Messuring volts with an integrator and binary counter, or analog; peak detector

[ELS122]

Could I just use an integrator for measuring volts. Then when it reaches some level it resets the integrator as well as parses the value of a binary counter that was counting since the last reset. Then the value representing the time the integrator took to charge to whatever level. Feed integrator capacitor with an V/I converter.
This way you could also measure very low currents or high resistances.

Or for an analog output, use a precision peak detector from the output of that integrator, resetting it together with the integrator.
Ranges could be changed by changing the V/I ratio. and/or changing the integrator capacitor.

[TimFox]

That is one traditional way to make a digital voltmeter, usually called "dual-slope integration".
https://www.eeeguide.com/dual-slope-integrating-type-dvm/
Done properly, many errors cancel out (such as the tempco and tolerance of the input resistor and the integrating capacitor, along with the clock frequency), to perform an accurate measurement of the ratio between the input voltage and the reference voltage (so long as the above parameters do not drift during the total time taken by the two slopes).

[Swainster]

I believe the OP is describing a single slope ADC, which requires a very stable and accurate resistor and capacitor. I have a nixie tube DMM which uses this method. There is no voltage reference required (at least not in the ADC). Typically this methods limits accuracy to 3 digits or less, which is still an improvement on most of the analogue meters that it replaced.

In a multimeter, modifications to measure very low currents or high resistances is usually done in the signal conditioning stage, before the ADC. Some specialist instruments may integrate the input signal directly - e.g. some ion counters.

[jonpaul]

used rundown ADC in 1968, 711 comparaor and RTL logic in 20 kHz 500 pt RTA

j

[bdunham7]

I believe the OP is describing a single slope ADC, which requires a very stable and accurate resistor and capacitor. I have a nixie tube DMM which uses this method. There is no voltage reference required (at least not in the ADC). Typically this methods limits accuracy to 3 digits or less, which is still an improvement on most of the analogue meters that it replaced.

Wouldn't a single-slope ADC have a variable reading interval, with inputs near the bottom of the scale taking a very long time?  I don't see how this would work out in practice, do you know how yours does it or can you post the exact model so I can read about it? 

[bdunham7]

Could I just use an integrator for measuring volts. Then when it reaches some level it resets the integrator as well as parses the value of a binary counter that was counting since the last reset. Then the value representing the time the integrator took to charge to whatever level. Feed integrator capacitor with an V/I converter.
This way you could also measure very low currents or high resistances.

The problem I see with a too-simple approach is that very low readings would approach infinite integrator time.  I suppose you could use this to improve accuracy at very low readings without scaling down, but your input and related circuitry would have to be very good--low leakage, etc--to make that work.  The typical 'dual-slope' technique is to charge the integrator for a fixed period of time using the binary counter and a gate, then discharge the integrator using a precision fixed reference and count the time that it takes to get back to zero.  This gives you a very linear system almost without any effort because a lot of things cancel out.  I have an old Nixie-tube display 4.5-digit DMM that implements this system entirely in simple logic chips (DTL + some TTL) with remarkable accuracy.  It charges the integrator for 10,000 counts and then discharges it with a precision reference current derived from a tempco-controlled zener for as many counts as it takes to cross zero. 

[Swainster]

Wouldn't a single-slope ADC have a variable reading interval, with inputs near the bottom of the scale taking a very long time?  I don't see how this would work out in practice, do you know how yours does it or can you post the exact model so I can read about it?

You are, of course, correct - I dont know where I got the idea that it was single slope... it may be something I aquired before I got more interested in how they work rather than how they look... mmmm nixies. Although it turned out to be a red herring, the model is the Fairchild (Systron Donner) 7050.

[iMo]

In LT datasheets there are at least 2-3 examples of relatively high resolution V/F converters. One simplest (from the LTC1043 datasheet) you may see below. As a teen I built myself a voltmeter with a simple V/F converter and a 3 digits LED counter (556 being the clock and control)

[iMo]

.. and even better one V/F converter from the LTC1052 DS..

[bdunham7]

You are, of course, correct - I dont know where I got the idea that it was single slope... it may be something I aquired before I got more interested in how they work rather than how they look... mmmm nixies. Although it turned out to be a red herring, the model is the Fairchild (Systron Donner) 7050.

Nice.  Mine is a Fairchild 7000A, 10k counts.