Eliminating Noise on a PCB Board.

[Stanfi]

I've got A PCB board. It has circuitry that supplies 12VDC to a load cell, and then amplifies the 0-30mvdc signal to 0 to 10VDC. This board is housed in a steel box. When I close the door of the enclosure, my 0 to 10VDC signal will increase .15VDC, then opening the door the singal will return to it's orginal value.

I believe this is due to EMI interferance. How can I elimated this problem? any advice would be appreciated.

 

[buzzp]

Sounds like you have some parts located inside right next to the steel enclosure, too close. That or you have something generating noise in your design (fast switching devices or signals or maybe not so fast).
I would try moving your device outside or some other location and see if the signal increases still. If it don't then it sounds like your enclosure is capacitively coupling with something in the design that may be too close to the enclosure. If it still increases then something needs to be done in your design, if the 0.15 Volt increase is an issue.

 

[Stanfi]

I can place a capacitor across the + and - terminals for the incoming mv singal, and it does help decrease the fluctuation.

 

[Comcokid]

Your problems are very simliar to the following active thread:
thread240-107822

Most load cells are like most bridge pressure transducers - a Wheatstone arrangement made of strain-varying resistors.

I once troubleshot a load cell offset problem that turned out to be a unreleated sensor whose ground return path included the circuit board on which the load cell interface circuit. Make sure your box doesn't have a interlock switch or other item that shares a power or return path with your load cell interface circuit.

 

[Stanfi]

I have .5" from my pcb board to the edge of my enclosure.. Is this to close???

 

[Stanfi]

I thought maybe I should tell a little about my circuit. It uses a LM723 Volatage regulator to supply the 12VDC excitation voltage. An AM411 chip that amplifies the 0-30mv signal to 0-10VDC. and and NMR106 dc/dc convertor that provides an isolated voltage across an couple of resistors and 49.9K and 500 ohm pot to create a 0-30mv offset voltage for the incoming 0-30mv signal. The offset circuit allows me to offset "dead load" on my load cell signal.

Thanks

 

[buzzp]

I missed the part about opening the door and then the problem goes away. Sounds like there is something generating trash (DC/DC converter? or some other device in the same enclosure) that is being picked up before your amplifier or after for that matter. When the door closes, this trash is trapped inside and when you open it, you let the garbage out.
Thats not to far of a reach in my opinion.
I dont think anything is too close to your enclosure since I missed the door closed part of your post.
The fact that you used a cap across the 30mV signal and it helped would indicate noise being rectified and amplified. What type of cap and what was its value?

 

[Stanfi]

The cap is 2.2uf ceramic type. I am in the process of building one of these boards circuit by circuit. After completing each phase I test to see if the noise appears. I will post what I find.

Thanks

 

[BrianG]

In addition to previous posts about the door conducting noise around the box from some otyher source, there is a possibility that source is your on-board 12V power supply.

The LM723 can only supply about 150mA directly and needs a separate series pass transistor to boost its current range beyond this. Have you checked with a 'scope to see that the power supply is not oscillating at a very high frequency (can sometimes be 10-50MHz)? This this can occur with many linear regulator designs if the layout is poor or there is inadequate h.f. decoupling.

If it is oscillating then moving door can behave like a "proximity detector", producing big changes in frequency or amplitude which get into the load cell and amplifier input via the 12V power supply rail.

 

[Higgler]

This might be simple dc currents flowing across the door. Possibly try opening the door and connecting some aluminum foil across the door opening. Measure the voltage as you place more and more aluminum foil across the door. Might be informative, and it's a simple test. Maybe cover it all up initially, then cut slots in the foil.

Another thought is that RF coupling between leaky/radiating objects increases dramatically in a cavity compared to an open box. Energy from a switching regulator would flow around the inside of the box much more when the door is closed.
kch

 

[Tofflemire]

Hi,

I am assuming you have connectors on your box; we have had similar problems in the past. We improved the noise by installing filtered connectors; these cost a little more, but are still cheaper than trying to find RF noise. Another thing to check out is if you have wires coming off the door, the position, twisted vs. shielded all have effects on this noise. We have also installed ferrite beads to improve the noise.

Do not get me wrong, these two options are not fixing the problem, but may help in the short term.

Tofflemire

 

[nbucska]

I had a case when the cabinet acted as a cavity and
when closed, an part of the circuit oscillated and
an other part rectified this generating a huge offset.

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[Stanfi]

Thanks to everyone who has responded to this thread. The problem asssociated with the voltage rising turned out to be related to the DC/DC Converter.

Also couple with this problem was that the output singal (0-10V) after a would deadily decline. To make a long story short, and I eventually got the chip back on a bread board and it still drifted. Supply constant exiction with a regulated power supply, and had a constanct mv load cell input. The output of the chip drifted from 3.44 to 3.22 in an hour. This is not acceptable with my application.

I am now redoing my desing with a INA125 chip.

Thanks again to all.