Proximity switch repeatability?

[phil@philsfoils.com]

Has anyone found an inaccuracy with their proximity switch(es)?

My situation is that I zeroed the machine, then cut a groove down the center of the table (at y = 24) to insert a T-channel for a defined centerline, needed when flipping work over to mill double sided.

I find later when re-zeroing the machine, and then moving to y = 24 again, it can be visibly off-center from the channel, by as much as 1/8".

Any hints here?
THANKS
Phil

www.philsfoils.com (http://www.philsfoils.com)

[gerald_d]

Do you approach the proximity at the same speed every time?

(I don't use them - one of the reasons being that I believe they would be sensitive to speed of approach. No prox. switch supplier, that I know of, actually gives a spec on repeatability of sensing distance, other than to say it will switch inside 1/8" or 1/4")

[phil@philsfoils.com]

I always zero with the "C3" command, which goes at jog speed until it hits the proximity switch, then backs up a bit and re-approaches the switch slowly.

Only thing I can think of offhand is that I have had the router out of its mount, but re-inserted it in the same orientation (and to me it looks perfectly round where it clamps into the mount anyway... Makita RF1101).

Phil

[erik_f]

It could be possible that saw dust sitting on top of the bolt or collected on the bottom side of the switch itself is acting as a shield and blocking the indcutance needed to make the switch work properly? Just a thought.

Erik F.

[Scott Smith (Unregistered Guest)]

I've bumped my sensors out of adjustment before. Only had to adjust the values in my zero program to the new distance from zero. The setup procedure is quite easy and you could set it from your groove in the center of the table. (center the bit in the groove, Z2, do a move command to hit the sensor, note distance, subtract 24, enter new zero offset in the zero program)
Other than that my sensors have been very repeatable.

[ckurak]

A trick that I learned from Chris Burns at ShopBot: Add a touch of silicone to the side of the sensor to keep it from wiggling.

[peter_hill]

Phil - Using the C3 calls the file S_home2.sbp. This file just uses the current speed settings and if you had previously run different files with different speeds then the move approaches to the proximity switches would be at those different speeds. I have inserted a move speed setting into the S_home2.sbp file (MS 1) to ensure that it will always approach at the same speed.

[Sheldon@dingwallguitars.com]

We've done what Peter suggests. It works reliably.

[phil@philsfoils.com]

Thanks for the tips.
Phil

[slendon]

I'm still having trouble with repeatability using the prox switches:

I have a very recent AlphaPrt. The latest version of control box software downloaded from the website, and the carriages run very sweetly on the x and y axis.

So far all other moves are accurate and as expected. JH from anywhere on the table is pretty much spot on.

The problem is if I use the prox switch.

I choose a physical location and drill a hole, then set zero in both axis. Using keyboard and move (at same speed as the homing prog uses i.e 1)until the prox switch stops. Then pull away from the switch using the keyboard(JH wont work directly off the switch,even after waiting the 5 seconds indicated in manal)Once off the switch use JH and the machine, physically zeros a few millimetres further along the positive direction of the axis.

The same thing happens if I alter the S_home routine files with the negative amounts, and run the homing program.

I had thought it was to do with the speed of approach to the switch, as per some of the above posts. But I have tried many speeds, and also made sure that I have matched keyboard speeds to JH speeds (and homing routine speeds)

The error occurs on both axis (X and Y)and is always repeatable.

Physically the machine seems to stop with (at) the prox switch in the same place every time.

Anyone got any ideas?

Steve Lendon

[paco]

Is your gap at sensor/head screw right? Is it set as suggest by manufacturer? How much is it off position, average? Is the sensor dirty?
Hmmmmm..... get me thinking that I may avoid this type of "swichtes" for this purpose...
Be aware that the control software is... still developing... a little "buggy"... as from my observations though...

[fleinbach]

Steve,

I just tried to see if I could duplicate your problem. I Zeroed my table using the prox switches via the C3 command. Next I used the M2 command and drilled a 1/4" hole at X12,Y12 I then used JH to home tool. Next I moved several times with both the M2 and J2 command to position 12,12 and back to JH. After manuely moving Z down into hole I could freely spin the spindle by hand in the hole.

Then I repeatedly moved to any location on the table and reset zero. Next I used the C3 command to rezero to the prox switches. Then I repeated as above. I hit the hole every time perfectly.

I have aproximatly 1/16 gap between my prox switch and target bolt. This appears to give me great accuracy in setting Zero/Zero.

[slendon]

Paco,
Sensor gap is set at 1.8mm as per instructions. The switches and bolts (triggers) are clean and new.

Frank,
Thanks for duplicating the test. The machine seems to stop, using prox switch, in the same physical place every time.

The errors look like this:

(from a 0,0, point in the middle of the table, move back by keyboard to prox switch, move off switch and JH home. Then repeat at same speeds)

First move from o,o to prox= -54.871

2 -55.035 difference= .164
3 -55.166 = .131
4 -55.361 = .195
5 -55.557 = .196

Any ideas?

Steve

[bill.young]

Hey Steve,

What happens if you give it a J2,0,0 command instead of JH? How about an M2,0,0?

Also how about zeroing the X and Y axis at the usual 0,0 corner of the table (so that it doesn't have to move so far) and then use the keypad to move each axis in a negative direction until the prox switches contacts and the tool stops? If you do this a bunch of times is the position readout in the red panel consistent and show that it's triggering at the same position each time, or does it vary?

On last thought. I've had the tiny little bolt work loose that mounts the prox switch to the carrige...it's awfully small. That caused all kinds of headaches until I figured out that the prox switch was moving a little as the tool moved.

Bill

[fleinbach]

Steve

I tried your method and I completly agree. Using any command either J or M will keep resetting zero when running back to the prox switches. I don't know why this occurs, but I'm also not sure why you are doing it this way.

If you use the C3 command to set your zero possition (and it doesn't matter where even in the middle of the table if that is where you want zero to be) It will remain accurate as I described earlier. As long as you don't hit the prox switch again, and the only reason you would should be accidental.

Of course setting zero in the middle of the table will take a while due to the slow speed needed for the C3 zeroing.

[slendon]

Thanks for the posts Bill and Frank, I will be back in the workshop tommorrow, and look forward to trying these ideas....will post the result.

Cheers
Steve

[slendon]

Well, after a sleepless night, I got up early went down to the workshop........

Bill..Over the past three days I've tried most everything including your suggestions, still the same errors.

Frank..You win, using C3 command,( after pulling off the prox switch with the keyboard )zeros the machine spot on.
The reason I have been using the JH command is because the manual says so !!(cavaet:Unless I'm missreading it, which is quite possible)(under the Assembly Accessories tab, prox switch installation, page 6 par 8 ). I have had several e mails and a call with support over this but they were unaware of problem/solution.

Would be nice if shopbot set up a scheme for rewarding people who identify repeatable software errors (that they haven't spotted yet), maybe a free cutter, or two if they provide the solution.

Frank, I owe you a beer or three, you'll have to come and collect them though!

Cheers

Steve

[fleinbach]

Let's see. Paking, Air fare. Nope, I believe beer is a bit cheaper here. But thanks anyway.

I have reported about half a dozen software problems so far and most have been fixed.

[fleinbach]

By the way 3.2.17 was posted yesterday.

[ted]

On Homing with C3:

Here's some info for those who are hassling with this.

First, let's separate for the moment PRTs and PRTalpha. Ultimately the core control software code will be the same for both tools. At the moment, as indicated by version numbers, the PRT software is lagging the PRTalpha software (this is probably a good thing for PRT users).

As far as we know, the -PRT- software for homing with C3 and also the operation of JH works just as it should. Problems that are happening may be related to issues discussed in the postings above. You should be able to reliably position with to within a couple of thousandths with the prox switches and the C3 routine. When you hit the prox switches manually, because speeds and distances may vary, there may be a little more variation in how the prox switch acutally triggesr. So if you are using prox to zero your tool (as you should), then use C3 or some routine you create that works on the same type of principle. Call us if you need some help on this.

Now ... let me complicate things a bit and talk about the situation for -PRTalpha- tools. As we began to get more and more experience with the alpha tools, we realized that to optimally benefit from their speed and performance the acceleration and deceleration ramping needed to be enhanced from the existing PRT Windows software. We have added a 'softer' curve for initial stops and starts, and to the look-ahead functions, we have added the ability to slow down for curves and corners to a graded degree. That is, look-ahead ramping now anticipates how much to slow down, rather than just whether to slow down. For a PRT, it was adequate to just reduce speed (say 1.5in/sec) down to ramp speed (say .6) if a sharp corner was coming up. Medium corners could easily be handled at full speed (1.5). For alpha tools, cutting fast (say 8 or 10 in/sec) additional slowing options needed to be in place for partial slowing rather than to just slow to the full (low) ramp speed. This advanced ramping is now implemented in the current version of the software. I believe that you will find ramping and cornering very smooth, even with very fast cutting speeds. (Earlier version slowed much more than was necessary in some situation, and in others did not slow enough.)

Unfortunately, in the course of getting these changes into the sofware, we broke a couple of things ... in particular the response to hitting and "S" stop and then resuming (particulary bad when it happened in a JH).

I believe that the version posted yesterday for alpha tools (3.2.17_PRTalpha) deals with the majority of these issues ... though in a few situations, the stop you get after hitting the "S" is not yet as smooth as we would like it to be (a little more work should resolve this).

All this means that I believe that zeroing the tool with C3 should now work just as you expect, automatically and with a lot of precision. In the alpha software, C3 sets the prox approach speeds to a constant speed and then re-contacts at an even slower speed to give a very reliable measurement of prox location. At the end of C3, your speeds are automatically reset to what you had them at when you started C3 (you will need to reset the speeds if you break the C3 routine before it finishes).

I think Steve's use of JH for 'homing' the tool may result from the confusing way we have used the word 'home'. In one case, when we say JOG HOME, we mean return the tool to the 'current home position' ... and this does not involve using the prox switches. In the case of C3, we use the 'homing' routine to 'SET the home position'. Yes ... this is too confusing. I'm thinking we should refer to C3 and the routine it calls up as something like: 'Set the X & Y zero'. This would be more like 'Z_Zero' which does the equivalent thing in setting the Z zero location (for some reason, we seem to have fortunately avoided saying 'home the Z').

It is a confusion we have become more and more sensitive to because we have become aware of several ShopBotters who have tools with prox switches and who have not realized for months that there was a routine for automatically and reliably re-zeroing the X and Y axes that made use of the prox switches (that is, that they were much more than just limits).

Other suggestions ??

-Ted

PS. The issue of zeroing brings up the somewhat challenging matter of also making full use of the 'TABLE BASE COORDINATE' system, which enables file and move checking for verification and making sure that table limits are not violated. While most PRT users do not make use of this system, such checking is increasingly important at PRTalpha speeds. It is covered in the manual, but Bill Young has prepared an additional little tutorial on how to easily incorportae using this feature of the software (all versions) into your operations. We'll post it on the web site in the next few days.

[paco]

"Would be nice if shopbot set up a scheme for rewarding people who identify repeatable software errors (that they haven't spotted yet), maybe a free cutter, or two if they provide the solution."

Can you imagine Frank?!!

Noting about wrong about this Steve but it really got me rolling on the floor!
How about you Ted?

Anyway, I think that the answer is complete! Thanks Ted. I'm eager to test 3.2.17

Bits for problem report!?...

[ron brown]

Paco -

Are they going to take bits away when someone asks a stupid question that has already been answered 20 or 30 times?

Ron

[paco]

Ron -

To who are you refering to?! Or to what "stupid" question?

Paco

[Brady Watson]

The real reward of providing feedback and reporting errors is knowing that you are helping the ShopBot community, and in turn, yourself.

-Brady

[ron brown]

http://www.talkshopbot.com/forum/messages/312/4524.html?1093952294

From this thread Paco, you were looking for proximity switches on August 26th - I think - Did you install some sort of locator switches? If so - how did the installation go?

What sort of accuracy did you get? What did you use to measure the accuracy? Inquiring minds want to know.

Thanks,
Ron

[simon]

Hi
I do not use the prox switches at all. For repeatable homing I have drilled oversize holes, and insulated a bolt using a short length of PVC pipe and tap washers. The bolt is then connected to an input. No 3, I think, from memory.
Same deal on both the X and Y axes, bolts on both rails. I have adapted the homing routine to run fast at the bolts, which enables the carriages to square up and zero, before approaching again slowly for accurate homing. First x, then y, and then it homes on a wide aluminium plate for z.
I think this method is cheaper and less failure-prone than proximity switches.
Simon.

[paco]

Ron -
Yes, I am still looking... Is it stupid to you to be looking for something? And trying to get info from others?
Only stupid is stupid, I believe... and no questions are! If my, or others, questions get you annoyed, just don't bother joining it!!!

No, I did'nt installed anything yet because I have'nt made my mind on a reliable, accurate and cheap "system". But I plan to use both Z-zero methods (to material surface and spoilboard surface) depending on the needs and probably use micro switches to zero X and Y. Later, when I get more reliability from Alpha control...
If I can get 0.015" accuracy ALL the time I'll be happy! And I think it will be and maybe even better. I'll be using my eyeball to appreciate accuracy... as; do inlay part fit tight? do different zeroed bits depth cut are about the same? does it look accurate and good? Things like that... I don't machined aviation parts.

Now, Simon, is it the carriage that hit the bolt head??? The first approach at high speed interest me... please tell us more from your method. You got me curious...

By the way, the laught I had from a previous post in this thread was only a thought I had... nothing about anybody. Just an idea I had...
I think like Brady about the rewarding part...

[beacon14]

Here is a pic of my old shop-made prox switches. It was nothing more than a carriage bolt through a block of wood, with a spring to keep it extended, but allow a "cushion" to prevent damge to itself or the carriage upon contact. I chucked the bolt end into my drill press and "turned" the head round with a file, so that any rotation of the bolt in the hole wouldn't matter. There is also a little "ceiling" built into the wood block over the nut and washer, to prevent dust from falling in and affecting the setting. The bolt length and spring were selected so that contact would be made when the carriage was 1/2" or so from the hard stop, so the spring would retract enough that in case of operator error, the hard stop would catch the machine before it tore the whole switch assembly off.
I wrote a program to contact, pause, move away, and repeat, so I could check repeatability. It never varied more than .002 (the machine's own resolution) in over 100 hits.
By the way, my first attempt involved microswitches (biggest advantage: they're cheap), but I was lulled into false sense of security as they would work well for a time, then all of a sudden (usually in the middle of a run) they would change by several thousanths. That's what led me to develop the bolt method.

Regards,
David B.
3688

[paco]

Thank for sharing your design David! Seems great!
Did you find out why you observed "error" from microswitches?
Would'nt the "default" X and Y zero routine (home.sbp) work good with your design?

[slendon]

Ted,and all.

So let me get my head straight about whats happened so far. ( and I do appreciate its a tuff time at the moment with so much new stuff to be integrated )

I insatalled prox switches supplied by shopbot, I followed the directions in the supplied shopbot manual, they failed to work, so after a day of going over everything my end I contacted support at shopbot, who were unaware of the problem. I spend another two days back and forth with support, finally guys on this forum stepped up and Frank provided the answer.

Next the boss steps up and says that they did know about the problem, and I must have been confused by the manuals teminology.

Seems like I've lost three days because of very poor communication within shopbot.

Would it not be common sense to post all known bugs, as soon as they are spotted, on a seperate thread, so that we can all look it up, before spending time trouble shooting.

By the way I do still feel guys who spot bugs, should be rewarded. I applaud the generous ones who post here, but the world being what it is, there are probably plenty of other clever guys out there who need a little incentive to benifit us all, shopbot should provide the incentive as it is one of the quickest ways to de bug the software.

For my part the biggest single issue that I had to come to terms with when I bought the machine was concern over the capability, bug free nature of the software, and the quality of the support. The downloads from the web site for control software and part wizard, before purchase, are great ideas...but the anomolies and missmatched help files nearly put me off buying.

So far my fears were well founded


Please understand this post is not "putting the boot in" It's from a desire for things to only get better with the use of constructive critiscm. After all I have a large vested interest in the long term health of both shopbot and the forum.

Steve

PS.
Anyone know how to get the Zzero prog to work in mms?

[paco]

Steve, I fully shared your opinion. I wish I could be rewarded for reporting "bugs"... BUT, the only thing I am looking for as doing this is to get the "bugs" fixed so my tool get more reliable and to enjoy sharing thoughts and ideas with other users. I think now that it could be faster to get "support" by using both forum AND SB support at the same time as this might focus more people about one problems. But if one is really hoping to get rewarded by reporting problems to SB, he might just get a REAL suprise!...
I have use other CNC machines and NEVER get any kind of good support from any. Must I had that thoses machines were priced between 200 000$ and 1 500 000$us... So I think that SB support is way great and the forum even better... but not perfect! One have to dig to get is questions answered... sometime.

I just use 3.2.17 today and Ted, you got homework! But IT IS getting better.

Now about Zzero Steve, had you try to get all the distance JOG and MOVE at metrical size in the Zzero file(like changing the thickness from 0.125" to something like 3 mm)? Don't forget that the tool too should be set metrical... I don't notice anything in this file that related it only as a standard settings. Keep us inform of what you find out!

[beacon14]

Paco,

I think the microswitches were pretty accurate for what they were, but they are simple mechanical switches, and they only had to change by a few thousandths to throw off the accuracy of what I was cutting at the time.
Yes, you can use the standard XYZero routine, the program doesn't know or care what type of switches you use, it just looks for the circuit to be completed.

I have also experienced the frustration of being told 2 (or 3) different things by each successive person I spoke to at ShopBot. To some extent that is an inevitable outcome of a company employing more than 1 or 2 people, especially when they are busy making improvements, keeping up with changes, growing, and trying to keep track of all the earlier model configurations that customers have purchased over the years. I wish there were a simple solution. All I can suggest is a method that much larger companies use (or should use), where each customer has a record, and whenever you make contact with a customer service rep, they annotate your record with detailed notes - so the next service rep can scan your record and know your history and what you have beed told previously.
No company (or person) is perfect, of course, but ShopBot gets good marks for making an active and honest effort to diagnose and solve problems. I will be interested to see how they continue that record as their company continues to grow and their customers become more numerous.

David B.

[sevy (Unregistered Guest)]

Hi everebody. I'm a CNC operator, who likes computers and knows a little bit of electronics! My dream, one day is to build my own CNC machine...but until then, i draw almost everyday what my circuit board will look like and of course, it will be driven by a PC!
I always knew that a CNC machine had at least one limit switch each axis to tell the controller when the Home position is reached but didn't know exactly how ii works until the day I asked a tecnician who came repair the machine i operate everyday. I told him about my simple project and that it won't use any scale to know the right position, So, he gave me a simple solution to stop an axis accurately.
First, i need to install a limit switch to tell the controller to slow down axis feed(after a HOME REF. command is issue). A disk with/or a cam should be installed at lead screw end that will look for a proximity switch, and stop! I didn't try it yet but it looks like a good idea!

Try it out...Hope that helps...

Bye

Sevy

[paco]

Hello Ron Brown!


Just want to let you know that I installed, some week ago, microwave door microswitches; one on both X and Y axis. I have installed 'em with double side automotive tape on side of the rails so the gantry is the machanical actuator (the rounded corner of the vertical part of the gantry)... Sure I plan a stronger installation but this is now prototype being test...
Here pics...

3686
3687
So far the accuracy suit my needs but I did'nt check it with a accurate tool like a dial... On parts; I can't see/feel a "play" when I ZERO the X and Y axis then do some JOGS/MOVES then CUTS then JH then RE-ZERO AGAIN... and so on... Just to zero thoses axis is a major improvment both at reliability and tool safety...

[paco]

What's wrong with this forum going down then up?!?!
Note that this post should be at the bottom...!!

[ron brown]

Paco,

Congratulations - soon you will be wondering about the accuracy of those switches!

BTW - I too wonder about the wandering of the proper locations for posts these daze.... it confuses anyone with less than half a brain such as myself.....

Ron

[hummmmmmm (Unregistered Guest)]

I tough alpha never lost any step???

[paco]

Ron-
Did you experienced something wrong with this type of switches? Or did you meant that they can actualy be as accurate as other type? Anyway, it's a start!


Hello hummmmmmm! It should'nt... but it wont respond me when I ask...! And I sure hope it will not!!

[hummmm (Unregistered Guest)]

I still dont get it why the proximity switch with alpha since they are close loop???????

[paco]

Hummmmmmm...
The Alpha use close motor/driver system but it is'nt yet made out of gods flesh... meaning that it's better to rely in two system (or more)... to make the tool safer/accurate after an event (power lost, collision and such...) $$$$$ tool do too use such device to "calibrate" at start up and after events...
Hope this help...

Maybe SB tech. member could join this thread to explain/detailled this topics in more "scientific" words... maybe it's already been covered...

[ron brown]

Paco,

Many mechanical switches have a large operating range. Some have a small range. A microwave just needs to know if it is open or closed. I might or might not worry about it depending on what was found when the switches were statically tested. After all, most folks don't have the tools to measure anything as accurate as a ShopBot, even the old cable models.

I wouldn't bothering answering folks who don't identify..

Ron

[paco]

Sure Ron; I'll keep checking'em reliability... Since no spacerocket engineer approach me for machining 'em parts, I'll keep appreciate this accuracy with my eyeball and finger touch!

Since I have to carefully plan my buisness expenses ("real" proximity switches cost about 8-9 time thoses microswitches), this setup is a fine entry one... I believe... Thanks anyway for the warning!
But what about you? What do you have tested and currently use?

"Hummmm" sound like a shy personnality... but I can answer some easy one...

[paco]

Here better views of my setup...

3684
3685
...I did'nt yet test this with a dial but when go back and forth touching the switches it's pretty much always within 0.002"!!! Sure this need to be done at slow approach... I have perform some testing to see if it was reliable between cuts and power failure/"collision" and I just can't see any play/differences I both cut that were made at different Z height...

[jthelen]

I designed and built my own machine a couple years back. I just used N.O. momentary contact switches.
My home sequence send all axis's back to 0,0,0 at 300 inches per minute.
After it hits all three switches I have it move away 1/2 in on all axis's then go back to the switches at 10 inches per minute before moving back to the homed position. I don't know what kind of rapid move speeds you guys are using or if this would work but it might be worth trying.

My top is a 1"x4'x8' sheet of aluminum. Into this I machined 1/4 inch deep x 3/8 wide street and avenues over the top in four sections for my vacuum base. This was a slow process and I would cut a section then come back the next day a finish cutting. I did not notice any deviation in the cuts. As you can see I am not done yet.


3683

[richards]

Yesterday I finally installed the proximity switches that came with my alpha last July. A few days ago a job come in where both sides of the material has to be cut accurately and pin registered to make everything fit properly. After searching the forum, to find an easy way to xy-zero the machine (I built an xy-zero device last year, but it is cumbersome and slow), I saw that the proximity switches were supposed to be dead-on accurate (within one or two thousandths).

First trials yesterday were extremely disappointing - absolutely no repeatability. I checked and rechecked gap distance, ground the targets flat, rerouted the wires away from the stepper cables, varied the move speed. Nothing worked reliably. Finally, after several hours, I decided that I would have to build my own device after all.

Just before stripping the proximity switches off the machine and starting over, I reread this thread and finally read Frank Leinbach's post where he wrote about just using the C3 command. It works just as Frank said it would with total repeatability. Thanks Frank for your post.

[fleinbach]

You're welcome Mike, I was just about to give you the answer again until I read the last paragraph.

I find the repeatability to be incredibly accurate using the proximity switches.