Analog input card voltage source

[RyreInc]

We have a linear pot that precisely measures position. The pot is fed from a precision 10VDC source, and is fed into a ±10V analog input card (14-bit resolution; only 13 bits are utlized since the signal is 0-10V).

My idea is to do away with the 10V source and instead use the pre-existing 24VDC power supply source, running the pot output to a 0-30V analog input card (14-bit). This card has two channels--the second channel, currently unused, would have the 24V line running in to its analog input directly. This way I could take the ratio of the pot output to the 24V signal and remove any fluctuations in line voltage.

This should result in a 60% increase in resolution, plus do away with the 10V power source.

The 10V and 24V supplies both have a 0.5% line regulation, and 0.7% and 0.5% load regulation respectively. Of course the load on the 24V supply will vary much more greatly.

Both cards are differential input, although that shouldn't really matter.

Is this a bad idea? What are some caveats I am not thinking of?

Thanks!

 

[itsmoked]

The biggest one is that the power dissipation in the pot is going to increase significantly. Remember it will go up with the square of the voltage change. You need to do the power calcs for the pot and check with the pot data sheet to see what the maximum dissipation can be. I suggest you stay about 50% below it.

Another is that while the voltage can now be ratio'd that only works if you're taking the samples closely and then it works not at all for noise related issues which are very fast events. If you can supply some filtering that will carry noise on the 24V across both samples, then it should work fine. The filter would consist of an R between the 24V rail and a bulk capacitor to the rail return. Feed the pot from the the R/C capacitor. This will actually drop some of the 24V but since you're now ratioing it that won't matter.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[RyreInc]

Thanks Keith.

The pot's recommended maximum input voltage is 42 VDC, and has a resistance of 10kΩ.

The analog input card has a first order low pass filter at 5kHz. Is that low enough? Also what if we did software filtering instead of hardware--the sampling rate does not need to be super fast.

 

[itsmoked]

SW filtering doesn't work well here since there is no correlation with the sw filtered results and the near simultaneous readings.

5kHz on the inputs suffers from the same issue. You can have an instantaneous noise spike that occurs on only the pot reading or only the reference reading and there is no way to resolve it after-the-fact, but it will change the result of your measurements in a manner having no relationship to pot position - something I image is not desired. You have to keep the reference voltage stable during the short time between adjacent readings.

How fast are you going to be reading the these two inputs?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

a precision 10V reference like REF102:

http://www.ti.com/product/ref102

is pretty cheap. Seems hardly worth the bother to create a complicated filtering and SW processing scheme to save a couple of bucks.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529


Of course I can. I can do anything. I can do absolutely anything. I'm an expert!
There is a homework forum hosted by engineering.com:

http://www.engineering.com/AskForum/aff/32.aspx

 

[RyreInc]

This is for industrial machinery so chip-based solutions are out, unfortunately. My reasoning behind this was to get rid of the ($100) 10V reference source and utilize a spare channel on the A/D card. Adding software filtering is essential free, since this is a product we sell repeatedly.

For the same reason adding an RC filter is out as well, unless it has industrial, probably DIN-rail, packaging, which will considerably increase the price.

I'm not 100% sure on the input scan rate, but it is between 15-50ms.

 

[djs]

Unfortunately if you are using the 24VDC PS to supply analog power to other devices connected to your controller, then voltage variations in the 24VDC voltage will be in sync with your controller, and no amount of filtering will eliminate them. What's more the period of these variations can be in the second or minute range due to the sync with the controller. At this point how precise does your measurement need to be?

 

[RyreInc]

Well I need at least 12, maybe 13, out of the 14 bits to get the desired resolution.

At this point I'm thinking this idea will not work. Thanks everyone for walking me through this--I kinda figured there were some aspects I wasn't considering, and I appreciate the feedback. Good stuff!

 

[itsmoked]

RyreInc; I'm not following your reticence... I'm talking about a single capacitor and a single resistor. Both can be tagged in between the 24V supply and your use rather easily, dare I say trivially?. It could be done right at a small terminal block, or in the newer din rail connector blocks, or possibly right at your analog card's terminals... It's so trivial it completely negates the hassle of doing any software.

Keith Cress
kcress -

http://www.flaminsystems.com