Advantest R6581 8.5 digit DMM mini teardown/repair

[eurofox]

It looks like one byte was changed (apart from stepping revision numbers).

This byte is a part of 32-bit floating point number, used in calibration math procedures.

Is this a "real" firmware upgrade with eventually bug fix? 

Strange with only one byte change 

[eurofox]

Replace binding post .

I will post here how to replace the binding posts on this instrument.

[eurofox]

What do you need:

Aluminium plate 1,5mm
6 spacers M3 and 8 mm long
3 red and 3 black microvolt bindings posts

On the last picture due to optical effect it look like the bindings posts are not centered but they are drilled with a precision of 0.01mm

[Kleinstein]

It looks like one byte was changed (apart from stepping revision numbers).

This byte is a part of 32-bit floating point number, used in calibration math procedures.

Just having one constant changed is odd. I would not expect any really serious floating point constants in the ROM that would effect the final reading. The main constants I would expect would be something like limits for the self tests and maybe a set of default cal values to get started, when the there is no valid calibration available.  So I would guess the self test to get more or less stringent at one point - kind of a small bug fix.

[TiN]

Just did another performance verification test with A03 firmware on my 6581T.
No difference in results. 6581T was calibrated to this same source 1 week ago, so all specs are relative 24 hour, not absolute.

DCV test



DCL 10V range test



4W-resistance test, everything over 1Kohm is too low, 19K-19Meg is -270ppm off.



DCI test. Good except 10uA and lower ranges.

[eurofox]

Report of R6581T that pass all tests:

[Kleinstein]

I am not that surprised to see the -1000 V test fail: with the high impedance of the input divider the input current spike from AZ switching can easily cause trouble.

For the specs of the meter, I would expect slightly different values: there usually are parts relative to range and relative to measured value. It looks like the relative to range part is not included.

For the Ohms ranges, it is odd to see the large error for the step going from 10 K range to 100 K range. The earlier tests with just a resistor had trouble between higher ranges. The higher ranges still look like they are consistent. It could be a problem with the ACAL step from 10 K to 100 K. If it is an ACAL problem, it might be worth checking how repeatable the ACAL results are for an 10-50 K (in 100 K range) or similar resistor are: so measure the same resistor with a few (e.g. 5 ) ACAL calls in between. This would at least show random contributions and maybe drift type errors (e.g. ACAL getting worse at higher temperature).  The problem with the Ohms also corresponds to the small current ranges to be off.

For the INL test it is rather difficult to look at the data in a table. A graph with INL deviation would be more practical. However the data still look good and well within the source specs numbers given in the table (not sure how much of this is scale factor and thus not relevant to INL though - so for the INL test the source might be better than shown).

[TiN]

Actually, +1000V also failed, I just spotted error in my math, its false pass on table. .
I did some test earlier multiple CALs and value of 100Kohm was about same off every time.

Also 10Kohm drifted a lot last week from last ACAL, over 14ppm.



Purple is 6581T over 10K (5450A).

I have 3458 running same test suite over same source, will see data in few hours. That will allow direct unfair comparison between the two DMMs.

[Mickle T.]

Report of R6581T that pass all tests:

God, what a mess!
-100V (Short) Tolerance- is -0.300063V It's a joke?
All of DCV (and other) tolerances span are 10 times bigger, than DMM's spec. DMM fails almost all

[eurofox]

Report of R6581T that pass all tests:

God, what a mess!
-100V (Short) Tolerance- is -0.300063V It's a joke?
All of DCV (and other) tolerances span are 10 times bigger, than DMM's spec. DMM fails almost all

Don't get your instrument calibrated or checked by those guy's.

My bad, I did not check the data, just found it and supposed coming from a calibration company that it is correct and reliable. 

[Kleinstein]

That test mainly showed the calibrator is by far not up to the task. Even if the test results are outside the meters specs, it is well possible that the calibrator that cases the error.

The large drift of the resistance readings of the 6581 could be due to the meter not been used for quite some time before. This can cause quite some drift for the initial phase. The ohms circuit is not build for low drift and may rely on regular ACAL cals.

I think TiNs cal check table has quite some odd points with the uncertainties. Somehow values relative to full scale and measured value are mixed / not combined.

Though not part of the official specs and test procedure, it might be a good idea to check the input current at different levels of input voltage. I kind of understand that they don't have that as part of the specs and cal procedure - this is expected to be one of the weak points of the meter. Still I would consider checking input bias an import point, as quite some faults (e.g. ESD damage, leakage at the LM339) and contamination might cause excess input current.

[TiN]

That test mainly showed the calibrator is by far not up to the task. Even if the test results are outside the meters specs, it is well possible that the calibrator that cases the error.

That refer to eurofox data, I'd believe .

The large drift of the resistance readings of the 6581 could be due to the meter not been used for quite some time before. This can cause quite some drift for the initial phase. The ohms circuit is not build for low drift and may rely on regular ACAL cals.

I agree on drift point, as 6581 indeed was collecting dust unpowered for months before. But I'd still expect ACAL to perform in 24hr-spec accuracy transfers for all functions/ranges.
Obviously my 6581T have ACAL broken/faulty as such transfers are impossible here. That was my only conclusion from data.

Here is promised data from 3458A on same source

I think TiNs cal check table has quite some odd points with the uncertainties. Somehow values relative to full scale and measured value are mixed / not combined.

I used worst-case sum value for all 24 hour specs, not RSS.
In case of DCV 10V range 10V value math is :
10.0 VDC (10.00 Range) = Source 10.0000000 V vs DUT result 10.000002 V,
MFC spec = 1.47 ppm + DUT 3458A 24h 0.55ppm = 2.02ppm.
Measured deviation is 0.219 ppm or 10.84 % of 2.02 ppm.
Since calibrator is adjusted to same meter (or other 3458A I have), only relative specs are used.

Here is absolute calibration report from Jan 7, 2017.
Generated from same meter using actual in-cal 732B/SR104/SRL1 <1ppm absolute.

Though not part of the official specs and test procedure, it might be a good idea to check the input current at different levels of input voltage...

Noted, will do, but it's low in priority chain.

[eurofox]

Discharge a 5.6NF capacitor starting with 10V, 1 PLC and no AZERO

10V to 9V in 2353 sec
10V to 8V in 3067 sec
10V to 7V in 5325 sec


I would like to compare to a 3458A.

[Kleinstein]

The discharge is really very slow:  around 1600 seconds for a 1 V drop and thus around 3.5 pA.  This is well in the range of random leakage currents. This is an exceptional good value - nearly to the point of having some doubt in the capacitor value. Typical DMM specs are more like < 50 pA. The current does usually depend on the voltage and could be different with a negative voltage.

The more tricky range is the negative sign, especially negative voltage with AZ mode, as there likely is some extra current pulse from AZ switching.

[eurofox]

The discharge is really very slow:  around 1600 seconds for a 1 V drop and thus around 3.5 pA.  This is well in the range of random leakage currents. This is an exceptional good value - nearly to the point of having some doubt in the capacitor value. Typical DMM specs are more like < 50 pA. The current does usually depend on the voltage and could be different with a negative voltage.

The more tricky range is the negative sign, especially negative voltage with AZ mode, as there likely is some extra current pulse from AZ switching.

I agree, my DMM got a very high impedance.

I think that many R6581 sold on second hand have problems, I think I'm lucky to have one that spend probably most of he's life on a shelf.

The checked the capacitor, picture from my LCR meter in attachment.

It show as wel that isolation from my new bindings posts are good 

[Mickle T.]

I've finally found it
Advantest R6581T, as well as ADCMT 7480T, are not true metrology instruments. They are part of Advantest T2000 system mainframe.

[szszjdb]

Hi,Mr.Kleinstein and Mr. Mickle T.,
Thanks a lot for the help on the reparing of my R6581!
Really learning a lots from both of you!

Merry Christmas and Happy New Year
szszjdb

[branadic]

After waiting for almost decades my R6581D received today. Bought it on 07.02.2019 on ebay, it was shipped on 08.02.2019, arrived in Frankfurt on 09.02.2019 and made a long road trip through Germany, was lost meanwhile and arrived without any further announcement today. Luckily I had enough money in my pocket to pay the customs.

Code: [Select]

    2019-02-26 22:35

    Importzentrum, Germany, Shipment on hold
    2019-02-25 16:35

    Shipment has left pick-up location
    2019-02-25 15:48

    Frankfurt, Germany, The item has left the import center
    2019-02-23 06:05

    Shipment has arrived at delivery location
    2019-02-23 01:47

    Shipment has left hub
    2019-02-23 00:51

    HUB Nürnberg, Germany, The item has left the import center
    2019-02-23 00:25

    Shipment has arrived at hub
    2019-02-22 22:29

    Shipment has left pick-up location
    2019-02-22 22:29

    Leipzig, Germany, The item has left the import center
    2019-02-22 16:07

    Shipment picked up in
    2019-02-22 06:47

    This shipment has been delivered to the consignee’s local customs office for clearance. The consignee will be notified with instructions by the Deutsche Post, in writing, including where the shipment can be collected. Further information: ww.zoll.de. Select: Postal Consignments and then Procedures
    2019-02-13 11:39

    Shipment delivered
    2019-02-13 06:28

    Stuttgart, Germany, With delivery courier
    2019-02-13 05:32

    Shipment has arrived at delivery location
    2019-02-13 00:04

    HUB Staufenberg, Germany, The item has left the import center
    2019-02-13 00:04

    Shipment has left hub
    2019-02-12 21:15

    HUB Staufenberg, Germany, The item has left the import center
    2019-02-12 21:11

    Shipment has arrived at hub
    2019-02-12 13:22

    Shipment has left pick-up location
    2019-02-12 09:50

    Importzentrum, Germany, Shipment is in customs
    2019-02-11 20:48

    Importzentrum, Germany, Shipment is in customs
    2019-02-11 20:25

    Importzentrum, Germany, Shipment handed over for Import
    2019-02-08 22:47

    The shipment has arrived at the export parcel center
    2019-02-08 22:03

    The shipment will be transported to the destination country and, from there, handed over to the delivery organization. (Homepage / online shipment tracking: )
    2019-02-08 16:11

    The shipment has arrived at the parcel center of origin

Herkunft : Südkorea - Abfragezeit: 201 ms
    2019-02-23 06:05

    FRANKFURT, Arrival at Office of Exchange
    2019-02-22 06:47

    FZ0 ZB0 0, Unsuccessful delivery, Reason : Payment of charges
    2019-02-13 11:39

    STUTTGART, Delivery complete
    2019-02-12 13:22

    DEFRAA, Departure from inward office of exchang
    2019-02-11 20:54

    DEFRAA, Ready for customs clearacne
    2019-02-11 20:25

    DEFRAA, Arrival at inward office of exchange
    2019-02-09 16:20

    FRANKFURT, Delivered to Destination Post
    2019-02-09 15:50

    FRANKFURT, Airrival at Destination Airport
    2019-02-09 12:10

    INCHEON, Departure from the Airport
    2019-02-09 01:06

    INCHEON, Received by Air carriers
    2019-02-09 00:41

    INCHEON, Handed over to Transport companies, Flight number : OZ541
    2019-02-08 22:47

    INTERNATIONAL POST OFFICE
    2019-02-08 22:03

    INTERNATIONAL POST OFFICE, Ready for dispatch, Dispatch number : 71, Flight/Ship No : OZ541
    2019-02-08 21:27

    INTERNATIONAL POST OFFICE
    2019-02-08 21:07

    SUWON MAIL CENTER, Handed over to Transport companies
    2019-02-08 20:13

    SUWON MAIL CENTER, Handed over to Transport companies
    2019-02-08 18:06

    SUWON MAIL CENTER
    2019-02-08 17:19

    SAGANG, Handed over to Transport companies
    2019-02-08 16:11

    SAGANG, Posting/Collection, Posting office zip code : 18550, Transit or Destination country : GERMANY

Bad enough I found the meter packaged very very poor, just put diagonal into a much to small cardboard box with some bubble wrap on top and bottom. The handles, rack mount adapters and screws were just put loose into the box without any plastic bag. I wondered that they were still there, but sure enough they are the reason for some additional scratches in the chassis coating. 
Due to the bad package the meter was slightly damaged at one corner. Could have been prevented if the box was fitting the size of the meter and the bubble wrap put around the meter.

A very quick eye on the inside of the meter revealed nothing obvious, all caps were looking fine, everything seemed to be in place, so I changed the transformer taps from 100V to 240V and replaced the T500mA fuse with a T250mA as suggested by the marking on the back of the unit.
A check of the battery shows it's completely dead, so needs to be replaced. Comparing the pictures of R6581T on the web with the meter in front of me revealed a few differences, that I will present soon.
Time to power the unit up to see if we got a winner. 
Well, maybe... the meters powers up to a certain point, the fan is running, but the display, which is slightly dim but still readable, shows:

PANEL CHECK
+291,"Communication error (from main to panel)"

Time to contact the seller and ask for some refund as the meter was declared as used and not as defect.
Most likely a power rail issue or some connection problems? Any suggestions what to look for? Turns out to become a repair thing instead of get it, service it and use it.

-branadic-

[Kleinstein]

Because of the rough trip, I would check for loose connectors, including the optical fiber.  There might have been quite some force to cause the damage at the corner - so even loose parts or other mechanical damage is possible.

[branadic]

Well, checked for that already.
My japaneese is a little bit rusty, but the manuel reveals: "Communication from the main body to the panel at power up when there is an abnormality in the check"
Wow, that's a lot of new info 
I think I need to check the boards for some indications of a crack and measure power rails.

-branadic-

[TiN]

That packing is among the worst I've ever seen. I'd not rest until seller refunds at least half of the unit  .

As for FP issue - I would try to release FPC cable from frontpanel/CPU boards and reseat it back again, to see if it's just the connector got loose.

[branadic]

Tried that, but without any change in behaviour. Currious to note is, that cursor left and cursor right button are working. Will check the rails and solder joints tonight as all the connectors have been disconnected and reconnected.

-branadic-

[BFX]

Poor meter 
 

[branadic]

What I can say for now, it's not an issue of the flexible cable and their connectors, as I was able to check the connection from one board to the other. Next step is to identify the power rails and check if they are any good. Silly me, I forgot to pick up the ESR meter so I can't check the caps in the unit today.

-branadic-

[Andreas]

what a pity.

Good luck