Take the 2-minute tour ×
Electrical Engineering Stack Exchange is a question and answer site for electronics and electrical engineering professionals, students, and enthusiasts. It's 100% free, no registration required.

I'm designing an homebrew RF spectrum analyzer.

Controlling the analyzer will be done using USB which is converted to UART using an FTDI chip.

My concern is that noise (probably common-mode) will be injected from the PC (USB/UART) and affect the SNR of my spectrum analyzer. I know that PCs are very noisy.

My idea was to use some separation(probably an optocoupler) to completely separate the PC from the spectrum analyzer.

Does anybody know whether this is necessary or not?

share|improve this question
    
Check out the ADUM4160. I dont know whether isolation is necessary, so I cannot provide an answer, but if you do, take a look at this chip. (analog.com/en/interface/digital-isolators/adum4160/products/…) –  cksa361 Dec 7 '11 at 20:45
1  
It's probably more important to shield your highly sensitive receiver circuits than to specifically isolate your digital control circuits. After all, there could be lots of external interference sources (CRTs, fluorescent lighting, ...) that could be just as problematic as your USB circuit, and that you have no control over as you design your instrument. That said, if you do choose to isolate the USB circuit, I'm not sure an optocoupler is meant to do exactly what you want --- I think of them more for dealing with high voltage (100, 200, 400 V) transients, not ordinary interference issues. –  The Photon Dec 8 '11 at 2:47
1  
@ThePhoton - an optocoupler is workable (provided it's on the serial side, not the USB side). I recall seeing a receiver design that used visible LED/transistor pairs for most of the internal control signals, both to reduce coupling and because it "looked cool". –  Chris Stratton Dec 8 '11 at 21:35
add comment

2 Answers

I definitely would apply galvanic isolation, for the reasons Steven mentions. Especially ground loop currents can create offset voltages over all your circuitry.

It may sound logical to have the isolation on the UART level, because Rx and Tx are unidirectional, but you can also have the isolation on the USB. If you want to go that way, don't DIY this, buy a commercial product:

enter image description here

This one is USB2 Full Speed and costs 34 EUR.

share|improve this answer
    
If you're making your own board, the ADuM4160 is a single-chip USB isolator. They're really easy to use, and likely what's actually inside the commercial product above. –  Connor Wolf Jun 14 '12 at 21:06
    
Note that this means you will still have the high speed USB signal downstream of the isolator. –  Chris Stratton Jun 15 '12 at 16:55
    
@Chris - er, yes, is that a problem? And after the FTDI USB-UART bridge it will be high speed UART. Sorry but I fail to see your point. –  stevenvh Jun 15 '12 at 16:57
    
@stevenh USB is likely a higher data rate than your UART, and is not quiescent even when nothing is being exchanged. On the UART side, you can lower the baud rate if needed for sensitive measurements, and use feedthrough caps to limit the edge speeds. –  Chris Stratton Jun 15 '12 at 17:27
add comment

While noise from the PC may be an issue you could argue that adding some form of galvanic isolation (e.g. opto-coupler) is good for two reasons.

  1. Isolation of noisy GND and circuitry: If there is noise being generated by the PC or its supply then the ground is likely how said noise will couple from the the source to your sensitive analyzer circuitry. Isolation limits this method of coupling.
  2. Ground loop: This is the big one and if not handled properly may lead to communication problems for your UART interface. If you share the PC GND and the inevitable external GND that you will require for your quiet PSU to power your analyzer circuitry there will be a ground loop. When this happens both sides of the UART interface may end up referencing different voltage levels because one GND can be higher or lower in potential than the other. The result may lead to a broken or intermittent UART interface as the logic levels no longer correspond due to the different reference.

I suggest you implement your galvanic isolation using some opto-couplers on your UART interface. This ensures that your PC GND and analyzer GND are not connected. As a result there is no noise and no ground loop. Once that's done you can implement a very clean power supply for you analyzer that is powered externally (i.e. not from the PC PSU).

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.