Probe issues, especially the first few times after launching UGS

I consistently have issues with probe functions right after launching UGS, but sometimes I discover probes fail at other times as well.

I have been meaning to post the console message I often see. This one was after a successful XYZ probe but the coordinates were not applied and it threw this error. @chrismakesstuff any idea what is causing this? On UGS from August 4,2019:

[Error] Error while processing response <<Idle|MPos:-587.000,-730.000,-92.000|FS:0,0>>

-Jeff

Hi Jeff, few questions:

  • is the probe behaving an any particular way that’s outside what you’d expect when you get this error?
  • is the error you’re getting the exact same one every time?

When UGS has an error the specific error code pops up in the bottom, right-hand corner so if you could screenshot that the next time you have the issue then that would help narrow down what’s going on here.

There are two specific behaviours that I’ve observed, although not at the same time.

One is that, when doing a Z probe after starting up for the day, the probe fails. You unlock and reprobe and it may or may not work the second time. The third time is charming.

The other one is that the Z appears to go off fine and then it crashes in to my block when it is attempting the X probe. I am not currently using my Sienci block, I’m using a stepped block that has a lower edge 12.7mm down from the top. If I unlock the machine and return to the start and run it again it works fine the second time.

Actually there is a third behavior I have seen now that I say that. The third one is that the Z probe looks ok, the X probe looks ok and the Y probe looks ok. It finishes with no errors. When I go to remove the block I discover that the Y probe is sitting in the second (slow approach) position, right against the block and is stopped. The coordinates are not updated in the system but all appears ok. Very confusing. Running it again works fine.

It’s been reliably unreliable during the first few probes and then is usually ok for the rest of the day, but I can’t trust it. My setup block as a circle in the corner so before I remove it I now always test a Return to Zero and make sure it is centered in the circle over the actual stock corner. That’s how I’ve caught it a few times where it appeared to have worked but the data wasn’t applied. It’s like a timeout issue between UGS and the controller, perhaps?

I’m running windows 10 on a Core i7 laptop that is about 6 months old. It’s basically dedicated to Fusion 360 and UGS, so nothing unusual about it, nothing extra loaded or things changing.

-Jeff

@gwilki has made similar observations about issues with probing after first startup.

That’s all very odd. There’s got to be a pattern here. I can’t find anything in the UGS github issues regarding probing failure on startup but I can give it a try when I’m back in the office on Monday.

I’d say in the meantime make sure that you’re starting at a set location every time you probe. The probing module is programmed to ‘seek’ a certain distance before it throws an error because it thinks that something has gone wrong. If you’re initiating the probe that puts the bit outside this ‘seek’ distance then the probing cycle will fail. This is the purpose for the circular logo engraved onto our Sienci touch blocks.

Also if you have a multimeter handy: ensure that all your wiring is well secured. Try to check for continuity while tossing the wires about to see if you can get it to lose contact. I can’t be very certain with your setup here since you’re using another block.

By the way, can you satisfy my curiosity and let me know what the perks are to this other block you’re using?

Hi Chris,

You can read more about it here: XYZ Zeroing plate and tool diameter

I originally had it made for the ShopBot at MakerLabs here in Vancouver. When using a shared machine under time pressure, the ability to reliably set up your work and recover from issues is really important. Especially when doing full sheet jobs at 4x8 I found the probe and the scripts that go with it and then did a slightly modified spin on it. The scripts Ron Olson wrote for it also put the bit in the center of the circle and prompt for confirmation from the operator that it is correct before retracting.

The biggest benefit, day to day, is the size and weight of the block. The Sienci supplied unit is so small and light that it can easily be pushed off square, this thing isn’t so easily misaligned. Also, with two Z height options, in certain circumstances it is helpful to be able to probe Z on the lower portion. I had to do it one day last week although the exact use case escapes me. Of course you need to adjust your height accordingly since the offset at that point is zero not 12.7.

Ron and I didn’t implement it but we were discussing the fact that the code could be improve to actually detect bit diameter so you didn’t need to worry about remembering to set the correct bit diameter. That’s something that would be welcome. I find the placement of various probe functions (well a lot of functions, actually) in UGS to be quite unintuitive compared to the ShopBot software (which is far from brilliant but makes more sense in some areas).

So really it’s size (weight/stability), the ability to see the corner position centered in the circle and some flexibility with the lower edge. It isn’t something I bought to replace the Sienci one (which I included in my order) it is something I had made while waiting for my machine to use with a specific ShopBot script so I could improve reproducibility and recoverability when using the PRS-Alpha at MakerLabs.

It’s become very clear to me in the past week that recoverability and consistency require homing/limit switches and a solid probing solution - combined with a grid/doghole alignment strategy to position work on the table. I’m getting closer to having my arms around all of this. Next step is playing with the limit switches but I noticed mine are labeled 6V-32V I’m not sure if there is enough juice to drive them off the 5V labeled line on the Longboard. I’m going to pay with them on a Raspberry Pi setup and see what I can do with a bench power supply. If that works and I need to I may just power them separately and only connect the signal pin. Any reason that won’t work?

-Jeff

I’m assuming that you have normal mechanical limit switches (AKA micro switches). They don’t require power. they just switch it. Are you sure that the markings aren’t 6A-32V? These would be the maximum limits. Anything lower should be fine. They should have no problem switching 5V.

@jwoody18 @chrismakesstuff I’ll chime in only because Jeff mentioned that I had issues with probing in UGS. I am risking glitching it by typing this, but since I went to a 2020 release of UGS, I have had no issues with it at all. The probe function works fine and no freezes. If anything happens later this afternoon, I’ll blame Jeff. :wink:

1 Like

@gwilki I’ll give the updated version a try hopefully that fixes it for me.

@chrismakesstuff I remembered what the use case is for the stepped block. Both the top surface and the stepper surface are identical in thickness, in my case 12.7mm. So this means you can either corner probe and get the Z or, say you are in the middle of a tiling job :slight_smile: and you no longer have your corner to measure, you can simply set the corner finder atop the work piece and probe at the lower step without having to remember to change any settings and you’ll get the same reading for Z9. Handy.

-Jeff

Paul,

Mine are the electromagnetic type, I think they are called Hall sensors? I have one here, let’s see if I can find a link.

LJ18A3-8-Z/BX is the mouthful of the Chinamazon part number. They are quite long, maybe I simply ordered the wrong thing? There was precious little information or guidance on the Sienci pages about what actual parts could work so I tried to dig around and this seemed ok.

According to the company name this should be my best option. :wink: https://www.amazon.ca/URBEST-LJ18A3-8-Z-Inductive-Proximity-Cylinder/dp/B06XRSV6XG/ref=sr_1_1?dchild=1&keywords=LJ18A3-8-Z%2FBX&qid=1591560235&sr=8-1

@jwoody18 Jeff: Re: your touch plate. If I am understanding you, I do the same thing with the Sienci plate. When using it on the corner, it is placed on the corner with the printed side up. When using it in the middle of a piece, it is placed with the printed side down. The settings in UGS stay the same.

Well that makes sense. All good tips and trick to add to the doc wiki?

And maybe some info about these sensors and how to make them work? :slight_smile:
-Jeff

@jwoody18 Jeff: Paul will likely reply, but FWIW, I don’t believe that is what you want for limit/homing switches. They are proximity switches and from your link, detect out to 8mm. I believe that we are better off with mechanical switches such as these

These are just an example, not necessarily what you want for your machine. I bought micro mechanical switches that can be either normally closed or normally open. There are scads to choose from.

@jwoody18 Jeff: I’m not trying to pull your chain, but here is the relevant paragraph on the Sienci site: :wink:

To find only the top of your material in the Z direction (Z),

Flip your touch plate upside down.
Place your touch plate on top of the material that you want to find the top of. Position the machine so that the end mill is above the touch plate.
Go to the ‘Z’ tab and press “Initiate Probe”.

(slinks away in shame)

1 Like

I think the ones I have will work well, they are probably just overkill.

I found this detailed page on how to wire switches to GRBL: https://github.com/gnea/grbl/wiki/Wiring-Limit-Switches

@chrismakesstuff on that page they detail a $2.30 board they published the design for that substantially improves the noise resistance of the limit sensor pins if placed in front of the Arduino inputs. It seems like a really good thing to include inside the Longboard the average Joe CNC buyer doesn’t need to think about it, or at least somehow offer it as part of a limit sensor kit for the Longmill?

As I’ve gotten more in to this over the past few weeks it’s become increasing clear to me that to really unlock the repeatability and accuracy of the machine (of any CNC) it is critical that you avoid crashing it and causing (even micro) changes in alignment and coordinates. Moreover, if you are doing complex, multi-part cuts (including tiling or cutting on both sides) you really need that accuracy to be maintained.

Recognizing that cost/price is always a consideration as well, it would still be good to have a check box item to add to the cart to add these recommended components that are known compatible with the Longmill and make it easy for new purchasers. For the time I’ve spent frustratedly re-aligning things and re-zeroing to try and get back to normal in the middle of a job, as well as the stock I’ve chewed up while learning about this, I would have happily paid $100+ to have had it all up front. Had it been properly explained to me.

I sometimes think the tendency is to over focus on price when promoting products, especially ones with elements of open source, instead of valuing the whole equation of price vs. time and materials. Obviously it’s different for everyone, but you may be surprised how many people would part with another $100 if the risk/reward was clearly outlined. It’s nice to have options…

I’ve got my switch behaving just like I would expect here on my “bench” (currently a large plastic tote sitting on a moving dolly with additional work space provided by the leftover box from my shop vac. :slight_smile: Even so, I was pleasantly surprised to see those switches I bought have status lights on the back so you can instantly see, even with just power connected, that they are triggering on approach of metal.

This gentleman provided a wonderful overview of the switches and various resistor values to use to create voltage division to provide appropriate voltage for your microcontroller.

@chrismakesstuff is it correct that the pins are normally 5V high when the limit sensors are connected and they should be pulled love to 0V when they are triggered? It appears to be the case but there isn’t any detail on that here: https://sienci.com/dmx-longmill/adding-limit-switches/

Suggest adding a bit on the different types of switches, maybe a link to that GRBL page and a few details about the specs of your ports for limit switches. The switches I have report requiring about 300ma of draw but I think I saw in a post from Andy that (at least one port, maybe the PWM port?) is only good for 300ma total. How many amps could you draw on the 5V line for the limit switches? I suspect powering them separately or from the 24V supply outside of the Arduino is a better idea? I should go look at my Longboard, but do you expost 24V +/- anywhere on the edge connector?

-Jeff

For the benefit of new users (like me) who discover this thread at some distant future date - and who don’t have a soldering iron/soldering skills - check out Wago connectors as noted in this thread: Wago Wire Connectors

Just set up two of the switches in parallel on a small 12V power supply I have and they seem to work a treat. Will play with resistor values now, thanks to the hints Martin left in the video I linked above, to get the voltage down to 5V for the trigger value for the Arduino. If you don’t drop the voltage (and you’re powering them with more than a 5V supply) then you will likely damage the Arduino controller inside the Longboard. Don’t do that.

-Jeff

I agree, Grant . Simple mechanical micro switches.

After getting it all laid out on the table and getting it working for three axis (per my multimeter), I’m inclined to agree on the mechanical option. These look great but are a lot more expensive and don’t have any particular benefit that I can discern, other than I learned quite a bit and enjoyed it.

Unfortunately, a lot of conflicting information out there and conflicting opinions but the best argument for Normally Closed (NC) switches is that when switches fail they fail open. You can put the NC switches in serial and then you will know immediately if a sensor line is down, whereas with the NO (like the ones I have) you don’t know the sensor isn’t operational and in the event of a failure you will have a crash.

There are both mechanical and proximity (magnet) ones for NC and NO. The simple mechanical NC ones seem like a good option for most, although it seems like quality of the various ones reviewed on Amazon leaves a lot to be desired but they are super low cost.

I’m considering returning mine and going the NC route.

-Jeff