AD623 min voltage in single-rail vs dual-rail supply mode

Hi,

Yes, the amplifier doesn't have GND pin and the supply voltage applied at -Vs determines if the amplifier has single or split supply.

From the data sheet, table 4, single supply operation range is 2.7 to 12V.
And yes, the 12V corresponds to the potential difference of the max split supply of +/-6V.

However, the minimum split supply is +/-2.5V and this doesn't match the minimum single supply of 2.7V.
1st page and page 24 of the data sheet specified that the amplifier can work with 2.7 to 12V single supply, same as on table 4.

I tried to simulate AD623 using Diamond Plot Tool and the model works for as low as 2.7V single supply as well as +/-1.35V split supply.
Having said that, I think the amplifier would operate using +/-1.35V supply but you have limited input signal which the amplifier will accept negative input signal only.
The headroom of input voltage range from the +Vs is 1.5V.

For the optimization of the part using dual supply and to be able to accept positive input signal, it is recommended on the data sheet to use +/-2.5V dual supply.

In addition, to optimize the performance of the part and have the expected result from the specification table, please use the specified test condition. From the table single supply uses 5V and dual supply uses +/-5V.

Thanks and regards!

Thanks for the reply! I understand about half of it, but other parts are still not quite clear.

My basic question is very simple. Physics says that voltage is defined as the difference between two electrical potentials.

If the potential difference between the V+ terminal and the V- terminal is 2.7V, then some users may denote this as V+ = 2.7V and V- = 0V (which the datasheet says is legal); other users may denote this as V+ = +1.35V and V- = -1.35V (which the datasheet says is illegal). However, without having access to a pin defined as 0V, the AD623 has no way of distinguishing between the two cases -- they both merely show 2.7V delta between the two supply pins. Physics says that the chip cannot behave any differently between the two cases. And yet -- the datasheet does say that one is legal and one is illegal.

My suspicion is that there are assumptions built into the datasheet that I'm unaware of. E.g., perhaps a minimum size of the operating region given some "reasonable" assumption of what Vref is?

/Joel

Hi,

As tried on the Diamond Plot Tool, the AD623 was modeled for as low as 2.7V single supply or +/-1.35V dual supply. However, the data sheet says minimum +/-2.5V dual supply.

My understanding on this is that you can use the part for as low as +/-1.35V but since the headroom needed for the input voltage range is 1.5V from the +Vs, this means your input signal will only be negative signal (-1.5V to -0.15V) and the amplifier cannot accommodate positive signal on this case.
Related to the headroom needed by the input voltage range, I would think that this is the reason why data sheet recommends +/-2.5V supply so that the amplifier can accommodate bipolar signal which most of the applications would be.

May I know some of your application requirements? Are you limited to single supply or you have the capability to use dual supply? What is the expected range of your input signal?

Thanks and regards!

Thanks -- that makes sense now.

So to summarize... in theory, +-1.35V is indeed indistinguishable from +2.7V/0V supplies. Any input that will work for one case will also work for the other. But for the dual-rail interpretation, the legal input range would not span any positive voltage values. For that (arguably somewhat arbitrary) reason, the data sheet calls the dual-rail +-1.35V use model illegal.

And for your question about my use model -- it's to detect/amplify a surface EMG signal. The instrumentation amp then feeds a filter/amp to drive a ADC in an embedded microcontroller. The ADC runs of 3.3V single rail, and any input >3.3V can damage the ADC.So the easiest thing is to run the in-amp and the amplifier/filter both on 3.3V single rail. Option #2 is to run the in-amp on a higher dual-rail voltage, run the filter/amp on 3.3V, pick the in-amp gain so that it "should" never exceed 3.3V voltage swing, and then pick an op amp for the filter/amp that would not be damaged by an occasional input overvoltage.

All of this is for for a freshman-level Intro to Engineering course that I'm putting together for this fall that will focus on bioelectricity. The easiest approach would likely be to use a more integrated solution like the AD8232 (and I will indeed do that for ECG readings), but having them play with the individual parts for the EMG seems like it would be educational also. The problem is that while I have 30 years of experience in CPU design and post-silicon debug, it's all in digital circuits and not in analog.

Thanks,

/Joel