On Friday 08 October 2021 21:15:12 John Dammeyer wrote: > I disagree. Grounding won't fix what is inherently a bad design. > Might make it appear less often if that indeed is the reason but it's > still a bad design. > HAving taken a third look at that link, I have to agree. That family of chips have a proclivity to set off a substrate scr, which can muck up the logic or even crowbar the power supply when an input pin is pulled above the supply line, a possibility given that the encoder is running on 12 volts and the 74hc4050 is on 5 volts. Bad dog, no biscuit.
> Way back in 1993, before Philips had even released the P82C250 CAN > driver, I worked on a system that had about 12 nodes on CAN bus. > Although the resistance between the CAN wires was 60 ohms with the > properly terminated network the input impedance on the receivers > relative to ground was still in the 10K ohm range. > > It was on a 30 ton fly wheel press that used a stepper motor to close > the frame to hold the forming tools. When the press ran without any > leadframes there were no issues. When leadframes were formed and > trimmed occasionally a node would report enough errors to cause the > entire machine to fault. > > Although the bus looked noisy from a single ended perspective the > differential was clean. I then put the scope on the TTL side of the > bus receiver circuitry and it was also very clean. Until the press > hit a leadframe. Then very narrow pulses that were normally below the > detection threshold appeared randomly on the other side. > > Turns out that the SloSyn stepper motor that closed the Tool Holder > frame under load would deflect just enough to cause the constant > current into the stepper windings to increase enough to prevent the > motor from turning further. That extra bit of 18kHz electrical noise > was coupled onto the CAN bus resulting in the occasional failed > message. > > CAN works as follows. For each transmitted message that fails, a > counter is incremented by 8. For each successful message the counter > is decremented by 1. There's a warning at 96, and then a mode called > Error Passive at 128 and finally if the errors continue the node goes > bus off at 256. The bus off was causing the system to stop. > > Eventually a hardware solution involving a brake and disable of the > stepper solved the problem but the interim period was handled with > dual core ferrite cores at each point where the network entered a > node. One loop of each side of the CAN bus through the core. > Looking at the signals showed a remarkable rounding of the corners as > this massive amount of inductance on the CAN bus caused considerable > loss in high frequency response. But it was also enough to > completely suppress the stepper motor chopping frequency and no more > partially completed or damaged Pentium-66 processors which at $600 > each or so was expensive. > I won't go so far as to call it a no brainer John. I will freely admit I've been caught out by it, in fact the 7i90HD running my Sheldon has a couple blown gpio pins, forcing a hal rewrite to use different stepgens. I didn't take my own advice, pure and simple. So I had 3.5 volt signals with 80 volts of sub-microsecond noise, kind of hard on the fpga's in the mesa stuff. Redoing the grounding as I have described, and adding 3 each 7i42TA's to protect the fpga's, brought the noise on that same wire down below 100 millivots. At about that scopes 100 megacycle rated bandwidth. I've since bought a much faster scope, siglents best. I first wrapped my hand around a scope probe a few days prior to my 17th birthday, 70 years ago. It has been at hand since. Very educational tool. > So as always, try and remove the source of the noise if you can find > where it's coming from. Good grounding is a no brainer and of course > needs to be done. > > But running a high speed edges, single ended from 0.4V to perhaps 12V > into a load that draws at most 5uA is bad design when the driver is > capable of driving 10mA or more. Thats a cmos chip, any current flow after the gate capacitance has been charged to the new value is prima-faci evidence the chips gate oxide has been punched, and the chip needs replaced. That whole design needs a diode such as a 1n914, connected to the input pins of the chip such that the actual voltage on that pin cannot rise to more than about .6 volts above the 5 volt line. > Not to say that the DC load may be > brought down to 1K with an opto-isolator input while the impedance due > to wiring might still be at 50K Ohm at the interfering frequency. > Which is why other types of filtering like ferrite cores around the > wire are also useful solutions. To lower that impedance and block the > signals. Differential instead of single ended so common mode noise is > cancelled out. > > But running that 12V signal through a 1K resistor into what is > essentially a 2.4M ohm load won't stop noise from being picked up and > transmitted through the 74HC4050. Absolutely John, throw in the substrate scr problem and its going to bite, somewhere, sometime. > My two cents. > John > > > -----Original Message----- > > From: Chris Albertson [mailto:albertson.ch...@gmail.com] > > Sent: October-08-21 4:52 PM > > To: Enhanced Machine Controller (EMC) > > Subject: Re: [Emc-users] Fwd: Rogue Index Pulses > > > > I'll say what Gene said. I'd bet on a groubing issue. The schmatic > > has an error, a groubd wire ismissing and also shows random crossing > > grounds. Run everything back to the power supply minus post. > > > > Also you be much better if the sheild was grounded to the controller > > end. Never us a machine and it's mounting bolts as grounds. > > > > > > -- > > > > Chris Albertson > > Redondo Beach, California > > > > _______________________________________________ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users > > _______________________________________________ > 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, 1940) If we desire respect for the law, we must first make the law respectable. - Louis D. Brandeis Genes Web page <http://geneslinuxbox.net:6309/gene> _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
