Adjusting a 3456a to a 34401a

[saturation]

You're welcome.  Yes, a voice humanizes the bland name, isn't it?  I like the dark as it covers a lot of my lab's dirt  but it also focuses my attention to whatever is lit.  The videos are a bit darker than life due to my cellphone camera. 

I just uploaded this video I just took to show the nV resolution of the 3456a:

Thanks for the video saturation. I can now put a voice with the name.  BTW I usualy work in the dark also

[robrenz]

What portion of that SD is the voltage source and what portion is the 3456?

Just for fun I fired up the 8846A and a PD2010 to see what it looks like after 11400 samples. (in the dark of course )

We will have to wait about 5 hours for the picture at 100NPLC

10nV resolution but SD is 10 times worse with the PD supply as a standard

[saturation]

Good question, difficult to say, the meter is at/near/over  it spec sheet limits, and the voltage source, MV106, is over its spec sheet; however the MV106 has a claimed resolution of 10nV which I've yet to see given the limits of my equipment.  However, good designers should report a conservative value in spec sheets for gear.  Would be interested in your report.  I've also tried different PLC settings on the 3456a, and it does improve on SD values a little bit, suggesting some variation is still from the 3456a.

Nevertheless, the video hopefully shows a 3456a still provides excellent capability as a DC DVM despite its being 30+ years old and that acceptable DIY calibration can be done with some perseverance.  Depending on what is costs in your country, 3456a is often a low cost DVM that provides stable 1uV resolution easily and can be accurately calibrated by owners, also for low cost.  Not to mention general ease of repair, should it happen, making the TCO, fairly affordable at least to USA owners [ because I'm most familiarly with the going prices].

What portion of that SD is the voltage source and what portion is the 3456?

Just for fun I fired up the 8846A and a PD2010 to see what it looks like after 11400 samples. (in the dark of course )

We will have to wait about 5 hours for the picture at 100NPLC

[king.oslo]

How could you calibrate against the geller board? How did you get the 100V?M

[quarks]

10nV resolution but SD is 10 times worse with the PD supply as a standard

That's interesting. I just try to do the same with my (new/old) Knick JS300. It is quite difficult to hit 10,000mV with the single turn pot. Now I think of pimping it with a 10-turn pot.

Attached is the actual data. Is this worth letting it run longer?

edit: it does not seem to change much (see second + third picture), I stopped it after that

[robrenz]

Have you done a long term drift check of your meter with shorted inputs to characterize its behavior? Without that you dont really know what is meter and what is calibrator.

[quarks]

I stopped the other test. Now I have shorted the inputs and started it again. For how long would you let it run?

[robrenz]

I have done it for a week and I take pictures at 1, 2, 4, 8, hrs and each 24 hr period after that. I use 6.5 digit 100NPLC.  Make sure your lab temp is reasonably stable.

[quarks]

Uh that is a long time. I don't think I can do it that long soon, because I need the meter every day. What was your result after all?

I will let it run and have a look tommorrow morning. Right now this test already shows SD around 50n.

[robrenz]

Min. Max. span of 1.08µV after 152 hours.  You only need to test as long as care about. If you want to check a power supply stability for a week you need to know the meter stability for a week. If you only care about an hour check it for an hour.

[quarks]

I now had it running all night. That is long enough for now. The temperature dropped at night about 1.5 °C in my lab room. Min - Max difference is around 0.89µV and SD about 78.9n.

[saturation]

The Geller SVR provides 10.000 00 VDC, but is divided by a precision resistor divider to check or adjust the lower ranges of the 3456a.   You need at least a calibrated ohmeter to calculate the accurate ratios of the divider.

Once the lower ranges are calibrated, I used the 3456a ratio mode to take a stable 100.000 0V source [ or are best as you can obtain close to 100V, but best stable to 100uV] and ratio it against the Geller board.  The ratio output is then scaled to <= 10V, which uses the just calibrated range of the 3456a.   Using this value x 10, you can now adjust or compare the 100V range of the 3456a.

I initially used a linear PSU, but found stacking 9V batteries x 10 or 11 provides a stable ~100V output down to 100uV.

https://www.eevblog.com/forum/projects/poor-man-calibration-facility/msg97350/#msg97350

How could you calibrate against the geller board? How did you get the 100V?M