Is it possible to design an isolation voltage measurement circuit up to 200kHz?

[onemilimeter]

Wish to have an isolation voltage measurement circuit with bandwidth up to 200kHz.
Hope that the phase response is ~ 1 deg.
Input voltage: +/- 5V. (able to measure AC and DC)
Gain=1

Do you think this circuit performance can be achieved?

Cheers.

[Zero999]

With an analogue opto-isolatior such as the HCNR200
http://datasheet.octopart.com/HCNR201-000E-Avago-datasheet-30849.pdf

[onemilimeter]

Thanks. It's a good solution...

[DaveW]

Another option is something like this,
http://www.analog.com/en/specialty-amplifiers/isolation-amplifiers/ad215/products/product.html

(I know it doesn't have enough bandwidth, but it just happens to be open already for a project)

This particular one is expensive as it's isolated up to 1500v, but there are cheaper ones with the bandwidth you need. Opto's are probably cheaper though

[onemilimeter]

Another option is something like this,
http://www.analog.com/en/specialty-amplifiers/isolation-amplifiers/ad215/products/product.html
(I know it doesn't have enough bandwidth, but it just happens to be open already for a project)
This particular one is expensive as it's isolated up to 1500v, but there are cheaper ones with the bandwidth you need. Opto's are probably cheaper though

Thanks. But this solution costs much more than my target budget.

[onemilimeter]

Wish to have an isolation voltage measurement circuit with bandwidth up to 200kHz.
Hope that the phase response is ~ 1 deg.
Input voltage: +/- 5V. (able to measure AC and DC)
Gain=1

Do you think this circuit performance can be achieved?

I'm sorry as I realise that the specifications I placed earlier are too high for my application.

In my application, the correct specifications should be as follows:
Input voltage: +/- 5V (or scaled to be lower using voltage divider)
Phase Shift @ 1kHz: less than 1 degree
Gain: 1 (or 2) (flat from DC to at least 1kHz)
Input to Output Propagation Delay (90%-90%): less than 5us

[Zero999]

Does it still have to work down to DC?

[onemilimeter]

Does it still have to work down to DC?

One of them will be used to measure the input dc-link voltage to a 3-ph inverter. It's a DC voltage and some "ripples" may ride on the DC voltage.

Thanks.

[NiHaoMike]

How about performing the A/D on the primary side and isolating the output?

[alm]

That would seem the obvious solution to me, digital signals (especially something with unidirectional lines like SPI) are much easier to isolate.

[Zero999]

Digital would definitely work but then the question is what level of accuracy is desired?

Going from your response, I assume the signal is just positive i.e. monopolar? Bipolar would require a precision rectifier and transmitting the sign separately or biasing at DC which would need some zero trimming.

How about transmitting the AC and low frequency/DC signals separately? 20kHz to 20kHz can be sent across an isolation transformer and all information below 20kHz could be sent digitally at a high resolution using an audio ADC.

[onemilimeter]

How about performing the A/D on the primary side and isolating the output?

Thanks. This solution is promising too...

[onemilimeter]

Digital would definitely work but then the question is what level of accuracy is desired?

Going from your response, I assume the signal is just positive i.e. monopolar? Bipolar would require a precision rectifier and transmitting the sign separately or biasing at DC which would need some zero trimming.

How about transmitting the AC and low frequency/DC signals separately? 20kHz to 20kHz can be sent across an isolation transformer and all information below 20kHz could be sent digitally at a high resolution using an audio ADC.

I wish to build four voltage measurement channels.
One is to measure the dc-link input voltage to a 3-phase inverter (the voltage is 0~50VDC, unipolar).
Another three are to measure the phase voltages of a 3-phase induction machine (bipolar).

I think 12-bit resolution is enough for my application.

Do you mean "20kHz to 20kHz" or "20kHz to 200kHz" can be sent across an isolation transformer. By the way, what is the maximum frequency that an isolation transformer can take without distortion?

Cheers.

[Zero999]

Yes, it was a typo, I meant sending 20kHz to 200KHz through an isolation transformer and a lower frequencies via some other means.

I think a 10-bit accuracy at 200kHz (a minimum sample rate of 400kHz) will be expensive.