Repair worklog : Old rusty HP 3458A

[KJDS]

I nearly bought a 3458A in an unknown state earlier in the week.

I'm now glad I stopped bidding when I did and left it to someone else.

[TiN]

These instruments are not for weak-hearted people, I can tell you that..

Here's python I'm running now, which samples DCV, calculates ppm deviation from hardcoded value [n]level[/b] and prints it on meter's display.
It also creates three CSV files (one with measured voltage, another one with temperature logs, third with ppm deviation calculations).
Don't ask why not single file with three values together .

Temperature is read on every 25th sample (as it clicks relay to take TEMP reading, I don't want kill relays too quick).

Code: [Select]

# xDevs.com Python test GPIB DeviationPPM app
# http://xdevs.com/fix/hp3458a/
import sys
import Gpib
import time

with open('10v_oldadc_d5_nplc200.csv', 'a') as o:
    with open('10v_oldadc_d5_nplc200_temp.csv', 'a') as t:
        with open('10v_oldadc_d5_nplc200_ppm.csv', 'a') as d:
        inst = Gpib.Gpib(0,3) # Instrument GPIB Address = 17
        inst.clear()
        inst.write("PRESET NORM")
    inst.write("OFORMAT ASCII")
    inst.write("DCV 10")
        inst.write("TARM HOLD")
    inst.write("TRIG AUTO")
        inst.write("NPLC 100")
    inst.write("AZERO ON")
        inst.write("LFILTER ON")
    inst.write("NRDGS 1,AUTO")
        inst.write("MEM OFF")
        inst.write("END ALWAYS")
    inst.write("NDIG 9")

    min = 0
    tread = 100
    temp = 0.0
    level = 10.0000340
    ppm = 0
            while min <= 10000000:
min+=1
tread = tread - 1
if (tread == 0):
    tread = 25
    inst.write("TARM SGL,1")
    inst.write("TEMP?")
    temp = inst.read()
    print(time.strftime("%d/%m/%Y-%H:%M:%S; < TEMP") + ("%d;%4.1f;\r\n" % (min, float(temp) ) ))
    t.write(time.strftime("%d/%m/%Y-%H:%M:%S;") + ("%d;%4.1f;\r\n" % (min, float(temp) ) ))
inst.write("TARM SGL,1")
data = inst.read()
ppm = ((level / float(data))-1)*1E6
d.write (time.strftime("%d/%m/%Y-%H:%M:%S;") + ("%d;%16.16f;\r\n" % (min, float(ppm) ) ))
inst.write("DISP MSG,\"%12.4f ppm\"" % float(ppm))
time.sleep(1)
print time.strftime("%d/%m/%Y-%H:%M:%S;") + ("[%8d]: %16.16f , dev %12.4f ppm" % (min, float(data),float(ppm) ) )
o.write (time.strftime("%d/%m/%Y-%H:%M:%S;") + ("%d;%16.16f;\r\n" % (min, float(data) ) ))

[TiN]

Same stuff, ran away 16ppm in just about 16.73 hours. +0.96ppm/hour 
I added temp readings too, just to be sure. It's very well unrelated...

[Dr. Frank]

Hi TiN,

did you also monitor the + / - 12V reference voltages directly at the A/D hybrid, if these run away?

Frank

[TiN]

Not yet. I moved 3245 (10V source), 3458 (sick bastard), K2001 (control meter) into other room, connected only to Raspberry Pi running simple Python app , to capture data and send to FTP in realtime, to test this over this week.
Then will be testing one part at a time. Good idea about references.



Here's new test rig.

EDIT: On photo click - realtime datalog chart , updated every 7 seconds.

EDIT2: So far same -0.98ppm/hour. K2001 stays within 1ppm about 10.0000125 VDC, just to make sure that 10V 3245A source is stable (it was measured with K2002 before as well). I have wild idea to freeze bad ASIC on second A3 board with -196°C to cycle it few times. Maybe it will help?
Lol, this is most popular thread in repair section here by big margin. Would never know fixing dead 9K$ DMM would be so popular?

[BFX]

TiN you are crazy     
Good Job

[VK5RC]

TiN=ABSOLUTELY crazy. Still waiting for my DS1220, otherwise my little baby (3458) is up and running, getting a Rasp Pi, and learning a bit more code (Arrrrrgggh)

[TiN]

What is crazy? I don't even know Python,  .
Obviously there is issue, as it script hangs sometimes. I suspect frequent FTP-connections cause of it.

[Dr. Frank]

 I have wild idea to freeze bad ASIC on second A3 board with -196°C to cycle it few times. Maybe it will help?

That's really crazy. What positive effect should that have??

If you want to destroy the resistor array inside completely by thermal shock of liq-N2, causing cracks in the resistor film and its top layer protection, just go ahead.
Maybe the outer case will also crack.

Frank

[TiN]

That ASIC not working properly anyway. And ceramics usually not so bad with slow temperature gradient.

Anyway, lets read reference voltages on Mr.Drifty. its -26ppm now, just from sitting on , not even sampling half a day.

K2001 is only 1ppm up. Take that, 3458A

[nfmax]

This is a crazy way to learn Python!

[HighVoltage]

TiN,
It is amazing how much work you have put in to this broken 3458A and we all have learned so much from you.

There have been several 3458A been sold on ebay France in the last few weeks for very low prices and some of them in good working condition:
http://www.ebay.fr/sch/i.html?_odkw=3458A&LH_Complete=1&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR0.TRC0.H0.XAgilent+3458A.TRS0&_nkw=Agilent+3458A&_sacat=0

I think I will wait for another one of these to come along, since I don't think I would have the patience to repair one like you did.

[TiN]

Well, I did not discover much new, just wish I could get working instrument though. It's a time black hole..
I guess law of balance is in works, due to 3245A which was "repaired" in one minute 

[Zucca]

...was "repaired" in one minute 

and "improved" with a killing LTZ1000ACH in 10 minutes... . A master piece.

[TheSteve]

Lol, this is most popular thread in repair section here by big margin. Would never know fixing dead 9K$ DMM would be so popular?

Even if most of us never own a 3458A I think we all respect it and desperately want you to be successful, so of course we're following your efforts closely!

[Theboel]

Are this ASIC not available from agilent ?
Sorry my experience with agilent so far are replace 20pcs tantalum cap from my 8590L and replace fuse from my 34401A

[Monittosan]

TiN,
It is amazing how much work you have put in to this broken 3458A and we all have learned so much from you.

There have been several 3458A been sold on ebay France in the last few weeks for very low prices and some of them in good working condition:
http://www.ebay.fr/sch/i.html?_odkw=3458A&LH_Complete=1&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR0.TRC0.H0.XAgilent+3458A.TRS0&_nkw=Agilent+3458A&_sacat=0


I think I will wait for another one of these to come along, since I don't think I would have the patience to repair one like you did.

My old meters in that link 

Tin I think there are a lot of people out there that just love the idea that ultra high end equipment is repairable. Give hope to owing equipment you otherwise would not be able to justify.

[Macbeth]

Are this ASIC not available from agilent ?
Sorry my experience with agilent so far are replace 20pcs tantalum cap from my 8590L and replace fuse from my 34401A

Nope - http://www.keysight.com/my/faces/partDetail.jspx?partNumber=1NC1-0017

However the complete A/D board appears to be available for GBP 898 with the old board traded in.

http://www.keysight.com/my/faces/partDetail.jspx?partNumber=03458-66513

[crispy_tofu]

here is another 8.5 digit for the "picking"
http://www.ebay.com/itm/Solartron-Schlumberger-7081-Precision-Voltmeter-/121661653123?hash=item1c53998083:g:NuAAAOSwstxVZIEd

Stop tempting him! 

[TiN]

You guys can have it, I'm done with dead stuff for now  .

Hooked up 2002 to scan those A3 voltages. Here's first hour result:



Legend is on graph. All plots are normalized to 2ppm scale.

Setup itself:



All signals wired to 2001-TCSCAN inside 2002. All shields and covers are installed in place, to keep same thermals. 

[TiN]

Also poked around with thermacam.

That buffer on A2 AC board is cooking... It was like that even before though, it does not affect DC path, meter working even if I remove it (chip is in the collet socket)...

[TheSteve]

Sadness - time for a new ASIC.

[lukier]

Sadness - time for a new ASIC.

Probably 

What about other possible suspects? Integrator capacitors, opamps and comparator? Crystal oscillator (stability) that is used to clock multi-slope ASIC?

If this is the ASIC then maybe desolder it and measure slope resistors with 2001+scanner for a day or two.

[TiN]

OK, run overnight results:

+10Vin -14ppm
+12VREF +/-0.6ppm 
-12VREF -5ppm 
+5VREF -4.2ppm 
LTZ ZR +/-0.5ppm 

CSV-source file for you, data-nuts.

Next step - isolate ASIC resistors for -12VREF and +5VREF opamps and hook opamps to error114 ASIC resistors instead.

EDIT:



Here's spicy stuff. I'll isolate pins 2,3,64,9,15,16 and try to use resistors from another ASIC to see if can change anything on -12 and +5 refs.

[Dr. Frank]

TiN,


What's strange is the continuous negative drift.
If the resistors R4, R5 (10k) inside ASIC are not correctly matched (any more), I'd expect the +12V reference and 10V reading going up and down with temperature.
It may also be possible, that U165 is damaged, like excessive input bias, or so.

But I lost overview, which ASIC is now on which PCB..so if this failure went with the ASIC, it's not so probable.

Frank