One way to deal with negative analog input voltages

[GTBecker]

I need to accept two analog signals (from an exotic positive-ground instrument) that range between -0.8 and -8.0v. In the past I used a bipolar +/- 10v-input 12-bit ADC ISA board in a laptop. $450 for the card, as I recall.

The function the laptop performed will be done with a ZX-24n this time. The 10-bit conversion is adequate, but I still need to accept the negative signals. At the risk of being obvious, one solution is simple; divide the input and subtract it from the 5-volt rail. Code can then right and rescale it.

Edit: I added a diode to block an accidental positive input, and changed a value to reflect the source impedance.

[twesthoff]

First two questions.
1) What is the output impedance of the neqative voltage signal?
2) How accurate do you need to be?

Tom W


On 9/17/2011 10:56 PM, General wrote:

[GTBecker]

[twesthoff]

I'm traveling now, sitting in a hotel room, so I can't test this first, but an op-amp summing circuit should give you a voltage the A/D converter can read, and invert the value as well.  You could use a single supply bias if you don't have a negative power supply voltage handy, then use a value of summing resistor for the RF field strength input that is high enough (>10K) so it won't load the input source.  Don't use the capacitors on the inputs so it works for DC.

In this case you will be summing a positive voltage ((+5V)with a negative one (RF signal strength).  By choosing the correct resistor values, you can get the output of the op-amp to be the input range of your A/D.

At this link see 2.3
http://www.eng.yale.edu/ee-labs/morse/compo/sloa058.pdf

Tom

[GTBecker]

Yes, an opamp could be used to scale and invert , but what advantage would that provide over simply subtracting the scaled voltage from the supply (and ADC reference) rail with two resistors as I drew? My only concern is noise on the +5v rail, but I think a smoothing RC is all that might be necessary to keep the noise below ~2.5mV, half a bit at 5v/10bits.

[twesthoff]

No advantage. Sorry I was getting the e-mail version and it did not show the diagram you drew. I logged in to the forum and saw your diagram now.
Either way should be about the same. As you say the 5V could be filtered or readings averaged to get past that.