Matt Shaver wrote:
> On Thu, 2008-06-05 at 00:18 -0500, Jon Elson wrote:
> 
>>I don't think there is truly ONE way to do things.  You pretty 
>>much have to use safety-rated controls from Pilz, Crouzet,
>>Schaffner, Faulhaber, etc., and then set them up so they can 
>>function correctly to stop motion, even after a component failure.
> 
> 
> Yes. In fact, that's what I've learned so far! The most recent control
> I've built uses a commercial safety relay in the estop circuit, as well
> as an EN418 qualified (anti tease, positive acting) estop button.
> There's also some redundancy of power control elements and monitoring
> circuits that prevent resetting estop if a contact has welded closed.
> 
> What I'm interested in is what I've missed! I learned all those
> techniques from manufacturers literature that refers to these standards,
> but I've never actually seen the standards documents themselves, and I'd
> like to...
> 
The problem is they all cost several hundred $ each, and every 
document references something like 100 other documents, with no 
indication how important that one is.
> <...>
> 
>>pretty expensive.  Then, you get into calibrated wideband 
>>antennas, and on and on.  Finally, you have to test every 
>>different configuration, so you generally have to test each 
>>specific installation, if you are doing one-off machine retrofits.
> 
> 
> All true of course. I guess I put too much emphasis on the "accuracy"
> part of my request. I don't doubt that elaborate test equipment and
> facilities would be required to certify performance to a particular
> standard or spec.
> 
> The most basic test for RFI is to tune around the AM radio band on a
> receiver with the antenna close to the equipment under consideration,
> listening for changes in the level of "static". I'm just hoping to
> improve on this; to take it to the next step. Maybe build some sort of
> "hobby standard" wide band RFI detector useful for comparison purposes,
> rather than a calibrated measurement instrument.
> 
The problem is even a hand-held 7 transistor radio is way too 
sensitive.  Try it with your home computer, it will pick up all 
sorts of fairly loud "birdies", like the disk drive spindle 
motor drive circuit, the main power supply, etc.  All the CPU 
stuff will be several orders of magnitude above the AM band, so 
you won't hear any of that, except sub-harmonics from memory and 
I/O data patterns.  You'll be fairly surprised at how BAD the 
average PC is, no WAY they'd pass a real FCC class A test, but 
they come in, anyway.
Even the cheapest radio has an AGC circuit or AM wouldn't work 
at all, but this makes it impossible to compare one buzz against 
another.  They will all appear as loud, as the AGC's JOB is to 
make them all the same.  It might not be real hard to hack a 
small radio from the 1970's to disable the AGC, it is probably 
all in one chip today.
> For example, if I put a ferrite bead on a cable, did it help reduce the
> RFI, or was there no change at all? Right now, I'm operating in the
> dark, installing filters, chokes and beads in an anticipatory,
> prophylactic way ;). What I need is a relative indicator to detect
> electrical interference, locate the source, and compare the effect of
> different mitigation methods.
> 
You might have some success tuning the radio to listen to local 
stations, and compare the noise WHILE tuned to the station, that 
will lock the AGC to the station's power level, so you can maike 
relative comparisons.  But, of course, that only works on 
wide-band emissions.
> 
>>Most machine tool controls have had totally uncontrolled 
>>emissions, because they were mostly exempt as heavy industrial 
>>equipment.  Note the lack of output filters on most PWM servo amps.
> 
> 
> Your own being an exception to this!
Well, yes.  And, I had some prior hints that people were having 
real trouble with various drives that messed up encoder and 
step/direction signals, so I thought it would be wise.  Having 
100 ns risetime pulses of 50 -100 V flying about systems that 
need to correctly sense 5 V signals which also have fairly fast 
risetimes, so you can't massively RC filter them seemed a bad idea.
> 
> 
>>Devices to inject disturbances into the line and equipment 
>>chassis are not as bad as the spectum analyzer type gear, and 
>>you can even make much of this yourself, like with auto ignition 
>>coils and pulser circuits.
> 
> 
> I did make a vibrating relay type noise generator which helped recreate
> an intermittent problem with a spindle speed control board. Once they
> were able to reliably produce the problem, the manufacturer of the board
> was able to add hardware filtering and do some software modifications
> that eliminated the problem.
> 
VFDs, of course, are pretty powerful noise generators, too.


Anyway, for noise suppression, the best thing is shielded cables 
tightly bonded to the cabinet, a cabinet free of large holes and 
unsecured slots, and a good power line filter module.  Adding 
filtering to PWM outputs is also a big help.  Slots formed by 
covers that overlap a basic box defeat all the shielding, so you 
need a screw every 3-4 inches along the edge to clamp the metal 
together.

Jon

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to