Delta preferred tower end stop switches?

Hi, I have a bunch of switches already, but want to select the best, cheapest endstop solution for my micro delta design. I have read that you do not want mechanical switches with arms because of insufficient accuracy.

I have some optical switches, are these the best option? For my printer they are a large part, so if there is no accuracy difference, a microswitch without an arm may be smaller, better and cheaper.

something like this:

small mechanical limit switch without lever on Ebay

The optical switches I already have that are larger that I like are these:

Optical Endstops, Ebay

also, I found some super small and cheap magnetic switch items. I think they are too unknown for me to use, but has anyone seen these? could be something to develop for the future.....

Cheap Micro Magnetic Induction Switch, BangGood

Optical or magnetic switches are superior to mechanical switches.
The simple reason is, the carrier can pass the endstop after homing without being blocked by the switch.
That is important because you will want to use the full print radius even at full print height. Especially on small printers this can happen often.
-Olaf

It doesn't really matter what type of endstop switches you use. I use ordinary microswitches, which take up 20mm of build height. If maximising build height is very important to you, a different sort of sensor might take up less space as o_lampe suggests, depending on your carriage geometry - although you could achieve the same thing by mounting the microswitches on one of the other the sides of the towers instead of on the inside.

Edited 1 time(s). Last edit at 12/08/2015 12:37PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I use simple mechanical switches with lever. The carriage touches between the lever hinge and the switch pin (i.e. the lever magnifies the carriage movement when it is transferring it to the switch pin).
Their precision is around 0.0125 mm, probably a bit better, definitely not more than 0.025 mm. I know because I modified the firmware so that homing does touch the endstops more times to checks whether the touch position is the same. In about 70% of cases it is exactly at the same microstep and in the rest of the cases the difference is at most one microstep.
That is good enough. I do not see any need to replace them.

Quote
patrickrio

also, I found some super small and cheap magnetic switch items. I think they are too unknown for me to use, but has anyone seen these? could be something to develop for the future.....

Cheap Micro Magnetic Induction Switch, BangGood

Reed switches are not particularly accurate and repeatable. This thread shows how to build an accurate magnetic endstop/homing switch.

Rather than reed switches, I'd look into something like these:

[notanumber.net]

Even though they're discontinued, the A3144 is pretty easy to find.

I am not worried about losing build height with mechanical switches. As dc42 suggests, it is easy to change the design around so that the microswitches do not rob you of print height. Most important is repeatable accuracy. Is the consensus here that small mechanical switches with short levers are just as accurate as the leverless mechanical switches???

within one microstep seems pretty good..... +/- 1 microstep would be an error range of about .01mm on my design.(30.48mm/rev, 400x16=6400 microsteps/rev)

I used a micro switch with a lever to rig up a simple z-probe and it consistently achieves a repeatability of no more than 2 steps and most of the time it's 1 or less. That's with 1/32 microstepping.

I use 20c hall sensors. This guy mentions the part numbers.

[www.thingiverse.com]

The switches are actually very very accurate. Check out my video on them. [www.youtube.com]

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

These?

[www.ebay.com]

So the switch ends up being the cost of one of these, cost of a connector to connect to one, and a magnet?

You can use those sensors, with two caveats:

1. The minimum supply voltage is specified as 4.5V. So if using them with 3.3V electronics, you need to provide 5V power to them,

2. They behave like normally-open switches. This means that when used with electronics such as RAMPS with high value pullup resistors and no filtering on the endstop inputs, the printer electronics will be highly susceptible to triggering from capacitively-coupled electrical noise. So either add filtering, or use shielded cable to connect them to the electronics.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

For me the cost was sunk. They are more acurate than the mechanical switches I had as well.

I spent like 5bux per end stop in the past.

However the cost breakdown would be.

~1usd for the switches.
~3-4USD for a cable/shrink sleaving.
~1USD for a set of 4 neodymium magnets from the dollar store.

~6usd for a setup and still have some parts left over as spares.

Still cheaper and more acurate than the end stops I had before.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

Hi,

there is a gap, cheapest & best (i assume you mean accuracy). For end stops i would not go for cheap, they have to get repeatable accurate measurements. I preffer opto endstops as they do not wear out. Hall effects are cheap as well, but after a couple of years the magnets loose their power (not significant but they do). Mechanical ones (china stuff) i had that had a repeatability of 0,1mm ich is pretty bad IMHO.

For cartesians the Z endstop only has to be "repeatably accurate". On deltas if one of your endstops trigger earlier, then while bed leveling, you will get bad print adherison.

Just my 2 cent

Hall E i use on my delta (8€ each):
[max3dshop.org]

Edited 1 time(s). Last edit at 12/11/2015 07:24AM by Neutrino.

For me the repetability was higher than what my probe could read. 1/1000th of an inch or 0.0254mm. That was good enough for me. The magnets will "wear" however I do not forsee them wearing out before the machine does. Even if they do, a recalibration does not take more than 15-20min for me to do it manually. The only down side I had with these is that they are not linear. If you approach from a longer distance they will always triger, but they will not unhook until I exceed the approach. Meaning, if I am at the last 10mm of travel they will trigger at 9mm for arguments sake, but for them to go into off state I have to move the carriage out to 15-20mm before they register as "off" This is not necessarily a con in 99% of how I am using them, however it could be a con for someone somewhere in some rare circumstance.

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11

Quote
jaguarking11
...The only down side I had with these is that they are not linear. If you approach from a longer distance they will always triger, but they will not unhook until I exceed the approach. Meaning, if I am at the last 10mm of travel they will trigger at 9mm for arguments sake, but for them to go into off state I have to move the carriage out to 15-20mm before they register as "off" This is not necessarily a con in 99% of how I am using them, however it could be a con for someone somewhere in some rare circumstance.

Some firmwares including RepRapFirmware and Marlin home the printer once at high speed, then move all the carriages down a little and home them again at a slower speed. So all you need to do is increase the amount that the carriages move down if necessary so that the sensors are guaranteed to be not triggered. For RepRapFirmware this is easily done by editing the homedelta.g file on the SD card.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

My firmware moves once quickly. Then it backtracks and then moves slowly 5 times. The measurements are ordered. The maximum and minimum are rejected and an average is taken from the rest. Then information is printed how many microsteps all 5 measurements deviated from the average. I never had it more than 1 (ie. 0.0125mm). That also indicates that rejecting the maximum and the minimum is probably not needed.

Anyway I like one thing about this approach. I will know in advance when endstop fails to be precise enough.

Edit:
Here is a typical output (values in microsteps):

Tower 1:-1,0,0,0,0
Tower 2:-1,0,0,0,0
Tower 3:0,0,0,0,0

Edited 1 time(s). Last edit at 12/12/2015 07:33AM by hercek.

Quote
dc42
You can use those sensors, with two caveats:

1. The minimum supply voltage is specified as 4.5V. So if using them with 3.3V electronics, you need to provide 5V power to them

I know that's what the spec sheet says-- but I'm getting the A3144 to trigger on a Due/RADDS at 3.3v. YMMV, caveat emptor, contents may settle, etc., and I haven't got a holder I'm happy with, so I haven't done much more than proof-of-concept testing.

I have ordered levered mechanical, leverless mechanical and the A3144 sensors. Testing will ensue upon receipt.

Quote
dc42

Quote
jaguarking11
...The only down side I had with these is that they are not linear. If you approach from a longer distance they will always triger, but they will not unhook until I exceed the approach. Meaning, if I am at the last 10mm of travel they will trigger at 9mm for arguments sake, but for them to go into off state I have to move the carriage out to 15-20mm before they register as "off" This is not necessarily a con in 99% of how I am using them, however it could be a con for someone somewhere in some rare circumstance.

Some firmwares including RepRapFirmware and Marlin home the printer once at high speed, then move all the carriages down a little and home them again at a slower speed. So all you need to do is increase the amount that the carriages move down if necessary so that the sensors are guaranteed to be not triggered. For RepRapFirmware this is easily done by editing the homedelta.g file on the SD card.

I have this on by default on marlin. This is not what I was hinting at. What I am hinting at is when you are printing to the max height of travel, you will not get the maximum size of printer space as the sensors trigger before max is reached. In either case it is irrelevant to me as these sensors have been one of the best upgrades I have done to my 3d printer. After 1000+ hours of printing the mechanical switches which were not of great quality to begin with showed wear and were basically reaching their end of life, at least as far as I am concerned. The hall effect sensors have been much cheaper than anything else I could replace them with and have been very reliable.

Two things to consider.

Some have mentioned that these sensors are susceptible to signal noise. I have not experienced this as the signal is routed through cat5e cables dedicated to the sensors. For me this is good practice for any signal cables that work with low voltage. I also braided my motor wires to try to reduce the noise emitted.

The other thing is make sure you point the proper polarity towards the sensors. The magnets in reverse will not trigger, while I have no idea if its south or north, I simply tested them and then glued them on the carriages.

-Bruno

---

My Personal Blog. Build blog.
[engineerd3d.ddns.net]

Modicum V1 sold on e-bay user jaguarking11