Is there any reasonable way to assess the quality of grounding at a power receptacle with limited resources? (by which I really just mean a fair quality multimeter)

The situation is that we have several prototypes of a new machine and we've just set up two of them at a trade show. The two at the show are now regularly exhibiting problems that we'd seen briefly early on and thought we had already fixed. None of the machines (including the two at the show) have had this problem since we made changes and I'm wondering if something unique to the trade show environment (weird voltage, bad grounding, etc.) could be causing trouble.

The problem is this: the machine includes 220V (meant to accept at least 208-240 without trouble) power, including switching 220V pump motors on and off, along with several microcontrollers running on a 5VDC supply. The microcontrollers communicate via RS485 (a rather poor implementation at that: data lines are in a ribbon cable instead of twisted pair, and are not shielded.) Fairly often a message that gets sent exactly when a motor gets switched on isn't received properly. We were already planning to improve grounding practices within the machine and add snubbers to the motors (based on conversations in an earlier thread here) before we discovered these data problems, so we made those changes and haven't had a problem with any of 4 separate machines since then... until we got to the trade show.

I imagine significantly higher mains voltage would mean more rf/emi, but we already run around 246VAC in several of our test bays, so it's hard to imagine that the convention center is giving us too much more than that (but I'll check tomorrow.) The only other thing I could think of is rotten grounding, but I don't really know how to go about testing it. Any ideas?

For anyone interested in more background, this is closely related to an earlier thread of mine:
http://forum.allaboutcircuits.com/threads/tracking-down-dealing-with-voltage-spikes.100777/

 

Just to clarify the timeline: we built the first prototype many months ago, discovered the tweaky RS485 communication problem maybe 6-8 weeks ago, and then changed our grounding scheme and added snubbers to the motors maybe 5-6 weeks ago.

After making those changes, we never had another problem with that machine. Since then, we've built three more prototypes to the new specs and all of them worked perfectly for several weeks in our shop. Now we've set two of them up at a trade show and been having issues quite regularly there for both days we've been there.

 

Just a quick thought before I run out the door for a family visit...
Did you try a transient suppressor on the mains power outlet where you feed all your machines from?
Might be that the mains is crapped with transients, if loads of people draw off it with whatever they have.
You can buy finished mains "protectors" that kills the worst off - or if you have a UPS that you could run your machines off.

 

Let's set aside the issue of grounding for now.

What is the RS-485 baud?
What is the length of the RS-485 cable?
Is there a common GND in the RS-485 cable?
How are the RS-485 lines biased and terminated?
Show us the RS-485 circuit schematic with bias and termination resistors.

 

Here is a $5 item that will give you an instant view of the situation.
http://www.guitarcenter.com/America...4@ADL4GC-adType^PLA-device^c-adid^57619015002

Google: "Ground Fault Outlet Receptacle Tester"

 

Just a quick thought before I run out the door for a family visit...
Did you try a transient suppressor on the mains power outlet where you feed all your machines from?
Might be that the mains is crapped with transients, if loads of people draw off it with whatever they have.
You can buy finished mains "protectors" that kills the worst off - or if you have a UPS that you could run your machines off.

The tvs is a good thought, although I'm afraid I don't have time to choose and install anything in time. Today's our last day at the show and whatever testing I do needs to happen in the next 3 hours, before the show opens again.

As for UPS, that would be tough. Forgot to mention, but the machines also have high power heating elements and so they draw between 20 and 40 amps, depending on which model you're looking at.

 

Let's set aside the issue of grounding for now.

What is the RS-485 baud?
What is the length of the RS-485 cable?
Is there a common GND in the RS-485 cable?
How are the RS-485 lines biased and terminated?
Show us the RS-485 circuit schematic with bias and termination resistors.

Sorry I don't have answers to most of those questions.
Baud is unknown, but I could probably get an answer early next week.
Cable length is between 5-6 feet, with one connection to a master control unit which does all the talking, and then 2 to 3 receiving units spaced roughly 1 foot apart at the opposite end.
Yes, a +5VDC supply and the ground wires for the receiving devices run through the same 10 pin ribbon cable that carries the RS485.
I don't know about the biasing, but should be able to get the schematic early next week. For some unknown reason the EE who set all this up determined that termination resistors weren't really needed here, so there is nothing being done for termination.
I'll get a schematic up when/if I can. I'm not even sure we have a schematic of the receiving devices yet... we got the pcb layout files to send out for production, but I don't think we ever got proper documentation otherwise.

Anyway, aside from the myriad things that are less than ideal about our RS485 implementation, I'd still like to know why simply reducing the motor switching noise and improving our grounding seemed to be a 100% fix in our shop, but our machines are still struggling so much at the trade show.

 

Here is a $5 item that will give you an instant view of the situation.
http://www.guitarcenter.com/America...4@ADL4GC-adType^PLA-device^c-adid^57619015002

Google: "Ground Fault Outlet Receptacle Tester"

Unfortunately, these issues are on 240VAC outlets ranging from 20-50 amp supply. So, whatever I check will have to be done with my multimeter.

 

I guess for now I'm just going to check whatever I can with a multimeter in the hopes that something obvious turns up.

I'm thinking:
1) Check mains voltage to see if it's outside of our expected range
2) Check each hot leg vs ground
3) Check for voltage first, then continuity between our 240V plug's grounds and nearby 110V plug's ground to see if any of our grounds have voltage on them or have high resistance back to the proper ground.
4) and of course verify our own wiring from the point where the convention provides us power to the points where our machines plug in. That much should've already been checked, but I haven't seen it with my own eyes, so I'll verify.

Anything else worth testing on site with a multimeter? I've only got a couple hours. After that I have to be out of the way for the machines to run publicly again. As soon as the doors close tonight this will all get disconnected and brought back to our shop.

My fear is that after this show we won't be able to reproduce the problem in our shop and we'll never know if the problem is something our machine shouldn't ever have to deal with (like an open ground) or if it's something we'd really like it to be capable of handling (like slightly noisy AC.)

 

Anything else worth testing on site with a multimeter? I've only got a couple hours. After that I have to be out of the way for the machines to run publicly again. As soon as the doors close tonight this will all get disconnected and brought back to our shop.

I can't think of anything else worth checking, given the limited time available; when the curtain goes up, you're going to have to run with what you've got unless you can beg/borrow/buy/steal a really heavy-duty power line filter.

My fear is that after this show we won't be able to reproduce the problem in our shop and we'll never know if the problem is something our machine shouldn't ever have to deal with (like an open ground) or if it's something we'd really like it to be capable of handling (like slightly noisy AC.)

Best not to count on having only "slightly" noisy AC. Far better to design things on the assumption that your input power, and the EMC environment generally, will be simply horrible.

 

Best not to count on having only "slightly" noisy AC. Far better to design things on the assumption that your input power, and the EMC environment generally, will be simply horrible.

Agreed. Most of that work was done by real EEs and was meant to have been done to handle harsh environments, although I'm beginning to wonder...

That's why I'd like to figure out what's causing our problem here. I don't really care that it's not perfect for the show... we've pitched this all a experimental features and product in development, so glitches at the show are pretty acceptable.

If these issues are simply due to imperfect AC power, then I think our EEs need to figure out how to make this system more robust. If there's something bigger going on, like faulty grounding, then I don't think any redesigns are warranted, because it's fair to insist that the machine never be run without proper grounding (we have to insist on that anyway for safety reasons.)

 

Unfortunately, these issues are on 240VAC outlets ranging from 20-50 amp supply. So, whatever I check will have to be done with my multimeter.

And I'm sure they make one for 240VAC as well. My point was to introduce the concept.

 

Agreed. Most of that work was done by real EEs and was meant to have been done to handle harsh environments, ...

As opposed to the fake EEs being graduated by mail order universities.

 

... just reread my last post in the context of the earlier questions about RS485 configuration. So clearly, with the lack of shielding, twisted pair wiring, and terminations on our RS485 system, there is room for improvement in terms of handling harsh environments. Sadly, this is now the second major redesign of communication systems in the machine, and for whatever reason the second team of EEs (contractors in each case; we don't have in-house EE or software capabilities) chose to continue with ribbon cables and lack of terminations. The first version has run successfully for many years now, albeit with some kludgy software workarounds to beef up the RS485 communication. It would be a hard sell to scrap all the work that's just been done and go back to the drawing board yet again, but if that's what it would take to kill these glitches we may have to consider it.

 

As opposed to the fake EEs being graduated by mail order universities.

I just meant that although I'm learning everything I can, I'm not anything close to an EE (obviously) and had nothing to do with any of the design work, other than the recent changes to the grounding and adding the snubbers, both of which seem to have improved our situation, although perhaps not enough.

Maybe I could get one of those fake, mail order degrees!

 

And I'm sure they make one for 240VAC as well. My point was to introduce the concept.

That's a good point, although I think in this case I don't have time to get one and check things out. Maybe next time.

 

That's a good point, although I think in this case I don't have time to get one and check things out. Maybe next time.

Next time you go looking for a house to buy, having one of these things will clue you in to the presence of a do-it-your-selfer as opposed to a licensed electrician. Even they make mistakes however.

 

Well, no major obvious wiring problems at the trade show.

One oddity: When I tried measuring resistance from the ground wire of one of the 240V receptacles to the ground wire of a nearby 120V receptacle, the number danced around a bit between 1.3 to 2.8 ohms. I would've hoped for a lower number and can't really explain the jumping around (this meter had shown very stable readings on everything else I've tested with it) but I'm not sure it tells me anything profound.

I guess tomorrow we'll get the machines back to our shop and see if we can recreate the glitches and get to the bottom of why they're happening again.

 

One oddity: When I tried measuring resistance from the ground wire of one of the 240V receptacles to the ground wire of a nearby 120V receptacle, the number danced around a bit between 1.3 to 2.8 ohms. I would've hoped for a lower number and can't really explain the jumping around (this meter had shown very stable readings on everything else I've tested with it) but I'm not sure it tells me anything profound.

Could simply be oxidation on whatever you were attaching your probes to and the "tips" just weren't getting down to bare clean metal..
Or it could be loose connections or even a long ground wire run between each..

 

Mystery solved!

It turns out each of the prototype machines had snubbers on every motor but one. We sent two prototypes and one current production machine to the show. The current machine didn't have snubbers and somehow their motors got all mixed up so we were running snubbers where they weren't critical on a current machine while we were running prototypes that desperately needed the snubbers without them.

An embarrassing mistake, but I'm glad it turned out to be something so simple and not something fundamentally wrong with our new designs!

Thanks for all the the great ideas.