chassis ground

[Simon]

I have a design with significant filtering in it. I have several capacitors that will return common mode noise with respect to the chassis to source. I have put a chassis ground plane across the bottom of the board to help screen it and to have convenient points of access for capacitors. But is this a good idea?

My stack up is the chassis on the bottom, then negative power then two layers of signals.

[T3sl4co1l]

What's on top, more signals?  More ground/supply?

More grounding is usually a good thing, but it can be used incorrectly.  As with your other EMC question -- the only way to figure out is to test, or to provide enough details that an analysis can be performed.  (I'm guessing the former, since you've been quite mum on details for anything so far?)

Tim

[David Hess]

Tektronix did a lot of this with their oscilloscopes which used separate assemblies and it shows where baluns and other filtering circuits were used between assemblies to remove common mode noise.

[Simon]

What's on top, more signals?  More ground/supply?

More grounding is usually a good thing, but it can be used incorrectly.  As with your other EMC question -- the only way to figure out is to test, or to provide enough details that an analysis can be performed.  (I'm guessing the former, since you've been quite mum on details for anything so far?)

Tim

Well the summary is that i have these motors rated for 24V. The manufacturer insists strictly 24V and that the parts inside are rated for 40V Abs-max. These motors are brushless and shit! they don't do closed loop speed control and they don't have soft start so belt straight into whatever speed and if there is too much resistance/inductance in line they fail to start or oscillate mechanically in an attempt to start.

I have a linear voltage regulator to protect the motors fram any over voltage and a basic µC based system to read a tacho feedback and change the speed drive to keep a consant speed over any voltage.

Testing has shown a lot of noise in the 80MHz range. We suspect common mode from the motor driver coupling to the chassis of the motor and system. So I put filtering on the motor, a bit of a potshot guess using CMC and some differential filtering for good measure. This removed much of the noise but there is still noise at 85-105MHz (yes it appears to have shifted while partially dissipating). So I am building a new ECU section that has basic power input fitering with a CMC followed by capacitors to the chassis ground plane and then additional filtering on a seperate voltage regulator PCB that will have further differential and CM filtering, again with capacitors to chassis ground.

It is the ECU PCB that I wantet to put the chassis ground plane on. Obviously this will create a capacitance with the negative supply rail, so will this cause a problem, maybe if it couples before the CMC rather than after it.

[T3sl4co1l]

Ah, still wrestling with that stupid motor?

Hope your project budget doesn't run out

What's so bad about chassis ground? Is it supposed to be (galvanically) isolated?  What's wrong with RF-bypassing to it?  (I guess that was the question.  The answer is, probably nothing.)

Tim

[Simon]

Yep, still that stupid motor, even after i produced a study of this one versus another. We are isolated from the chassis so any noise coming aut on neg or pos relative to chassis so radiating out.

[coppercone2]

yes I thought of this and even to use multilayer boards to make distributed capacitance to chassis ground with interlapping fingers (pointless I think but interesting idea).

The problem is if you have chassis ground like that, and the case ground is disconnected or poor, you can introduce alot of noise to the circuit with close coupling. My idea was to make it and test it with a walki talki (seems like a plausible interaction, someone just installed equipment wrong and went on a walki talki to inform their boss its done).

If its well screened and lined up and everything I thought there would be no problem since its not much different then the coupling caps, but I think typically they do additional L after the chassis reference C to raise the impedance.

I mean you have mains filters like LCLC or LCL or LCLCL (I think I saw this in milspec version of a line filter) or just LC not sure if they use C first (when it comes to earth ground referenced C). If you have the C connected through a L you are isolating anything in your chassis by the filter L. If you have a C as the end node you get a direct low impedance path from earth ground to your PCB ground and PCB rail but you might still have some inductance from traces. You are directly wiring it into the entire PCB as a giant distributed capacitor so its a good capacitive coupler.

Maybe float the plane and connect it to a RF gen to see what happens?

[Simon]

i have common mode LC filters so that any CM noise coupled to the case in the motor returns to the poawer lines rather than bounce the chassis all over the place. The there is a CMC so that there is a high impedence between all the noise and the power input. So my only concern is that too much capacitance between the chassis and negative allows coupling after the CMC.