FW controlled range switching for ADC input

[Mark]

If you were designing a data acquisition system with the following specification:
Ranges (full scale): 1.414VRMS, 10VRMS, 150VRMS, 300VRMS
Sample rate = 30kSPS
BW: DC - 7.5kHz
Accuracy = 0.1% of reading from 3%-100% of range. 
ADC is 24 bits, differential input +/- 2.5V centred on 2.5V common mode. 

What would be the best way to implement FW controlled range switching? 
Analog mux/Analog switches/Electromechanical relays?   
In the past we just used jumpers, but now it should be FW controlled. 

Thank you for any ideas or solutions! 

[David Hess]

Are there any requirements for input impedance or resistance?  I assume Vrms means AC, positive, and negative inputs.

How much overload protection is required?  Does the 1.414 Vrms range need to survive 300 Vrms applied to it?

What supply voltages are available for the input signal conditioning circuitry?  Is there +/- 15 volts available?

I would tend to use a high input impedance divider, like that found on a digital voltmeter or multimeter.  This allows for exceptional overload capability, but requires capacitive compensation to achieve a 7.5 kHz bandwidth.

If the input resistance/impedance can vary with range, then there are some simplifications which can be made with the input circuitry.

[dobsonr741]

Design the input protection so when the autorange parks in the lowest range it can safely take the highest voltage you expect (round up to 500Vrms ->700Vpeak)

As said in the previous reply, define the input impedance your DUT and accuracy requirements tolerate. Somewhere in the range of 100K...10Mohm for practical reasons. It will drive whether a CMOS/MOSFET switch can survive the stress as above or you need a mechanical relay.

There were many posts here on the forum about bootstrapped input buffer and protection diodes for analog frontends.

Do frequency compensation for the divider, segment the first resistor to multiple ones to tolerate the max voltage, or use a ready made and high voltage spec'd divider, like a Caddock.

[1audio]

This thread may give you most of what you need: https://www.diyaudio.com/community/threads/autoranger-for-soundcards.299635/ Its based on the input circuit of the Boonton 1120 audio analyzer. It addresses most issues.

[dobsonr741]

The audio autoranger targets 1V or so output, so a sound card can be at its optimum SNR.

In data acq/instrumentation the ranges are not so small, usually decades wide, not down the the dB as the autoranger project implemented.

I did not find schematics or source code of the audio autoranger, did the project ever publish it?

[1audio]

I believe he has shared all of it. And he would share the source, Gerbers etc if they help. Your application is different but simiar enough that it could get you pretty far. PM me and I'll connect you.

[David Hess]

I was thinking about it and for a simple design which does not require high impedance, and has bipolar supplies available, I would use an inverting amplifier with gain selection controlled by changing the feedback resistor.  This moves all of the gain switching to the low voltage side, and makes the common mode range of the operational amplifier zero.  This configuration is much simpler than capacitively compensated high impedance attenuators as would be found on a digital multimeter.  Bandwidth varies with gain selection but this could be roughly adjusted with feedback capacitors across the gain resistors.

If a differential input is required, then there is a similar circuit which uses a high common mode range instrumentation amplifier.