ATC module for ER20 spindles

[Daz]

Hello everyone,

I have developed an Automated Tool Changer module for ER20 spindles. It works on the impact driver principle and delivers 35Nm or more of nut-tightening torque, meeting the ER20 spec. Happy to chat and answer any questions. Let me know what you think.

Here is a video of its operating principle:
https://www.youtube.com/shorts/eBymr9Z7TvU

And here, when I used it to mill an aluminium vacuum holding plate, doing some decent cuts with an 11mm, 2 flute end mill.
https://www.youtube.com/watch?v=FH3Ue6h8MHE&t=1s

[EddyCurrent]

Daz,

I've posted a link to this on the CamBam forum, it is very interesting. https://cambamcnc.com/forum/index.php?topic=10783.msg80626#msg80626

[Daz]

Hi, thanks! Much appreciated! Let me know if you have any questions, happy to chat.

[Daz]

Thanks, much appreciated! Let me know if you have any questions, happy to chat.

[needleworks]

No idea how it works mate but it certainly looks good !!

[EddyCurrent]

Daz,

Someone on the CamBam forum asked if there were any issues around cross threading of the collet nut.

[Daz]

Hi, thanks. I was trying to register on the CamBam forum, but the registration is disabled. I have reached out to the support for help.

With regard to the question about cross-threading, I have performed about 300+ tool loading and unloading cycles during development and subsequent use and didn't have a single cross-threading problem. The cross threading can happen if the spindle doesn't have a good concentricity with the tool holder, or the tool holder gets dirty with swarf, and the nut doesn't sit straight in the holder.

The tool holder can take quite a large degree of concentricity misalignment, but if a machine has crap homing, and one of the axes is off by a few mm from the previously set position, then I expect the cross threading can happen.

[EddyCurrent]

Daz,

There have been ongoing problems with the CamBam forum but in the past month or so a new top admin has stepped in to sort things out, his handle is "kneave".
Our stalwart admin is David, handle "dh42", he has had limited access to perform top level stuff but has worked hard on the things he could do, I've alerted him to your case.

300+ is a good record for no cross threads. I have found in the past that it's mostly the nut at fault rather than the spindle nose so that's a good thing at least.
As you point out in the video, shielding the tool holder rack from swarf etc. is key.

[Daz]

My project this weekend is to upgrade the tool rack cover. I will post a video once I have it running.

[dh42]

Hello

Send me a PM with the login, email and password you want and I'll create an account for you on the CamBam forum.

I can't find how to send you a private message ...

++
David

[Daz]

Hi, I have sent you a PM with the details, thanks.

[dh42]

registration done ;)

++
David

[JohnHaine]

Daz, that's very cool! Could you show more of its workings please? Will you publish the design?

[needleworks]

I'd love to know what software you are running to achieve that ! eg, how do you get a 20k spindle to run slow enough to start the threading process of the collets ?

[Daz]

Hi JohnHaine, yes, I will share some more details. I have a nice slow-motion video of the impact mechanism in action. I will upload it to YouTube over the weekend.

Re your second question, I won't be sharing the detailed design, as I am making and selling the modules.

[Daz]

I'd love to know what software you are running to achieve that ! eg, how do you get a 20k spindle to run slow enough to start the threading process of the collets ?

I am running LinuxCNC with a Digital Dream EC300 controller that was originally for March3, but I have converted it to LinuxCNC with Remora firmware. I have 2.2Kw spindle that goes up to 24k RPM. I don't remember exactly the RPM resolution, but it is better than 50RPM around the 1k RPM range I am using it for tool changing. The inverter is Huanyang 220v 2.2kw. With this setup, I can run the spindle backwards and forwards, from I think around 300 RPM (I would have to confirm that) all the way up to 24k RPM. After moving from Mach3 to LinuxCNC, the setup is rock solid. I had lots of problems with Mach3.

[JohnHaine]

Well I could be interested except I use ER16 on my mill.

[Daz]

Well I could be interested except I use ER16 on my mill.

It is very straightforward to modify the ATC module for ER16. Just a simple change to one part, and I would be happy to do it.

The main challenge will be around the torque your spindle can generate. This is mostly a problem for the unlocking step where the spindle needs to overcome the module's initial resistance from the standstill. In case of my spindle, and with the current module setup, I had to increase the acceleration setting on the VFD to give the spindle a bit more torque, as it was stalling from time to time.

If you could let me know your spindle model, I can check its specs and let you know if it would work.

[JohnHaine]

I have a Novamill which uses a DC commutator motor, about 375W. Max speed is 5000 rpm. At the moment I only have manual speed control and no reversing so that would need to be added. The table is also quite small so at the most only room for 2 modules - what size are they?

[Daz]

The module external dimensions are: 95mm X 95mm footprint, and 88mm height.

Checking your spindle, I am afraid you most likely won't have enough torque to operate it. I might consider making a lower torque version in the future, but I will need to modify some internals.

[Daz]

I have done some more digging and since it is a DC motor, you will have completely different torque curve, with some significant torque available from the start. Theoretically many times higher to what I can get on my 2.2kw spindle at very low RPM, assuming your spindle DC power supply can handle it, it might actually work!

You would need to sort out the manual speed control and the reverse operation, but other than that, it should work.

[JohnHaine]

Thanks for the extra information Daz. If the motor could generate the torque it would need a very different controller, the current one is KBE thermistor type. Something that would generate stall current at low volts would be needed, and anyway as it's a PM motor it might be field limited. Also the height would severely restrict me and my table is only 125mm wide! Shame it's a very nice concept.

[magicniner]

That's as cool as a Hippy Penguin's Jesus Sandals!

[magicniner]

Thanks for the extra information Daz. If the motor could generate the torque it would need a very different controller, the current one is KBE thermistor type. Something that would generate stall current at low volts would be needed, and anyway as it's a PM motor it might be field limited. Also the height would severely restrict me and my table is only 125mm wide! Shame it's a very nice concept.

The whole point of an Impact Driver Mechanism is that it isn't reliant on motor torque, it releases the drive in response to pre-settable torque demand, allowing the spindle to accelerate, it then uses spindle system angular momentum against a braking mass to apply the required torque, repeatedly.

[Lee Roberts]

Very cool!

Cant wait to see your first cassette of tooling in play and in production :thumsup:

[Daz]

Very cool!

Cant wait to see your first cassette of tooling in play and in production :thumsup:

Here is my first job using the ATC modules. The cover needs some improvements to keep the collets clean.

https://youtu.be/FH3Ue6h8MHE?si=oE4Sxmk5fvD7b8i3

[magicniner]

That is Totally Ace!

[JohnHaine]

The whole point of an Impact Driver Mechanism is that it isn't reliant on motor torque, it releases the drive in response to pre-settable torque demand, allowing the spindle to accelerate, it then uses spindle system angular momentum against a braking mass to apply the required torque, repeatedly.

That's true when tightening but loosening is a different matter as the spindle will be stationary.

[Daz]

That's true when tightening but loosening is a different matter as the spindle will be stationary.

It is the same for tightening and loosening. The operating principle doesn't change. The average torque required to spin the module is low, but the impact torque resulting from the hammer hitting the anvil is high. The impact happens two times every full revolution. The impact torque depends mostly on the spindle RPM.

The main difference between tightening and loosening is that for loosening, the spindle needs to overcome the initial module resistance from the standstill. The resistance comes from the need to overcome the hammer return sprint in order to depres the hammer. Once the hammer is depressed, the anvil (top of the module) can slide over it and make the rotation.

[JohnHaine]

Aha! Right, I didn't spot that in the video.

[magicniner]

That's true when tightening but loosening is a different matter as the spindle will be stationary.

Impact driver mechanisms are bi-directional

[Daz]

If anyone would like to try out the ATC module, let me know. I have a few additional units.