Other Parts Discussed in Thread: OPA1637, THS4551
Tool/software:
First, I would like to thank TI for releasing this high-voltage, low-speed precision fully differential amplifier (FDA). It is very convenient for use with classic +/- 15 V bipolar supplies (or even at a bit less than that). I have a few successful design with it that seem to be performing better than their predecessors built with two or more discrete op amps.
-
The amplifier datasheet, however, seems to suggest that THP210 would also work with low-voltage unipolar supplies down to 3 V.
There are also examples, like Figure 9-10 with a 30 kHz low-pass filter (third-order Butterworth) with a gain of 5 on a unipolar 5 V supply.
The example does not specify the voltage at VOCM node, but for any valid value, this example, however, seems to violate the specified input common mode limit (VICM > VS- + 1 V) for all possible input voltages with a valid output voltage through an ideal FDA with gain 5.
The PSPICE model for the amplifier also confirms this misbehaviour, and that it is due to VICM limits.
-
Is the example a bad example of the use of this FDA, or are the VICM and the PSPICE model too pessimistic about the valid operating window of the amplifier?
In the example, at VOCM = 2.5 V, the VICM would be just 0.4 V when the input is in the middle of the valid range. This would mean that the VS- would need to be at most -0.6 V for guaranteed correct operation. And, even at the highest valid VOCM (which leaves no room for actual signals with the shown VS+), one would not get past 0.8 V VICM which would still need a negative supply of at least -0.2 V.
This is with a +5 V supply, which is quite a bit above the minimum specified supply of +3 V.
-
Which FDA would you recommend for such low-voltage unipolar application where dc precision and good low-frequency noise performance would be highly advantageous, and where the signal bandwidth need is less than 100 kHz?
I have a few ideas of drop-in replacement from your catalogue of low-voltage FDA, but they all seem to have quite a bit more noise at low frequencies.
As, alternatively, I think, the only valid solution with THP210 would require all signals referred to a virtual ground near the mid-supply level. This of course greatly reduces the versatility of having an FDA, and would have very marginal headroom for operation at, say, a hypothetical 3.3 V application.
