My first 3458A

[70bytes]

A few weeks ago i won the auction of a HP 3458A and a General Resistance RDS74-M Resist-o-stat II decade resistor. I paid EUR1368 for the HP and EUR38 for the decade resistor.
There were only 10 other bidders for the HP. Maybe because the auction house did not know what they had. It was listed as "Hewlett Packard 2823A Multimeter".
This week i picked them up at the auction site.
Both were ex US-Army equipment. The decade resistor has a cal-date sticker of feb-2020 with a due date of feb-2023. The HP has no cal-sticker.
The HP is missing the fuse-cap and the plastic protection of the display. It has minor damage on the left front side.

What have i done so far :
- visual inspection of the pcb's. All good. DALLAS chips have datecode 9734. Lets hope they are ok.
Looks like this meter is from around the year 2000. The housing has date stamps of Jan 23 1998. Chips have dates of 9750, and 2046.
- used an autotransformer and powered up the meter (left it on 120V). We have 230V here in the Netherlands. Didn't want to risk an explosion of the shaffner mains filter.
Started of with no ERR. What a surprise!! Display is looking good.
- made a backup of the calram.
- a few quick measurements.
It is about 1.5mV too high when i measure 10V (compared to my Keithley 2700 and HP3457A). 10kohm is about 80 ohm too high.

To do:
- Order replacement parts (caps and a new mains filter).
- Order replacements for the DALLAS chips.
- Find a new or used plastic display cover.
- Check if this meter has the drift problem.

[CalibrationGuy]

Whenever we get an old meter like this, we leave it powered up for a month, unless something starts smoking. Then we check all functions. Bulging electrolytics always replaced first. Since you already have the covers off, I would replace them right off the bat before aging and then re-testing. Good luck.

TomG.

[TERRA Operative]

I have a spare Agilent branded display window somewhere here if you need it. You can have it for cheap. IIRC Keysight charge about $60 or so for them.

I would first up replace the Rifa capacitors, and any visually bad capacitors as CalibrationGuy mentions and let it stabilise. My thread for my 3458A has a list of parts.
https://www.eevblog.com/forum/metrology/agilent-3458a-i-joined-the-club/

If you have a chip programmer etc, you could also update the firmware to v9.2, the file is in my thread too.



Let it sit powered on with the covers on for a few days, then run a full ACAL sequence.
Note: You can turn off the display to help prevent burn-in and dimming by pressing the blue <Shift> button, then the <C>/<Auto Cal> button, then pressing the <Down Arrow> button until you get to 'DISP', then press <2> followed by <Enter>, I also set NPLC to 200 to reduce the flashing 'SMPL' annunciator too.)

Once you have done that, pull the cal constant 72 as per the attached document then wait at least another week with the meter powered on the whole time and to the ACAL procedure and pull the constants again. Run them through the formula to see what your drift is like.

Then follow

[picburner]

If the instrument works well you have made a great deal considering the current market prices.
I have two and they cost me much more and required expensive repairs.
If you are interested I still have a HP (03458-69302) and a Agilent (03458-4933) display windows in NOS condition (in original packaging).

[70bytes]

Yesterday i found your post with the list of caps to replace. I will order them today.
I also found the mouser part# for the fuse cap. It is 693-0031.1666. I used the cap from my HP 3478 for testing.

I have a spare FN1608 DIP that will be used to replace the DS1220Y. I ordered 5 pcs a few years ago for my pinball machines. Need to order the conversion pcb (found the gerbers in this youtube video https://youtu.be/cAA1TCvwhuw?si=Hmcr1Bd5VZo6D-ZI)

Also found this https://github.com/JaredCabot/3458A-Calibration-Downloader/ for extracting the calibration constants. It works fine with my GPIB-adapter from  Xyphro https://github.com/xyphro/UsbGpib

[TERRA Operative]

That calibration downloader script in mine. I'm glad it actually works for someone else besides me.

I also have some FRAM adapter board designs too, but the DS1220 one needs to have the FRAM programmed with the cal constants before being soldered to the adapter, and from there it operates exactly like the Dallas chip once installed in the 3458A.

youtube.com/watch?v=uNbI3SUZ7oI

https://www.pcbway.com/project/shareproject/Dallas_DS1220_to_FRAM_FM16W08_adapter_2bd23c10.html

https://www.pcbway.com/project/shareproject/Dallas_DS1230_to_FRAM_FM18W08_adapter_c5d2a792.html

[70bytes]

Thank you for the downloader script.
Just ordered the caps,mains filter and fuse cap.
Ordered the pcb for the FM1608 at JLCPCB.
Also ordered some HM62256 , FM1808-70-PG on aliexpress. Lets hope these are no fakes.

Now wait for the orders to arrive.

@picburner : I am very interested in one of those display windows. I will PM you.

[aronake]

I like the insightful title on this thread "My first 3458a", yes they tend to grow in numbers over time.....

Good you have a GPIB adaptor, that will be very useful.

Seems you already know some python, if not good to learn some basics to be able to control and get data from the 3458a.

You can then do a program that run ACAL say once very 2 hours, and collect relevant data like CAL? 72, TEMP etc.

Here is the messy code I use. It supports multiple 3458a so you are future proof Parts for using a 34470a for ambient temperature as been commented out. I use an Agilent E5810a as Lan to GBIB interface (highly recommended solution) so will need some edit for what you have. And need to install python-vxi11.

I typically run it a couple of times per week, so for recurring measurements this need to be added as well.

Code: [Select]


import vxi11
import sys
import time

print ("ACAL collector meter started")
   
Instr= [vxi11.Instrument("192.168.0.60", "gpib0,11"),vxi11.Instrument("192.168.0.60", "gpib0,12"),vxi11.Instrument("192.168.0.61", "gpib0,13"),vxi11.Instrument("192.168.0.60", "gpib0,14"),vxi11.Instrument("192.168.0.60", "gpib0,16"),vxi11.Instrument("192.168.0.60", "gpib0,15")]

#hp34970a = vxi11.Instrument("192.168.0.60", "gpib0,17")
#hp34970a.write("SYSTem:PRESet")
#hp34970a.write("INSTrument:DMM ON")
#print(hp34970a.ask("MEASure:TEMP? THER,10000, (@101)"))
i=0
print ("assigned instruments")
for ThisInstr in Instr:
    #print (i)
    i = i+1
    ThisInstr.write("END")
    ThisInstr.write("PRESET NORM")
    #ThisInstr.write("DCI")
    ThisInstr.timeout = 30
    ThisInstr.write("BEEP")


print ("ACAL stat collect")

print ("Instrument initialized")
print("TEMP = " + Instr[0].ask("TEMP?"))
InstNo =1
for ThisInstr in Instr:
    print(f"Inst= {InstNo}",end =" ")
    InstNo=InstNo + 1
    CAL72 = float(ThisInstr.ask("CAL? 72")) * 1000
    CAL175 = float(ThisInstr.ask("CAL? 175"))
    TEMP = float(ThisInstr.ask("TEMP?"))
    AdjShort = float(ThisInstr.ask("CAL? 58"))
    Adj10V = float(ThisInstr.ask("CAL? 59"))
    Adj10K = float(ThisInstr.ask("CAL? 60"))
    Dif10V = TEMP - Adj10V
    print(f"C72= {CAL72:.6f} TEMP= {TEMP:.1f} Acal Temp= {CAL175:.2f}  ADJ: Short= {AdjShort:.2f} 10V= {Adj10V:.2f} 10K={Adj10K:.2f} Diff 10V = {Dif10V:.2f} ")

print("Running ACAL")
print ("Start time: " + time.strftime("%d/%m/%Y-%H:%M:%S "))
StartTime = time.time()
for ThisInstr in Instr:
    #ThisInstr.write("ACAL DCV")
    ThisInstr.write("ACAL")
time.sleep(120)   #150 sec is probably shortest for ACAL DCV
time.sleep(800)   #Extra wait of doing ACAL ALL

for ThisInstr in Instr:
    MySTB = ThisInstr.read_stb()
    while not (MySTB & (1<<4)):
        MySTB = ThisInstr.read_stb()
        time.sleep(2)

print ("Done time: " + time.strftime("%d/%m/%Y-%H:%M:%S "))

EndTime = time.time()
AcalTime = EndTime - StartTime
print(f"Time to do ACAL = {float(AcalTime):.1f}")


print (time.strftime("%d/%m/%Y-%H:%M:%S "),end =" ")

o = open('F:\\SyncServ\\Dropbox\\dokument\\electronics\\MesData\\ACAL_stat.csv', 'a')

o.write (time.strftime("%Y-%m-%d"))
o.write (";")
o.write (time.strftime("%H:%M:%S "))
#o.write (";")
for ThisInstr in Instr:
    CAL72  = float(ThisInstr.ask("CAL? 72")) * 1000
    CAL175 = ThisInstr.ask("CAL? 175") # DCV acal temp
    MyTemp  = ThisInstr.ask("TEMP?")
    print ("; %9.6f" % (float(CAL72) ),end =" " )
    print ("; %4.1f" % (float(MyTemp) ),end =" " )
    print ("; %8.2f" % (float(CAL175) ),end =" " )
    o.write(";%9.9f" % (float(CAL72) ) )
    o.write (";%4.1f" % (float(MyTemp) ))
    o.write(";%8.5f" % (float(CAL175) ) )
    o.write(" ; ")

for ThisInstr in Instr:
        CAL58 = ThisInstr.ask("CAL? 58") # Short adjust temp
        CAL59 = ThisInstr.ask("CAL? 59") # 10V Cal Temp
        CAL60 = ThisInstr.ask("CAL? 60") # 10K R Cal Temp
        CAL71 = ThisInstr.ask("CAL? 71") # 71 dcv gain 1 V
        CAL73 = ThisInstr.ask("CAL? 73") # 73 dcv gain 100 V
        CAL75 = ThisInstr.ask("CAL? 75") # 75 ohm gain 10
        CAL78 = ThisInstr.ask("CAL? 78") # 78 ohm gain 10 K
        CAL87 = ThisInstr.ask("CAL? 87") # 87 ohm ocomp gain 10 K
        CAL99 = ThisInstr.ask("CAL? 99") # 99 dci gain 100 mA
        o.write(";%8.5f" % (float(CAL58) ) )
        o.write(";%8.5f" % (float(CAL59) ) )
        o.write(";%8.5f" % (float(CAL71) ) )
        o.write(";%8.5f" % (float(CAL73) ) )
        o.write(";%8.5f" % (float(CAL75) ) )
        o.write(";%8.5f" % (float(CAL78) ) )
        o.write(";%8.5f" % (float(CAL87) ) )
        o.write(";%8.5f" % (float(CAL99) ) )

AmbTemp= 1# hp34970a.ask("MEASure:TEMP? THER,10000, (@102)")
print ("; %5.3f" % (float(AmbTemp) ),end =" " )
o.write (";%5.3f" % (float(AmbTemp) ))
o.write ("\n")
o.close
print("Done!")
   



[70bytes]

Thank you for the python script. Will try to use it after i replaced the caps and mains filter. I don't want to keep it powered up 24/7 before those are replaced, just to be shure.
I will change the script for pyvisa.

If i have some time this weekend i also want to try to change control and datalogging software for a HP3457A. See this github : https://github.com/Niravk1997/HP-3457A-Software
This is very nice software that i use for logging and graphing for my HP 3457A and Keithley 2700.

[TERRA Operative]

Ooh, that software, but for the 3458A would be nice.
I use the averaging and histogram functions on my 34461A a lot, and do miss them on the 3458A.

[Dr. Frank]

It is about 1.5mV too high when i measure 10V (compared to my Keithley 2700 and HP3457A). 10kohm is about 80 ohm too high.

Congrats for 1st impression on this purchase!

These big deviations, providing:
your DMMs themselves are not too far off
you've properly warmed up and made an ACAL All
used 4W Ohm and OFFSET COMPSENSATION

are 150ppm for 10V and 8000ppm for 10kOhm, much too big for the usual drift.

This indicates, that the cal constants are both lost, that would fit to the age of the nvRAMs. 
You can read the 40kohm and 7V reference values by CAL? 1,1, CAL? 2,1.
Maybe these are set to their initial values of 40,000k and 7,2000V.

If you used 2W Ohm only, these +80 Ohm might indicate the usual contamination of the front/rear switch, which you can check by engaging multiple times, and by using the back jacks.

Fingers crossed, that you got a stable one (U180).

Frank   

[IanJ]

Nice price!, fingers crossed U180 is stable. Here's what I would do:

1. Extract the Calram data, if you are on Windows you can easily do this using WinGPIB.
2. Power up the 3458A for a couple of days in a stable temperature environment
3. Run ACAL ALL and then with a precision voltage source, monitor it over a few days. You can log this data using WinGPIB.
4. If the temperature is stable throughout and the logged voltage data stays stable then repeat over a longer period, try a week.

Don't confuse U180 drift and a varying lab temperature (if it's quite big).

U180 drift can be very severe and quick to deviate, I had a 3458A years ago that visibly deviated overnight, but it can also be very slow.......so you might not see an issue unless you log for weeks.

If you come out the other end with a stable U180 then you picked up an extreme bargain!

Ian.

[70bytes]

@Dr. Frank :
Those offset where before CAL ALL. I used 4W without offset compensation.
After CAL ALL the ohms measurements is almost spot on. DCV is still too high. No change after CAL ALL.
CAL? 1,1 = 39.9995310 E3
CAL? 2,1 = 7.08252832

@Ian :
Thank you for your software. I extracted  the calram data. When the capacitors and mains filter are replaced i will do the stability test.

There were also 2 Fluke 5720A calibrators at the auction. Both were sold for EUR 9000 each. I wanted to bid on these also but was a bit too high risk for me. Those were also ex US-Army.

[70bytes]

An update of my 3458.
Last week i replaced the mains filter and the caps. Replaced the calram with a a FM1608, added 4 pcs HM62256.
Meter has been powered on for 6 days now.
Yesterday i received the NOS HP display window from picburner. He has two NOS, 1 HP and 1 Agilent. Thanks again picburner!
I also received 3 FM1808. They are used but tested OK. I don't replace the DALLAS chips until they fail.

Because the VDC reading was 1.5mV too high i calibrated this one myself with my Fluke 343A. None of my meters have a recent cal.
Have been logging TEMP? and CAL?72 for 3 days now. Temperature was stable until yesterday. We have a heatwave for a couple of days. The outside temperature went up from 20C to 28C. This will last till sunday. Room temperature is up 1 degree.
Not so good.

[70bytes]

These are the results of the CAL?72 test. Looks like no drift problem.
I started the test on 11-jun. After 2 days the outside(and inside) temperature started to rise from 20C to 25C room temperature because of a heatwave. The last 8 days where more stable temperatures.
I added the log graphs of my 3457A and 3458A. Both are connected to my new LM399. This new LM399 has been powered on for about 7 weeks now. I used the circuit of https://www.circuitvalley.com/2025/01/TEN399-ADR1399KHZ-LM399-10V-reference-calibrated-DIY-PCB.html

I have a question about the LM399.
The LM399 has spikes down and spikes up. The up-spikes are only visible with the 3458. These are not visible with the 3457, i think because of the high internal noise? See graphs Last5hours.
Do these spikes disappear over time?

[Kleinstein]

The jumps are typical for a LM399. There are likely not going away over time. The hight of the jumps is often similar (~0.5 ppm) hight, but how often the jumps happen can vary and a few lucky ones may not have  them very prominent. The 3457 likely does not see the jmps due to the higher noise. This includes comparable jumps from the 3457 internal LM399 reference.
The general drift of the 10 V external reference may well go down. I would suspect that the triangle peaks are a thermal effect at the external 10 V reference.

[arvee]

A zener circuit can sink large current but it cannot source more than its bias current.

My primary reference is an unbuffered LM399.
It had similar symptoms with small positive spikes and large negative spikes.
The cause is long wires that act as antenna.
The problem was solved with an RC low-pass filter between reference and output.
The C was a low-leakage tantalium capacitor salvaged from a laptop, 10 uF.
The R was 10 ohm.
Any low-pass filter will do.