Analog isolation circuits: any experience on this?

[mfilippa]

Just a quick question:

Regarding the chopping frequency, for a bandwidth of 1MHz, I selected a chopping frequency of 5MHz. I figured that it has to be greater than the bandwidth to avoid transformer saturation.

Unfortunately, I wasn't able to find any analog switch (in DIP flavor) fast enough to switch at 5MHz. Any suggestions ? How about using fast mosfets as analog switches ?

7roots51: I have been using isolation products before going into this (analog devices) but the problem is that their bandwidth is limited and the cost is generally high.

I've taken appart a digital oscilloscope input board and I found that they were not using any isolation products at all. They are using transformers and lots of signal conditioning circuitry. I also believe they must be chopping the signal too (I guess from the ICs involved in the board). I assume it is still the cheapest and best solution.

Mariano

 

[artem]

May be this is little bit stupid idea , but what is about PWM modulation .
The main question could PWM be used for freq up to 5 MHz ?

 

[mabbri]

Use HCNR200 Linear Analog Optocoupler


Low Nonlinearity: 0.01%
Wide Bandwidth - DC to > 1 MHz

 

[mfilippa]

Thanks mabbri for the suggestion, but unfortunately I've been playing around with HCNR201 for a while without getting good results.

I could never design a circuit with HCNR201 and opamps that operate at 1MHz. I could do it with transistors, but the accuracy was not good. I gave up on this isolator.

UPDATE: I've been looking hard for transformers that are able to couple signals in the MHz range, and found that the LAN transformers (the ones found inside ethernet cards) are suitable for this app. Generally there are two or more small transformers inside, so I can couple the signal and also build a feedback loop to compensate for non-linearities.

I requested samples of these transformers from Pulse Engineering (www.pulseeng.com). They operate with 10BaseT ethernet cards, so they can easily handle signals of up to 10MHz.

I'll give them a try as soon as I get them and post the results.

Mariano

 

[hifni]

Hi all...
Try to use ISO122/ISO124 from TI.... They have a good isolation for analog signal....
They also have an isolated power supply for this application..

Rgds,

 

[cherrytart]

I haven't played with these, but they claim to be faster than optoisolators, and I have no idea what the cost is.

https://www.nve.com/products/isolators/isproducts.php

 

[mfilippa]

I finally gave up on optoisolators. I am trying to do AM modulation using LAN transformers. LAN transformers are cheap, small and offer an excellent isolation.

Any ideas on AM modulation?

I chopped the signal (DC to 1MHz) with a 10MHz square wave (carrier), then did a low pass at around 15MHz to get a sinusoidal wave suitable to couple through a transfomer.

Questions:

1) should I do chopping (alternate between signal and zero volts) ? or should I alternate the signal using a multiplier (multiply by +1 / -1 ) ??
2) how can I demodulate ? remember that the ouput might be DC, positive or negative polarity.

Thanks !
Mariano

 

[Regnum]

mfilippa,

As flatulent previously stated, ADC and digital isolation are the way to go.

...I believe it is a very expensive method for signals up to 1Mhz.

Anyways you will need a high-speed ADC, it doesn't care what kind of isolation you're using nor where the isolation is.

You haven't told us about the accuracy you need, but high-speed and high-accuracy analog isolation devices/subsystems always results in higher costs than a simple fast digital-isolator (or just a couple of them).

Regards

 

[muko]

I recommend 6N187-186optocouplers which works fine up to 10 MHz signalsss!!!

 

[hifni]

BW of ISo124 only 50 KHz...