Getting a metrology'y feel(quasimetrology)

[nnills]

Hi,

To imitate the idea of real metrology(multi k$ equipment) I plan on building 10 bad voltage reference so that my 34401A can measure a good amount the drift. The refences will consist of a 12V regulator and a tl431 with a divider of 3k:1k to get 10V(and a pot for trimming). To *really* get the metrology'y feel I want to buid a scanner and program my RPI to scan, measure and plot the values as they drift.

The question is: are 8:1 analog muxes any good for (in my case)quasimetrology?

Nils

[tggzzz]

Hi,

To imitate the idea of real metrology(multi k$ equipment) I plan on building 10 bad voltage reference so that my 34401A can measure a good amount the drift. The refences will consist of a 12V regulator and a tl431 with a divider of 3k:1k to get 10V(and a pot for trimming). To *really* get the metrology'y feel I want to buid a scanner and program my RPI to scan, measure and plot the values as they drift.

The question is: are 8:1 analog muxes any good for (in my case)quasimetrology?

Nils

That's like asking "are opamps any good for quasimetrology?"! The specification details and (especially for metrology) implementation details matter.

[guenthert]

  Get one additional fairly good 10V reference (something with an LT1021 or AD586, AD587, AD588 in it -- don't waste your time on the cheap AD584 ones) and a ADS1256 module off fleebay.  Then you can measure the difference to four of your 'bad' references simultaneously with sub-microvolt resolution w/o having to power and occupy your 34401A for weeks.

[Conrad Hoffman]

I'd suggest it's more useful to build 3 or more decent references as described above, then compare them with each other.

[nnills]

Thank you for your answers. I do however have 2 decent voltage reference(LT1236A and LM399 with 10V gain). The "problem" is that these only move a little, the lm399 only moved from 10.00003 to 10.00002 and back over 4 weeks. The LT1236 moves more. I could use that ADS1256 ADC with the one of those but it only accept up to 5V for Avdd.

I was always under the impression that analog switches had no offset voltage, that's why I didn't feel the need to specify a part.

[iMo]

Thank you for your answers. I do however have 2 decent voltage reference(LT1236A and LM399 with 10V gain). The "problem" is that these only move a little, the lm399 only moved from 10.00003 to 10.00002 and back over 4 weeks. ..

FYI - this is how an LM399 10V fluctuates (in ppm). Measured with 34401A today.
Mind the 34401A is using an LM399 as the Vref too..

PS: the dip at the beginning is because I opened the box with the 399 Vref to see the temperature change response.

[nnills]

How did you measure the fractional ppm's. I know that the 34401A outputs way more digits on the rs232(and probably GPIB) but are these in any way useful?

[iMo]

You get 1 digit more, and it seems to me it is useful. With Rpi connected to RS232 you may process the incoming data (ie adjust for DMM's offset, gain, temperature, and apply various filters)..

[tggzzz]

I was always under the impression that analog switches had no offset voltage, that's why I didn't feel the need to specify a part.

That will entirely depend on the device - read the data sheet, or for more general information The Art of Electronics or, the Tektronic Low Level Handbook.

If your analogue switch was a relay, then clearly there are metal junctions and hence the Seebeck effect will be relevant.

[Kleinstein]

The usual CMOS switches and MUX (e.g. DG201B, DG408 ) are quite good in the DC way.  They have very low offset voltages and for the modern versions also low leakage.  Like with mechanical switches there can be thermal EMF if there are temperature gradients. As the CMOS switches are low power the internal gradients are small.

Just keep in mind the limitation that the voltage at all pins must be within he supplies. It can become tricky if a part is fully isolated with no defined common mode voltage.

For the references the TL431 is really low quality, more like ok for a power supply but hardly for a measurement. It could still be interesting to measure a few for learning, but don't expect too much.

Trimmers are notorious for drift and high TC. So if used at all try to limit the effect they can have, e.g with series / parallel resistors.

[nnills]

I am aware of the very limited quality's of the TL431. But that is exactly why I chose it. It is only meant for learning. Graphs, uncertainty's and more. I expect it to drift a LOT.
As for the ground potential of the analog switch, I intend to have all the references connected to one supply.

[Theboel]

I am aware of the very limited quality's of the TL431. But that is exactly why I chose it. It is only meant for learning. Graphs, uncertainty's and more. I expect it to drift a LOT.
As for the ground potential of the analog switch, I intend to have all the references connected to one supply.

why not use plain zener diode ?

[nnills]

I am aware of the very limited quality's of the TL431. But that is exactly why I chose it. It is only meant for learning. Graphs, uncertainty's and more. I expect it to drift a LOT.
As for the ground potential of the analog switch, I intend to have all the references connected to one supply.

why not use plain zener diode ?

I could do that... Maybe using 10V zeners or even 6Vsomething. But in what respect are they better?

[Theboel]

I am aware of the very limited quality's of the TL431. But that is exactly why I chose it. It is only meant for learning. Graphs, uncertainty's and more. I expect it to drift a LOT.
As for the ground potential of the analog switch, I intend to have all the references connected to one supply.

why not use plain zener diode ?

I could do that... Maybe using 10V zeners or even 6Vsomething. But in what respect are they better?

bigger TC