Companion High Speed Spindle, LinuxCNC

[@kentvandervelden]

In case it helps others, the Norbar FAQ corrects two mistakes that I've been making. Their torque wrenches: 1) should be stored long-term with a slight load; and 2) can be used to loosen if the capacity of the wench is not exceeded. (I've always stored torque wrenches with no load, and never used them to loosen.) www.norbar.com/News-Events/Blog/ArticleID/129/The-ten-things-you-should-know-about-your-torque-wrench

[@Dancer148]

Browsing your video's found this HS spindle addition for "also" dual head CNC milling. After James Clough42 his mill-spindle conversion I thought of doing the same but you convinced me. Addition is better than conversion, thanks!

[@kentvandervelden]

To increase capacity, addition may be best; if optimizing for aluminum or plastic or wood swapping may be best. Seeing how well the high speed spindle works forced me to realize how poor the bearings are in the main spindle and consider the bearings or spindle cartridge with a BT30 spindle. Have a look through the comments the suggestions from others such as mounting the secondary spindle on dove tails to raise it out of the way when not in use. Best of success to you!

[@Dancer148]

@Kent VanderVelden Indeed, better bearing with the same dimensions is a good upgrade. Your aluminum adapter plate for a dove tail construction should then be of steel, I think. Metal EndStop so that the Spindle can't fall. Good to add a steel strip inside for bolt force spreading. Also to keep it simple maybe the HS spindle casing flange could be machined to directly be the dove insert. All this with a minimum offset distance to the main spindle centre. Any thoughts?

[@kentvandervelden]

​@@Dancer148 Everyone who comments on my videos have superior mechanical skills and creativity :) Making the casing of the spindle into the dovetail insert is interesting. All of the surfaces of mounting flange that I received were rough, so leave some means to adjust for misalignment. Do believe there there's some problem with having a spindle hang off the side of the head? The forces exerted into the head during machining are probably small, but maybe not. The main spindle is centered between the column dovetails, in some way, but the secondary spindle is not. Again, not sure if it makes any difference, but it's something that was an unknown concern. I thought I would come back to this when making an extension for the mill head, to improve the usable Y travel.

[@Dancer148]

​@@kentvandervelden I already made a sketch of the dovetail plate construction that needs more info to complete. But I consider that your issues with rough surfaces while milling could be resolved. It seems like the HS spindle does have some runout that could be measured like you did but pushing with the hands some side forces to the spindle. If that play is too high the adapter plate bolt tensions should be checked. Have you tried milling the part with a much smaller endmill to lower the side forces for a smoother surface finish?

[@kentvandervelden]

I've not tried smaller than 1/4". Maybe side forces are not too much at the higher speed? Spindle doesn't bog down. The surface finish from the high-speed spindle is much better than the main spindle, but machining with the high-speed spindle seems more sensitive to speed. May be seeing resonance? I did get a surface profiler to take the measurements. Maybe a video next month?

[@gordonjones1516]

As you have a nice selection of collets you may be interested in purchasing a "Bearing nut". The ER bearing nut has a bearing incorporated into it allowing the collet to seat more uniformly and precisely. The nut is also torqued to 95-100ft/lbs and holds the tool with substantially more force. Depending on the size of the collet they range from 20-40$. It's a great way to get more precision and from your tooling. There are a couple nice vids demonstrating the gain here on the tube.

[@kentvandervelden]

Thank you Gordon. Maritool has a ER20 bearing nut. I'll need to get a different torque adaptor, and that's a special order. Should be pretty interesting to try though.

[@MrKurdishFreak]

Happy to see another video

[@kentvandervelden]

Thank you :) Yes, I'm neglectful of KZbin. We all could use more hours in the day. Merry Christmas and Happy New Year!

[@MrKurdishFreak]

@@kentvandervelden Merry Christmas and Happy New Year to you too.

[@exol511]

Great videos and commentary! I did the exact same mod on my bf20 as you did here on your machine (water cooled 2,2kw spindle) and I am not going back to the slow main spindle for aluminum ever again. Finish is superior, MMR is superior and chip evacuation is a breeze above 12kRPM.

[@kentvandervelden]

Wonderful to hear! With the HS spindle, do you prefer full depth or shallow cuts at higher speeds? Thank you

[@exol511]

@@kentvandervelden My machine/hfspindle (slapdash conversion which I never fully polished as I wanted to make stuff on it not just mod it, which I kinda regret in hindsight) prefers generally lighter/medium depth of cut with smaller endmills (8mm or less). Specifically case of that would be for Aluminum 5754 15kRpm, 2m/min feed, doc 5mm and 3mm woc. It can ofc be driven even more aggressively, but for this test it worked fine. Here is a demonstration of the cut kzbin.info/www/bejne/b3amd62gobWto9E (there are some problems in the X direction with chatter (the Z axis gib loosened a bit which I only recently fixed, as well as not having a second mist unit causing spots where coolant can not get at the cutter to cool it which is one of the main reasons as to why I am not doing 10mm doc with the same woc) Edit: I do actually much prefer milling (roughing and finishing) with the HF spindle. It handles itself much better on a small machine like this with the loads present in high speed machining. It is only for face milling and drilling that I ever go back to the old spindle.

[@kentvandervelden]

@@exol511 Thank you, this is all really great insight. I left some questions on your video about the coolant. You've given more to think about :)

[@wiredodger60]

Just found your site. Great info! I’ve wondering how a high speed spindle would perform against a lower speed /higher torque spindle. Impressive. Best

[@mc-eschx8972]

Awesome, Kent! Saving this

[@kentvandervelden]

Let me know if you do something similar. Check the comments, there are many great suggestions.

[@danielminskey]

love the wrench idea ...great channel ...great video...ohh insanity..i dint realize you were going to use them at the same time...its like having an atc..but not :)

[@kentvandervelden]

Thank you Daniel! I've been considering what to do next, and your support means a lot.

[@gordonjones1516]

I've been pondering this same addition to my mill. It might be worth your time to build a second linear guide with ball screw that allows you to retract the second spindle above the primary and possibly lower as well. Something with a smaller range 4-6" seems reasonable.

[@kentvandervelden]

Hi Gordon, that's great idea! Would get a lot of utility with the linear rails and a solid locking mechanism alone. Since the tool length must be measured each time a tool is changed, no harm is not having super precise vertical positioning.

[@DanielHeineck]

I had thought about this myself as I'll be going dual spindle (albeit with a very small NSK precision electrospindle). I'd be recommending either box or dovetail ways (either made or rescued from a cross slide on a dead lathe?) over linear rails because you only need to move them between two stops (unless you have extra long tooling that creates a specific condition) and then lock. Pinching the gib against the ways is an excellent way to make sure nothing moves (some clever hydraulic setup?). Then you're not reliant on a lead/ball screw and the attendant backlash. That said, it wouldn't allow you to swap in-flight.

[@justinwhite9535]

I just stick the tool into the collett with the nut hand snug. lightly grab the nut in the bench vise with the holder pointing the tool down and tighten the assembly by torquing a wrench on the flats of the holder body. I've honestly never found the need to come up with a better way to do it because it just works and just a common 6" vise is more than sturdy enough that it wont flex. Maybe some smooth steel jaws in the vise if the finish on the nuts is a concern but I never tighten the vise or the tool holder enough to mar the nut.

[@kentvandervelden]

Thank you Justin, that is a really slick solution!

[@JesseSchoch]

if you are doing pcbs keep an eye out on ebay for the NSK atc spindles. I got one for $200. they are really great and you can use the 24k spindle to drive them up to the 50k the NSKs can do. The ATC adapter is pretty easy to make, it is just a m5 thread and I just adapted that to a 1/4NPT threaded connector.

[@geekoid183]

Interesting, they seem to grab 1/8 tool shank directly. It's not an ISO, SK or HSK holder

[@JesseSchoch]

@@geekoid183 yes, the downside is the collets are expensive.

[@kentvandervelden]

Very interesting! Is the output on the NSK 1:1 with the input? If you have a video showing up to set one up, please put it in a comment here. There are some exotic looking electric NSK units on eBay.

[@JesseSchoch]

@@kentvandervelden this video describes the driver adapter, I can share the fusion file if you want. kzbin.info/www/bejne/pHvHiqBtlrxgn9U

[@kentvandervelden]

Thank you, that's perfect!

[@durangotang1681]

Nice

[@kentvandervelden]

Thank you :)

[@durangotang1681]

@@kentvandervelden still working on the electronics for my Pm25mv, but wishing it was as plug and play as this ender 3 pro I’m putting together as I write this. I couldn’t wait to give it to them tommorrow. 😂

[@joels4208]

Very nice video Kent.... Just curios is you are happy with the Drewtronic probe?

[@kentvandervelden]

Thank you Joel. I am happy with the Drewtronics probe, it's higher quality than I need. Simple to install in along with the tool length probe connected to the sample input. Drew has been great, and even advised away from buying something from him that was not necessary. Maybe send him an email with your application and see what he recommends.

[@joels4208]

Thanks Kent.

[@msh4805]

At 15:00 this is normal due to the difference in milling direction : climb milling vs conventional milling its obvious that one have a better surface finish so set your finishing pass to be in a single direction to eliminate the roughness. texture on the wall is due to deflection take lighter final cut at half or third the feed rate and you will get a much smoother finish. Thnks

[@emmanuelperez962]

Great tips. I will only add that with the notched finish/rougher, a small vertical offset final pass to cut the leftover lines vertically. Or simply do multiple vertical passes that are different from the notch spacing should improve your finish.

[@alexandern8hgeg5e9]

8:48 When you run very small diameter tools the runout matters more and the "flex-force" is lower.

[@kentvandervelden]

Absolutely! Thank you

[@CalvinoBear]

Depending on a few things the alternating surface finishing will come from the climb vs conventional milling. If you only climb mill of the facing pass it’ll come out much more consistent even at higher feeds.

[@kentvandervelden]

Hi Calvin, do you have a suggestion for automatically assessing and quantifying surface quality? Of course, quality is pretty subjective. I've looked at roughness testers, but they cost as much as my first car.

[@CalvinoBear]

@@kentvandervelden Automatically I am not sure. I have only used two methods. First is the small almost record player needle-like drag needle that Mitutoyo makes, but of course that is in the two kilobuck range. Then there is the old hand scratch method, with a comparative set of rougness gauges you feel and see and use to compare to what your part feels like. That is more in the hundred dollar range. I have used both the Mitutoyo device and the hand scratch device to decent effect in a production environment. With that in mind Perhaps you could come up with a way of using machine learning and a well tagged data set (machine method, ie end milling, side milling, turning, grinding, etc) with known Ra values and approximate the roughness value from an image, but I see many potential issues with that, such as the rainbow reflectiveness seen in the video. Or maybe using an actual record player needle and strain gauge could work. Whatever the case may be it is certainly not easy to do. Maybe some researchers somewhere out there have some better ideas.

[@CalvinoBear]

I should also add that some machining methods leave dominant geometry on the surface which will almost completely define the surface roughness and therefore be calculated in theory. For example if you knew the geometry of the feature actually left by a a side milling cut and the feed per tooth you could mathematically predict a roughness average. Looking at a high quality side milling cut in a microscope is a great demonstration, the roughness in theory would come between the peaks and valleys of the chips cut off the material. A real practical example of this is in the single point diamond turning industry. By calculating the peak height desired between the feed per revolution of the turning passes, you can achieve a particular surface finish and therefore a particular optical transmissibility. However for finishing operations in metals with milling tools (especially face milling) I imagine it gets much more complicated with that as there could be very complicated mechanics with tool rubbing/friction/heat generation/cold working etc at the cut.

[@kentvandervelden]

@@CalvinoBear I really enjoyed reading these. If you've not, I hope you can go through graduate school if you have the interest and find a great lab. It is that gage you mentioned that I was thinking of, and the 1k$ knockoffs, and did try my parent's 60-yo phonograph. I've been tinkering with accelerometer measurements and perhaps those combined with your suggestions would help considering the flex of my mill. For machine vision, I tried enhancing the hills-and-valleys using side illumination and polarized light, but when that works the surface is obviously rough. For manual quantifying, I've been measuring with a mic before and after lapping to a smooth appearance.

[@routercnc9517]

Nice work Kent. Could some of the surface finish be due to machine rigidity ? Interesting project anyway. When I designed my CNC machine I allowed for an interchangeable DIY spindle to fit a low speed high torque unit but that is a distant project (!)

[@kentvandervelden]

Probably nearly all of it :) And, in the main spindle there's 2 thou of runout, but only when the spindle is clocked in one place. Bad bearing or missing bearing? I debate between replacing the bearings and replacing the spindle. That's an interesting direction. Most people look for more speed, but you want lower. It's probably not practical at such high speeds, but something like a CVT or cone pulleys, optimizing pulley ratio for torque and speed could be nice.

[@westweld]

Good thing you didn't have to square that plate in the Y direction haha.....are there torque specs on all collets? I never considered it could effect tool run out that much.....cool video Kent

[@kentvandervelden]

Hi Aaron, would need to use quite a stack of shim stock :) The torque spec is on the ER nut. Needed for pullout prevention and runout. Merry Christmas!

[@westweld]

@@kentvandervelden Merry Christmas 🎅

[@kentvandervelden]

Ran across this torque chart: www.cnc-tool.com/uploads/1/0/3/3/103363/torquechart_4.pdf

[@westweld]

@@kentvandervelden awesome thanks Kent thats the stuff I...... think I typically over torque it looks like

[@bookyle21]

Could you switch to only using the high speed spindle if you were only doing aluminum and other soft materials? It's surprising that the high speed spindle won't have enough usable torque under 9000 rpm.

[@kentvandervelden]

Absolutely! If only machining aluminum, plastics, wood, etc., I would have no hesitation replacing the entire head with a high-speed spindle. The upgrade would be nearly as much of an improvement as the CNC conversion. I would only think carefully about using a spindle like this or spending more to get one with BT/ISO or an ATC. Regarding the 9000 RPM minimum, a few reasons come to mind. My best guess is that because the motor is relatively small for the power, the inductance is lower than big motors, more current will be drawn at lower speed, and the heat can't be discarded fast enough. The higher speeds would keep the current down. Might be able to get away with running a little slower for a short while. Maybe water cooled spindles would have a lower minimum? Torque is probably not effected too much, but power will be. Related to the comparison, I have no idea of the bearings used here, but many high-speed spindles are not meant for axial loading, and that's big difference between the two spindles.

[@bookyle21]

​@@kentvandervelden I seen some forum experiments with 4 pole high speed spindles and in theory could provide more low rpm torque. Maybe that could be a good compromise. And I definitely agree the loss of drilling and rigid tapping would be another feature to consider.

[@kentvandervelden]

Hmm, searching four and six pole spindle motors on Amazon returns hits. I'll remember that. Thread milling is awesome. Pilot holes could be started with interpolated helix and finished later. With a proper enclosure maybe small diameter drilling would be OK. Calculated speed for 1/8" drill in aluminum is near 9000RPM, but so sure that's so safe. The water cooled spindles like mine might go to 6000RPM.

[@durangotang1681]

Hello Kent. I’ve finally started putting my control box together. Forgive me for this question. I’m absolutely certain at this point how the wires connect to my acorn board, but....I can’t seem to find much info on what wire size, or wire type, to use for the different connections. I’m using longs closed loop motors on my pm 25 that seem to replicate the lead shine motors you were using originally. I see they have 18 gauge wire coming out of the and 26 for the encoders. So that’s settled. My problem is I don’t know what size wire to use for things like limit switches, the power connections, connections to the acorn board. If you could throw anything even remotely comprehensive my way I’d truly appreciate it. It seems to be the one thing that nobody ever covers. I’ve resorted to watching videos at .25x in an attempt to zoom in on what wire size they are using. Not working out so well. Save me.

[@kentvandervelden]

An educational channel with emphasis on cabinet wiring is @corvetteguy50. Vince has a lot of great suggestions in his videos and sells premade components. My $.02: I think UL508A is the industrial cabinet standard. Given that may be hard to find, maybe look around for Siemens and Schneider guides. If I run across one of the cabinet wiring manuals that I have I'll try to get you a link. Keep in mind, these are big industrial cabinets, and where the thinnest wire may be required to be 16AWG but in our case, we're carrying 10mA at 24V from a limit switch. Most of my stuff is 12-18AWG for power, size depending on maximum current, power lines are fused, and digital stuff is on 22AWG. In general, try to keep high-voltage, high-current, or high-noise components separated from low voltage logic. Verify the current carrying capacity of each wire. Consider the location of the wire and cable, including: is it external or internal, does it need extra protection, does it need shielding. Consider a VFD cable... it's expensive, and excessively thick considering the current that's carried... but it's also has the requirements of limiting noise radiation and surviving being banged around. Just compare 300V vs. 600V cable at the hardware store. One is way thicker and useful for abrasion resistance. Use ferrals to contain the wire strands. Get a good labeler such as a Brady. Have a good wiring diagram that's kept up to date with each wire labeled on the diagram and in the control box. Those are the suggestions that come to mind... take these recommendations more so than what I do. I experiment a lot, and carefully. I would do things differently if a permanent installation. I hope this helps.

[@durangotang1681]

@@kentvandervelden thank you so much. I know your audience appreciates your involvement. Not to mention the ultra fast replies. I’ll go ahead and order those spools now.

[@kentvandervelden]

@@durangotang1681 If you don't have a favorite source, checkout AutomationDirect for bulk wire and cable. For smaller gauge, I get spool assortments from Jameco. (I don't have any relation with either.) The big problem with cable for us, is the good stuff is typically sold in huge rolls. AutomationDirect will do cuts of 20-50' though, which is a small enough amount that it'll eventually all get used. They will also have data sheets and answer questions. And I get ferrules from them. I started with an assortment pack from McMaster but get refills from AutomationDirect for something like 1/10th the price.

[@durangotang1681]

@@kentvandervelden I just ordered a ten pack from Remington industries of the 22awg. I’ve got a lot of power cords and and switches and stuff from my excessive scrapping habit. I’ll probably go pick up some 18 gauge from the store this week.

[@joels4208]

Great video... Thanks for sharing. I am setting up a new machine shop in my garage. I notice that you are using the Tormach TTS tool holding system. I have heard that some folks have issues with end mills pulling the tool holder out of the spindle. Have you experience this kind of problem?

[@kentvandervelden]

TTS is a great convenience. There's a spindle power limit, but I have no trouble today with holder pullout with a 1HP spindle. Originally I did, but the drawbar needs to be tight with TTS. Use a torque wrench until getting a feel. Tormach lists the torque requirements. I have no trouble recommending TTS for a small CNC mill. Thank for the comment and have a great weekend!

[@joels4208]

Thanks for you reply... Here is an interesting video that happens to provide tips along this subject. kzbin.info/www/bejne/o2rVfpWJgd50rK8

[@kentvandervelden]

Thank you, Cliff is a very talented professional machinist

[@davidtaylor6124]

I have had endmills pull out of TTS collets, and one came out so far it cut through the sacrificial material under the stock and through the T slots on the table :( I wasn't in the shop at the time so didn't catch it in time. First few passes seemed fine so I left it to go have dinner. Costly dinner.

[@kentvandervelden]

@@davidtaylor6124 Through the t-slots!?! I'm lucky only to have lost a few parts and jaw plates. Did you continue using TTS and torque the down super snug or stop using TTS all together?

[@JesseSchoch]

wondering why the bottom height of the er colllet was not the same on both?

[@kentvandervelden]

Mostly laziness? :) Probably I was overthinking it, but I had two reasons. I didn't like the looks of hanging the spindle so far, less supported, probably doesn't matter. The other reason is, the tool holders in the main spindle have greater variation in length. For instance, an R8 collet, TTS end mill holder, TTS ER20 holder, TTS modular. So, I settled on having the high-speed spindle up, and out of the way.

[@spoonvalleyracing]

With higher rpm/feeds/sidestep/depth comes stability problems. You might have to consider at some point that your main body of the machine has limits.

[@kentvandervelden]

There's no truer advice than yours. Fun or frustrating, trying to correct or workaround problems is a teacher.

[@spoonvalleyracing]

@@kentvandervelden So true.......so true but sometimes frustrating as you say. My frame is a little stronger but no where near complete..

[@garylarson6386]

the high speed spindle has less torque at low rpm's because it is probally only 1/5 to 1/10 the HP of the main, I have both and my main is 5HP and my high speed is only 3KW , I always remove the tool holder from the one I am not using (clearance)

[@kentvandervelden]

Thanks Gary, I'll make a couple plugs to fill the empty spindle, to keep the chips and coolant mist out of them. Keeping tools in both spindles made for a more suspenseful show :)

[@garylarson6386]

@@kentvandervelden I learned the hard way

[@garylarson6386]

your going back and forth, for a good finish only climb mill

[@kentvandervelden]

On the facing operation? It's strange that Fusion does that by default, probably use a 2D pocket operation to produce a cool looking rectangular facing pattern. Interestingly, just slowing feed down a little improved the facing operation regardless of direction.

[@shirothehero0609]

@@kentvandervelden while we're talking finish - I randomly tried the "Speed Tiger" variety of endmills off of Amazon and OMG. Their 3 flute .5" bright endmill gives a brilliant facing finish in every direction I've tried. For the price they are epic.

[@kentvandervelden]

@@shirothehero0609 Is the end mill model IAUE? Thank you

[@shirothehero0609]

@@kentvandervelden that's it!! Just snapped a pic - photos.app.goo.gl/GGhSk6tNCPDit2g48

[@shirothehero0609]

@@kentvandervelden tried to find a recent cut example, but everything laying around is all scratched or cut up. I'll try and see if I can find or better, run a quick face op on some 6061 tonight. I'm running a Tormach 770 with a fogbuster and Qualichem TRIM Mist. Ive been blown away by the sidewall and bottom finish. Bought one then ended up getting the 5 pack.

[@stephan9976]

Happy New Year. Nice to see you again Now you also have an HS Spindle and now only one thing is missing.(Linear rails) Then everything is taken out of the little machine.Or Maybe another grinder.As you saw with me. Anyway, nice to see you again. best regards to you and your family greetings Stephan (ps. Something has happened to me. Unfortunately, I lost your email address. Can you write me again then I can tell you a bit and send you pictures of the exciting things)

[@kentvandervelden]

Hi Stephan, thank you for the well wishes. Email sent to you :) We'll stay in better touch this year.

[@garylarson6386]

I have friends with shops machining aluminum at 6,000 rpm's and wonder why they arent getting bids and making money

[@xillianto]

try to slow that finishing pass abit more, i belive i am down to around 9000 rpm and 40 IPM on the finishing pass, with 0.2mm radial cut and a spring pass, leaves a very good finish, close to mirror Even your spindle should handle that rpm, it doesnt have to run at high rpm all time ;)

[@kentvandervelden]

:) Will give it a try. Your most recent video you use 4300mm/min for roughing, 24kRPM, then slowed down for finishing?

[@xillianto]

@@kentvandervelden 12600 rpm, 10mm depth, 0.8mm radial and 4300mm/min, my spindle only goes to 18000, but mostly run it around 12 and down, sometimes 15000 the endmills can handle more than most know :)

[@xillianto]

Sorry, was in a christmas party (we have little counts of Covid-19 around me) Yes i do main cut, and then i leave 0.2mm on the walls for a finishing cut + spring cut at around 1000 mm/min (just below 40 IPM) and that has worked great. I have also used some DLC coated endmills lately as finishing tool. and that has given great surface on walls.

[@KyrychenkoAnton]

Hey, found your video, but bit late I'm afraid) I wonder have you considered "speedy head" speed multipliers, this one can multiply 6 times, so your native 5000 rpm spindle can go 30k with it - kzbin.info/www/bejne/n2mxo6WKh9eMd9k , or even cooler - there are "air turbine" spindles, that works from air and can go from 15 to 50k rpm - like this one kzbin.info/www/bejne/gpyzi2Svo8-latU . I also wonder if you think placing spindle aside from main axis of pressure - so that dovetails are pressured not in center but on a side, can somehow decrease accuracy? Thanks!

[@kentvandervelden]

The speeders are pretty neat. The spindle that I used is pretty inexpensive and solid. The speeder depends on the stock shaky mechanical power transfer. On a mill this size and rigidity, I believe that machining soft metals (maybe steel), replacing the stock head with a spindle similar to what I used, making small fast cuts, is ideal. The air spindles are neat, perfect for micro tools, but the efficiency is pretty low (compression-expansion of gas vs electrical) for regular size tools. Starting over, I would replace the head with spindle with ATC and use small carbide tools. An expensive route, but the cost is paid once, and everyday solves problems and saves time. Hope this helps.

[@KyrychenkoAnton]

@@kentvandervelden Thanks, did you experienced some sort of rigidity issues while machining with high speed spindle, since it mounted sideways and not at the center of milling table?

[@kentvandervelden]

@@KyrychenkoAnton With the stock spindle it's easy to be tricked into using big tools beyond limit of the machine. The high speed spindle leaves a far nicer finish than the stock spindle. No perceived rigidity problems, but the chip load is typically lower (and the spindle, bearings, power transfer is tighter) unless running the tool feed really fast.