Is a current sense oamp the same as a precision rail to rail?

rwgast_logicdesign

The past week or so I have been hitting the analog electronics hard! It has been a great learning experience! Basically I started designing this solder station, along side that I have been absorbing all kinds of EEVBlog videos. It all started because I needed to build a variable dummy load to test out some transformers, and I found some information about using a precision opamp with a logic level N FET on the EEVBlog, to do just this. This has also led to me watching Daves 8 part series on designing a precision linear power supply. All this stuff will be very useful when building this solder station.

Anyways here's the deal in a lot of his videos he uses precision rail to rail, or instrumentation amps. I don't have any of these on hand so I just use an LM358, this is fine as I haven't needed PRECISION equipment. But let's take the dummy load as an example... using an lm358 I cant adjust my dummy load all the way to 0 amps. Im guessing this has to do with the swing voltage across the LM358 rails. I don't need to go to 0 amps but it would be nice.

Yesterday I was playing with some precision current amps, from touchstone semiconductor. I planned to use these in my Robot to implement a current feedback loop using a .005 shunt. I have both the 200v Gain and 25v Gain models. The math for a 200V gain works out perfect for my bot, it will amplify 500uV across my shunt to 100 millivolts which works out to 1v per 1 amp when measuring my motors.

Anyways I have a ton of these things and realized that, hey these are precision amps, they reject noise, have a low 100uV offset etc. So are these the same as a general purpose precision amp, the only difference is that they have a fixed gain? Would it be feasible to use these in projects that need precision amps and then just use 1% resistors to form a voltage divider on there output to get a more reasonable value. Then I would be able to get my dummy load to 0 amps. A better example would be reading a photo resistor, I use the 25v gain amp, between a photo resistor and ADC, then use a voltage divider to set the voltage output back in line to 5v, instead of 25. Or is the johnson noise and impedance from the divider going to totally negate any better precision gained over a standard lm358?

These are the amps i am using,
http://media.digikey.com/pdf/Data%20Sheets/Touchstone%20Semiconductor%20PDFs/TS1101.pdf
Much better than there maxim equivalents.

RDL2004

Dave used an LM324. The LM324 is intended for single supply operation. It will only go to a volt or so below the supply voltage, but will operate down to a within few millivolts of "ground". I don't know if the 358 can do that from a single supply.

Tracy Allen

The watchword for using the TS1101 is "high-side". In the description of the chip you will find this phrase, "feature a wide input common-mode voltage range from 2V to 25V". It will not work to sense voltages near ground, rather, only voltages that are in the specified range. Within that range it will sense small differences in voltage, usually measured across a shunt resistor of low ohms value. It is a specialized circuit, not meant at all for use as a general purpose op-amp. That is not to say that it wouldn't be useful in constructing a variable dummy load. It can measure the current for you.

By the way, gain is expressed as for example 25 V/V, not just 25V.

An LM358 can swing down to within about 10mV of ground if the output does not have to sink more than 1 microamp. But the output can't get nearly that low if it has to sink current, at 100 microamps it can only get down to about 0.5 volt.

rwgast_logicdesign

Ok, so that means if I try to put less than 2v across the rails, the ts1101 will not work? Im just now starting to learn how an opamp really works, and trying to figure out what differentiates them.

For instance lets go back to the dummy load, I have some of these on the way http://ww1.microchip.com/downloads/en/DeviceDoc/21669D.pdf The data sheet says they are rail to rails and operate with a single supply voltage down to 1.4V. Is this considered a precision op amp? I also ordered some lm324s (http://media.digikey.com/pdf/Data%20Sheets/Fairchild%20PDFs/LM224.pdf) today, because that is what the original schematic uses, I had forgot I had these microchip amps on the way. Which one is better for the dummy load, how about for reading a low voltage sensor output?

Now I also have a bunch of these, http://www.ti.com/lit/ds/symlink/opa134.pdf which are designed for "Audio" and that is how I use them. Would these make bad general purpose opamps? It seems as if these would produce less noise than an LM358....

Maybe someone could help explain the differences in these three amps?

jmg

rwgast_logicdesign wrote: »

Ok, so that means if I try to put less than 2v across the rails, the ts1101 will not work? I

Not quite - the wording "feature a wide input common-mode voltage range from 2V to 25V" means the INPUTs must be held above 2V and below 25V - you can sense current on a positive supply. but not in a ground lead.

rwgast_logicdesign wrote: »

Maybe someone could help explain the differences in these three amps?

Look at the data, the details to check are words like Rail-to-Rail In / out / In/Out.
Also look for Common mode, and check Vos vs Common mode.


An LM358 should sense to GND, but not to Vcc as its common mode range does not include +ve Rall (it uses PNP darlington in).
The LM358 is simply the Dual version of LM324.

These days, it is simplest to get Rail-Rail IO with the lowest Vos you can afford, for general use.
Very low noise opamps tend to have poorer common mode, and higher bias currents and often limited Differential tolerance.
(that low noise comes at a trade off)

Tracy Allen

"The data sheet says they are rail to rails and operate with a single supply voltage down to 1.4V. Is this considered a precision op amp? "

There are at least three separate ideas there. "rail to rail", "single supply down to 1.4V" and "precision". Those and a host of other things go into the data sheet of an op-amp and account for the fact that there are so many kinds and classes of op-amps available that feature one aspect or another. The "ideal" op amp has it all.

The audio op-amp is stellar in terms of performance for its intended purpose, high bandwidth and able to accurately reproduce audio inputs. The Microchip part on the other hand operates on less than a microamp, and has a bandwidth of only 14kHz, so it is intended specifically for those purposes that are featured in its data sheet. Note in its data sheet that there is a circuit for using it as a high-side shunt amplifier, which it can do due to the fact that its input common mode range includes the high rail, and that circuit is pretty much the same as your dedicated shunt amplifier. The LM324/LM358 are general purpose. They have been around forever and won't do those special things that the others do, but they are a workhorse.

If you hear an op-amp advertised as precision, you usually expect to see a low offset voltage and low offset current, and low drift factors in both, and they are usually meant for amplifing DC or relatively low frequency signals. None of the ones you listed are precision in that sense. Their offset voltages are around 3mV. Precision unsually means something less than 0.1mV, even down to a few microvolts. If you use a precision op-amp, you can put in a tiny signal from say a thermocouple or a shunt, and pretty much ignore the offset voltage, because even though it is amplified along with your signal, it is small enough to be ignored. It is quantitative, it is relative to the range of the signal you want to measure and how many bits of precision you expect to get out of that range. If the offset is stable, you may be able to calibrate it out, so it becomes (with that extra effort) a question of accuracy vs precision, but if the offset drifts with time or temperature or has a lot of noise, that will put a hard limit on the measurement.

The audio op-amp is precise in a different sense. The 3mV DC offset and slow drift are not important for its intended purpose, rather, its precision depends on its ability to give a true representation from input to output at audio frequencies, free of added noise.