VistaCNC P4-SE pendant stopped working

Yeah, I understand.... There are some real areas for noise here... To be honest, I'm not entirely convinced that the cable even *has* a shield, but even if it did, I don't think it is connected to anything. And yes, in that case, it isn't going to do its job.

I thought about some suppression elsewhere on the estop lines themselves. I was going to read up on that, but you have proactively answered my question. I'll likely try the ceramic cap trick. easy to try.... Note, however, that I don't have a pc case. My "pc" is a motherboard that is mounted into the cabinet. So, I guess I'll tie to the enclosure.

Do you have a latching relay setup like the one I have in my schematic snippet that is attached to an earlier post from me?

No, I have always intended on installing a latching relay to have a fully mechanical E-Stop but haven't done it yet. Great reminder.

I did some probing with my meter and found that the shield on the RJ45 plug on the pendant is connected back to my cabinet ground through the shield or wire that connects the metal on each end of the USB cable together. Of course the EE_box is not shielded so there is an entry point. I have some 3/4 inch copper braid that I thing might fit over the USB cable. I might try sliding that over the USB cable and shielding the EE_Box.

One thing I do not fully understand:
You write : pendant "goes back to the LinuxCNC screen". Do you mean that is written on the LCD? If yes, why and where from come these letters? Do you configure the pendants cpu somehow to be in a linuxcnc mode???
Mike

Hi Mike,
Yes, there is an LCD on the pendant, along with a polycarbonate membrane keyboard, pulse encoder, etc. You can see a picture here:
www.vistacnc.com/a01_index_home/pic_a01/p4s_side.jpg

It actually is quite a nicely thought out pendant, which is why I like it. Not too much information to overload you, etc.

Johnny:
God I love this forum! In some of the most innocuous comments one can find new information. One of the reasons why I try to point out everything I've done and what my machine is like is that you never know when someone might stumble on an idea... In this case, you inadvertently told me something that I hadn't thought about before....

When I replaced my RJ45 connector, I used a pure plastic one. I actually never really knew that there was a grounded one, but your reference to grounding in the EEBox could only mean one thing (because I know the ground isn't carried through one of the 8 pins)... the connector. After a little searching, I learned about the shielded connectors. Great find, thank you! I would still swear that I *never* saw any foil or drain on the cable, but I must have missed it when I installed the new connector...

There is another point here that is a possible cause of problems. Remember that reference to a bulkhead connector? That is there because my EEBox is inside the cabinet. Well, first, the jumper between the connector and the EEBox has plastic connectors (so obviously not passing a shield), but the connector itself is almost certainly not contacting a metal shield on the wire.

I've ordered some shielded connectors, I'll have to buy a 6" jumper with shielded wires and determine if the bulkhead connector is going to be problematic for me. Great finds! I know you didn't point me directly to all of that, but the discussion definitely did.

[update]
Whew... The bulkhead connector *does* have the ability to pass the shield along... I have to say that I've been pretty happy with this design... The following is what I used ("CNLINKO RJ45 IP67" is the ebay search term for readers who come along long after the link expires):
www.ebay.com/itm/CNLINKO-Circular-Ethern...of-IP67/122258060427

Mike_Eitel wrote:

One thing I do not fully understand:
You write : pendant "goes back to the LinuxCNC screen". Do you mean that is written on the LCD? If yes, why and where from come these letters? Do you configure the pendants cpu somehow to be in a linuxcnc mode???
Mike

When the PC is powered up but before LinuxCNC is started the LCD first says "VistaCNC V120" then switches to "LINUXCNC". Once you start LinuxCNC the display changes to E-Stop. When you click in the AXIS E-Stop button or press F1 the display Changes to Machine Power OFF. Then when you press F2 or click on the power button the LCD show the axis location etc.

If you shutdown LinuxCNC the display goes back to LINUXCNC.

Yes, there is firmware that makes the pendant work with LinuxCNC.

When my pendant bricks, disconnects or whatever it sometime goes back to the LINUXCNC display or display complete garbage that are not even letters or numbers.

To elaborate a bit... The "LinuxCNC" screen is basically the default screen when the pendant is waiting patiently to be activated. Obviously that is after power up but, in my case, it is also if there is a hiccup in the communications between the LinuxCNC application and the pendant.

After you purchase a pendant from VistaCNC (the manufacturer), you use a loader to install the correct firmware for the CNC system you are using. Mach, LinuxCNC, and another (BeagleBone?) all have different pendant firmware.

The second step is to run a makefile on your computer. Among other things, it installs some files in a usr directory and also an executable in the LinuxCNC config directory. It also places a prototype HAL file which, frankly, was a somewhat weak starting point for me but still appreciated.

That pretty much sums up the pendant. It actually is pretty nice. I've not seen an XHC one, but I think I like the balance between the cleanliness of the VistaCNC and its capabilities. I don't actually want a 4x24 character display. I do have to say that the jury is still out on the membrane keyboard versus rotary knobs though. And the fact that I'm having problems getting the pendant to work is a real drag. There again, though, the jury is out as to whether this was self inflicted or if the pendant is overly sensitive.

OK, for me your explanations points 90% to a restart of pendant cpu. I do not belive that the pc part goes to a reset or loop or similar. Minor possibility would be that you stall the sw in pc or give a constant reset condition there. Is there a hal signal to initiate the pendant and have you halscoped all pendant signals.
Mike

Yes, those seem to be reasonable conclusions.... LinuxCNC is obviously multi-threaded so it is difficult to say whether or not I've hung a thread that is dedicated to the pendant. I guess it is all kind of academic though as, in the end, the final result is that "it doesn't work." But, I can tell you that there are no meaningful error messages that I see on the console (from which I start LinuxCNC).

HAL signals are an excellent idea. I'll have to investigate that (which is a nice way of saying, "No... I haven't looked at that"
Alan

So... Got my grounded RJ45 plug but it didn't make a difference. Tried adding some ceramic caps to ground, ditto... I even jerry rigged a brand new, isolated 24VDC power supply, and I've got the same behavior.

I finally took that new supply and wired up a 24VDC relay so that the pole of the estop switch was switching that relay on and off and nothing else. So, supply pos to estop, other side of estop to relay positive, relay negative to supply negative..... Yep, switching that relay on/off bricks the pendant.

Adding a diode improves things, a bit, which is interesting.... But it doesn't eliminate the problem. I'm using a giant rectifier diode though. I'm wondering if maybe a signal switching diode might be faster? Interesting thing is that there is a diode in the relay (although maybe it isn't working). Second, The problem can occur even when there is no relay in the circuit at all (essentially no load), which is really strange.

UPDATE:
I tried using a 30V zener diode too, because it was easily available. Really no difference in behavior than with the rectifier diode. Better than before, but still not great.

OK, that is really strange and gives me the impression that I would open the pendant. Unplug the cable and measure it completely. Resistance of each wire and between wires. Have also a look if the estop cirutrie completely from the rest of the electronic, or better, cable direct.

One completely different thoughts: Is in the original VistaCNC mode the estop eventually read by the pendant controller to show estop. That way the button need some hw connection inside and that would explain some side effects when you use other supply source. So forget warranty and look the cabling.
Mike