Filter 8 RTD channels with one DFB

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
MiSt_296941
Level 4
Level 4
25 replies posted 10 replies posted 10 questions asked

We currently have a product that uses the PSoC3 to measure 8 RTDs (or TC or mV).  We use Correlated Double Sampling at 2x the line frequency to try to reduce line frequency noise.  We have some customers who have experienced too much noise due to the harsh environment so I was thinking that maybe we could improve our product by using the DFB with a notch filter.  The DFB has 2 channels so on the surface it would seem impossible, but I read an EDN article by Kendall Castor-Perry who used to work for Cypress that specifically talked about doing this, but gave no details.

How would we run 8 channels through the DFB in real-time?

Mike.

0 Likes
1 Solution
odissey1
Level 9
Level 9
First comment on KBA 1000 replies posted 750 replies posted

Mike,

I remember that paper which you are referring to. Few years ago I also tried to contact the author about it to no avail. But in your case of temperature measurement using the DFB might be an overkill, as sampling rate is likely very slow. I believe that filter can be implemented in code. To better understand the nature of the noise I recommend to test one DFB channel with various filters to see what works. it may happen in the industrial environment that the noise is sporadic (not normally distributed), then the linear filter (such as DFB) will not do. In such case a median filter can help:

MedianFilter: sliding window median filter component

This median filter works well for rejecting random spikes. It is implemented entirely in software, so should run OK on PSoC3 (or you can copy/paste the code)

In case that the source of the noise is harmonic (60Hz), this notch filter may help:

Harmonic Notch Filter - Mike

/odissey1.

View solution in original post

0 Likes
2 Replies
odissey1
Level 9
Level 9
First comment on KBA 1000 replies posted 750 replies posted

Mike,

I remember that paper which you are referring to. Few years ago I also tried to contact the author about it to no avail. But in your case of temperature measurement using the DFB might be an overkill, as sampling rate is likely very slow. I believe that filter can be implemented in code. To better understand the nature of the noise I recommend to test one DFB channel with various filters to see what works. it may happen in the industrial environment that the noise is sporadic (not normally distributed), then the linear filter (such as DFB) will not do. In such case a median filter can help:

MedianFilter: sliding window median filter component

This median filter works well for rejecting random spikes. It is implemented entirely in software, so should run OK on PSoC3 (or you can copy/paste the code)

In case that the source of the noise is harmonic (60Hz), this notch filter may help:

Harmonic Notch Filter - Mike

/odissey1.

0 Likes

odissey1,

Thanks for the response.  I actually am having a conversation with Kendall via LinkedIn.  We will see where that goes. 

Our issue is that this occurred in a plant half way around the globe from us and we don't have a way of replicating what they are doing.  They were measuring Motor Stator Temperature with 3-wire RTDs (probably on some giant 3 phase motor).  We know two things for sure - when the motor was off there was no noise, so that pretty much shows that it is line frequency.  Secondly, at one location when only one RTD was connected (we built a 16 channel mux using 2 PSoC3's) the reading was stable.  As soon as the second RTD was connected then the reading was noisy.  This points to a channel to channel isolation issue which is very difficult to do anything about since we are using the PSoC as a mux and all IO pins have ESD diodes on them - the one to ground being the biggest problem.

Mike.

0 Likes