Difference Amp.

I've used them in various test jigs and small instruments (usually the PMI/AD AMP03 because I had a bunch of them). They're very convenient and the resistors track well, which is one advantage, but I'm usually using them for strange stuff such as very accurate x1/x2 amps.

One problem is that you have to drive all three inputs from super low-Z sources or you'll ruin the CMR.

Cheers

Phil Hobbs

(Just getting up from a bout of pneumonia due to a virus I brought back from vacation.) :(

I've used the INA128 a few times, no complaints about it.

They are pretty common and there are lots of equivalents. I think they are here to stay.

A difference amp, with tracking laser-trimmed resistors inside, is fairly cheap. One cool part is AD8130, 270 MHz, interesting architecture.

Right. These parts aren't going to disappear. Most of the popular versions shown in our 1989 2nd edition are still going strong, and will for decades to come.

The high-frequency gain and good CMRR is nice, but since it uses the cancellation of two differential BJT g_m stages as its input, it isn't a precision part, like a fine op-amp + 0.01% matched resistors.

Yeah, this is just inside a slow analog control loop, so opamps or such driving all the R's. I don't care too much about CMR... (I think.. noise and interference can tricky,) I'm adding and subtracting analog signals, I was thinking how much easier it would be with an Int amp, rather than two opamps to sum and then invert. And then why not a difference amp. (which I only learned about from AoE3.)

Then again I've seen analog bits go away. (I've got a few UAF42?? (burr brown SV filter IC) in my parts box... went away just when I built my first SV filter...(did I mention I often feel like I live on the trailing edge of technology?) We can save a little by buying Int amp's in bulk... they will have a longer lifetime than Diff amps.

Then again again, I can see all sorts of nice ways to use, opamps w/R's, saves space... there're many vendors.

George H.

Yeah, this is just inside a slow analog control loop, so opamps or such driving all the R's. I don't care too much about CMR... (I think.. noise and interference can tricky,) I'm adding and subtracting analog signals, I was thinking how much easier it would be with an Int amp, rather than two opamps to sum and then invert. And then why not a difference amp. (which I only learned about from AoE3.)

Then again I've seen analog bits go away. (I've got a few UAF42?? (burr brown SV filter IC) in my parts box... went away just when I built my first SV filter...(did I mention I often feel like I live on the trailing edge of technology?) We can save a little by buying Int amp's in bulk... they will have a longer lifetime than Diff amps.

Then again again, I can see all sorts of nice ways to use, opamps w/R's, saves space... there're many vendors.

George H.

Thanks, I'm gonna order some. (someone has to use these parts.)

George H.

Seems like an expensive way to save four ordinary resistors if you don't care about CMR. The AMP03's DC CMR is 100 dB typical, which is pretty good for a part that old.

Cheers

Phil Hobbs

Until it's the last low-volume product that won't be transferred onto the next wafer size or a different fab.

robert

Hmm maybe. I was replacing an Int amp, so it's cheaper than that. I do care about the gain accuracy... I'm taking the difference between a set point and temperature, to make an error signal. (As well as other even less critical stuff.)

OK, I'll stick with the Int amp, that is in stock and make the other diff amp from an opamp and R's.

Say, Here's a somewhat related question. I'll often use an inverting amp to make input connection to the outside world. (With input R's ranging from

George H.

Diodes from summing junction to ground will make it very robust.

(May need small resistance in series with each for stability.) ...Jim Thompson

No, because

a) most of them are trimmed to way under 0.1%

b) I had the misfortune of needing 0.1% resistors recently for a specific application, and four resistors definitely cost more than a single differential opamp

c) I think while they're trimming them, they are also trimming some extra resistors that play in the compensation network and you can't do that outside the chip

d) Those resistors stay at the chip temperature and the nulling circuitry takes the tempco into account

Anti-parallel. Is that called back-to-back or belly-to-belly ?:-}

...Jim Thompson

I think TI / Burr-Brown have already done such a transfer for these parts. Precision difference amplifiers are visible high-profit items.

Thanks Kamen, I've been using these susumu 0.1% R's.. they use to be about $0.2 each (at hundreds quantity) but now you can get them for about $0.10.

George H.

This is what we have in stock:

ICS DIFFAMP 18V SO8 1X INA117 200KHz ICS DIFFAMP 40V SO8 1X INA154 G=1 ICS DIFFAMP 44V DIP8 1X SSM2017 OBS ICS DIFFAMP 36V DIP8 1X INA105KP G=1 ICS DIFFAMP 36V SO8 1X INA105KU G=1 ICS DIFFAMP 36V SO8 1X INA157 G=0.5/2 ICS DIFFAMP 34V MSOP10 1X AD8253 PGA ICS DIFFAMP 50V SO16W 1X INA103KU INST ICS DIFFAMP 36V SO16W 1X INA111 FET VG ICS DIFFAMP 36V DIP8 1X INA114 VG OVP ICS DIFFAMP 36V SO16W 1X INA115 OVP VG ICS DIFFAMP 36V SO8 1X INA129 VG ICS DIFFAMP 36V SO8 1X INA141 OVP ICS DIFFAMP 36V SO8 1X AD8221 800KHZ 400PA 8NV ICS DIFFAMP 3V QFN16 1X ADA4950-1 ADC DRVR ICS DIFFAMP 5V QFN16 1X ADA4960 ADC DRVR ICS DIFFAMP 26V SO8 1X AD8130 270MHZ ICS DIFFAMP 15V MSOP8P X1 THS4505 260M ICS DIFFAMP 5V QFN16 1X LT6402-12 ADC DRVR

We don't use the DIP parts much any more.

Oh and a colleague pointed me to these in through hole.

GH

Not only kosher, it's an excellent idea, as long as (a) it doesn't trash the noise performance, and (b) you label the inputs with Rin as on an oscilloscope.

The resistor provides current limiting, and since the summing junction stays more or less still, you can add external protection networks easily.

Cheers

Phil Hobbs

Yup. Back-to-back is inverse series, as in "back-to-back zeners".

Cheers

Phil Hobbs