HP 3458A (and other HP and later DMMS) ratio mode

[calmaster]

Hi together,

this is my first post in this forum, but I'm reading since a long time already.

I always wondered how exactly the ratio mode works on meters like the 3458A and other HP meters. Even the 34401A has such a mode.

What is the range of the ratio input? I read it is limited to 12V, does that mean it is fixed to a 10V range? Or does it change the range automatically as soon as I connect only 500mV or 50mV?
What about the specifications of that range? Is it similar to the standard ranges? I asked, because I would like to calculate the uncertainty of such a ratio.

And is the transfer specification of the 3458A also valid for such a ratio measurement?
If that would be true, than I could simply connect my 732 to the ratio input and whatever I would like to measure to the standard input. No need for unplugging cables and so on.

Cheers
Rainer

[Kleinstein]

There are 2 ways how the ratio mode can be implemented. With many modern meters the inputs for ohms sense are used as the reference channel and the meter does 2 measurements: one for the reference and one for the voltage to measure from the normal inputs. Often this are are actually 4 conversions of the ADC, one for each side of the 2  voltages.

Some older meters did a ratio measurement in a way that the reference voltage to the ADC was actually derived from an external reference and than only one measurement is done.

Quite often the protection at the ohms sense inputs is a little more primitive with more added resistance and thus noise for protection. So the reference reading can add quite some noise.  The added resistance may also add a separate offset - though this should be corrected.  The range is also usually limited, as there is obvious no divider for a higher voltage. One may be able to use a lower reference, like 1.18 V -  this is more a software question. Obvious a 10 V reference would be preferred, as the amplifier in the DMM can add errors.  To a large part the DC specs should also apply to the reference input via the ohms sense input. However expect a little more noise (not many references are so good that this would be a real problem) and maybe a little higher bias current (should not matter with a reference). Details an extra specs  are rare. So this is kind of a gray area.

[MegaVolt]

For DMM7510:

[MegaVolt]

For 6500:

[MegaVolt]

As you can easily see, not every multimeter has a range switching for the Sense input. The 7510 has it. The 6500 has only one range.

[Dr. Frank]

...

I always wondered how exactly the ratio mode works on meters like the 3458A and other HP meters. Even the 34401A has such a mode.

What is the range of the ratio input? I read it is limited to 12V, does that mean it is fixed to a 10V range? Or does it change the range automatically as soon as I connect only 500mV or 50mV?
What about the specifications of that range? Is it similar to the standard ranges? I asked, because I would like to calculate the uncertainty of such a ratio.

And is the transfer specification of the 3458A also valid for such a ratio measurement?
If that would be true, than I could simply connect my 732 to the ratio input and whatever I would like to measure to the standard input. No need for unplugging cables and so on.

Cheers
Rainer

Hello Rainer,

the 3458A is the mother of all later DMMs, like the 34401, 410A/411A and 46xA/470A.
That means, that the Ratio function works always in a practically identical manner.
Your questions are all covered in the different manuals, also the Ratio specification, but anyhow I'll try to make these descriptions more transparent to you.

The Sense jacks act as the reference signal.
This is always autoranging DCV, i.e. 100mV, 1V or 10V, max input +/- 12V_dc

The Input jacks might be Autorange, or Manual, from 100mV to 1000V_dc on all DMMs.
Only the 3458A allows to also measure ACV signals in relation to the reference DCV signal, but that's a special case, which should not be discussed here.

The potential between Sense GND and Input GND jacks must be less than 250mV, that means that you somewhere have to tie both signal grounds together.
If you forget that, the ratio measurement will not work.
Also pay attention, if both your sources do not have independent ground signals, e.g. when they are both powered from the same supply.
In this case an additional connection between both ground pins might create a loop, or other problems.

The Ratio mode (two DCV signals) makes four consecutive measurements, i.e. it measures the voltages at Sense+, Sense-, Input+ and Input-, and then purely mathematically calculates the Ratio:

Ratio = (I⁺ - I^-) / (S⁺ - S^-)
     
From this formula it's evident, that the DMM does not need to do an additional internal Auto Zero cycle, and that it takes the same time as making 2 separate DCV measurements with Autozero.

Ratio has its own specification, which is worse than the transfer specifications, i.e. its:

± (Input error + Reference Error)
Input error = 1 x Total Error for input signal measurement function (DCV, ACV, ACDCV)
Reference error = 1.5 x Total error for the range of the reference DC input

That means in first instance that you would have to use the appropriate 24hr, 90day or 1year specifications.
If you'd do an ACAL before the RATIO measurement, you can at least use the 24hr specifications.
But I strongly assume, that in this case you could instead use the DCV transfer specification, for your case of fixed 10V ranges that would be  2x (0.05 ppm of range + 0.05ppm of reading).

There was a good white paper from KS dealing with these different transfer techniques, and how to optimize this ratio measurement .. I will look for it and link it down below.

I could have referred directly to this document, I think, as it's much better explained there 

Frank

[calmaster]

Hello all,

Thank you very much for the explanations!
With that information in mind I went again through the manuals and found the autorange of the sense input. That is very interesting, because a 1V to 10V comparison with a 34401A should then be done with 1V on the input and 10V on sense, otherwise one cannot guarantee that the same range ist used for both sources.

@Frank: Your explanation added at lot to the linked paper. Therefore, it was worth every line 

Btw: The HP 3456A is also able to do ratio between AC and DC. But it is older than the 3458A and therefore maybe not the line of instruments you spoke about.

Cheers
Rainer

[notfaded1]

@Dr. Frank sometimes your layman's terminology explanation really helps to make sense of some docs!   Especially the gotcha stuff you usually list along the way. 

Bill