EMI issue

Technical Discussions on any InterfaceKits
danhoskin
Phidgetly
Posts: 45
Joined: Wed Jan 16, 2008 6:05 pm

EMI issue

Postby danhoskin » Tue Jun 08, 2010 12:25 pm

Hi,

I have had my project sitting in a corner for a few weeks now due to an issue with losing the connection with the Advanced Servo controllers.
This is the current setup and a description of the environment that is operates in:

-Win XP PC with VB6 program
-USB extender rated at 100 ft. limit with 50ft cat6 cable
-3 Advanced Servo Controllers
-6 individual 3 DOF arms built on paralleled Hitec HS-645MG servos for shoulder and elbow joints and single HS-645MG's for the wrists.
-servo leads are about 4 ft. long and are made from a sheilded cable, grounded at the cabinet end only (where the servo controllers are)
-arms are mounted about 6" apart

The PC is in a cabinet 10 feet away from a steel paint booth where the arms are installed on a DC motor-driven carrier running along an "I" beam 20 feet long. The arms purpose is to position small paint heads that spray material 16 feet long of many different contours.

The issue that I am having is that whenever I run the fan on the paint booth the controllers will occasionally disconnect with from the PC. When they disconnect they no longer not show up in the Phidgets Control Panel or respond to the software.

I have tried putting a power filter at the 110v source in the cabinet above the paint arm that house the Phidgets, hoping to eliminate any EMI from that end but this causes the arms to jump around as soon as I turn the booths fan on. I have even tried using an SBC wirelessly at the paint head end to eliminate any possibility of EMI through the network cable.

The question (at last) is, how can I reduce / eliminate the EMI from the booth fan and other sources that are causing the unit to disconnect from the PC? I have read about the EMI issues that stem from long servo cables on RC models, but have not found anything that will resolve this issue.

Any thoughts on what I could try? The thing runs flawlessly on the test bench, complete with the 50 feet of USB extension. Everything is the same on the test bench as when it is installed except the fact that it is installed on the steel beam, is in the paint booth and is next to the drive motor for the carriage that it is mounted on.

I am really stumped, any suggestions would be appreciated.
Thanks,
Dan

Robert

Re: EMI issue

Postby Robert » Tue Jun 08, 2010 12:47 pm

Placing a filter on the fan's power likely isn't going to fix anything. The interference is likely coming from the windings in the fan motor itself.

There are several places that EMI can be entering your system, but it's likely coming in either on the long servo wires, or the long USB cable. The first test would be to disconnect the servo motors, and see if the device still drops out. If it does, it's likely a USB problem. If the problem goes away, the problem is likely with the servo cables.

The quickest and easiest solution is to by a few large ferrite beads, and place them on each end of the USB and servo cables.

If that doesn't work, you should look into the shielding that you have on the servo cables. Having the shield connected only at one end could cause it to act like a giant antenna, funneling the electromagnetic energy into your ground plane. Try connecting the shield at both ends, or removing the shield completely. We've found that having an open-ended shield like that is almost always worse than no shield at all.

danhoskin
Phidgetly
Posts: 45
Joined: Wed Jan 16, 2008 6:05 pm

Re: EMI issue

Postby danhoskin » Tue Jun 08, 2010 12:55 pm

Thanks for the quick reply, I'll try unplugging the servos first and see what happens.
Dan

danhoskin
Phidgetly
Posts: 45
Joined: Wed Jan 16, 2008 6:05 pm

Re: EMI issue

Postby danhoskin » Thu Jun 10, 2010 8:00 am

I unplugged the servos and it still disconnects, especially when I put my EMI source (110v drill switching rather abruptly between forward and reverse) at the USB extension cable. Reconnecting the servos did not make it any worse, even with EMI right at the servo leads.
So, I ordered some ferrite beads from Digikey and will see if that helps. I am also going to make a seperate aluminum cabinet for the Phidgets and their individual power supplies to help spread things out a bit.
One thing I was going to mentiuon that I found unusual was that one of our guys plugged a vacuum cleaner into the same outlet that the project is plugged into and the arms jumped like crazy, but when the vacuum was plugged in on the other side of the room (vacuum itself in the same place) it didn't affect the arms at all. Does that tell us anything?

Robert

Re: EMI issue

Postby Robert » Thu Jun 10, 2010 10:21 am

It definitely sounds like you USB cable is where the noise is being introduced into your system. What model USB extender are you using?


As for the vacuum cleaner, if the motors are jumping around when it's on, the noise from the vacuum cleaner is being introduced into your servo cables, corrupting the position signal. It's a very simple signal, and can easily be changed by interference. There's no sort of error checking or noise rejection. USB on the other hand is a little more robust. It's very difficult to introduce errors into the USB bus that will change position. Either you end up with good USB data, or no USB data.

danhoskin
Phidgetly
Posts: 45
Joined: Wed Jan 16, 2008 6:05 pm

Re: EMI issue

Postby danhoskin » Thu Jun 10, 2010 10:43 am

I am using a 'StarTech USB 110' extender.

As far as the arms jumping around and the vacuum, I am curious why it doesn't do it unless the vacuum is plugged into the same outlet as the Phidgets. The vacuum itself is sitting in the same place, only with the cord plugged in across the room.

Robert

Re: EMI issue

Postby Robert » Thu Jun 10, 2010 11:13 am

Because the noise is probably be coming in on the power cable through the ground plane, to the servo motors. A ferrite on the power cable, or a better isolated power supply should fix it.

danhoskin
Phidgetly
Posts: 45
Joined: Wed Jan 16, 2008 6:05 pm

Re: EMI issue

Postby danhoskin » Thu Jun 10, 2010 12:36 pm

OK, thanks again. That makes sense now.

richfiddler11
Phidgeteer!
Posts: 55
Joined: Mon May 24, 2010 2:05 pm
Contact:

Re: EMI issue

Postby richfiddler11 » Thu Jun 10, 2010 8:00 pm

Just my 2 cents -- I worked at the CMU Robotics Inst.for 10 years where I created and solved all kinds of problems like this. These can be tricky problems to solve.

If you have the possibility to run your servo signals differentially, that can cure a world of problems. If the signals are PWM rather than straight DC, then you can sometimes put a balun (balanced-to-unbalanced transformer).

Also, system grounding is something you should look at. A few general rules of thumb are

- single point of grounding if possible

- devices at the end of a long line should float w.r.t. earth ground, if possible, to avoid ground loops

- run signals differentially whenever possible

- use separate power supplies for noisy vs. sensitive system partitions.

Good luck.

Rich

fraser
Engineering
Posts: 254
Joined: Thu Nov 19, 2009 4:41 pm
Contact:

Re: EMI issue

Postby fraser » Fri Jun 11, 2010 9:03 am

We have lately been focusing on the reduction of board level EMI, expect improvements in the future.


Return to “InterfaceKits”

Who is online

Users browsing this forum: No registered users and 3 guests