[pminmo]

If the chips are socketed, you might be able to fix the problem simply replacing the 7486 with 74HC86 parts. The 74HC86 parts will have more noise immune thresholds than the 7486 parts, although you need to your PC parallel port putting out logic HI's of 4V or better. I don't believe you would need to change the 7474 parts, although the 74HC74 would actually source more current than the 7474 to the TIP120's and suspect the circuit is such that a 74HC86 shouldn't have problems driving the 7474.

[cbcnc]

Would the HC series chips still use 5v, and be pin for pin compatible with the TTL chips that are there?
Unfortunately, the chips aren't socketed so it would mean unsoldering the existing chips.
I'm tempted, but that would mean tearing the controller up.
I think that before I go to that extent I will explore other alternatives.
Do you think that the noise is RFI , EMI or both?
Seems like if I can't eliminate it by switching circuits and filtering the power to the router that it would be RFI.
I have had thoughts of going crazy with the aluminum foil duct tape to completely shield all of the wiring. As well as installing more filtering caps where I can.
I gotta play around some more and try some different things.

Thanks for your thoughts,
Chris

[pminmo]

At 5V the 74HC would be a drop in replacement. They have different thresholds, so the concern would be what the output of the PC's parallel port high levels are. Your post seems to indicate symptoms of emi coupling in the step direction lines. Whether it's radiated or capacitively coupled or a combination of both is probably irrelevant, as in this enviornment they both exist. Magnitude varies, but it's there. Tack a 1000pf cap between ground and the step lines and see if that has an effect. Probably the easiest is on the 7486 chip itself.

[cbcnc]

There already is a 470pf cap on the step line to the 7486. Is it possible that it isn't the right value?

Chris

[Mariss Freimanis]

Classic straight-74 and 74LS logic is notoriously famous for being noise-susceptible and difficult to filter for noise. This holds particularly for clocked devices (D-Flops, counters, etc.). It is very difficult to tame these devices unless they are mounted on multi-layer printed circuit boards (top, bottom copper circuit connections, inner-layer ground-plane and supply distribution).

You will have much better luck with 74HC logic. Use 74HCT devices for interface where the inputs are either TTL or 3.3V logic; use Schmitt-trigger devices here if you can (74HCT14, etc). Use 74HC everywhere else inside. Use 74AC series for output interface if you need decent output drive current (+/-24mA). Never use a 74AC device that has a CLK input (no D_Flops, counters; gates only). Bypass every 74HC, 74HCT and 74AC package (Vdd to Vss) with a 100nF multilayer ceramic capacitor as close as you can get it to the device. Use fat, short circuit traces for ground and +5VDC; learn to hate the color of FR-4 green. If you see it, there is not enough copper on your pcb. Pull-up every interface input to +5VDC with a 10K resistor. Filter every interface input from pull-up resistor to CMOS logic input with a 10K, 100pF low-pass RC (1uS delay). If you are seriously anal-retentive, use a series Schottky diode clamp to GND and +5VDC at the CMOS logic input. Then not even +/- 50V spikes will hurt your CMOS logic. Use a 20 to 30 Ohm 1/8W resistor in series from CMOS logic to every interface output; you goof-up a connection, the resistor burns, not the logic.

There you have it in a nutshell. You will now live a long, happy and serene life. Your dog will love you because you will stop kicking him in frustration after dealing with 74LS noise issues and bad pcb layouts.

Mariss

[cbcnc]

Thank you, Master.

Chris

[neilw20]

Where ever possible have your limit switches Normally Closed
That way, when in normal operation they short out induced noise.
Put a 100n capacitor across all low voltage operated switches, so that the switch contacts get a high current zap from the capacitor. This helps to keep the contacts clean, and makes the wetting current much closer to what the manufacturer specifies. The capacitor needs to be close to the switch, so the pulse current is not lost in the wiring, and stops these current pulses interfering with other circuits.
DO NOT USE the 100n capacitor method with anything other than mechanical contacts. It can cause opto-slot switches to not work properly, or even damage them.
Using NC switches often means you must put a number of them in series, so trouble shooting can become more difficult. At least the machine will fail safe if a wire breaks.

[electomek]

Hello Chris,
One other thought, check your power supply, check the AC ripple with an O-Scope, and see if can catch any glitches when you turn the power on and off.
Thanks, Joe

[electomek]

Chris,
I correct myself on the quick reply, check to see if you get any glitches when turning the spindle on and off.
Thanks, Joe

[cbcnc]

I'm using a "Todd" 600 watt, 24v switching supply that I picked up as surplus. Also I have a little power block that I am using as a 5v supply.
I know I checked it on my scope when I first bought the supply and the ripple was minimal.
I will do it again.
But, see, you guys jogged my brain again. I will also start probing around to see if I can isolate where the noise is coming from.
My scope is packed up and I have been resisting unpacking it.

Chris

[electomek]

Hello again Chris,
When you first checked the power supply, did you check it under the present load conditions? If not, try it, hopefully during the problem. Also, it wouldn't hurt to put an MOV across the AC in put to the power, and one each side to ground.
Joe

[electomek]

Chris,
Again a correction, that is put the MOVs on the AC input of the power supply, and each side to ground.
Joe

P.S. I just fired my editor!

[pminmo]

The noise is comming via the AC supply on your router, you pretty much determined that via your note on the speed control effect. It could be coupling back through your AC power distribution system, but in all likely hood it's radiating via the AC wiring and capacitively coupling to your system wiring. The fact you went to a metal Z has probably effectively created more antenna radiator area. Simple test, disconnect your cable back the the PC parallel port at the piker board. Turn your power on, then turn the router on. Do the motors start jumping like you described in your first post? Do the same thing but with the cable connected to the board but disconnected at the PC, and compare results.

BTW, I've always though your router build was "simple elegance". Well thought out and executed.

[cbcnc]

Thanks for the compliment. Check this out though.
I just uploaded a picture of my new antenna. Just increased the area by about 120 sq in. Maybe that's why I'm having problems.

My new Y axis.


Chris

[cbcnc]

Thanks to all for the help.
I think that I have solved my problem!
I have done some rerouting and reorganizing of my 120v wiring.
It turns out that I had a couple of ground loops.
I have two power supplies; one for 5v and one for 24v that I had powered separately with two power cords, thinking that I didn't want them to interact with each other. I had hooked the ground wires for both to the common ground in my case.
So by eliminating the dual power cords and having just one 120v connection I think that I have solved the glitching problem. I'll be on the lookout for problems under heavier loads but it's certainly way better than before.

Chris