Forums

Ground currents blowing up my CM choke

Started by Allan Herriman October 20, 2011
Hi,
    I recently designed a product with an RS232 port on it.  I put a 
common mode choke on the ground and data lines to reduce EMI.  Two ground 
lines, Tx data and Rx data go through the choke.

The choke I chose was a tiny eight terminal 0805 thing with an 80mA 
rating.
At the time I thought that would be adequate for RS232, but we've managed 
to destroy a number of them through regular use in the lab.  In all 
cases, the failure is due to the ground lines through the choke going 
open.

The failure is obviously due to exceeding the rating of the choke due to 
ground currents through the RS232 cable.  The RS232 connection on the 
computer on the other end of the cable may be grounded (e.g. a desktop 
computer) or it may be ungrounded (e.g. a laptop computer with a 
charger).  Similarly, my board may be grounded or ungrounded (depending 
on whether it's in a chassis or just connected to a bench supply).

I have not determined whether the fault happens due to a turn on or off 
transient, or whether it's more of a steady state thing.

Questions:  What current should I allow for?  Are there any standards or 
guidelines that apply to my problem?

BTW, My product will need to be certified for installation in most parts 
of the world.

The RS232 / V.28 standard doesn't help at all, merely saying that 
connections to protective earth should comply with local regulations.

There are a number of possible fixes.  The most likely fix is replacing 
the CM chokes with pieces of wire then retesting EMI.

Thanks,
Allan
On 20 Oct 2011 14:54:12 GMT, Allan Herriman <allanherriman@hotmail.com> wrote:

>Hi, > I recently designed a product with an RS232 port on it. I put a >common mode choke on the ground and data lines to reduce EMI. Two ground >lines, Tx data and Rx data go through the choke. > >The choke I chose was a tiny eight terminal 0805 thing with an 80mA >rating. >At the time I thought that would be adequate for RS232, but we've managed >to destroy a number of them through regular use in the lab. In all >cases, the failure is due to the ground lines through the choke going >open. > >The failure is obviously due to exceeding the rating of the choke due to >ground currents through the RS232 cable. The RS232 connection on the >computer on the other end of the cable may be grounded (e.g. a desktop >computer) or it may be ungrounded (e.g. a laptop computer with a >charger). Similarly, my board may be grounded or ungrounded (depending >on whether it's in a chassis or just connected to a bench supply). > >I have not determined whether the fault happens due to a turn on or off >transient, or whether it's more of a steady state thing. > >Questions: What current should I allow for? Are there any standards or >guidelines that apply to my problem? > >BTW, My product will need to be certified for installation in most parts >of the world. > >The RS232 / V.28 standard doesn't help at all, merely saying that >connections to protective earth should comply with local regulations. > >There are a number of possible fixes. The most likely fix is replacing >the CM chokes with pieces of wire then retesting EMI.
It sounds like you have a ground problem in your lab. There shouldn't be *that* much current flowing in grounds that are that close to each other. Other than fixing that problem, the obvious solution is to isolate the grounds/signals. Either an optical isolator or one of the ADI isolators would be good. RS-422 might be called for, too.

Allan Herriman schrieb:

> The choke I chose was a tiny eight terminal 0805 thing with an 80mA > rating. > At the time I thought that would be adequate for RS232, but we've managed > to destroy a number of them through regular use in the lab. In all > cases, the failure is due to the ground lines through the choke going > open. > > The failure is obviously due to exceeding the rating of the choke due to > ground currents through the RS232 cable. The RS232 connection on the > computer on the other end of the cable may be grounded (e.g. a desktop > computer) or it may be ungrounded (e.g. a laptop computer with a > charger). Similarly, my board may be grounded or ungrounded (depending > on whether it's in a chassis or just connected to a bench supply).
Hello, a MC1488 driver is current limited to 10 mA, 20 mA ground current should not be a problem. I suspect there is a ground current of more than some 100 mA caused by different ground potentials on both sides. Maybe the choke maximum current rating of 80 mA was not specified for each of the four lines, but for all four together. I would try to measure the ground current, but in a noisy environment with strong currents over the power lines the resulting ground current through the RS232 cable may be much higher than the measured value. Bye
Allan Herriman wrote:

> Hi, > I recently designed a product with an RS232 port on it. I put a > common mode choke on the ground and data lines to reduce EMI. Two ground > lines, Tx data and Rx data go through the choke. > > The choke I chose was a tiny eight terminal 0805 thing with an 80mA > rating. > At the time I thought that would be adequate for RS232, but we've managed > to destroy a number of them through regular use in the lab. In all > cases, the failure is due to the ground lines through the choke going > open.
I'd get one of those 3-prong outlet testers at the hardware store and check all your outlets and power strips in the lab. I got bit once in my office while connecting test equipment that was plugged into a different outlet. I brought in my tester from home and found one wall outlet had an open ground. Some older test gear had really BIG capacitors in the line filter and would deliver a lot of current if the ground pin was open. Some of our power strips get open grounds from time to time and that outlet needs to be marked with red tape. Jon
Allan Herriman wrote:
> Hi, > I recently designed a product with an RS232 port on it. I put a > common mode choke on the ground and data lines to reduce EMI. Two ground > lines, Tx data and Rx data go through the choke. > > The choke I chose was a tiny eight terminal 0805 thing with an 80mA > rating. > At the time I thought that would be adequate for RS232, but we've managed > to destroy a number of them through regular use in the lab. In all > cases, the failure is due to the ground lines through the choke going > open. > > The failure is obviously due to exceeding the rating of the choke due to > ground currents through the RS232 cable. The RS232 connection on the > computer on the other end of the cable may be grounded (e.g. a desktop > computer) or it may be ungrounded (e.g. a laptop computer with a > charger). Similarly, my board may be grounded or ungrounded (depending > on whether it's in a chassis or just connected to a bench supply). > > I have not determined whether the fault happens due to a turn on or off > transient, or whether it's more of a steady state thing. >
Most likely both. The PE connection on your box and that of the box at the other end of the line seems to have a slightly different potential. It shouldn't, but sometimes it does. Even if this is only a volt or two, when you connect that with a short piece of serial cable there can be hundreds of milliamps. I have seen several amps. Then the cleaning crew plugs their big hoover into one of the outlets, turns it on ... ZZZONK ... *PHUT* Seriously, a client has seen an expensive laser controller come to grief that way.
> Questions: What current should I allow for? Are there any standards or > guidelines that apply to my problem? >
One problem will be that any such current will likely saturate the core of your CM choke, thus making the choke inefficient as an "RF scrubber".
> BTW, My product will need to be certified for installation in most parts > of the world. >
Ooooh, that raises a red flag. For example, in Germany I have some painful experience with legacy installations of what they call "klassische Nullung", loosely translated as "classic earth protection". In those cases there are only two wires going from outlet to outlet, phase and neutral. No PE. The PE-connection in each outlet is then bridged to the neutral wire. So now when there is a major load towards the end of the string of outlets you'll see a significant voltage drop. If the load is a power supply with large electolytics and no inrush protection you'd see a spike of several volts when someone turns that on. Vacuum cleaners are also lots of fun there. And, of course, if the neutral comes off at some outlet early in the string .... bzzzzt .. ouch ... *POOF*
> The RS232 / V.28 standard doesn't help at all, merely saying that > connections to protective earth should comply with local regulations. > > There are a number of possible fixes. The most likely fix is replacing > the CM chokes with pieces of wire then retesting EMI. >
The real solution for a worldwide product would be to provide truly isolated RS232 at your box. If you mean the ground that normally goes to pin 7, running that outside the choke will wreak havoc with your data lines. It puts an inductor right into the signal path, meaning a lowpass. But worst of all if TXD and RXD go through the same choke you will couple TXD into RXD. Static signals will look ok but the faster the data rate the more data errors you'll see. -- Regards, Joerg http://www.analogconsultants.com/
Sounds like you have a power mains grounding/earthing issue. If the
equipment on both ends has a proper power mains ground, you won't have
this problem. You'll probably find one side is properly earthed, and
the other side isn't. If you go barefoot and lightly rub equipment
(bare metal) with the back of your fingers, you'll quickly find the
evil doers.

I have seen this in a lab where they removed all the earthing on the
power plugs thinking this was a good thing to do. Unfortunately, their
anti-static, slightly conductive mats were sitting at half the mains
voltage causing them to blow out all sorts of stuff. I rewired the
offensive bench I was working on, but next time I came back, they
mangaged to make it unsafe again. They also had a heap of RS232 boards
which were blown out due to earthing issues.
On 10/20/2011 9:54 AM, Allan Herriman wrote:
> Hi, > I recently designed a product with an RS232 port on it. I put a > common mode choke on the ground and data lines to reduce EMI. Two ground > lines, Tx data and Rx data go through the choke. > > The choke I chose was a tiny eight terminal 0805 thing with an 80mA > rating. > At the time I thought that would be adequate for RS232, but we've managed > to destroy a number of them through regular use in the lab. In all > cases, the failure is due to the ground lines through the choke going > open. > > The failure is obviously due to exceeding the rating of the choke due to > ground currents through the RS232 cable. The RS232 connection on the > computer on the other end of the cable may be grounded (e.g. a desktop > computer) or it may be ungrounded (e.g. a laptop computer with a > charger). Similarly, my board may be grounded or ungrounded (depending > on whether it's in a chassis or just connected to a bench supply). > > I have not determined whether the fault happens due to a turn on or off > transient, or whether it's more of a steady state thing. > > Questions: What current should I allow for? Are there any standards or > guidelines that apply to my problem? > > BTW, My product will need to be certified for installation in most parts > of the world. > > The RS232 / V.28 standard doesn't help at all, merely saying that > connections to protective earth should comply with local regulations. > > There are a number of possible fixes. The most likely fix is replacing > the CM chokes with pieces of wire then retesting EMI. > > Thanks, > Allan
You are blessed with an opportunity solve potential end user problems right now. Assume your customer has this problem. Don't change anything in your test area. Fix your product to accommodate. Cheers, John S
On 10/20/2011 1:21 PM, John S wrote:
> On 10/20/2011 9:54 AM, Allan Herriman wrote: >> Hi, >> I recently designed a product with an RS232 port on it. I put a >> common mode choke on the ground and data lines to reduce EMI. Two ground >> lines, Tx data and Rx data go through the choke. >> >> The choke I chose was a tiny eight terminal 0805 thing with an 80mA >> rating. >> At the time I thought that would be adequate for RS232, but we've managed >> to destroy a number of them through regular use in the lab. In all >> cases, the failure is due to the ground lines through the choke going >> open. >> >> The failure is obviously due to exceeding the rating of the choke due to >> ground currents through the RS232 cable. The RS232 connection on the >> computer on the other end of the cable may be grounded (e.g. a desktop >> computer) or it may be ungrounded (e.g. a laptop computer with a >> charger). Similarly, my board may be grounded or ungrounded (depending >> on whether it's in a chassis or just connected to a bench supply). >> >> I have not determined whether the fault happens due to a turn on or off >> transient, or whether it's more of a steady state thing. >> >> Questions: What current should I allow for? Are there any standards or >> guidelines that apply to my problem? >> >> BTW, My product will need to be certified for installation in most parts >> of the world. >> >> The RS232 / V.28 standard doesn't help at all, merely saying that >> connections to protective earth should comply with local regulations. >> >> There are a number of possible fixes. The most likely fix is replacing >> the CM chokes with pieces of wire then retesting EMI. >> >> Thanks, >> Allan > > You are blessed with an opportunity solve potential end user problems > right now. Assume your customer has this problem. Don't change anything > in your test area. Fix your product to accommodate. > > Cheers, > John S
I meant to add: I would go with Joerg's suggestion and opto-isolate the link.
On Thu, 20 Oct 2011 13:21:39 -0500, John S <sophi.2@invalid.org> wrote:

>On 10/20/2011 9:54 AM, Allan Herriman wrote: >> Hi, >> I recently designed a product with an RS232 port on it. I put a >> common mode choke on the ground and data lines to reduce EMI. Two ground >> lines, Tx data and Rx data go through the choke. >> >> The choke I chose was a tiny eight terminal 0805 thing with an 80mA >> rating. >> At the time I thought that would be adequate for RS232, but we've managed >> to destroy a number of them through regular use in the lab. In all >> cases, the failure is due to the ground lines through the choke going >> open. >> >> The failure is obviously due to exceeding the rating of the choke due to >> ground currents through the RS232 cable. The RS232 connection on the >> computer on the other end of the cable may be grounded (e.g. a desktop >> computer) or it may be ungrounded (e.g. a laptop computer with a >> charger). Similarly, my board may be grounded or ungrounded (depending >> on whether it's in a chassis or just connected to a bench supply). >> >> I have not determined whether the fault happens due to a turn on or off >> transient, or whether it's more of a steady state thing. >> >> Questions: What current should I allow for? Are there any standards or >> guidelines that apply to my problem? >> >> BTW, My product will need to be certified for installation in most parts >> of the world. >> >> The RS232 / V.28 standard doesn't help at all, merely saying that >> connections to protective earth should comply with local regulations. >> >> There are a number of possible fixes. The most likely fix is replacing >> the CM chokes with pieces of wire then retesting EMI. >> >> Thanks, >> Allan > >You are blessed with an opportunity solve potential end user problems >right now. Assume your customer has this problem. Don't change anything >in your test area. Fix your product to accommodate.
Don't worry about frying anyone in your lab. Power faults just aren't worth fixing.
John S wrote:
> On 10/20/2011 1:21 PM, John S wrote: >> On 10/20/2011 9:54 AM, Allan Herriman wrote: >>> Hi, >>> I recently designed a product with an RS232 port on it. I put a >>> common mode choke on the ground and data lines to reduce EMI. Two ground >>> lines, Tx data and Rx data go through the choke. >>> >>> The choke I chose was a tiny eight terminal 0805 thing with an 80mA >>> rating. >>> At the time I thought that would be adequate for RS232, but we've >>> managed >>> to destroy a number of them through regular use in the lab. In all >>> cases, the failure is due to the ground lines through the choke going >>> open. >>> >>> The failure is obviously due to exceeding the rating of the choke due to >>> ground currents through the RS232 cable. The RS232 connection on the >>> computer on the other end of the cable may be grounded (e.g. a desktop >>> computer) or it may be ungrounded (e.g. a laptop computer with a >>> charger). Similarly, my board may be grounded or ungrounded (depending >>> on whether it's in a chassis or just connected to a bench supply). >>> >>> I have not determined whether the fault happens due to a turn on or off >>> transient, or whether it's more of a steady state thing. >>> >>> Questions: What current should I allow for? Are there any standards or >>> guidelines that apply to my problem? >>> >>> BTW, My product will need to be certified for installation in most parts >>> of the world. >>> >>> The RS232 / V.28 standard doesn't help at all, merely saying that >>> connections to protective earth should comply with local regulations. >>> >>> There are a number of possible fixes. The most likely fix is replacing >>> the CM chokes with pieces of wire then retesting EMI. >>> >>> Thanks, >>> Allan >> >> You are blessed with an opportunity solve potential end user problems >> right now. Assume your customer has this problem. Don't change anything >> in your test area. Fix your product to accommodate. >> >> Cheers, >> John S > > I meant to add: I would go with Joerg's suggestion and opto-isolate the > link.
If cost is not super-critical or rolling their own is not desired for some reason, here's the canned solution for the board level: http://www.analog.com/static/imported-files/data_sheets/ADM3251E.pdf If it's a hindsight "oops, got to fix it fast" situation, maybe something like this: http://www.serialcomm.com/serial_rs232_converters/opto_isolated_and_surge_suppressors/rs232_isolator/rs232_3_wire.product_specifications.aspx Haven't tried either though, I've always rolled my own. -- Regards, Joerg http://www.analogconsultants.com/