5700A OHM troubleshoot

[dl1640]

Hello,
Our 5700A have been observed not stable output especially on and above 10k Ohm. On 100k Ohm and above it is off specification and not stable. This model is Series II and is made around 2003 or 2004.
Any idea for troubleshooting this particular issue is welcomed. Thank you.

[Echo88]

I dont own a 5700, but assume its the following problem:

I assume that one or more relays on the assembly need to be changed, since it may be contact problems or the Resistor-Network which contains the 10k is faulty (which is unlikely i think).

https://xdevs.com/doc/Fluke/5700a/5700A_old_sm.pdf Page 538 lists the relay-configurations for the Ohms-Main-Assembly. Check which Ohm-Ranges are working properly and which not and find the faulty relay by looking at the relay-config-table.

But it might also be something else, since i just took a guess. 

[TiN]

Mmm, tasty thread about my nemesis units.

What calibrator see by itself, running from calcheck report? How you test resistance error?

[dl1640]

At first, this issue is reported by a Fluke repair center when the unit was calibrated. But later they said after a few days of constant run, the issue was gone, so calibration report came with all items passed. This is about on Nov.2017.

Now in our lab, I will check it by DMM, mostly with 3458.
I will update after I find more..

[dl1640]

Echo88

Thank you for your info.
I would consider the relays if output is not stable, but if it is resistor network, I think Fluke is asking for money..

Today I use another cable to hook up 5700 and 3458, the result is good. This cable is diy 2 pairs + 1 shield twisted in one jacket.

However I think Fluke use good cable, if they observed off spec and instability I would believe it does..
 Then I changed the cable to seperate 4 cables for 4w measurement, again the result is off spec and not so stable.

I shall do more check to find what is wrong, 5700? or inadequite cable?
Fluke repair center is not good service?...

[TiN]

There is set of two resistance boards from 5700 on ebay now, in case you want to skip on troubleshooting .

[dl1640]

Will do a full self test..

[Dr. Frank]

More details, please!

If the reading is ok in 2W mode, then that might be a relay problem.
You can check that. Set 3458A in 2W mode, 5700A in 4W mode (if selectable), and measure these unstable ranges in different configurations on the 5700A, i.e. Sense or Output Jack at Hi and lo.

A defective relay should be identified that way.

But pay attention, the 3458A is about 10 times less accurate in these high Ohm ranges, and might not be well suited for verification.

There might be also a problem, that these old relays are leaky. I have found that out on my >20years old 5450A resistance standard, which is very similar to the 57xx family in aspect of Ohm topology.
See my story here: https://www.eevblog.com/forum/blog/eevblog-544-fluke-5450a-resistance-calibrator-teardown/100/

So, I would like to ask you, that you do some 2W/4W measurements in these ranges, and publish your 3458A settings (NPLC, 2W/4W, DELAY, ..),the values found, with statistics,what the FLUKE calibrated values should be, and an indication, what is going wrong, to your opinion.

Frank

[TiN]

It's probably useless for the matter, but here's last week test result of 5700A (mark I, calibrated in 2013) vs my 3458A_2 (calibrated in Jan 2017 to SR104 known to 0.55ppm).
Reference values are the numbers reported by CalCheck procedure prinout of 5700A. Settings are OCOMP with DELAY 3, NPLC 100 for all values except 100 Meg.

[dacman]

Try to find a way to exercise the relays (such as switching between DCV and Ohms, 2-Wire and 4-Wire, changing Ohms ranges, or performing the cal check, toggling OPR and Standby).

[dl1640]

As Dr.Frank suggested, will have some test next week.

BTW, TiN
What is the CalCheck printout values? The "true" value of nominal Ohm? At the time it was adjusted?

[TiN]

CalCheck is the procedure used to calibrate 5700A , but with only difference that it use internal 10 KOhm and 1 Ohm resistance references on A9 board and bi-LTFLU reference value instead of external artifacts. You can start this procedure locally on calibrator, but you need RS232 or GPIB interface to readout calibration check results. It shows what are true values (determined by running full external artifact calibration) and how much each function and each range deviate from original calibration. It gives you ppm deviation and gives you also % of spec change. This can be used as substitute to performance verification to build up confidence of the calibrator performance, if you don't have means to do external verification using calibrated standards. Running such artifact calibration allow to bring MFC output source accuracy within 24 hour specification, and also provide drift information if such reports stored and tracked periodically. I'm surprised you have 5700A and do not know of this.

So while it is not a replacement for "traditional calibration", it is swell tool to maintain 57XXA confidence in between the external calibrations.  I have touched this stuff a little before, and also recommend great article An evaluation of the 5700A Series II Artifact Calibration from three NMIs in Europe. Performance points and methodology of function traceability of 57XX covered there in great detail. It works little different than 3458A's ACAL, but idea of use internal higher-stability standards and accurate ADC (in terms of 5700A a null-meter ADC, as reference point can be adjusted unlike the HP 3458A) is in general same.

[dl1640]

TiN

Sorry, I have no deep study 5700, normally it is a voltage generator in our system, not act as a standard.
I reviewed its service manual, I will run CalCheck thru GPIB and see if there is surprise..

A calibration check is similar to the calibration, with the primary difference being that
no changes are made to the stored constants, and the internal check standards are used as
the reference points. A calibration check produces a report similar to the normal
calibration report, and shows any proposed changes.
This procedure can be performed from an external computer, and can be set to run
automatically, with no assistance (there is no need to enable the CALIBRATION switch,
since no constants are changed).
You can use the calibration check at any time to confirm the integrity of the calibrator
without connecting external standards. The calibration check is also useful for collecting
a performance history.

[TiN]

Here's piece of code I use to run CAL ZERO (internal) and CAL_CHECK over GPIB:

Code: [Select]

intf.write("CAL_ZERO")
    f5700_dormant(180) # take a little nap, zzz
    intf.write ("CAL_CHK")
    f5700_dormant(3700) # take a long nap, ZzZ
   
    f5700_cal_report (cal_report_check_after, "CAL_RPT? CHECK")
    f5700_cal_report (cal_report_cal_after,   "CAL_RPT? CAL")
    f5700_cal_report (cal_report_const_after, "CAL_RPT? RAW")
    f5700_cal_report (cal_st_cal_after,       "CAL_SLST? CAL")
    f5700_cal_report (cal_st_shift_after,     "CAL_SLST? CHECK")
   

[dl1640]

I don't think that CalCheck bring the MFC to within 24 hrs spec, it doesn't change any constants. The set of internal standards used for CalCheck is the one used in normal operation or not? I think CalCheck use exclusive standards and they will be adjusted if an external calibration ( adjustment) is executed, so do the internal standards for normal operation.
The 24 hrs spec is achieved only after a external calibration, the ArtCal.

Though a little off the topic, if it is wrong, please correct me. Thanks.

[TiN]

Internal standards resistor and AC calibration reference sensor are never used in normal operation, only the LTFLU DC reference is same for both external cal/internal cal. I shouldn't say it bring MFC to spec, but rather shows what is deviation from internal standards and if that still fit 24 hour spec (if you set MFC to use 24hour spec) of the unit or not, given assumption that internal standards are not drifted since external ACAL.

[dl1640]

Ok, the internal resistance standards (1? 10k?) for CalCheck should be fine since last ArtCal at Fluke is a few months ago.

[dl1640]

After cooking some code, I get CalCheck results of Fluke(last external adjustment at Fluke), and me(after CalCheck run by me, today).
The Fluke result has zero drift, I'm not sure if they did run it..

It looks Ohm is fine by my result. I'd run it once more before I cook some code to measure it by 3458.

[Pipelie]

it probably has something to do with leakage of the PCB, caused by the humidity. and the issue will be gone after a few days of constant run. you can put the ohm board in the dry box to make it dry.
the later version of this board, fluke changed the material of PCB, it looks like PTFE PCB.

[dl1640]

Fluke did not open the box to have a look after they found such issue.
They sent us a quote to adjust the box again, which is meaningless.

Thank you, I think humidity is a killer.
Still try to find more

[dacman]

A Cal Check does not adjust anything.  It is a sanity check, and results will be lost once the unit is powered off (so if Fluke ran a Cal Check, the results will not be saved inside the unit). If a Cal Check report is obtained without performing a Cal Check (since last powered on), the reported changes will be all zeros.

Results of an Artifact Cal are saved in memory (if the results are stored).  The Artifact Cal report will show the results the last time the results were stored, or the proposed changes if an Artifact Cal was just run prior to selecting Store.

To eliminate the obvious, check connections on the rear of the binding posts.  (I believe they have screw terminals that are made of copper and snug should be ok.)

[TiN]

Rear terminals on 5700A are optional. Only one set of terminals is connected at the instrument internally, either front or rear.

[dacman]

I mean the back side of the binding posts that are on the front of the 5700A.  Connections are held with nuts that are connected to the binding post screws, not soldered, to the binding posts (IIRC).  I would check them to make sure that is not the problem.  I'm not just coming up with this.  I've seen a loose post on a 5725A that was causing 10A checks to trip.

[dl1640]

True.
CalCheck data are lost after power-off. However CalArt report is not lost, and internal standards shifts can be printed out.
But why there is no show of 10k Ohm internal standard??

------------------------------------------------------------------------
                                 STANDARDS USED
    ------------------------------------------------------------------------
    STANDARD         NUMBER             CERT. DATE

    VOLTAGE      ______________       ______________      10.000091 V

    10 kOhm      ______________       ______________      10.000599 kOhm

    1 Ohm        ______________       ______________      1.0000025 Ohm


------------------------------------------------------------------------
                    EXTERNALLY CALIBRATED INTERNAL REFERENCES
    ------------------------------------------------------------------------
        REFERENCE            Most Recent          Previous       Shift (ppm)
    6.5 V DC Reference          6.894766          6.894771             -0.64
    13 V  DC Reference         13.841004         13.841022             -1.31
    Resistance Reference       0.9999773         0.9999760              1.27
    1.0 Ohm Resistor           1.0000749         1.0000690              5.91
    1.9 Ohm Resistor           1.9000607         1.9000560              2.46

[TiN]

True.
But why there is no show of 10k Ohm internal standard??

Because you did not RTFM long enough  .
Use CAL_CONST? CHECK, RS10K command.