Homemade DMM again, with a scaled back set of specs.

[Marco]

Most designs are just slightly tweaked copies of designs unchanged for decades ... if Jim Williams says he can get a LTC2400 to 1ppm I think you can get a LTC2400 to 1ppm.

I think the "inherent" linearity of multislope is a bit of a joke considering the amount of voodoo which goes into dielectric absorption compensation.

[macboy]

LM399 will make charging such a meter rather usual thing, with  it's ~20mA consumption.
It's way too much for just a 4.5-digit DMM.

The LM399 can be handled this way: the LM399 module is only turned on when an internal calibration is about to happen. The heater current is monitored using a sense resistor, and once the current draw of the heater stabilized to within 1 LSB of the built in ADC of the uC for about a minute, a measurement is taken. Since such an internal calibration is not performed only once in a blue moon to calibrate the long term stability of the secondary reference internal to the ADC, this can be performed only when the unit is powered from USB to save battery life.

Or I can always use ADR02B as primary reference, which is just as good as LM399.

No, the microcontroller's reference doesn't only drift over long periods of time. It will also drift significantly with temperature, variation in Vcc, even humidity. Definitely use a quality external reference. There are plenty available that will get you into the tens-of-ppm stability/repeatability without much expense. That will be orders of magnitude better than any uC internal Vref but not quite as good as the LM399.  The LM399 is overkill for 4.5 digits; it is used in most 6.5 digit (HP 34401A, Keithley 2000, etc. etc.) and even 7.5 digit (Keithley 2001, HP 3457A) meters.

[Kleinstein]

The LM399 is overkill for a typical 4,5 digit meter. It might be useful as an external reference source to check the meter and possibly others from time to time, like once a month or to check how bad the ref. inside the ADC chip (usually not inside a µC) is. For a 4,5 digit meter a simpler ref. might be good enough. The cheap ref chips in plastic case may have a problem with the influence of humidity, but this a rather slow effect over many hours and days. The same may also apply to many resistors.

Bringing down the power of the LM399 by insulation has it's limits. The LM399 even without the heater active needs some power of something like 10-15 mW. If thermal resistance gets to high, temperature stabilization will not work at high environmental temperature, or the temperature needs to be set a too high a value. So it will get hard to bring the typical power to less than about 3 times the minimum power. If the long time reference does not need to run all time, power dissipation is not that critical any more. At low power and if the requirements are not that high there are also good band-gap references - preferable in a metal case.

Good sigma delta converters can give something like 2-20 ppm INL. Some may even allow for correction in software, as the shape seems to be rather predictable and fixed. This is well good enough für 4,5 digits, 5 digits and may even work for 6 digits. At the low ppm levels linearity is not that simple - even resistors may not be that linear any more. At least some of the SMUs already use sigma delta chips instead of a multi slope converter at 6 digits.

The multislope mechanism should not need that much compensation for dielectric absorption: only a small fraction of the resolution is determined from the change in voltage of the capacitor. As the capacitor is much smaller (e.g. 1 nF instead of around 500 nF in a dual slope) it's easily possible to use capacitors with low DA (e.g. NP0, mica or teflon). Still linearity is influenced by similar effects as in the sigma delta chips: voltage dependent charge injection, leakage, voltage dependent resistance of FET switches. This is not a surprise, as there is not that much difference between high resolution sigma delta converters and multi-slope converters. The difference to the classic dual slope is larger.

If you want the option of running on battery (even if rechargeable) I would definitely not use a multi chip, multi slope converter - this part alone would likely use way more power than the LM399. The sigma delta converters are really low power in contrast (e.g. 1 mW for an LTC2400).

[technix]

The LM399 is overkill for a typical 4,5 digit meter. It might be useful as an external reference source to check the meter and possibly others from time to time, like once a month or to check how bad the ref. inside the ADC chip (usually not inside a µC) is. For a 4,5 digit meter a simpler ref. might be good enough. The cheap ref chips in plastic case may have a problem with the influence of humidity, but this a rather slow effect over many hours and days. The same may also apply to many resistors.

Bringing down the power of the LM399 by insulation has it's limits. The LM399 even without the heater active needs some power of something like 10-15 mW. If thermal resistance gets to high, temperature stabilization will not work at high environmental temperature, or the temperature needs to be set a too high a value. So it will get hard to bring the typical power to less than about 3 times the minimum power. If the long time reference does not need to run all time, power dissipation is not that critical any more. At low power and if the requirements are not that high there are also good band-gap references - preferable in a metal case.

Good sigma delta converters can give something like 2-20 ppm INL. Some may even allow for correction in software, as the shape seems to be rather predictable and fixed. This is well good enough für 4,5 digits, 5 digits and may even work for 6 digits. At the low ppm levels linearity is not that simple - even resistors may not be that linear any more. At least some of the SMUs already use sigma delta chips instead of a multi slope converter at 6 digits.

The multislope mechanism should not need that much compensation for dielectric absorption: only a small fraction of the resolution is determined from the change in voltage of the capacitor. As the capacitor is much smaller (e.g. 1 nF instead of around 500 nF in a dual slope) it's easily possible to use capacitors with low DA (e.g. NP0, mica or teflon). Still linearity is influenced by similar effects as in the sigma delta chips: voltage dependent charge injection, leakage, voltage dependent resistance of FET switches. This is not a surprise, as there is not that much difference between high resolution sigma delta converters and multi-slope converters. The difference to the classic dual slope is larger.

If you want the option of running on battery (even if rechargeable) I would definitely not use a multi chip, multi slope converter - this part alone would likely use way more power than the LM399. The sigma delta converters are really low power in contrast (e.g. 1 mW for an LTC2400).

I am aiming for a cheap, simple construction that works well enough for day-to-day use yet have better precision and construction than those El Cheapo meters, not volt-nuttery. The Sigma-Delta method is under investigation currently.

The LM399 internal reference's power consumption is limited by turning the entire portion of the circuit off when unused. Its usefulness is also extended by allow outputting this reference voltage from the meter (since I do support 4-wire resistor measurement, there is a set of output jacks that is otherwise unused) so I can use this reference to calibrate other things.

[Kleinstein]

Depending on the source where you get parts from, the LM399 may not be so much more expensive than another suitable reference in a metal case.
So it may be much better than needed and higher power, but still not overly expensive. Especially when using setup with a normal working reference at the ADC and an additional one to check drift, it's very easy to change the reference. So this is the least problem in the design.

The bigger questions are how to do auto-ranging and switching ranges.
Also the choice of the amplifier for the voltage ranges has several options. A supply higher (e.g. +-15 V) than found in usual hand-held meters would allow using CMOS switches for much of the functions, but lead to a little more power usage. With a rechargeable option this may be still viable.
Using the inverting amplifier setup may have trouble (due to leakage)  in the low voltage ranges
so if accuracy is important, I would still consider a relay to switch the input divider, similar to the input stages of high end DMMs.
Also switching current ranges has several options - if the ADC has enough resolution, or switchable amplification one might even use only 1:100 stepped shunts.

So the next step would be to draw a few general design possibilities.

[technix]

Depending on the source where you get parts from, the LM399 may not be so much more expensive than another suitable reference in a metal case.
So it may be much better than needed and higher power, but still not overly expensive. Especially when using setup with a normal working reference at the ADC and an additional one to check drift, it's very easy to change the reference. So this is the least problem in the design.

The bigger questions are how to do auto-ranging and switching ranges.
Also the choice of the amplifier for the voltage ranges has several options. A supply higher (e.g. +-15 V) than found in usual hand-held meters would allow using CMOS switches for much of the functions, but lead to a little more power usage. With a rechargeable option this may be still viable.
Using the inverting amplifier setup may have trouble (due to leakage)  in the low voltage ranges
so if accuracy is important, I would still consider a relay to switch the input divider, similar to the input stages of high end DMMs.
Also switching current ranges has several options - if the ADC has enough resolution, or switchable amplification one might even use only 1:100 stepped shunts.

So the next step would be to draw a few general design possibilities.

The higher +/-15V voltage is not likely to be used in the feedback switched amplifier matrix or I won't be able to do low-cost range switching using d-pots, which is effectively a tempco-matched resistor array coupled to a CMOS switch array on a chip, with surronding support circuitry like shift registers, decoders and latches built in. Also one of the amplifier choices, TLC2272, a CMOS RRIO op amp, won't really work with higher supply voltages.

[Kleinstein]

There are a view general circuit layouts to choose from: One is switching the feedback of an inverting amplifier. Here you don't need a higher voltage - something like +-5 V might be enough. However this setup is weak in measuring low voltages, as leakage becomes really critical - more than 3.5 digits in the 200 mV range might get challenging in this arrangement.  The switches in digital pots might not be a good option, as leakage might be to high. There is hardly a way around some high quality switches, like relays, good CMOS or possibly JFets or Photomos. This setup also needs at least two (usually more) high quality resistors.

The other arrangement is the one high end DMMs use. Here one usually needs one relay, one high impedance divider (e.g. 9.9 M and 100 K)  and a relatively high supply voltage so the amplifier can handle something like 1/100 of the maximum voltage. Lower Voltage ranges (e.g. up to 10 V) might have very high impedance. Usually the lower ranges are very good - the higher voltages that use the divider have slightly larger errors. The main disadvantage of this method is the need for the higher supply and thus higher power - often to much for a 9 V block but still viable with rechargeable.

In both cases a normal CMOS Rail-Rail OP is not a good choice, as they have a double input structure and thus more bias / noise than normal OPs. Also DC drift and LF noise of CMOS OPs can be a problem. The more realistic choices are auto zero OPs or JFET based OPs.