I have a DAQ with many inputs that can measure within +-10 V, and I want to go up to +-30 V. Using a resistor voltage divider could be an option, but this lowers the input impedance. I would prefer using an op-amp.

Is there any device that can divide all inputs with high impedance?

Like an array of 16 or more op-amps?

Or do I have to design it myself?

  • \$\begingroup\$ Use a (high voltage) analog switch/demultiplexer and feed the output to non-inverting an opamp? \$\endgroup\$ – Huisman Feb 12 '19 at 14:05
  • \$\begingroup\$ all channels are sampled in parallel, then I would need as many amps as channels \$\endgroup\$ – Rémi Baudoux Feb 12 '19 at 14:23

You're in fairly rarefied territory with voltages in that range, and you'll get some mediocre DC specs and high prices along with a lot of power consumption.

For example, if you use +/-36V supplies with 32 OPA454 chips, you'll have 4mV Vos maximum, and total power consumption typically in the 7W range (which won't help the offset voltage drift during warm-up). That doesn't count the power consumption of the dividers on the op-amp outputs. The LTC6091 is another possibility, but the chips alone would be in the $400 USD range.

Be careful of the loss in bandwidth with using dividers (with or without the op-amp) if that matters at all to you. For example, if you used 200K/100K dividers, divider input impedance would be 300K and output impedance of the divider 66K so the settling time would be in the 150us-200us range. Error due to that divider output impedance should be more than acceptable (66K/10G * 2^16 = 0.4 LSB).

You might want to consider just buying a more appropriate data acquisition subsystem.

  • \$\begingroup\$ indeed an op-amps array not a very fancy solution. I did not find any DAQ meeting these requirements unless assembling more than 5 modules in a rack, which is not very handy either. Thank you for your remark on the bandwidth, I forgot to consider that point \$\endgroup\$ – Rémi Baudoux Feb 12 '19 at 16:29

You have to design it yourself. Linear Tech (ADI), Burr Brown (TI), Apex (?) and others make high voltage opamps that will run on the +/-36 V or so that you would need for a +/-30 V output. Some are available in duals; don't know about quads.

Another approach is to use the DAQ input impedance to your advantage. Depending on the input circuit, it might be low enough and repeatable enough to form the shunt leg of an input attenuator. Do you know what the input impedance value and tolerance are? Also, what is the minimum load impedance your signal sources can handle without affecting accuracy?

  • \$\begingroup\$ I was thinking along the same lines but would the ADC input not go open-circuit during switching from one channel to another and thereby disconnect the "shunt leg" resulting in full input voltage being applied to the chip? \$\endgroup\$ – Transistor Feb 12 '19 at 14:11
  • \$\begingroup\$ The device is a USB-6353 (NI, ni.com/pdf/manuals/374592d.pdf ), it says input impedance > 10 Gohms, but no accurate value. I'll check for the minimum required impedance \$\endgroup\$ – Rémi Baudoux Feb 12 '19 at 14:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.