View Single Post
Old 10-19-2014, 04:31 PM   #1246 (permalink)
EcoModding Apprentice
Join Date: Nov 2012
Location: East Midlands
Posts: 180
Thanks: 13
Thanked 81 Times in 52 Posts
I would say you guys are working some NASA project

When I designed my encoder I used a 40mA current source to carry the signals from the tachogenerator to the control board. This "cleaned" any noises I previously had. I could also use differential signaling, but it ended up never being required.

I also made sure the frequencies from the inverter were sufficiently spaced from those from my encoder. This allowed a use of filters to further limit the noise. I am aware Paul setup uses much larger frequencies, all the way from zero, so this may not be possible

I am genuinely curious to what happens should the encoder signal be lost or corrupted. Was this simulated?

I can't really advise, in regards to the suggestion to keep the GND away from the I/O pins. In fact RF systems I worked had everything grounded until the output port to serve as shield. I reckon your best bet is to open a support ticket with the manufacturer and ask one of their engineers some advice regarding the layout. I'm sure there's a reason for that requirement but without knowing which it is hard to advice.

For the supply feeding the analog supply, a series inductor can be used in order to remove any (generally HF) noise. The exact inductor value will have to be simulated accordingly to the frequencies used. For low power systems, I generally just place a 4.7R resistor. Same as the inductor, but with a slight voltage drop (0.0xV, nothing major).

Last edited by cts_casemod; 10-19-2014 at 05:44 PM..
  Reply With Quote