I'm trying to use an ADS8504, a 12-bit ADC, and get as good a resolution as possible. The ADC accepts inputs in the range of +-10V.

However, my circuit is limited to only having a power supply of 5V via USB (I've already manage to produce a -5V source), as such, even though I'd want to amplify the desired signal to +-10V, the maximum I'll get is +-5V.

So my question is if there is any way to change the input range of the ADC so it will only consider +-5V as the conversion range. i.e: go from 10V being +2048 and -10V being -2048 to +5V being 2048 and -5V being -2048.

I've gone through the datasheet and haven't been able to figure out a way to do it. There is a part that talks about calibration, but I'm not quite sure if the trimming circuit provided does what I'm looking for or if it's for something else entirely.

If it is possible to remedy the same issue through software rather than hardware, it would also be good to know.

Any solutions are appreciated.

  • \$\begingroup\$ Welcome to the site. \$\endgroup\$ – mike65535 Oct 30 '18 at 17:46
  • \$\begingroup\$ Unfortunately, it seems like you can't do that with this ADC. Is there some reason you can't use a different one? \$\endgroup\$ – Hearth Oct 30 '18 at 17:51
  • \$\begingroup\$ There's not really a reason for it, other than cost and time constraints (I need to order them from the US and wait 1-2 weeks for arrival). So if possible I'd like to use this one which I already have. If it isn't possible, I'd still need 12-bit resolution and preferably parallel output. \$\endgroup\$ – R. Montalba Oct 30 '18 at 17:59
  • \$\begingroup\$ My mistake--it seems you actually can adjust this ADC's range that far, but it will no longer meet the linearity specification, meaning that it might not meet your needs. See Dean Franks' answer. \$\endgroup\$ – Hearth Oct 30 '18 at 18:03

If you use an external 1.25V reference voltage you will reduce the full scale input range to +/- 5V. Precision ADC references require more than a resistor divider and some of the performance characteristics might suffer with the reduced Vref and input range.

  • \$\begingroup\$ Actually, page 15 of the datasheet linked in the original question says that the range for the external reference is 2.3 V to 2.7 V, so it's more for trimming than anything else. \$\endgroup\$ – Hearth Oct 30 '18 at 17:50
  • \$\begingroup\$ My mistake--earlier in the datasheet it says that this is the range in which it meets its linearity specification. So it could probably be used outside that range, but with no guarantees on linearity. \$\endgroup\$ – Hearth Oct 30 '18 at 18:04
  • \$\begingroup\$ Yes, that is where you get into the "performance characteristics might suffer" zone. Chances are that the impact is not too high at VRef of 1.25V. \$\endgroup\$ – Dean Franks Oct 30 '18 at 18:54

There is one way to increase your resolution without analog changes, assuming that you (1) can afford an increased sampling rate, and (2) can afford the extra computational load. Do a search for "dithering" or "stochastic resonance" as the physicist call it.

Assuming a noisy input signal (mostly a given, but I have added analog noise in some of my applications), doubling your sampling rate and using an appropriate digital filter to decimate back down to your desired sampling frequency, you gain one bit of resolution. Quadruple your sampling rate, and you get two bits.

That gain assumes a proper filter, but something as simple to implement as an average or a first-order IIR would get you more than half the way there.

This is partially the way that Sigma-Delta ADCs operate and give you those 24-bit converters at a low cost.


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.