Electronics > Metrology

A very cheap 3458A

(1/6) > >>

The meter is one of the early units from around 1989 and upon arrival of course it did not work. It gave the typical RAM TEST 1 HIGH error message meaning that at least one of the NVRAMs had dead battery. Further, it also complained about 202 - HW FAILURE: SLAVE TEST (DC BOARD) meaning that the DC board had something in it that it did not like.

NVRAMs (from 1988) were easy enough to take care of, I used some old but still good ones from an another unit that had it's NVRAMs replaced with new ones, for now. CALRAM was ok, it had datecode from 2009 so obviously it has been replaced before. FW revision was 3, I put in version 8 in new EPROMs. The meter was also completely stripped down and cleaned of all the dust that had accumulated over the last 27 years.

Then to the DC-board error, root cause was a U1 (TI CD4094B) that was outputting only ones to the next in chain. Great, that ensured that all ohms ranges were selected at the same time and no surprise, ohms current source was faulty with CA3082 that had pin 15 (common collector) open and Q307 blown up.

Took some work to find out those faults but easy enough to fix, replacing those three made the self test pass.

At the same time I also removed all electrolytics as they had datecodes from 1988. That turned out to be unnecessary as all measured ok with only minor difference to new ones. Seems that the old electrolytics were made very well and they had also epoxy sealed bottoms so no leakage. Nevertheless, I put in new ones as the olds were off the board. The two Y-capacitors on outguard power supply board got replacements too as these had their casings cracking already. Line filter and fan got replaced too.

On the AC-board fuses F701 and F702 were 1/16A types. CLIP states these to be 1/8A types so I replaced them. I think this was a factory screw up as they definitely were not replaced before and even the spare fuses F001-F003 were also 1/16A types. F702 was also blown and it had a normal 20mm fuse bodged on it.

Has anyone encountered the same issue with those fuses?

The faulty DC board had obviously been replaced before as it was from 2007. The peculiarity in this instrument is that CALRAM has about 7.18V for the zener but I remember measuring it to be about 7.15V, albeit with a Fluke 87 while debugging the board.

The difference also shows in measurements, they are off about the same amount relatively. The meter is now hooked to a Fluke 732A and seems stable so far except for the gain error.

Now, what could have possibly caused the reference to change operating point?

In this case, I think nothing. The DC board has probably been changed from a newer unit where it failed. Then whoever changed the working DC board from this unit to whatever they had sent the fixed unit for cal and this unit made it's way to surplus store.

Well, thanks for the newer DC-board with VHP101 40k reference resistor. It looks like this meter needs only proper adjustment and calibration now.

Why is this message in metrology section? Plan is to see if this was really fixed and stays that way. That "SELF TEST PASSED" does not mean much - stability is of more concern.


Welcome. I liked your though train, well done. I assume you read my worklog  ^-^ perhaps.

--- Quote ---Nevertheless, I put in new ones as the olds were off the board.
--- End quote ---

Right there, earned a star. Not worth time to discuss reasoning for replacing 20 year capacitors in gear.  :-+

I'd watch out for mains EMI filter assembly. Some of them are Schaffner manufactured with issue, which make them time-bombs and fire generators!.

For zener voltage difference, it's likely that reference module (little A9 PCBA in far end of DC board) was swapped, and unit was not recalibrated since than.
If your 732A is in cal and you know it's output value, you can switch meter to DCV 10V range and run calibration at 10V output. Since your unit's calibration is invalid, you don't risk anything much at this point.

To get ball-park of stability - you should run unit at least a week 24/7 to get everything stabilized, and then run SN18 test procedure over another week (daily checks). This will give you approximate stability figures for DCV, which should be way <1ppm. Good meter/ADC should show <0.4ppm span over a week. Here's comparison of 4 units (one with good and bad A3 ADC PCBA) you can reference to from our member here.

P.S. If it's not a secret, how cheap was very cheap? :)

Dr. Frank:
As the analogue board A1 and the nv-CALRAM have been changed at different times, the date codes not being necessarily the exchange dates, I assume that the repair in total was not successful already at these dates.
So no new calibration had been done, and the 10V calibration value, you read from the nv-RAM, may refer to another reference board, which may have been exchanged on a different instance.

Anyhow, it's not possible, that the LTZ reference may change its value by that amount, only in case of a defect, and to lesser amount only, if someone "pimped" the oven temperature to lower values. But latter should be easily visible on the LTZ board.

You might also compare the 10k calibration to an external reference; if this value also disagrees strongly, then definitely, several  boards may have been mixed.

If you have a 732A, it's very easy to re-calibrate your 3458A to the uncertainty level of this 732A


I my case pimping A9 PCBA got original value 7.074225xx to 7.180754xx. Just for reference here.

I should have mentioned that the reference board was also from 2007 and has most likely been changed as a pair with the DC board. I still think someone swapped this faulty DC-board along with the ref board from another meter.

I got bored yesterday evening and already adjusted the meter to my Fluke 732A. That cured the reference voltage difference and "CAL? 02" reports now 7.15407265V.

My 732A is not calibrated as it is fairly new to me but I measured it with my other 3458A (ok, well, in that sense it is now actually calibrated) and used that value for 10 volts. Final cal will have to wait until I send the 732A in office to national lab sometime in the autumn. There is also a brand new 3458A in office, I just unboxed it last week. So, I think I am covered in getting the volt to home lab.

The other 3458A has not been on much and is likely still very near what it was a year ago when the company 732A came from visit to national lab. I then adjusted the 3458A to it within 24 hours.

About the EMI filter, as I mentioned the old one took a flight and there is now a brand new Schaffner there. Whenever I get a new box I always change those filters. And all yellow cased X- and Y-class caps too.

And as for the cheapness of the meter, it cost $95. Add shipping and repairs and it is already at several hundred dollars now. But still a very good deal, assuming there is nothing else wrong with it. If it starts drifting, then, well...


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version