 LO (is routed to COM starpoint via R021 8k2 and connected to LO terminal via K004 relay);
U001-8 +7.2V from Master Reference output;
U001-7 DCV (routed to HI terminal via R008-011 2k2 protection resistors, K008, K006 and K001-front-rear relays). I'd like to note that R008-011 replacement surely means a huge voltage overload in past.

I can't understand why Ext DCV calibration has actually been ignored without a sign of error on the display  :(

P.S. It makes sense to take hi-res photos of the main PCB.
« Last Edit: January 13, 2018, 11:24:52 am by Mickle T. »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #67 on: January 13, 2018, 11:52:53 am »
Thank you very much!
The site limit the size of the attached to 2M, let me try.

How about the picture attached.
« Last Edit: January 13, 2018, 12:02:47 pm by szszjdb »
 
The following users thanked this post: Mickle T.

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #68 on: January 13, 2018, 12:41:16 pm »
If the DCV calibration does not change anything (not even a fraction of a ppm), this would be more like a software / digital problem. Ideally the readout of the cal constants should tell about that. To force a visible change one might even try an cal to an intentionally slightly different (off) value. If could be a kind of extra check that it does not allow a DCV cal if the Zero offset is bad or maybe just to soon after turned on. However I normally would have expected an error message for this case.

A residual error of 9 µV is still way to high for the offset.

How is the zero reading in the different ranges, including the 100 V and 1000 V and maybe a current range ?

One more possible error source would be a relay (or similar) at the input - there needs to be some way to interrupt the direct path from the input when the higher voltage ranges with input divider are used. It should be possible to measure continuity here, just in case of bad contacts. If it is the input relay the 100 V and current ranges should be better.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #69 on: January 13, 2018, 01:16:56 pm »
Hi, Mickle T and Kleinstein,
I had found something strange with the oscilloscope connect at the ADIN point of PCB ,I can view the  10Vp-p square wave and the amplitude euqal to to the input DCV. when get into the external calibration menu and select DCV and press YES to perform the calibration, the  amplitude of the square wave change to 7v, that must be of overrange of the calibration limit.
What might cause it ?

Best Regards,
szszjdb
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #70 on: January 13, 2018, 01:42:58 pm »
7V p-p square wave is a second part of the calibration of Master Reference. But 6V p-p ringing in the LTZ1000 output voltage is very strange :o
« Last Edit: January 13, 2018, 07:43:51 pm by Mickle T. »
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #71 on: January 13, 2018, 01:52:20 pm »
Hi, Mickle T and Kleinstein,
I find out what wrong and the external DCV calibration working now. The front port of my 6581 is damaged and so I use the rear port. But the rear port does not support to pass the input 10v to ADIN, so the calibration is stopped because no signal found. I reconnect the front port and calibration done successfully.

Now the last issue is the zero point .
Before external calibration the day before,the  zero offset is  0.19806 mv , too large offset error.
After external zero calibration ,the  zero offset is  0.00962 mv(yestorday 0.002xxmv) , much better, but still large. In my 34401, it's 1uv.
Must I connect the guard port together with the HI,LO,HI SNESE, LO SENSE port?

What the zero offset should be and could you told me the number in your 6581?
How to find out the issue point? Would you please give me some advice?

Many thanks to all of you!
Best Regards,
szszjdb
« Last Edit: January 13, 2018, 02:07:44 pm by szszjdb »
 

Offline Mickle T.

  • Frequent Contributor
  • **
  • Posts: 463
  • Country: ru
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #72 on: January 13, 2018, 08:14:55 pm »
Must I connect the guard port together with the HI,LO,HI SNESE, LO SENSE port?
What the zero offset should be and could you told me the number in your 6581?
Guard must be connected. Have you seen this post? https://www.eevblog.com/forum/repair/advantest-r6581-8-5-digit-dmm-mini-teardownrepair/msg1397797/#msg1397797
My R6581T have a max 300 nV offset (measured on 100 mV range after 2 hours of preheating).
A possible sources of DCV offset are:
1) an excessive ADC noise / shot-term instability,
2) large leakage current or damage of FETs in MUX U001, ultra low leakage protection diodes D003-D004, short pulses protection capacitor C008, ultralow input bias current opamp U009 (AD549) in the Overvoltage Protection Unit,
3) non-Ohmic contact or large thermal EMF in the K001,006,008 nonlatching relays.
 

Offline Twoflower

  • Frequent Contributor
  • **
  • Posts: 735
  • Country: de
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #73 on: January 13, 2018, 08:36:24 pm »
Something I noticed in your picture. Can you check the solder joint I marked? It looks not well soldered.
 

Offline szszjdb

  • Regular Contributor
  • *
  • Posts: 208
  • Country: cn
Re: Advantest R6581 8.5 digit DMM mini teardown/repair
« Reply #74 on: January 13, 2018, 08:45:33 pm »
Hi, Twoflower,
Thank you for reply.
The point you mentioned is maked. But I have not test the function of the relay. I will check it.

Best Regards,
szszjdb
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf