Author Topic: Beginner's question on simple reference/DMM  (Read 446 times)

0 Members and 1 Guest are viewing this topic.

Offline nnills

  • Contributor
  • Posts: 14
  • Country: nl
Beginner's question on simple reference/DMM
« on: October 01, 2019, 08:53:04 pm »

I recently built up a super simple 10V reference circuit using a LT1236. It is connected using [name escapes me] clips at the end of cat5 networking cable to a screw-type 2 pin banana block. I measured for 24 hours on my HP 34401A(with unknown calibration) on the best settings(6Dig slow >10GOhm) with the integrated min max feature. Min:9.999934V Max:9.999991V dV:57uV But when looking at the specifications of the HP I saw that the 24H spec was 15+5 ppm... Does this mean I don't know anything about the quality my reference?

I think that the fact that the reading moves so little could have multiple explanations:

1) The two devices drift in the same directions with the same speed. However, the reference has a typical long term stability in the datasheet of 30ppm/1000h, much less than the spec of the DMM at 15+4ppm/24h...(but not much better than the 90 day spec at 20+5)

2) The multimeter is much more stable than the spec(which I expected. are the some figures on real meters?)

3) Coincidence of TC and thermal EMF due to the fact that there is no enclosure, bad connections and that the room temperature moves form 20C to 24C.

But feel free(better still, please) add your own suggestions.

Questions: Can I say something about the stability of my reference? Are there figures on the average stability of HP34401A's? (Is there a way that is not to expensive to have higher precision than the 34401A?)

Online imo

  • Super Contributor
  • ***
  • Posts: 1755
  • Country: 00
Re: Beginner's question on simple reference/DMM
« Reply #1 on: October 01, 2019, 09:27:12 pm »
The tempco stability of your HP is at least 5x better than of the naked LT1236 (statistically). You may also measure the ambient and internal temperature of the HP (I do) and make the regression (ie in excel) of the 1236 voltage against Tamb and against Tdmm. Also mind the HP34401 needs aprox 5-6 hours to stabilize its internal temperature, then Tdmm follows the Tambient changes with a propagation delay of aprox 20min - 1h based on the air speed around the box. Also my epoxy LT1021s (similar to the 1236) need several days (or more) to stop walking after soldering or after an IPA bath with subsequent hot air (hair dryer 50-60C) drying. Your 1236 must be mounted such the temperature or mechanical stress at its package is minimal (or none). No sockets. Best you will be logging the data coming from the HP's RS232 or HPIB and looking at the graphs - you may have voltage jumps there coming from various sources.
The 34401 is usually aged enough, therefore the HP's y2y stability should be much better than the 1236. Tempco of the HP34401's Vref is <= 0.5ppm/C.
Therefore - you can hardly judge unless you do the measurements carefully enough..

PS: below a measurement you may get with less luck..
« Last Edit: October 01, 2019, 10:41:31 pm by imo »

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 6219
  • Country: de
Re: Beginner's question on simple reference/DMM
« Reply #2 on: October 02, 2019, 07:47:11 am »
The accuracy specs for the DMM include more than just drift at a relatively constant temperature. The short time drift is only a small part of the accuracy limits. The meter drift must be much smaller than the 24 h specs. I don't have an 34401 (and have not even used one), but the expected drift and noise variation from the meter are more like in the  < 10µV range unless the temperature changes a lot or not yet fully warmed up.

A 50 µV difference from min to max  value is likely to a large part from the LT1236 reference. This can be both noise and drift with temperature. The actual curve and additional data on temperature can tell the difference.

Offline Andreas

  • Super Contributor
  • ***
  • Posts: 2385
  • Country: de
Re: Beginner's question on simple reference/DMM
« Reply #3 on: October 02, 2019, 06:09:57 pm »
10V reference circuit using a LT1236.

Which one exactly?
there are 3 packages (SO-8, DIP8 and LS8) with different performance and grades. (A-D)

of 30ppm/1000h, much less than the spec of the DMM

Be carefully with ageing specs: usually they are only valid for the "best package" (metal can case) which was ever available even if it is obsolete now.

The ageing specs for the reference is only valid for the first 1000 hours and at constant chip temperature of 25 deg C.
Whereas the DMM specs are either for 18-28 deg C or for 23+/-1 deg C ambient temperature.

My 5V LT1236 have a "typical" T.C. of -3 ppm/deg C. So if you want to have a stable reference you will need to heat it to a constant temperature.
Then after some few kHrs a buried zener reference at constant temperature will drift typically about 1-2 ppm/year similar to a well aged 34401A DMM.

with best regards



Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo