On Monday 14 December 2015 16:01:26 Ben Potter wrote: > > Note that those setup and hold times are relative to the riseing > > edge of > > the (step?) pulse, and the next line is not so specific, but I expect > it needs to be checked by a dual trace scope for both > > > conditions > > I'm very tempted to pull the G540 from the router and rig up a test > bench test to see if I can get my head around why this is happening. > Might get time over christmas. > > Would be nice to know why inverting the step output works for me, > rather than 'it just does' I'm starting to suspect that the 5i25 obeys > the set timings, whereas mach on a parallel port uses them as a > minimum guide, and frequently exceeds them. > > > > I have no additional opto-isolation in that path - I believe that > > > the G540 uses fast opto's on all the parallel port pins (from the > > > manual, I'm sure there was a post on cnczone detailed the choice > > > of isolater somewhere) > > > > Thanks for that , so it would have been DIRSETUP/DIRHOLD that I was > > concerned with. > > I managed to dig out the thread where Mariss of Gecko was talking > about opto circuits. > http://www.cnczone.com/forums/general-cnc-machine-related-electronics/ >23216- optoisolator-circuit.html > > Posts #2 and #6 are particularly interesting (not sure on cross-post > etiquette here), and possibly give a hint as to the root cause of my > problem. > > > > <snip> > > > > And that gives you far finer tools to tune with than software > > stepping > > ever will. There may better interfaces than the 5i25, but for us > stepper people, its the cats meow. Not bulletproof on its inputs, I > didn't have a good enough > grounding setup wrapped around my G0704, > and apparently blew the input I was using for a probe. > > > I don't know as Peter could tell me which chip to replace to rescue > > it. I > > AM a C.E.T. and in my dayjob before I retired in 2002/6/30, I could > have done it easily, but at 81, my hands aren't as steady as they once > were. > > They do pretty nicely on servo stuff too - I have a 5i25 on the > (servo) bridgeport, the (stepepr) lathe was converted earlier, and > that has a 5i20 on. > > > I can't comment based on experience with them so I won't, Ben. > > Other than > > to say that 98% of the opto stuff out there, will need a pulse in > excess of 1 microsecond on, and often pushing 2 u-s to turn off. I > understand there are > > > some Gas based optos about, but they'd have a serious effect on a > > BoB's > > price. One could turn a $15 BoB into a $100 BoB pretty fast. > > From the thread above it looks like Gecko use the HCPL-2531 in at > least some of their drives - price of $0.87 per 1k doesn't look too > bad for a fast dual channel opto. > > > However, one must bear in mind that when looking at it with a scope, > > we > > are watching the voltage, and in opto stuff, voltage doesn't mean near > as much as actually measuring the current. Why? Because when the > > > driving led in the opto goes out, which it might do at 1.85 volts or > > as high as 2.4 volts, the load is gone, leaving the voltage to decay > > toward (which ever rail its headed for) at a very leasurely rate. > > > Thats a fancy, CMA way to say the scope is lieing to you. So I tend > > to > > look at the decay with an eye toward seeing if I can define that point > where the down (up?)ward curve changes to a slower rate, then make the > "on time" timing > measurements to that point. That can be a difficult > point to define even for me, and I've "had a scope probe in one hand" > > > for 64 years. 64 years ago it was a 5mhz Hickock, today I have > > 100MHZ dual > > trace analog, and 200mhz dual trace digitals to go poking around with. > > > Cheers, Gene Heskett > > Good to know, I have a rescued tektronix scope which I rarely get to > use. Hints on those things that are only learned through experience > are always appreciated. > > I suspect spotting most of the ways phase shift can trip you up even > on simple circuits takes quite a bit of practice. > That, and ingenuity in figuring out how best to get at it with a probe. And sometimes you can't, you have to go well beyond and let experience guide one on figuring out what the heck is going on at the point of interest. I have even written software to excite the circuit, then asked it to play it back, then do a comparison of sent vs echo. In this case, a 4x4 fifo, used 2 wide for an 8 bite wide rubber buffer coupling in a video switcher, had one of the 2 4x4 fifo's that wasn't responding to the shift out command, about 2% of the time. Called the switcher maker, who wanted $1500 a copy for the chip. But I'm sneaky, I suspected it was an AMD product, called them, and they said yeah, we finally did get a JEDEC number for it, and its $2.10 a piece, how many do you want? I took a mental inventory of how many were in it, and ordered 2 sticks, then told the GM there would be a COD package from AMD. He had his usual litter of cows. I had no sympathy. When he finally got caught playing with the books, I told him as he was forcibly leaving in the company of a deputy, to be carefull and not let the doorknob hit you in the ass.
Good guess, I was out of them when they finally decided it was time to retire the beast for other reasons. It had a higher failure rate. None of the newer ones ever failed, just the OEM's. I believe there were 48 in it, and something over half of them failed in the next decade. Great switcher in terms of capabilities, advancing the state of the art & sold new for about $300k. We used up the one from the JCPenny production house in NYC, then finished off the one KTLA bought new. But that was another time & place compared to now. :( > Thanks > Ben > > > ---------------------------------------------------------------------- >-------- _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users 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) Some mill pix are at: Genes Web page <http://geneslinuxbox.net:6309/gene/GO704-pix> ------------------------------------------------------------------------------ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
