Isolation Amplifier Design for Universal Signal Conditioning System.

Thread Starter

Subhan9990

Joined Feb 12, 2022
3
I am trying to design a universal signal conditioning system in proteus. I need ideas regarding isolation amplifier. Can anyone help me designing isolation amplifier?
 

ErnieM

Joined Apr 24, 2011
8,294
I had a task last year where I needed to read a sense resistor with a DMM to pull about a 1V signal out of 200 of offset. The issue (I believe, not tested) was the high offset voltage was upsetting the remote interface (IEEE-488) to the meter, occasionally requiring a DMM power reset. Not good in a production test environment.

I was planning on using an AD8479 isolation amp for this when the project was canceled by the customer.

Anyway, the AD8479 has these specs:

  • ±600 V common-mode voltage range
  • Rail-to-rail output
  • Fixed gain of 1
  • Wide power supply range of ±2.5 V to ±18 V
  • 550 μA typical power supply current
  • Excellent ac specifications
    • 90 dB minimum CMRR
    • 310 kHz bandwidth
  • High accuracy dc performance
    • ±5 ppm maximum gain nonlinearity
    • ±10 µV/°C maximum offset voltage drift
    • ±5 ppm/°C maximum gain drift

https://www.analog.com/en/products/ad8479.html#product-overview
 

LowQCab

Joined Nov 6, 2012
2,058
That's a Single-Op-Amp Instrumentation-Amplifier,
it doesn't "isolate" anything,
but it might eliminate some "Common-Mode" Noise.

You still haven't explained what needs to be isolated from what.
.
.
.
 

MisterBill2

Joined Jan 23, 2018
11,913
Indeed LQC is correct. There is not really such a thing as a truly universal amplifier that would be useful.
When one size fits all, it fits none of them well.
What is required to be able to select or create an isolation amplifier we must know the signal voltage and the signal's source impedance, the common mode signal and it's source impedance, as well as the frequency of the desired signal. Capacitive isolation is one choice that in theory can provide adequate isolation, but in reality is a bit limited. Thransformer isolation can work fairly well if the signal is strong enough to be coupled thru an isolation transformer. But it does not work directly with DC signals.
Next is carrier modulation, which covers AC , lightwave, and magnetic field carriers. The signal to be isolated first modulated the carrier, which passes thru the isolation barrier, after which it is demodulated and restored to a representation of the original signal.
So there are manydifferent schemes each most suitable for a specific application, and each with different ranges of maximum voltage to isolate.
 

Thread Starter

Subhan9990

Joined Feb 12, 2022
3
Indeed LQC is correct. There is not really such a thing as a truly universal amplifier that would be useful.
When one size fits all, it fits none of them well.
What is required to be able to select or create an isolation amplifier we must know the signal voltage and the signal's source impedance, the common mode signal and it's source impedance, as well as the frequency of the desired signal. Capacitive isolation is one choice that in theory can provide adequate isolation, but in reality is a bit limited. Thransformer isolation can work fairly well if the signal is strong enough to be coupled thru an isolation transformer. But it does not work directly with DC signals.
Next is carrier modulation, which covers AC , lightwave, and magnetic field carriers. The signal to be isolated first modulated the carrier, which passes thru the isolation barrier, after which it is demodulated and restored to a representation of the original signal.
So there are manydifferent schemes each most suitable for a specific application, and each with different ranges of maximum voltage to isolate.
I don’t need something to fit each well but a universal one.
 

MisterBill2

Joined Jan 23, 2018
11,913
An isolation amplifier needs to cover some signal voltage range, some signal frequency range, and some common mode voltage range.
Without that rather basic set of requirements there is no reason to try to provide any help here.
Please keep in mind that you are requesting a great deal of engineering assistance, for free, and effort that would cost a whole lot if you had to pay for it.
 

Ya’akov

Joined Jan 27, 2019
5,669
Is DC to light bandwidth too much to ask for in these days of 9900mAH 18650 cells and 3W UHF handheld radios with 22 mile range (depending on terrain)?

As long as some loss in operation is acceptable very high isolation, >100dB, is easily achievable (though there is nearly 0dB insertion loss when not operating) the answer is simple:

859C9534-16FB-43E0-BE89-0192B23974CB.jpeg
 

MisterBill2

Joined Jan 23, 2018
11,913
View attachment 260592
There are many types of Isolation Amplifiers that I use every day.
Certainly there are, and none of those devices is "universal".
View attachment 260592
There are many types of Isolation Amplifiers that I use every day.
I have quite a few shunt resistors, mostly one hundred amps, that have attached isolation amplifier boards that use a very nice IC , the Burr-Brown 6502, I think is the number. It is unique in that the power for the device is isolated from both the input and the output sides. It runs on 15 volts, I have not tried it on 12 or 13.4 volts.
But while it provided excellent isolation and adequate frequency response for my applications it's isolation is only hundreds of volts, and the frequency response is only a few hundred hertz. So it is good, but not universal.
For close to universal you should contact the Rhodie and Schwartz company, who might come close. . For about $25,000.
 

MisterBill2

Joined Jan 23, 2018
11,913
Please define universal.
We have requested a definition of what the TS is seeking and that is what we get back. The fact is that R&S is able to pdesign and produce products with amazing qualifications, and so I suggested them because I think that they can come closer than others..

My definition of "universal?" for the crash test and other destructive testing industries: Stable drift free constant gain from unity to at least 1000x (60dB) over the range of DC to one megahertz, with a very high input impedance suitable for quartz sensors, and an output impedance of 50 ohms, with input isolation adequate for 500 volts common mode at any gain setting and any input voltage range.
For industrial production testing equipment? Stable gain from -40 dB to +60 dB, over a frequency range of DC to 100KHz, and an isolation ability for unaffected output with an input common mode voltage up to 1000 volts, from DC to 100kHz, with an input impedance of at least 1 megohm and an output impedance of 50 ohms. AND the gain and offset must be stable enough for no error for at least one year.
 

dcbingaman

Joined Jun 30, 2021
566
There are about as many ways to build an 'Isolation amplifier' as there are creative people to think them up. A 'Universal' Isolation amplifier (that is an isolation amplifier that would work for 'all' or 'most' applications (read that all frequencies for zero to Gigahertz and beyond with flat bandpass response), does not have a likely solution. An isolation amplifier that has say a fixed gain of 1 with a high bandwidth (0Hz to 100Ghz) or beyond would be suitable for most applications except the expense involved in creating such an amplifier would be prohibitive and push it out of most market applications. Amplifiers/Isolation Amplifier design is mostly specialized for a specific application/solution. Thus the 'Universal Isolation Amplifier' ends up being a fictitious device that a lot of physicist and others would just love to have.
Also, there is no such thing as a 100% effective 'Isolation' Amplifier. There will always be some energy transfer between the two sides of the amplifier that is fixed above zero from the information conveyed from one side to the other.
 

ronsimpson

Joined Oct 7, 2019
2,083
Stable drift free constant gain
I am using parts from many different places. Lets talk TI.com. Gain drift is 5 parts/million, the one I am using most is 30ppm.
Most amps are fixed gain. I use 1:1, or 5X or 10X. I have used a programmable gain amp ahead to make an isolated scope probe. That gives me 1x, 2x, 5x, 10x, 20x, 100x
What I am using has a frequency response of 100khz, 280khz. When we want faster there is a very different way.
Often an op-amp is on the output, and can drive 50 ohms if you pick the right part.
One of my problems is isolation transient. I need to hold off 1500V and 100V/nS swings. (1kv in 10nS)
A "universal" must do everything will cost big money.

p.s. If you see parts to do 100 or 300mhz I want to know. In a different project I need speed and not accuracy. (again 1kv in 10nS)
 

MisterBill2

Joined Jan 23, 2018
11,913
The scheme to deal with monster common mode spikes is a system with a llaser diode and a fiber optic link. Battery operation of the sending end is a cheap way to provide a lot of common mode isolation. That works OK for short tests. For long term operation a scheme similar to the wireless battery charging system for electric vehicles, with gaps of a few inches can be useful, although much bigger and a lot less efficient.
 
Top