THEORY-Clearpath Step Input Timing Error-Contributing Factors

All things related to the Centroid Acorn CNC Controller

Moderator: cnckeith

Post Reply
carboncymbal
Posts: 88
Joined: Tue Dec 15, 2020 11:12 am
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: 0035FF8A1900-10082038
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

THEORY-Clearpath Step Input Timing Error-Contributing Factors

Post by carboncymbal »

All,

I have a theory I’d like to present and would appreciate any feedback. I’ll also present my observations. Before I start, this isn’t meant to be an attack on Centroid at all, and I hope it doesn’t come off that way. I am simply trying to document my observations and experience. I have also made a number of assumptions, please correct me if I am wrong.

TL;DR- Proper panel design is important, including steps to reduce EMI. Clearpath is unique among systems because it actually senses and faults when there are step input timing errors AND Acorn may be more sensitive to interference than other systems.

Here are some observations to set the stage, please provide feedback if you disagree.

1) Clearpath servos are a popular option for DIY machines and there are years of example did people using them successfully.

2) Clearpath servos haven’t always monitored for the “step input timing error”. Apparently this is a feature not present in older versions of the Clearpath firmware.

3) because Clearpath uses current limited, opto-isolated inputs, they should be relatively immune to EMI issues on the control signal wires because the EMI produces very little current, even though it may introduce high voltage noise on the wires. According to Teknic, the motors are essentially blind to this type of noise. I assume that is because they inputs require a higher level of current to operate than can be introduced via EMI.

4) A quick search of forum posts across various CNC forums show that when clearpath error 73 or “step input timing error” is reported, it is almost always on an Acorn based system. I actually don’t think I have seen any reports of other controllers experiencing that particular issue. (Maybe they exist, I just don’t recall any).

5) Cenroid discusses the need for proper panel design and suggests many specific steps to limit that noise. They always advocate for using best practices to limit noise and treat this concept as a critical design requirement.

Now, my anecdotal evidence.

1. I could easily replicate the clearpath error 73 in my system by running low RPM on the spindle (~30hz) while jogging the machine. It failed every time, within seconds of jogging.(no resistors installed, VFD Cable shield not grounded at VFD)

2. I grounded the VFD cable shield at the VFD, no change (didn’t come grounded at VFD by my original
Panel builder). System failed.

2. I added the suggested pull-up resistors as Teknic suggests. The system failed.

3. I be added ferrites to the VFD output. I could not get the system to fail.


When I consider all of the above points together, I arrive at this conclusion. Teknic’s description of the noise immunity of Clearpath is correct, as evident by the very minimal noise issues reported generally, adding resistors is appropriate to help achieve a square wave signal due to the fact that acorn's 24v outputs do not drive the signal high, but only low.. A lack of the resistors can cause systems to float low for an extended period of time (effectively shortening the high part of the signal pulse), which is detected as a timing issue. If the Clearpath inputs are truly dependent upon current higher than can be generated by even extreme EMI environments, then EMI entering the control cables between Acorn and Clearpath isn't the issue. Yet, I have clearly seen where the VFD operating at low frequency has caused issues on my machine.

This leads me to believe that Acorn is actually outputting incorrectly timed pulses in these cases. I propose that this issues is caused by VFD created EMI affecting Acorn in some way. This may explain why some people report that installing a line filter as part of the steps that address the issue. I don't have enough understanding to justify that theory any further. It basically comes down to a process of elimination when I consider Clearpath's broad usage in the DIY and OEM CNC world, and the fact that these issues seem most common with ACORN systems.

Possible Countermeasures

-Install pull-up resistors as specified by clearpath
-Ensure that single + and - wires come back to Acorn in pairs
-Use best-practices for grounding components within the panel
-Install power line filters on the VFD
-Install ferrites on the output of the VFD (like the automation direct zero phase reactors)

Thanks,
Clay
Muzzer
Posts: 728
Joined: Mon Feb 19, 2018 2:52 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: 38D269594F9C-0110180512
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No
Location: UK
Contact:

Re: THEORY-Clearpath Step Input Timing Error-Contributing Factors

Post by Muzzer »

I'm not imagining any kind of special sauce in the Clearpath drives. I'm not aware of many drives that do NOT have opto inputs and the better ones have a resistor on the opto as std, so it's not clear why you'd need to "enhance" the Clearpaths by adding a pullup resistor in the first place.

Here's what you will find on a Yaskwawa Servopack, which I would consider to be a good example of professional / industrial practice. They have 4k7 resistors to help turn off the optos.
Servopack.JPG
This claimed error detection can only work if you have a continuous train of pulses and some are missing, in which case you most likely have an installation issue. It can't possibly tell you if individual pulses have arrived but not made it across the opto. Their marketing guy might think it sounds like a compelling USP and it seems to have gained some traction.

You should fit an EMC on the input of ANY VFD as a matter of course - unless there is one already. Generally they don't have them but that's only because the required filter will depend on the installation, not least whether it's domestic or industrial. Often in hobby forums people only only fit them to alleviate problems. Fit a fence at the top of the cliff, not an ambulance at the bottom!

Fitting ferrite beads on the outputs of the VFD may fix the problem in this particular example but it simply tells me the issue is down to poor noise immunity.

I'm not finding myself convinced by the "straw poll of general opinion on the internet" or somesuch. Nothing personal intended here but as a professional VFD / PSU product developer, I feel the theory needs some refinement!
Gary Campbell
Posts: 2164
Joined: Sat Nov 18, 2017 2:32 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: Acorn 238
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No
Location: Marquette, MI
Contact:

Re: THEORY-Clearpath Step Input Timing Error-Contributing Factors

Post by Gary Campbell »

Clay...
Agree with your observations, as they have been posted here and other forums previously

My anecdotal experience (thru a few customers translations) has been similar, but my testing order was different

I do not, however, agree with your surmised conclusions. Points of disagreement are as follows:

1) I was never under the assumption that EMI was "entering the cable". When EMI is present it is "in the system", often most evident on the ground. When it bleeds to LV com or any other signal wire, it usually has a negative effect.

2) I wholeheartedly disagree with the statement: "This leads me to believe that Acorn is actually outputting incorrectly timed pulses in these cases." When EMI is present it may shorten, lengthen, increase or negate a signal. Remove the EMI, the issue is gone. The Acorn board may benefit from some undefined improvement to make it more immune to EMI, most systems would, but no one has shown where this is needed.
GCnC Control
CNC Control & Retrofits
https://www.youtube.com/user/Islaww1/videos
carboncymbal
Posts: 88
Joined: Tue Dec 15, 2020 11:12 am
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: 0035FF8A1900-10082038
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: THEORY-Clearpath Step Input Timing Error-Contributing Factors

Post by carboncymbal »

Muzzer wrote: Tue Feb 23, 2021 11:04 am I'm not imagining any kind of special sauce in the Clearpath drives. I'm not aware of many drives that do NOT have opto inputs and the better ones have a resistor on the opto as std, so it's not clear why you'd need to "enhance" the Clearpaths by adding a pullup resistor in the first place.

Here's what you will find on a Yaskwawa Servopack, which I would consider to be a good example of professional / industrial practice. They have 4k7 resistors to help turn off the optos.
Servopack.JPG

This claimed error detection can only work if you have a continuous train of pulses and some are missing, in which case you most likely have an installation issue. It can't possibly tell you if individual pulses have arrived but not made it across the opto. Their marketing guy might think it sounds like a compelling USP and it seems to have gained some traction.

You should fit an EMC on the input of ANY VFD as a matter of course - unless there is one already. Generally they don't have them but that's only because the required filter will depend on the installation, not least whether it's domestic or industrial. Often in hobby forums people only only fit them to alleviate problems. Fit a fence at the top of the cliff, not an ambulance at the bottom!

Fitting ferrite beads on the outputs of the VFD may fix the problem in this particular example but it simply tells me the issue is down to poor noise immunity.

I'm not finding myself convinced by the "straw poll of general opinion on the internet" or somesuch. Nothing personal intended here but as a professional VFD / PSU product developer, I feel the theory needs some refinement!
No offense taken at all, I’m well out of my depth here. I appreciate the feedback.

Do you know if it is common for other makes drives to error out with step input timing errors or something comparable?

You are right that I’m basing this largely off of observations of postings on the internet, which is no doubt weak evidence. I suppose my main point is that that proper emi limiting design is important and may be a solution for the issues. The next question for me is what is the EMI actually harming? Is it the clearpath or the acorn?
Richards
Posts: 692
Joined: Sun Sep 16, 2018 9:01 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No
Location: South Jordan, UT

Re: THEORY-Clearpath Step Input Timing Error-Contributing Factors

Post by Richards »

Try to visualize an ideal step signal as a square wave with sharp sides and an error-prone step signal as a ramp with long sloping sides. Next think of what happens inside an opto-coupler. If we just look at the data sheets, we might see the forward voltage listed as 2-volts and the required current as 10mA. So, we might think that as soon as the voltage reaches 2V and the current rises to 10mA, the opto-coupler would turn on. In a perfect work, it might. What you also need to look at is the "rise time" that is acceptable for the opto-coupler. A fast rise time means that the signal will pass very quickly through the ramp. A slow ramp time means that the signal will pass through an "undefined" level slowly.

Electrical noise can be "induced" into a signal. Think of how a transformer works. One coil of wire is wrapped around or near another coil of wire. Voltage is "induced" into one coil from the other coil. Wires inside an electrical enclosure act much like those coils in a transformer. ANY piece of wire can pick up voltage from any other piece of wire. Good practice calls for routing high voltage wires away from low voltage wires and crossing them at 90-degree angles. Good practice calls for using shielded cable with one end grounded to a common ground lug. Good practice means using chokes and filters on known sources of electrical noise. Even after following good practices, some noise may still get through. Personally, I don't mount a VFD in the same cabinet as the Acorn. When I had a Shopbot, I mounted the VFD on a wall about 10-feet from the electrical box. I used input and output filters on the VFD. I did that mainly because I thought that the VFD was the source of sporadic noise. (It turned out to be a WiFi board in the computer that I had forgotten was even there. That board interrupted the CPU sporadically, causing my problem.)
-Mike Richards
Post Reply