[volt-nuts] Solartron 7150plus - Chasing AC gremlins

[Philip Pemberton]

On 24/12/12 16:03, Philip Pemberton wrote:
> On 24/12/12 15:54, David C. Partridge wrote:
>> Re: Drift of RMS converter - wait until after 24 hour warm up - does it drift after that? If not all is well. 
> 
> Well, after an hour of warming up I'm getting calibration errors again...
> 
> Lowpoint 0.01V RMS (10mV)
> Highpoint 0.2V RMS (200mV)
> Range 0.2V
> 
> Confirmed the input values with the DSO and a calibrated Agilent
> multimeter, but after entering these values the meter refuses to
> calibrate. Display shows "ERROR". :(
> 
> So it looks like I'm back to the original fault...

In other news, I've tracked down the fault!

Just to recap - the original fault was that the AC voltage measurement
would drift over time. My Agilent DMM reported 6.984V, the Solartron
reported 6.9540, then six minutes later reported 6.9620, an 80-count
difference. Switching from AC to DC and back again started the whole
merry dance again.

For bonus points, the RMS-to-DC converter was very temperature
sensitive, and would output utterly insane values if I put my finger
anywhere near it - touching the metal cover would make the output clip
to 13V.

Quick explanation of what I've replaced so far --

* IC21 (RMS-DC converter).
Didn't fix the drift, though its behaviour was weird enough that I
deemed it worth replacing for the sake of reliability.
I couldn't get the CERDIP version, so I bought the AD637BRZ (16pin wide
SOIC) and a Roth Elektronik RE932-05 adapter PCB. A 14-pin DIP socket
and some insulated solid-core "bell wire" converted it to match the DIP
footprint. It doesn't look pretty, but it works fine!

* IC18 (Analog Switch for AC buffer amp gain setting)
Original was a Siliconix DG211CJ, replacement is a Vishay-Siliconix
DG211BDJ. Replacing this didn't do a thing...

So what did fix it?

* IC5 (Analog Switch for AC signal routing and drift correction)
Another Siliconix DG211CJ. After I replaced this, the measured voltage
(with 200mV input) jumped from ~180mV to ~205mV. I suspect this part had
failed such that the on resistance was markedly higher than it should
have been. The current going into the A/D input was too low, resulting
in an erroneous reading which varied over time.
I suspect there's a "sanity check" in the calibration routine which
decided (quite rightly) that the calculated CAL factors were silly and
thus displayed the "ERROR" message - which can probably be translated as
"either your calibration source isn't working properly or there's a
hardware fault on this DMM."

On top of this, the reference voltage balance was badly out of
adjustment (spec is 300uV, I measured 3mV). This has probably impaired
the DC calibration too, as it sets the reference voltage for the A/D
converter. Given this and what I've done to repair the meter, it's
probably fair to say that the calibration on the DC ranges is
essentially worthless at this point.

I'm looking into the possibility of building a couple of high-stability
DC and AC sources to calibrate the meter (or at least check its
calibration against the other 7150Plus meter). I found a TI application
note (AN-263 "Sine Wave Generation Techniques") which shows how to do
this for AC using a Wien bridge oscillator, a valve output transformer
and a feedback loop. AC current is on the "need to figure out how to do
that" list.

Who knows, maybe I lucked out and the calibration isn't far off spec...
otherwise I need to find another £60 to get it recalibrated!  (erk!)

Thanks,
-- 
Phil.
philpem at philpem.me.uk
http://www.philpem.me.uk/