Precision odds and ends (feat. Moore, piezos, linear motors, and DTL updates)

[@StefanGotteswinter]

Cool Show&Tell!

[@cylosgarage]

🙏

[@JKKnudsen]

Cool stuff, I'm definitely making one of those dialed micrometers! Btw, if you know the gauge of wires in the STM, you can just buy the same gauge tungsten on roll. Then cut oversized, put the piece as the anode in an electrolysis cell, with the tip just slightly touching the surface of the electrolyte. Where surface-tension forms a meniscus around the tip. Set the voltage to just over activation, the cell will shut off, when enough of the tip has etched away to break the meniscus. Leaving a perfect tip fairly often. Sometimes ramming the tip into the substrate can fix a wonky tip. The characteristics of the tip depend heavily on the dominant tunneling orbital of the tip atom. So repeated ramming's may be necessary to get primarily a p or d orbital. I notice that the trans-impedance amplifier is still on top. Leave it as is, and try to make your data collection circuit around it's operating voltage. I also recommend running the amplifier on batteries, as even most linear regulators are too noisy. The gain is usually in the 1M-1G Ohm range, so any form of noise is amplified a lot. As for the piezo drivers, you can get pretty far with some low noise power op-amps (ex OPA455)and a microcontroller with at least 12-bit DAC's. But you may end up with some "large" capacitors to get a stable voltage output, resulting in longer scan times, and increase the likelihood of failed scans. And since microcontrollers don't like voltages in the double digits, some isolation of the high-ish(1-200V?) and low(3-5V) side, is probably wise. That said, if there is one part of this, that I might consider buying my way out of, piezo driver is it. Especially if I already had the trans-impedance amplifier. Or copy an existing circuit, but the open-source ones tend to use segmented bi-metallic disc piezos, operating on much lower voltages than typical tube-piezos. Regardless, hope this was useful, and not just stuff you already knew. Looking forward to future updates!

[@cylosgarage]

@@JKKnudsen thank you for the valuable advice

[@super8hell]

You might need something bigger than a dining table if there's more of those friendly strangers on the internet😂

[@Koptokaf]

Oh, wow. That scraper is like Excalibur! Or like the Kusanagi-no-Tsurugi of metalworking! The sacred blade. The true Edge of Precision. I'm so envious, i can taste it.

[@cylosgarage]

And that’s not even the cool one! The guy who sent it to me has one thats the same except it also has WRM stamped into it. As in, Wayne R Moore. 😳😳😳

[@Koptokaf]

@@cylosgarage *Head exploding noise*

[@stevenhavener7327]

I am so jealous I cant stand it !!! love the scraper !! best regards Steve

[@DiffractionLimited]

Really nice gear, thanks for sharing! I was quite impressed with the diameter of that optical flat!

[@kageyama8331]

Awesome! Very kind of you in bringing us some mn meatloaf! 👍

[@AABB-px8lc]

23:30 omg I just reach that chapter in Moore book "Holes contours and surfaces" that suggest such blocks for proper clamping in Jig Borer. Perfect timing, book copy have shit illustration but now i see it in detail. Miracle.

[@carltauber2939]

Great to see you back on the Tubes Cyrus! My take is that you'll have Spencer green with envy over that 14" toolmakers flat, Robin the same with the scraper, and you'll be solving the encoder noise in your day job! Best wishes.

[@cylosgarage]

Thanks Carl!

[@DUIofPhysics]

Awesome stuff! I love the scraper, very envious!

[@cylosgarage]

Thanks!

[@JerryBiehler]

Ive use a couple of those Parker linear stages to build a tiny X-Y table for laser cutting some prototype medical sensors with a UV laser. I had the stock compumotor drives for them which work really well. Ive got a much larger aerotech linear motor x-y stage I need to figure something to do with, I think 300x400 mm travel. Also about 10nm res.

[@rodfrey]

Inspiring stuff, thank for doing the video!

[@Konecny_M]

Wonder how many more people could not resist those Parker linear motors listing. Would be interested in the servo driver config if you mind to share.

[@joshuahuman1]

I picked one up off of ebay a few months ago. Mine came with a parker vix250 servo drive and a renishaw encoder interpolater but i haven't been able to get mine working. It mostly comes down to the vix250 drive being obsolete and not supported anymore. When I was trying to get it working the issue I was having was that the software requires windows XP and does not work in windows 10 compatibility mode. I could get the software to open in an XP virtual machine but had issues passing through the rs232 signal from a usb to rs232 converter. I probably just need to get a old pc with a dedicated rs232 port on it but if anyone has some tips on interfacing with these old drives I'd love to hear them. I bought the slide to possibly use on a diy wire edm using a RackRobo power supply. Id love to be able to use this slide as it would easily make it the most accurate piece of equipment I own (not a high bar). The project is kinda on the back burner due to some power supply issues RackRobo are currently sorting out. Id love to be able to switch to an Ioni drive after seeing what Cylo has been able to do with his but unfortunately granite devices who make the Ioni drives decided to have a minimum order quantity of 10.

[@cylosgarage]

Which parts?

[@Konecny_M]

​@@cylosgarage Misunderstood the first time, I actually have the original Parker drivers for it (and maybe one or two spares on top), not Ioni. I've had bad experience with Garnite Ioni drivers in past on some retrofitted system with brushed DC motors, took me three days to get semi decent response that was not too sluggish or on edge of oscillation. Also it has basically zero EMI filtering on the drive outputs, it needs to be implemented externally.

[@InductorMan]

Is that encoder stick-slip, or is it clamping? At around 8:56 we can see four conductors going down into the piezo actuator area. I know some of these stages use a separate clamping actuator to modulate the normal force and "let go" of the stage when stepping. Or maybe it's stick-slip, but with two parallel channels so only one needs to slip at once? Definitely curious why there are four conductors to the actuator.

[@lambda7652]

i would love to see Cylo's take on a speed benchy

[@hisuiibmpower4]

have u tried other commercial servo amp on that Parker crossroller stage to see if they can reduce noise? Most of those servo amp software has build in feedback oscilloscope at least u can see what's going on with the feedback

[@cylosgarage]

The software for my drives has a scope too. Noise looks to be electrical

[@shaunybonny688]

Way over my head, cool nonetheless

[@leonordin3052]

Where can I get such large optical flat?

[@xtevesousa]

Adventurous to put one flat on top of the other no? Enjoyed, as usual.

[@cylosgarage]

@@xtevesousa living on the edge

[@cnxunuo]

Made a 3 channels stick-slip driver GRBL compatible, or any pulse-dir compatible, can send you the design, simple mosfet RC bidirectional saw tooth, no high current opamp needed

[@cylosgarage]

Awesome!! I’d love that, shoot me an email

[@cnxunuo]

@ kzbin.info/www/bejne/d5rLYmmJnZh6oLMsi=mzD8qUzsbQOmct1D Cant find your email so made a video instead 😂

[@nicodesmidt4034]

18:51 Brownian noise ?

[@DUIofPhysics]

I think it's more to the encoder interpolation. - I've got one of these stages. Interpolating a sine wave to 65k segments is going to be very noisy.

[@super8hell]

You mean Brownian noise?😅

[@nicodesmidt4034]

@ sure I did 👍🤣

[@nicodesmidt4034]

Corrected

[@AABB-px8lc]

Spread spectrum - like algorithm, there is some base very high frequency it can drive, and mixed with uniform pseudo-random signal to increase accuracy beyond DAC capability. In theory, you get infinity resolution but if look from very long time interval ofc. Almost all modern drivers use it. Like dithering 8 bit image to 1- or 2 bit display in video or audio trickery when go from 24 bit to 16. Very cool trick.

[@BigBoss-rh7zq]

You do interesting things. I do similar in my lab. Are you doing this for fun or do you have some project for the future ?

[@cylosgarage]

All for funsies for now

[@nickhuwar7920]

What kind of number are you trying to g to get to? Optical flats are usually used to measure flatness of small surfaces. To measure the flatness of larger surfaces the modern method is to use two high accuracy levels. One is stationary. The other measures in a Union Jack pattern. But you talking g about 1/10000th of an inch. You should probably have a granite surface plate though.

[@cylosgarage]

@@nickhuwar7920 I’m not actually measuring flatness of anything with it, though I could. I plan to use it as a reference straightedge to calibrate the axes of the dtl

[@brandonb6164]

Mtb content pls

[@cylosgarage]

@@brandonb6164 if u do turbojet content first

[@camillosteuss]

Did anything ever come that replaced the jig grinder really? Jig borers maybe, but a jig grinder is simply unbeatable - EDM can deal with glass hard stuff too, but not with sick levels of precision that are part and parcel of a jig grinder - nothing really can... Honing relies on boring, as you would not be honing if you had a jig grinder large enough - so is there anything newfangled that beat the jiggy? I suppose that that water-stream laser tech from the Swiss could be something, but that was launched in like the 80`s or so - i love the old stuff, so i don`t pay much attention to modern ``advancements``... I guess that one could train their cnc mill to essentially hardmill as flat as a surface grinder would, but a true bore, and not just one, but a full set of bores(as in not easily doable on a grinding lathe(those can be mad precise)) - not a single one of which a mill could produce or bore out by cutting and simulated movements however precise - again, speaking in proper hard toolsteel or such, not in free machining steel or brass... And ``a cnc jig grinder`` is not an answer really, that is just a jig grinder with digital actuation not a ``modern advancement/replacement``... Also, that is a sexy optical flat... And i think that you over-cranked the milimess in + direction over the 10 by one full step, i`m not absolutely sure, but it seemed to me like the needle cried out for mercy and begged the Deus Mechanicus to strike you down... My tech-priest senses are rarely off - i can feel the agony of machine spirit when abused... Best regards! Steuss

[@cylosgarage]

I was just discussing this with Dave Arneson two days ago lol. The answer to this question in many cases EDM. For cases when edm doesn’t work, the fact of the matter is top of line modern cnc machines are more than accurate enough to interpolate holes as good as you could jig grind them. Think kern, röders, yasda, etc. Cool in their own ways, but the Moore style of making a machine of pure mechanical perfection may be a lost art these days. Also, yes, I did over travel the millimess, but that’s fine. They have a mechanical disconnect so when you overtravel an anvil simply moves away from a contact point internally. It puts no stress on anything in the indicator. You can overtravel a solid .125” before you actually crash something.

[@camillosteuss]

@@cylosgarage Fucking nice! My micron clocks are russian made jeweled bits, and while accurate, i do not believe them to be safe from over-travel... I never took apart a milimess... Speaking of such stuff, look up Mahr`s Feinproof Militron readout... It requires probes, but if allows for sick readouts on a massive scale(visually massive, as in huge and easily read)... Regarding the precision of the days of yore - no compensation system will ever be able to impress me as much as a machine made with such dedication to perfection which completely side-steps the requirement for any sensor feedback to achieve what it is designed to do... But sure, i was considering hardmilling cnc interpolation of geometries as a potential solution... The EDM as we agree - makes a lot of sense due to inherent precision of a contactless material removal setup, but EDM almost never beats a good jiggy... Cnc can compensate to achieve that, but cnc can`t offer the surface finish at that fine interpolation that jiggy offers(at least not from what i know... you can comp. the deflection and all the flex, but you have the issue of minimal depth of cut...)... Sure, there are great machines and makers out there, makers that still kinda live up to the quality of old, but their designs still are defiled by the FEA... Anyways, glad to see that that question is being thought about outside my own head... Best regards! Steuss