On Sunday 13 March 2005 23:14, Charles Lepple wrote: >On Sun, 13 Mar 2005 15:42:22 -0500, Gene Heskett > ><[EMAIL PROTECTED]> wrote: >> I don't have to have the 82c55, but if I did, I could maybe do two >> boards and let ngspice do some optimizing. > >There is an 8255A symbol in the micro subdirectory of the > geda-symbols package, but it's just a graphical symbol. That is, > you can draw a schematic with it, and if you have a generic 40-pin > DIP footprint, you can put it on a PCB. (This particular symbol > file has pin numbers associated with it, which lets you use it with > a generic footprint.) > >If you try to make a spice netlist, I don't know what would happen, >but I would bet that it's not what you intended. > >> How is ngspice at simulating analogue circuitry thats supposed to >> be running in pwm mode? > >Not sure about ngspice in particular, but in general, spice-like >modelling systems can easily handle simulating transistors operating >in saturation (as most PWM systems are designed to run). If you have >an integrated PWM driver chip, you probably won't be able to easily >simulate overcurrent limiting, but if you pick a transistor model > with similar output characteristics, you can model the non-limited > operating region.
The existing circuit I have uses an lf357 as a comparator, which when the sensing R's in the ground leg of the l298 bridge driver detect an over-current, pull the enabling input back down if it can. With an 82c55 pulling it up, and a 270k resistor between the 82c55 and the rest of the circuit including this pulldown, the darned thing is doing an analog regulation, so apparently I need another rolloff pole in the feedback loop to make it actually oscillate. Or figure out how to put a lot more hysteresis in the thing somehow. That may be the best bet in the long run, but my choices of pf sized caps here at home isn't much. I'll have to see if I can work that out in my head. I have done similar stuff before so maybe I can again. Humm, looking at the schematic, (I wish you had one too, so you could follow this one sided conversation better) the lm358 (each half) that is being used as a current comparator, has a 10 meg R from its output to its + input, which with a 10k R from the setting pot, also a 10k unit with the full 5 volts across it, means it has about a 4.5 millivolt positive feedback to establish some hysteresis. Watching it just now with a scope across both of the .5 ohm sensing R's, I can see its switching moderatly cleanly when only one coil is being driven, but goes a bit berzerk when both coils are on. But the switching frequency is something above 120khz too. When the source current turnon delay is about 1.25 usecs, thats a lot of time spent in the high heating transistions of the switching. The offtime delays are much faster of course at around .1 usec. So the waveforms are ringing pretty bad. When only one coil is on, it is about a 60% duty cycle square wave, from zero to an amp, but the top of the waveform isn't the usual ramp of ever increasing current one would expect to see when driving an inductance, its essentially flat. I'd think what I should be seeing across the sensing R's would be much closer to a sawtooth, but I guess its steering the freewheeling shutdown current not thru the chip, but thru that 8 pack of si diodes, type FR106's. Humm, google search on them puppies to see how fast they are. Cause they need to be fast for power diodes. Humm, 800 volts, but only 1 amp, fast. The curves go up to 3 amps though, but the killer to me is the >.5 usec reverse recovery time, no wonder there is a nearly 10 amp inrush for about 1.3 usecs when the bridge turns back on. No curves on the forward breakover time furnished but I'd bet its 100ns or more. They need to be schotkeys so they wouldn't need the 800 volt rating, 75 would be sufficient if they were fast enough. And to be current sensed correctly, those 4 diodes in the lower half of the circuit should return not to ground, but to the top end of the sensing R's so the motor current is being monitored full time as opposed to only when the bridge drivers in the L298 are on. That poor diode recovery time explains the square wave I'm seeing, with the ten amp leading edge spike. The spike is the reverse recovery time of the diodes. I'd bitch about the design, but SGS/Thompson shows it exactly that way in the data sheets. Monkey see, monkey do. A star topology for the grounds would have been nice, I wonder how much I'd have to cut & jumper to do that, no pcb artwork handy & the pcb's are all bolted down, and its getting late. That and lowering that 10 meg feedback to even as low as a meg would slow the pwm regulator switching speed by half I'd guess. I'll look at the actual voltages on a winding tomorrow and see whats going on there. One thing I don't like about this board is the lack of a centralized single point ground, radiating from the common point of the sensing R's. Theres at least a volt of noise sitting on the pin they call ground at the input header when compared to the ground at the power supply header. I think thats the first thing I'll fix on one board for effects, then play with the feedback R. And my thinking out loud has probably bored you by now, so I'll say goodnight. And I'll let you know what results I get for the cut & jumper operation to make a single point ground out of it. -- Cheers, Gene "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) 99.34% setiathome rank, not too shabby for a WV hillbilly Yahoo.com and AOL/TW attorneys please note, additions to the above message by Gene Heskett are: Copyright 2005 by Maurice Eugene Heskett, all rights reserved.
