Ultra precision air bearing stage with friction bar drive (DTL Z axis)

[@chuvvyyk]

There is so little information on youtube about this topic. Thank you for sharing your work with us!

[@ThermalWorld_]

This is insane 🤯 I always wanted a high precision stage for microchip scanning photography but with this kid of lathe you changed my mind how to think about it and how to design no backlash one.

[@elliotmlehman]

beautiful. First time I’ve seen a precision friction drive

[@hisuiibmpower4]

using Flexure to preload friction drive bar is a great ideal！heidenhain encoder can be interpolated by renishaw interpolator as far as encoder output is 1v vpp

[@ryanvanderpol7962]

Awesome work, smart design - thanks for sharing!

[@dsfs17987]

at around 7:00 mark talking about feed motor bearing precision, that would also contribute to error in Z absolute position even with infinitely divided encoder on the motor shaft, hence the choice of direct measurement with linear scale was the right thing to do

[@cylosgarage]

Correct

[@leonordin3052]

This is why ballscrew is better, it is easier to achieve a lower axial runout than radial in a bearing, I think.

[@stevendoesburg6555]

Looks great, it’s really starting to come along!

[@mitchwright1558]

Didn't think you could get the accurate with a brushed motor, impressive

[@TlD-dg6ug]

So you have sliding contact between two pieces of steel. #1-the (z-axis) bar and #2 The cam clamp thing you tighten to push the (z-axis) rod against the spindle V. Surely that won't be an issue later on.

[@Alexander_Sannikov]

does the motor controller really provide enough accuracy for precision machining needs? it looked mostly stiff when you touched it, but you didnt show its positional data when doing so.

[@cylosgarage]

Yes. Because of the nature of the control loop, the stiffness of the system is frequency dependent. However, in the diamond turning I’m doing, there are no fast movements of the axis, and no high frequency perturbations, at least none higher than experienced from an interrupted cut before the part cleans up all the way. And for these modest conditions, it performs very well. The following error for the X axis during a smooth continuous cut is less than 10nm

[@zyeborm]

Be aware of and don't underestimate the impact of the runout, bearing error etc in that motor will have on your system. I have a similar concept setup on my mill though at an order of magnitude or two looser tolerance lol. 0.001mm glass scales with spring loaded pair of ball nuts to give a zero backlash motion stage with a brushed motor direct driving the screw. The screw is slightly bent (as all screws are) about 3mm over a metre of screw. You can see the bend manifesting as periodic following error during rapid moves. The acceleration that would be required to take it out and give a smooth rapid velocity would be immense. So I have to take the hit to the tightness of the pid control at low speed to be able to keep the high speed motion. Not saying our situations are directly comparable given the scale differences and you can probably get around it by just capping your max feed rate. But it may be something to watch out for. Creep up on your speeds after tuning at low speeds. Be sure to check operation over a range of speeds, resonances spring up irritatingly. At your scale I'd be concerned about the periodic error that occurs as the balls in the bearings go past the peak load point. The differences between the balls themselves would likely be taken out without issue by the encoder system.

[@marcfaulk]

At this scale and speed of operation, would a bushing work instead of the ball bearing?

[@zyeborm]

@@marcfaulk the issue there is as soon as it starts wearing it'll get slop or backlash in it. The best solution to this I've seen is to actually build a linear motor into the slide itself. No backlash and no linkages between drive and motion component. Harder to get the force though and more stuff to make but it's a neat solution.

[@andrewphillip8432]

If the control loop can’t null out the brushed motor bearing noise then just need air bearing motor lol. Fundamental rule of precision engineering right? If the thing isn’t precise enough, it just means you didn’t add enough air bearings.

[@andrewphillip8432]

Also wow, 3mm seems like kind of a lot of bend for a 1 meter ball screw!

[@zyeborm]

@@andrewphillip8432 eh it was cheap and in an open loop controller you wouldn't have the issue. In terms of absolute position it's in the .01 mm range it's just a lot of noise added to the system.

[@YulehHabibi]

Tupperware dustshield? Vibe check: passed. ✅

[@billclark5943]

Nothing wrong with a brushed DC motor. If you want detent free, super smooth operation they are still a great option. In that application it may be worth playing with input voltage to quite the motor more

[@weatheranddarkness]

I expect the voltage will also have an impact on position stiffness.

[@nickp4793]

Wow, great project! Fellow Onshape user here.

[@cylosgarage]

Thanks, Onshape is the best!

[@lumotroph]

Incredible

[@DoRC]

Orphus compensated......naturally.

[@hereticswissery9010]

Yeah how do you couple a close loop with scale of this precision with a DC motor?

[@koharaisevo3666]

Just like any other servo control system I guess.

[@besenyeim]

Sometimes I want to give 2 likes to a video. I write this comment instead, for the algorithm.

[@cylosgarage]

Thanks man!

[@JimmyBatchelder]

Do you see any impact in position from the brushes sliding over the commutator? Like if you drove a full revolution, can you see where the brushes make and break?

[@cylosgarage]

Not thus far. Generally there is some amount of over lap in a brushed motor to prevent dead zones and jumps between commutator bars

[@weatheranddarkness]

@@cylosgarage must be a really nice motor! Who made it?

[@carltauber2939]

Please explain the meaning of the term "Orifice Compensation" in the context of the air bearing.

[@cylosgarage]

The restriction is just done via a small orifice on the bearing pad

[@janbeck8269]

Very interesting drive system.

[@janbeck8269]

Ah. And there are the IONIs :)

[@weatheranddarkness]

The friction drive is really neat. Since it's directly linked in terms of RPM and you're moving the Z such short distances, Is it difficult the drive that motor with the precision you want?

[@onurmemis3618]

Would it be accurate to say that the friction bar drive has zero backlash?

[@cylosgarage]

Well the backlash is on the order of the elasticity and hysteresis of the hardened steel itself, like down on the molecular level. I’ve not been able to measure it myself.

[@weatheranddarkness]

@@cylosgarage I guess as long as you don't drive the motor hard enough that it overcomes the metal on metal friction right?

[@jimmehdean012]

Looking really good! The scale is beautiful. I love the V-groove friction drive, that's clever. Is the air isolation table custom? Looks like a big trampoline :D

[@cylosgarage]

Thanks man. Table is old surplus.

[@Renvoxan]

Instant subscribe

[@DanielHeineck]

Gut instinct says that Seagate was using that for HDD manufacturing. Nothing wrong with a brushed DC motor!

[@agg42]

Thank the precision engineering lords you are not using laser interferometry for location/encoder

[@AABB-px8lc]

it is indirectly used in LIP 382

[@R.Daneel]

Subthumbcomment. Happy now, YT Algo?

[@cylosgarage]

@@R.Daneel thank ya sir