Can "digital" isolators be used to isolate analogue signals?

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Hi all,

This may be a daft question - but can you use the digital isolators, such as https://www.analog.com/media/en/technical-documentation/data-sheets/ADUM3200_3201.pdf to isolate a high voltage analogue signal? It seems as though maybe you can, since the document says they can replace opto-couplers with none of the side effects. But the "digital" in the name is confusing me. Maybe the way the chips isolate the signal is through a "digital" method, but it can still be used to isolate an analogue signal.

Reason I need the isolation: My DC/DC converter has a clamping capacitor which I want to keep below 400V, for fault detection. I may also use it for voltage feedback and control later down the line if needed. At the moment, I am just wondering the best way to isolate this voltage. I want to have at least 800V reinforced isolation and ideally it would be high speed since it is primarily for fault detection. It will be passed to a DSP with 15V supply voltage.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Why not? In the data sheet I attached, the document says that "The insulation lifetime of the ADuM3200/ADuM3201 depends on the voltage waveform type imposed across the isolation barrier. The iCoupler insulation structure degrades at different rates depending on whether the waveform is bipolar ac, unipolar ac, or dc. Figure 15, Figure 16, and Figure 17 illustrate these different isolation voltage waveforms."

To me, this would suggest that you can transmit DC across the isolation barrier?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Since you have chosen to employ isolators from Analog Devices (a wise decision in my opinion) search in their linear isolation amplifiers web page.

https://www.analog.com/en/parametricsearch/11062#/

But since you are going to post-process the information digitally, you could also perform the A/D conversion on the hot side, an use the digital type isolator.
Okay, this could be a good idea. Just as a side note, it is okay to use these in a switching supply? Would you be able to use the device purely as an isolator and do the compensation elsewhere like in the DSP or is that not a good idea?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Why not use an analog type optocoupler?
Thanks! I could use an optocoupler. On the website I was viewing, which actually I arrived to after looking at optocouplers, it states that there are a lot of issues with opto solutions such as non-linear transfer functions. But then again I only need the device purely for isolation of a signal anyway so it may not be an issue. I don't really want to have to have a lot of external components which I usually see when an opto is used, as I only really want it (at the moment) for fault detection and limit. What do you think? :)
 

OBW0549

Joined Mar 2, 2015
3,546
In the data sheet I attached, the document says that "The insulation lifetime of the ADuM3200/ADuM3201 depends on the voltage waveform type imposed across the isolation barrier. The iCoupler insulation structure degrades at different rates depending on whether the waveform is bipolar ac, unipolar ac, or dc. Figure 15, Figure 16, and Figure 17 illustrate these different isolation voltage waveforms."
The text you've quoted is referring to the voltage across the isolation barrier and its effects on the barrier insulation lifetime, and has nothing to do with whether the isolator is designed to transmit digital signals or analog signals across the barrier.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
The text you've quoted is referring to the voltage across the isolation barrier and its effects on the barrier insulation lifetime, and has nothing to do with whether the isolator is designed to transmit digital signals or analog signals across the barrier.
Okay brilliant that makes a lot of sense. Have you any recommendations then, for the best way to isolate between the DC voltage and my DSP at an iso voltage of about 1kV?
 

OBW0549

Joined Mar 2, 2015
3,546
Have you any recommendations then, for the best way to isolate between the DC voltage and my DSP at an iso voltage of about 1kV?
I'm not sure how to interpret your question. What do you mean by "between the DC voltage and my DSP"? The whole thing seems pretty obvious to me: analog is analog. Digital is digital. And power is power. Which of the three are you trying to transmit across that 1 kV potential difference?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
I'm not sure how to interpret your question. What do you mean by "between the DC voltage and my DSP"? The whole thing seems pretty obvious to me: analog is analog. Digital is digital. And power is power. Which of the three are you trying to transmit across that 1 kV potential difference?
Well, there is a maximum 400VDC signal that I need to isolate... from my DSP. I know that analogue is analogue, digital is digital, and so on. I am creating a document which shows every single isolation barrier between each zone and the low power region so that I know I have safe levels of isolation between every danger and myself. I am simply asking what the best way to isolate this 400VDC from the low power board is. Some have proposed the opto-isolator, could use another transformer, my question concerned a digital isolator but that won't work - I am thankful I know that now - I only aim to improve and learn every day as a PhD student. :) Another has proposed a linear isolation amplifier from Analog Devices. Well, there seems to be multiple isolation methods between these signals.
The P.D is not 1kV. I just need the isolation level to be a minimum of 2 times higher than the voltage I am attempting to isolate.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Is the high voltage on the input or on the output of the opto-isolator?
The input side. I will be taking the 400VDC from a clamp capacitor on the primary side of my isolated DC-DC converter. I wish to process this in my DSP for mostly fault monitoring, as my demo board capacitors cannot exceed 400V. I may also use it for feedback control, I do not plan to do this but I think it would be nice to have as a possibility down the line. My DSP has a 15V supply and obviously is unable to handle the 400V input.
 

crutschow

Joined Mar 14, 2008
25,395
Why not use a voltage divider to reduce the voltage to that required by the DSP?
Or is the isolation required for human safety considerations?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
Why not use a voltage divider to reduce the voltage to that required by the DSP?
Or is the isolation required for human safety considerations?
Yes, that is also the plan, but the idea is that with a simple voltage divider if a fault occurs in the chain then the full 400V may be seen by the DSP or presented to the operator. The same could be said for a 6kV output, for example - you can't just have a resistive divider without some kind of isolation added.
 

TechWise

Joined Aug 24, 2018
126
I have employed the Texas Instruments AMC1311 to do isolated voltage sensing on a DC bus. You divide down the input voltage using a resistive divider so you can select your resistor values to scale the 400V down to the 2V required at the input. There's also the AMC1302 which accepts +/-50mV on its input. It's designed for current sensing applications with a shunt resistor but I've also used it successfully to measure phase voltages.

I have two isolation barriers in my system. One barrier is achieved by using the above isolated measurements and isolated gate drivers, the other is achieved by using digital isolators for the JTAG interface between the DSP and my laptop.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
238
I have employed the Texas Instruments AMC1311 to do isolated voltage sensing on a DC bus. You divide down the input voltage using a resistive divider so you can select your resistor values to scale the 400V down to the 2V required at the input. There's also the AMC1302 which accepts +/-50mV on its input. It's designed for current sensing applications with a shunt resistor but I've also used it successfully to measure phase voltages.

I have two isolation barriers in my system. One barrier is achieved by using the above isolated measurements and isolated gate drivers, the other is achieved by using digital isolators for the JTAG interface between the DSP and my laptop.
Awesome response and thank you for the attached part - cheers! Will look into it.
 
Top