Instrumentation amplifier data sheet

[poiureza]

Hello, I'd need some education here ...

My understanding is that one major asset of instrumentation amplifiers is their CMR.
When I look at the data sheets (INA118, 126 etc ..) they have indeed high CMRR but only for low frequencies.
They alll exhibit a steep downward slope after a few Hz
@1kHz many "normal" OP amps have actually better values (e.g. OP282)

- What am I missing ?
- Do they trade off high freq CMR for better low freq CMR ? 
Then again, at low gain the instrumentation amplifier does not have better low freq CMR than the OP282 for instance ...
- Is high freq CMR actually a thing ? 
If I want to use it as mic preamp I suppose I would only expect DC and 50-60Hz + uneven harmonics to be present on both inputs, no ?
(If I'm not mistaken, the hum harmonic content is only really noticeable up to 1kHz or so ...)
- Am I right supposing that I need to use them with gain >1000 to actually take advantage of their CMRR ?

[Rob Strand]

Instrumentation amplifiers aren't opamps they are a circuit.   Input buffers gain setting resistors, programmable gain.

The key difference is they have matched precision resistors inside which let them have high CMRR when the gain is set.   An opamp circuit needs external resistors and if you use even 1% resistors the CMRR is degraded.   That's where the instrumentation amps win and why they are exist as a product.   Nowadays you can get precision resistors relatively easily so in theory you could roll you own.  However, then you might find the DC specs are need some care to compete with the instrumentation amps.  Things like input offset voltage, drift parameters.  That's another set of parameters the instrumentation amps offer.

[iainpunk]

there are more expensive INAmps that have higher CMRR range, but in most applications, its not needed.


i think that CMRR matters more for lower frequency's since its often a low frequency or DC used for things like phantom power. i did a project where we inserted a 4Hz AC phantom power on balanced signal lines for some remote sensing application, and DC is also quite common to find in similar applications. higher frequency CMRR is nice to negate whine and high frequency interference but when handling things like 50hz hum, low frequency CMRR is enough.

normal hiss/noise is also not induces the same in both wires of a balanced signal.

cheers

[Rob Strand]

Keep in mind audio is only one application,  there's other fields which use opamps and instrument amps like testing and measurement.

[poiureza]

Thanks for the replies
I'll try a couple of the cheaper ones and see how they work as balanced to unbalanced converters which is my actual need.

[iainpunk]

if its audio, you can get away with less than ideal CMRR, as long as the output impedance that drives the line is quite low.

if you want a better CMRR rating, you could look at rolling your own, and using a 20 turn trim pot in series with a sightly small resistor to fine tune the balance. for instance, use 12k for all resistors and replace one with 10k and a 5k 20 turn trimmer to fine tune it out.
those values can be changed to fit the application, ofcourse

cheers

[poiureza]

Thank you,

Yes I would use it as mic input.
And yes, I might end up rolling my own eventually but I wanted to check out if there are dedicated chip types that would fit the bill (and to understand why they would)


PS : Not sure I understand how the impedance at the other side of the chip affects the CMRR.
I don't see any reference to that in the spec sheets.  They only give "CMRR vs Hz" curves at different gain levels, but not vs load impedance.
I suspect the plots are shown for the ubiquitous 10kOhm load, but it's not stated.
Would you have some info here ?

[iainpunk]

Not sure I understand how the impedance at the other side of the chip affects the CMRR.

it has to do with the character of noise.
things like normal fluorescent light hum, or other mains hum are the same phase when picked up so when subtracted by itself, it zeroes out, when you have high source impedance, all sorts of other noise can be picked up by the long wire more easily, most of which is different between the two conductors so they get added for even more noise.
its not to much to do with CMMR as with general noise theory.

cheers

[poiureza]

OK you were talking about impedance between mic and IC (not the load after the IC)
Then I understand.