On Wednesday 23 August 2017 12:30:35 Jon Elson wrote: > Wow, I have made a big screw-up. My PWM servo amps have RC > snubber networks on the junction between the high-side > source and low-side drains (output terminals) of the > half-bridges. The current values are 10 Ohms and 4700 pF. > I have a guy who wanted to run 160 V DC supply. All the > parts are rated with enough margin to do it. (200 - 250 V > ratings on various parts.) I never worked out the power > dissipation in the snubber resistors. I used 3/4 W SMT > resistors and THOUGHT that ought to be big enough. > > Well, that guy kept having the resistors burn up, and then > the amps get flaky and trip falsely on overcurrent due to > too-high dv/dt affecting the sensing circuits. > > Well, I finally wrote up a tiny program to numerically > integrate the energy in the resistors, and was unpleasantly > surprised that I'd vastly underdesigned that part. At 160 V > DC, and 50 KHz PWM (that's 100K charging events/second) I > get over 6 W dissipation (in a 3/4 W resistor)! I already > have a 2 W resistor on order, however the space on the board > is not going to dissipate the heat all that much better than > it did before, so replacing the resistor won't help remove > the heat much better. The customer is getting a buck > transformer to drop the DC voltage to about 142 V, and if he > also drops the PWM frequency 20 KHz (from 50K) it will > reduce the loss in the resistor to 1.9 W. > > The problem happens on his Z axis, which is just sitting > there keeping the machine's head from dropping. So, that is > consistent, it sits for a long time on the same commutation > position of the motor, so one resistor at a time gets hot. > > Any comments? > > Jon > One thing I learned 60 some years ago Jon, was that in carbon composition r's, which yours no doubt aren't, a common half watter, regardless of its R value, also had a maximum voltage across it rating that was surprisingly low, like 25 or 30 volts, above which there would be microscopic intergranule arcs that would gradually lower the overall resistance until the power dissipation made white ash out of the composition tube it was in. After that occurred any thermal or mechanical stress and it fell apart. That usually broke the magic mirror and it didn't work anymore. The series resistor between the HV pulse coming from the horizontal output transformer to the plate of the keyed agc tube was commonly subjected to a 500 to 1000 volt pulse 15,750 times a second. But the designers were slow to catch on, and the bean counters buying parts were even slower, so they kept on using 330k 1/2 watt R's for many years in that position. Fresh and new, with a peak dc current in the sub milliamp range, no noticeable heat. But by the time the warranty (1 year in those days for the major brands like Zenith) ran out, they were cooked, and I was replacing them with 2 watters that lasted 2 years maybe, basically the same problem because even the 2 watters could only take 200 volts or so. Carbon film types when they came out were even worse if the spiral cut that trimmed them for accuracy was narrow enough to arc over under the paint. Metal film types were much better. You'll need high wattage R's with long leads to sink the heat and far enough away not to telegraph the heat to other semi-c parts.
So thats something (the resistors maximum voltage rateing) to keep in mind when you expect to close that cabinet door for 10+ years during which time it Just Works. Probably till the electrolytics die from high ESR. Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene> ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Emc-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-developers
