What represents, or what should I expect from, a well tuned servo system?

Thank you for the information on the index enable pin. It enhanced the description provided within the linked pages in my previous post.

It still feels very much like a one-shot (albeit a resettable type), but that is fine. I get what it is used for and why it was likely designed the way it was. I can get the information from it that I want.

I am curious about your comment that I can get the actual index signal. I didn't see any likely candidates in my list of signals when I started HALMETER (I assume the same selection is available within HALSCOPE). Using the actual signal should be a little easier than the enable because I would not have to reset the raw signal. Granted, it will be very short, but if I setup HALSCOPE with a rising edge trigger, then I can capture the pulse easily enough, even at speed.

I'm rather perplexed why my homing sequence failed on the Y. Normally I would say it was operator error (me mucking around with a config), but in this case I am pretty sure all I did was turn the machine on and then tinkered with the ACCEL and MAX_VEL parameters (per the latest conversations of this thread). As the index is only used in homing, I really don't want to touch anything in the cabinet, just in case I might disturb a broken wire, loose screw, or whatever. That is why I'm concentrating on software solutions using HALSCOPE or HALMETER, in addition to the fact that, frankly, they are simply easier to use and I should get practice with their peculiarities, etc.

The actual index signal will have a name like hm2_5i25.0.encoder.00.input-index

Note that you cannot reliably sample the index signal with halscope any faster than 24 RPM
(assuming 5000 PPR = 20000 count/turn encoders and 2 KHz base thread from your ini file)

This is why the "sets" method is better

Ah! I knew there was a good reason! Thank you for the explanation.

As for the actual index signal... I thought I looked at the pins too, but I don't recall seeing the input there. I was surprised at that but assumed it was due to the fact that the actual inputs were special purpose. I'll look again, though. I must have missed it.

Things have been going well. I have been tinkering with getting my INI values for the servo correct, etc. I've also got the VistaCNC pendant working reliably, etc. All that is good.

However, I've noticed a problem that cropped up recently. Often my HOME operation fails after turning on the power. The Y axis never finds the index pulse. I've watched the index enable signal (after a sets to set it to TRUE) and it doesn't change state). I'm not sure what is going on there.

While trying to figure that out, I noticed that my 5i25/7i77 seemed to hang once too. That was odd. But now I'm in even worse shape. I've not been messing around in the cabinet since it last worked, but all of a sudden, now I can't even start LinuxCNC. I get an error:

hm2/hm2_5i25.0: Smart serial card hm2_5i25.0.7i77.0.1 remote error = (8) Remote fault

I've cycled power and that doesn't make a difference. I've checked the 5VDC input to the board and that is fine too. I've also searched Google for information on what is really meant by a 'Remote fault' and that didn't yield any good results. I know there is some rudimentary connectivity because if I pull the 5VDC input to the 7i77, then LinuxCNC won't even start (says it can't find a pin, which makes sense).

This is certainly a regression from where I was before... Any ideas?

Not finding the y index could be a wiring issue or a dirty encoder.

The sserial error is very odd, is that the only error?
can you look at dmeg and post all errors?

Ignore

From dmesg:

[ 1466.439357] hm2/hm2_5i25.0: Watchdog timeout (50000 ns) is dangerously short compared to hm2_write() period (499888 ns)
[ 1466.439843] hm2/hm2_5i25.0: Watchdog has bit! (set the .has-bit pin to False to resume)
[

You have the watchdog setup with insufficient margin to run reliably
I would suggest at least a 10 ms timeout =10000000

Yeah, you beat me to it! I was hoping I could edit my post to this because I found the same thing:

How many ways are there to shoot yourself in the foot? Answer: I don't know... I'm still devising new ones!

Problem on the 5i55 was completely my own. I *had* made a change that I forgot about and the dmesg told me immediately. I had shortened the watchdog to something unreasonably short. I have to say that the *reason* wasn't totally bad though. I had left my computer on (not the servos) and I noted that there were some communications errors and resulting LinuxCNC fault when I came back 2 days later. That had me a little worried. I'd expect problems with my configurations of amps and the system, but not basic connectivity issues with the Mesa cards. So I shortened the dog.

I'm going to ignore that problem for a bit. Maybe it was just a power glitch... Don't know, but I'll monitor. Now that I can start my machine again, I'll delve into the index pulse issue. I noticed some irregularities with a logic probe (pulse LED was flickering even though there was even power to the amps). I need to get out a scope (difficult due to physical constraints). I'm not going to place much stock in the flickering yet though. It could just be a poor electrical connection between the machine and the probe power leads. I'll look at all that and post.

But, the question about the inability to start LinuxCNC (Remote Fault) error has been resolved.

You can probably diagnose the index issues with nothing but a voltmeter especially if you can manually move the axis to the index position (and measure the voltage across the IDX and /IDX pins in both states)

Quick update on the index pulse issue. Right now, my machine is behaving well. I can reliably home it over and over again. That is great except for the fact that I know there is an issue somewhere. I wish it would fail right now.

What I can say is this... The topology of my system is such that the 5VDC power for the encoder and limit switches actually comes from the Granite servo amps. The signals share a common ground that is connected into the rest of the system (Mesa) so the reference is there. I wasn't thinking about this when I hooked up my power for the logic probe and I used the system 5VDC (easier to access). Once I moved the probe to the same power as the encoder, my noise dominantly went away.... However, I still get some flickering of the probe's pulse light. Interestingly, however, this does not seem to affect the homing sequence (remember, it is working right now).

[edit because we were responding at the same time]
DVM might work. It is hard for me to position directly on the pulse though as I can't manually move that axis. I haven't calculated out the count of a 2000CPR encoder but my guess is that my 0.0001 resolution to jog will go over the index pulse, hence the reason I was using a logic probe.

Also, unfortunately I'm presently using the original wiring so I don't have differential. I hope to change that later but I wanted to get my machine running so I opted for single ended for the time being...
[end edit]