Frame ground vs. "neutral" for DC

For emi protection (which you sure as heck need) bolt everything to frame ground as often as possible, and hard ground cable shields at both ends. Multilayer boards with a solid internal ground plane help greatly, and don't even think about separating analog and digital grounds.

John

Got it, good idea about EMI. I will follow your advice exactly.

What do you mean about "separating analog and digital grounds", are you saying that my microcontroller should be grounded to the same ground as all other parts of my circuits?

I am installing all control pieces on one piece of aluminum.

Yes!

Good. Ground the boards to the plate with aluminum spacers if possible (some idiots actually *isolate* the pcb mounting holes, can you imagine that?) Now most EMI problems will be conducted in on leads, which can be managed if grounding is stiff everywhere.

John

OK, good, yes, they will all be grounded to the same machine screw in that aluminum plate.

OK, I am not sure how many boards I will have, maybe a couple of breadboards. Wil make sure to connect ground to frame ground.

Another related thought: my control will have several "outputs", such as 110V signals to the SCR firing system (mounted elsewhere), inhibit signal for same, some 110V commands sent to water solenoid, 110V output for water pump, high frequency arc starting, etc.

Even though my microcontroller has 110V capable relays mounted on it, I plan on making its outputs 12V only and connect them to 12V mechanical relays, which would in turn be connected to 110V power.

The idea that I have here is that this way, wires leading into the compartment with HF arc starting would be additionally isolated from the controller, by presence of mechanical relays not mounted on controller.

There seem to be 2 schools of thought on that last item.

Hello John,

Sssst.... quiet, you are destroying part of my business foundation here. What should I do if everyone does it right from the beginning?

That's a smart thing to do. Who knows what is riding on those mains lines when the thing starts welding. You aren't running this off 240V two-phase? Even then I often popped a 16A breaker with my (regular) welder.

Also, I'll second John on the ground philosophy.

Hello Richard,

And I enjoy some business because of that fact :-)))

Good. Thanks.

Well... My welder has two incoming power lines.

One is 120V single phase line for all control systems, little transformers, contactors, etc.

Another is 240V 3 phase line from my phase converter. Its ONLY function is providing power to my three phase main welding transformer.

When I weld at 200 amps, the current at the circuit breaker is about

Like Shiites and Sunis, Baptists and Methodists. All those other guys have to do is explain why they want a bunch of different planes at different RF/spike potentials. The *real* loonies want to run wires from all the planes clear back to a "star ground" at the power supply.

One of my customers makes excimer lasers, with everything bolted to a humongous plate aluminum "optical bench", and they insist on isolating grounds. Then the fill each box with plane-plane and plane-case bypass caps to make it work.

John

It couldn't hurt to provide a good EMI filter on that line, Corcom or something like that. With properly rated fuses, of course.

Oh, sorry. But they won't listen, anyhow, so not much harm done.

On Friday I had to make an emergency speed-limit-busting run down Interstate 280 to my customer. The CE inspection guys were there and my box had just "broken" after working fine for weeks. It connects to their probe gadget through two humongous custom cables, and they had removed the probe from its superconducting magnet and put it on a table to make it easier for the CE droids to poke around.

You know the rest: their people believe that a cable shield should only ever be connected at one end, their probe is full of emi filters, and I'm conditioning a thermocouple and driving a heater PWM. 48 volt spikes on a thermocouple input is way over my personal limits.

The fix: run a BNC cable from the monitor output of my box to the RF connector of their probe. I have no idea how those signals interact, but the shield is all we needed.

John

On just the single phase 120V control line? Why on it only?

I do have some EMI filters in my pile, and will definitely install them if you give me what "might be a good enough reason" (since there is no apparent harm, I would do it if I feel that they "might" improve something).

Hello John,

In the EMI world it's worse. Universities, part manufacturers, books and so on mostly advise to split grounds. So they are almost all on the other side of the Klondike. A few consultants know otherwise and so do the defense guys but those two groups have much lower visibility (for very different reasons).

Sometimes when arriving at clients we have to take a deep breath and bite the tongue.

from what i have been told is to gnd all at one point this will cut emi

From your description I assume that this is the mains line that supplies the juice to your micro controller board and all kinds of other control circuitry, via transformers. You don't want any arc spikes traveling towards those circuits. It won't get much past good bypass caps in the supplies but the cables have a chance to radiate some of it off before it gets there. Somehow that 120V line has to make its way into the control enclosure. That entrance point would be a nice spot for that filter, with its case grounded smack dab right there. That's how we design medical gear so it can function in very noisy situations.

Gosh, Joerg, you're giving it away!

But, like I said, most people will do it wrong anyhow.

My other pet peeve is optoisolators, like wearing a garlic necklace to keep vampires away.

John

Hello John,

Those are no fun on 65mph freeways. My first emergency run was from Cologne (Germany) to Paris. Almost up to the Belgian border no speed limit at that time. Same for the runs to Bavaria. Vroooom, yeehaa!

The only moments of fright were when thinking about that HP analyzer in the rear of the station wagon. Had I augered in that thing would have decapitated me.

Oh boy. Maybe you should use orange cables and attach some obscure warnings. IIRC nobody ever dares to touch anything painted orange under the hood of a hybrid. At least not without a good life insurance policy.

yes, and also note that the microcontroller takes 120VAC input power.

I will redo my entrance (into the control compartment) to work this way. Thanks.