Does anyone have any experience/thoughts on the robustness of a 3.3v CMOS Opamp configured as a differential amp with the inputs connected via 10m cables to a strain gauge bridge in an electrically noisy environment ? If I add clamp diodes from the opamp inputs to the rails is there any difference between a 3.3v device and a 12v device in terms of susceptability to damage from external transients, etc ?
The strain gauge is in a high temperature environment which is too hot for signal conditioning electronics. Also, the mechanical part doesn't allow easy location of a circuit board so we've elected to put the conditioning remotely. The signal can be heavily filtered as the sampling rate will be low. It seems there's a better choice of single rail opamps with good DC performance with low voltage rails than with higher voltage rails. Hence the question regarding the suitability of the low voltage amps in such an amplication.
A CMOS op amp is a horrible choice for a strain gauge signal conditioner, unless it's a chopper. Bad drift, bad offset, really putrid 1/f noise, electrically fragile, and you don't care about the high impedance because the strain gauge impedance is very low.
Thanks for your post. Looking at the opamp table on the Analog website, for example, sorting by Vos shows many CMOS/FET devices with very low figures for input offset, drift and noise but they tend to have low voltage rails. My concern was their robustness (but with external series resistor/clamp diodes) when connected to outside world signals and how they compared with higher voltage technology especially bipolar.
Thanks for your post. Looking at the opamp table on the Analog website, for example, sorting by Vos shows many CMOS/FET devices with very low figures for input offset, drift, noise but they tend to have low voltage rails. My concern was their robustness (but with external series resistor/clamp diodes) when connected to outside world signals and how they compared with higher voltage technology especially bipolar.
Are you sure you're looking at the guaranteed specs? For instance, the AD8603
formatting link
has typical Vos of 12 microvolts at 25 C, which sounds great except that the worst-case drift is *4.5 microvolts* per degree.
The Vos vs Vcm curves on page 6 show over 100 uV of offset change within about 0.2V of Vcm.
Also those noise specs are all in the flatband, whereas with strain gauges you're typically most interested in sub-audio frequencies. Even in the flatband, 22 nV typical is more than 20 dB worse than a good bipolar op amp, and I don't think you've looked very closely at the curves--which BTW are typical and not guaranteed performance. For instane, Figure 26 on P. 9 shows the noise voltage shooting off apparently to infinity at low frequencies--they plot it on linear scales to try to disguise how bad it is, but it goes up past 168 nV/sqrt(Hz) before they cut it off.
However, your application may be forgiving enough that these things don't bother you, in which case it might work fine. That part certainly wouldn't be in any strain gauge or thermocouple signal conditioner of mine, though.
"sorting by Vos shows many CMOS/FET devices with very low figures for input offset, drift and noise but they tend to have low voltage rails."
I've noticed this also, I'm far from a device guy, but I think that some of these specs might track with the supply voltage. Double the voltage and double the offset? So 36 V opaamps look like they have worse offsets. But hey if you are worried about siganl to noise then you want to crank the gain up to get the signal 'near' the rails. And then the relative error is the same.
Is the strain gauge a bridge? if so why not use a nice bjt instrument amp?
You may want to consider diode clamps from the rails back to ground. You can't insure a device is powered when a transient hits.
Lower voltage generally goes with finer geometry processing, but that doesn't necessarily translate to a more fragile part. Electrical overstress depends a diode doping profiles, assume a reverse bias. The gentler doping profile is less prone to hot spots. Some protection devices used a fet breakdown to snub the current. This could be better in finer geometry since the breakdown will occur at a lower voltage. Basically, I don't think you can make the blanket claim that a 12V chip is more rugged than a 3.3V chip. Generally when you scale a chip, the protection devices aren't scaled much at all.
Small strain gage signals and a 10M cable? Put your bridge amplifier right next to the strain gage, run the 10M cable to the amplifier output. Cables can be antennas, thermocouples, transformers, triboelectric sources (maybe not for a low impedance signal), etc. Don't do that to your precious not-yet-preamplified signal.
Call me weird, but there are fundamental reasons why a split supply opamp will always work better at DC than a single rail opamp. If you are willing to go to +/- 3.3 V or +/- 5 V for the first stage you can easily get much better DC performance. Much of it is a result of circuit topology. Just to push the point can you put the strain gauge bridge in the same thermal environment? It will save you much trouble if you can.
IME bipolar DC offset and drift of bipolar opamps kick CMOS opamps tush up and down the block. For really sensitive situations; bridge the strain gauge sensor inside the thermal environment, and use split supply opamps, perhaps driving the sense wire twisted pair shield with the common mode voltage. BTDT and i even had load cells (bridge included) to work with.
You're weird ;-) Your statement is only true if the OpAmp architecture (internal circuitry) contains a ground pin... such as my venerable MC1530.
Maybe you are arguing for getting the signals ground-referenced?
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine Sometimes I even put it in the food
Not really, the output from the bridge is inherently differential, and with split supplies and sits nicely close to 0 V for best CMRR. CMRR does degrade as the common mode voltage approaches the rails. Second i want the entire bridge in the same thermal environment to keep thermal mismatch as much out of the problem as possible.
It has been an issue with every opamp that i have ever used. If you have made some that do not have that problem, send me the data sheet and some samples. It will enlighten me. Just because i have not seen one or heard of one does not mean that they cannot exist, but that they are outside my experience. And i have seen testing of opamps where this not supposed to be a problem, but it still is.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.