#linuxcnc Logs
Nov 06 2022
#linuxcnc Calendar
02:16 AM Deejay: moin
02:43 AM travis_farmer: G'Morning
03:43 AM JT-Cave: morning
03:44 AM sensille: trying to get through the fpga sources ... there is a file etherhm2.vhd which looks like just some instruction ROM. is there an assembly source for that?
03:44 AM sensille: and a documentation for the instruction set?
03:44 AM sensille: the cpu seems to be d16w.vhd
03:47 AM jpa-: i agree on what the files are, but i'm not aware of the source code for that binary being published anywhere
03:49 AM sensille: so it's only half-hearted open source
03:50 AM sensille: i'm trying to port it to my RV901T
03:51 AM jpa-: sensille: actually, looks like the source might be here: https://github.com/LinuxCNC/hostmot2-firmware/blob/master/src/etherhm2.zip
03:52 AM sensille: nice :-)
03:54 AM jpa-: looks like assembled using TASM
03:57 AM sensille: tasm? borland turbo assembler?
03:59 AM sensille: ah, TASMD16.TAB
04:00 AM sensille: telemark assembler
04:01 AM sensille: https://github.com/kyleedwardsny/bergen
05:14 AM Tom_L: morning
05:37 AM JT-Cave: if you check and uncheck load config at startup the mesact.conf file is updated and I fixed the checkbox setting at startup
05:39 AM JT-Cave: going to test out the 6.0.0-3 kernel now
05:40 AM -!- #linuxcnc mode set to +v by ChanServ
05:48 AM JT-Cave: error finishing read
06:25 AM JT-Cave: building 5.10.153-rt76
06:53 AM -!- #linuxcnc mode set to +v by ChanServ
07:01 AM -!- #linuxcnc mode set to +v by ChanServ
07:18 AM -!- #linuxcnc mode set to +v by ChanServ
07:22 AM -!- #linuxcnc mode set to +v by ChanServ
07:46 AM -!- #linuxcnc mode set to +v by ChanServ
07:53 AM Tom_L: JT-Shop, get the float fixed writing a new config?
07:56 AM JT-Cave: I have not figured out where it comes from, building a new config does not create floats for those two spindle entries
07:57 AM JT-Cave: been putzing around building kernels this morning...
07:58 AM Tom_L: i see that
07:58 AM Tom_L: how do the backups work? should i look in one of them for a clue?
07:58 AM Tom_L: meh those are correct
08:01 AM JT-Cave: I just tried to update the config and no floats appeared so I have no clue how they got there now
08:01 AM Tom_L: i'll try another
08:04 AM Tom_L: load old config, save under a new name, floats still there
08:06 AM Tom_L: load newly saved config: loadini.py line 298 invalid literal for int() with base 10: '5500.0'
08:07 AM Tom_L: hand edit those 2 and loads fine
08:07 AM Tom_L: MAX_OUTPUT = and OUTPUT_TYPE =
08:10 AM -!- #linuxcnc mode set to +v by ChanServ
08:10 AM Tom_dev: http://paste.debian.net/1259736/
08:10 AM Tom_dev: grep result for MAX_OUTPUT
08:12 AM JT-Cave: hmm I got the float when I changed the name so must be in buildini.py
08:12 AM Tom_dev: i was looking at line 4 there but not sure what i see
08:12 AM Tom_dev: getattr(parent
08:13 AM JT-Cave: :.1f
08:13 AM Tom_L: bak in a few
08:14 AM JT-Cave: https://github.com/jethornton/mesact/blob/master/mesact/src/libmesact/buildini.py#L250
08:14 AM JT-Cave: ok
08:24 AM Tom_L: find it?
08:25 AM Tom_L: oh i see i think
08:25 AM Tom_L: need the .1f atribute added to those 2
08:28 AM Tom_L: wait.. my local copy has .1f on line 251 & 252
08:29 AM Tom_L: oh i see.. that ADDS the .0 in
08:30 AM JT-Cave: :.1f makes it a float
08:30 AM JT-Cave: .1 makes one decimal place
08:30 AM Tom_L: i edited my local copy to test
08:30 AM travis_farmer: like a printf statement i assume
08:31 AM Tom_L: probably
08:32 AM travis_farmer: (following for curiosity, but i don't know much about python...)
08:33 AM Tom_L: this is my first line of python sur
08:33 AM Tom_L: sir
08:34 AM JT-Cave: I pushed the fix
08:35 AM Tom_L: ok local copy worked too
08:35 AM Tom_L: i'll pull and build
08:36 AM JT-Cave: https://zetcode.com/python/fstring/
08:42 AM Tom_L: git copy tested good
09:04 AM Tom_L: JT-Cave, pr 48 up
09:26 AM solarwind: https://www.omc-stepperonline.com/ple-series-planetary-gearbox-gear-ratio-5-1-backlash-15-arc-min-for-nema-17-stepper-motor-ple17-g5
09:26 AM solarwind: Is the 15 arc min backlash on the input or output side?
09:27 AM Tom_L: between
09:29 AM roguish[m]: solarwind: it's the overall amount of 'slop' from input to output. only occurs when it's reversed.
09:34 AM solarwind: still having trouble understanding what that means in terms of input side vs output side
09:34 AM solarwind: if you hold the output side fixed, what would be the slop on the input side?
09:35 AM Tom_L: it would be the same set of gears inside with the same tolerances
09:35 AM JT-Shop: thanks
09:35 AM solarwind: but the play would be different on either side because of the gear reduction, wouldn't it?
09:36 AM Tom_L: it would be a collection of tolerance from gear to gear to gear
09:36 AM solarwind: so then what is the meaning of 15 arc minutes?
09:36 AM Tom_L: i didn't write the book, i just read it
09:36 AM solarwind: it's 0.25º, but where
09:37 AM Tom_L: find the guy that wrote it :)
09:37 AM solarwind: I'm just trying to understand it
09:41 AM roguish[m]: solarwind: https://www.automate.org/industry-insights/understanding-backlash-and-stiction
09:50 AM solarwind: https://apexdynamicsusa.com/blog/post/evaluating-gearbox-accuracy ok that answers the question
09:50 AM solarwind: According to that site, the input side is locked, and the output side is tested for play at a certain torque
10:05 AM QuakerOates is now known as wonderer
10:10 AM JT-Shop2: front O2 sensors changed... next up differential
10:13 AM Tom_L: fun
10:22 AM JT-Shop2: the front O2 sensors you have to either remove the intake manifold or the exhaust from the header back which I have off so not too bad to change them
10:23 AM JT-Shop2: one was tossing up a check engine light from time to time
10:24 AM roguish[m]: JT-Shop2: when you mail a mesa card out, what service do you use? and what's the approx average cost?
10:24 AM roycroft: my van is throwing codes, one of which indiates an o2 sensor issue
10:25 AM roycroft: but it does not specify front or rear, and there are two front and one rear sensors
10:25 AM roycroft: so i'll have to do some further diagnostics
10:25 AM roguish[m]: I have some stuff to sell on ebay, and am trying to figure the costs envolved.
10:25 AM roycroft: but there's also a map sensor code
10:26 AM roycroft: so the o2 sensor code may be a symptom, not cause, of the problem
10:27 AM acer: ve7it, if its the right type in a week or so I have an other one of the JUMBO sat's coming. If I get it I'll have those parts I was talking about to send to ya.
10:27 AM roycroft: ebay listing fees + seller fees + payment gateway fees + shipping cost + cost of shipping supplies + cost of time to prepare, list, package, and ship
10:27 AM acer is now known as _unreal_
10:27 AM roycroft: and potentially cost of answering questions from buyers
10:28 AM JT-Shop2: roguish[m], I offer USPS and UPS and the price depends on the service and location
10:32 AM roguish[m]: JT-Shop2: thanks, I am thinking small, prepaid USPS boxes. the stuff is all small and light weight.
10:33 AM travis_farmer: did something change with the buildbot repo for buster? i get an error that the buildbot repo is not signed, though i added the key with success "EF1B07FEE0EE663E" (setting up my test laptop the same as my production install, with buster)
10:35 AM JT-Shop2: usps flat rate is easy
10:37 AM travis_farmer: never mind... type...
10:37 AM roguish[m]: easy is what I like...
10:37 AM travis_farmer: *typo
10:42 AM _unreal_: man it kills me. I have this controller board that is amazing. SO many I/O has a 32bit main RISK-V chip an onboard FPGA secondary chip, 3 relays, 6 opto isolators. and an unknown number of other features?? would be an amazing marlin or GRBL, or hell maybe even mesa control board with proper code.
10:44 AM roycroft: priority mail flat rate boxes are really good for small items
10:44 AM roycroft: and you can purchase the postage through ebay and save a bit vs. usps pricing
10:44 AM roycroft: they will calculate the shipping cost for you
10:46 AM _unreal_: the control board is in this "WEG WFW 09" control board cc9
10:47 AM _unreal_: cc9.00 WEG
10:49 AM * JT[m] uploaded an image: (252KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/NkpVHAtWeZbmmPlvMAdGuuZw/20221106_104655.jpg >
10:56 AM sensille: jpa-: the rest of the riddle: https://www.forum.linuxcnc.org/27-driver-boards/47220-tasm-assembler-for-mesa-firmware#255208
11:45 AM ve7it: _unreal_, interesting control board... not much chance of reverse engineering that one! It sounds like it would be a real fluke to find the engineering drawings online
11:54 AM * JT[m] uploaded an image: (247KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/ZFDhFEgeAKSjNKtwqjJejXDc/20221106_115245.jpg >
11:54 AM JT[m]: Nap time
11:56 AM travis_farmer: same here, if i can catch any Zzzz
12:33 PM Tom_L: excellent choice
01:01 PM sensille: are leadshine and stepperonline somehow the same?
01:01 PM sensille: omc-stepperonline
01:03 PM sensille: or at least connected?
01:08 PM sensille: "Designed by StepperOnline®, Manufactured by Leadshine®"
01:27 PM solarwind: sensille no
01:27 PM solarwind: stepperonline rebrands them and resells them
01:27 PM solarwind: It's the exact same thing as the leadshine item, just a different brand sticker
01:28 PM solarwind: The ISV and T6 servo kits are amazing prices considering what you get
01:29 PM sensille: thinking about getting the DM556T
01:29 PM solarwind: I'm just about to order one to evaluate. The only annoying thing is the T6 drives are pusle/dir input only. I wish they had analog input at least
01:29 PM solarwind: sensille don't bother, get the closed loop
01:29 PM solarwind: the price difference isn't much. I have several of the closed loop drivers, both the CL57T and CL86T and they're _excellent_
01:29 PM sensille: for a cheapo 6040 CNC?
01:30 PM solarwind: yes
01:30 PM solarwind: exactly
01:30 PM solarwind: I upgraded my cheap 6040 machine with closed loop steppers
01:30 PM solarwind: total cost was like $400 CAD for all 3 axis
01:30 PM solarwind: Had to add spacers for the longer shaft on the new steppers but that took like an hour to make
01:31 PM solarwind: could also just machine the shaft down to shorter length but I didn't want to unnecessarily reduce shaft length
01:32 PM solarwind: But I'm going to get the ISV series integrated servos on my next order and repurpose the closed loop steppers for another project so the cheapo 6040 router is getting another upgrade next month
01:32 PM solarwind: $150 per axis is a steal for an actual BLDC servo
01:33 PM solarwind: get the closed loop steppers and power them with the highest voltage they're rated at, like 48V and you'll get a healthy amount of torque at 1,200 RPM
01:35 PM solarwind: my "Bridgeport" knee mill is getting EL7 series EtherCAT servo drives and and 750W servos, same with my 14x40 lathe
01:36 PM solarwind: And my 4 ton LeBlond Makino is getting 1.3kW servos. It has 1.1kW Fanuc DC brushed servos on it at the moment, from like 1986
01:36 PM sensille: CL57T + 3Nm stepper, 94 euro
01:37 PM solarwind: And today I just managed to test a proof of concept ethernet frame ping pong from an STM32F7 to my computer with << 20µs latency with DPDK
01:37 PM sensille: or only 2Nm?
01:37 PM solarwind: the STM32 was instant, latency was mostly on the Linux side
01:37 PM solarwind: sensille 2N•m is a lot for a 6040 machine
01:38 PM solarwind: closed loop drivers give you much higher torque at higher speeds than open loop
01:38 PM solarwind: especially if you drive them at max voltage
01:38 PM solarwind: like 2N•m is more than what you want on that class of machine because of how little rigidity it has
01:39 PM sensille: there's also a 1Nm kit
01:39 PM solarwind: no, get the 2N•m one
01:39 PM solarwind: that one has the overall best specs in terms of phase inductance and all that
01:40 PM sensille: but those are still steppers, right? how can they develop much more torque because of closed loop? because they can keep the load angle at an optimum?
01:40 PM solarwind: My last month has been entirely gathering specs and pricing for all these motors, drivers, servos etc
01:40 PM solarwind: sensille the driver has a signal processor inside
01:41 PM solarwind: if it senses a load causing the shaft to turn back a tick, it'll push more current through the appropriate phase
01:41 PM solarwind: open loop can't know how to do that
01:41 PM solarwind: and if there's no load, it barely uses any power so the motor stays cool
01:41 PM sensille: closed loop can keep the motor cooler i guess. but the amount of current it can push at speed depends on the voltage
01:41 PM solarwind: I have one right next to me being load tested with a microcontroller
01:42 PM solarwind: yes but also how it drives the phases
01:42 PM solarwind: field oriented control is a thing and it needs feedback
01:42 PM solarwind: open loop drives can't and don't do that
01:43 PM solarwind: these stepper drives are on a whole different level than old school open loop. They will not run open loop. Their entire control algorithm is based around that encoder feedback
01:44 PM solarwind: sensille https://www.youtube.com/watch?v=et3oXoA2Hqo here are some tests for an idea
01:44 PM solarwind: his testing is very crude but it gives you an idea
01:44 PM solarwind: he's also running them at significantly lower voltage
01:44 PM solarwind: and this is his next video on the ISV integrated servo: https://www.youtube.com/watch?v=clgpZJFoqIQ
01:45 PM sensille: ok, i can see it as an investment towards a full upgraed of that machine
01:45 PM solarwind: but he's being an idiot about the stiffness. He didn't tune it at all. Once you tune it, it behaves nothing like that. Obviously the default is going to be sloppy
01:45 PM solarwind: Well $400 CAD isn't much
01:46 PM solarwind: I had missed step issues all the time on that machine before
01:46 PM solarwind: I had to run it super slow. Now I can do rapids at 1200 RPM / 100mm/s no problem
01:47 PM sensille: too bad f360 doesn't allow rapid rapids in the hobbyist version
01:47 PM solarwind: there are so many ways around that
01:48 PM solarwind: just use a different CAM package
01:48 PM solarwind: or patch your G code
01:48 PM solarwind: The CL86T and NEMA36 steppers are even better
01:49 PM solarwind: the 8N•m ones I have only have 4mH phase inductance as measured with my Keysight LCR meter and the driver can be run at 100VDC
01:49 PM sensille: would be nice if patching the gcode could be automated
01:49 PM solarwind: I can run those at 3,000+ RPM
01:49 PM solarwind: sensille yeah, write a program lol
01:50 PM sensille: i mean, if f360 would annotate those moves as rapids, haven't checked. or is to g0 vs. g1?
01:50 PM solarwind: look for the pattern
01:50 PM solarwind: you can tell what rapids are based on the fact that they're rapid
01:51 PM sensille: hm. f360 sets them to the exact cutting speed
01:51 PM sensille: one more thing, there's a v3.0 and a v4.0 of the CL57T, is that important?
01:51 PM solarwind: That's why I use Inventor CAM
01:51 PM solarwind: or Mastercam
01:51 PM solarwind: sensille I don't know but email the guy, he responds within 24h
01:52 PM solarwind: I don't think there's much of a difference
01:52 PM solarwind: FYI the RATTM MOTOR on amazon is the same thing
01:52 PM sensille: looks like for the inventor cam money i can easily buy a f360 license as well
01:52 PM solarwind: I have that one too and they behave exactlyt he same
01:52 PM solarwind: LOL "buy"
01:53 PM sensille: :-/
01:53 PM solarwind: watch the two videos I linked above
01:54 PM solarwind: they'll give you a good idea what to expect. You'll get better performance actually because the guy isn't running them properly
01:54 PM sensille: v4: "Add S1 rotating switch, 5/24V selector switch, brake output."
01:55 PM solarwind: and never buy clearpath, they're a ripoff
01:57 PM solarwind: I honestly don't know why James is restricting himself to NEMA23 for his electronic leadscrew. He can just go NEMA34 and forget all his worries
01:57 PM solarwind: The cost is the same
01:58 PM solarwind: I'm not going any smaller than NEMA 34 for my lathe
02:02 PM CaptHindsight[m]: solarwind: did you ever get a hold of someone helpful there?
02:02 PM sensille: watching ... why should 3000rpm be a problem for a stepper?
02:02 PM solarwind: CaptHindsight[m] leadshine USA was the only pain in the ass
02:02 PM solarwind: the automationdirect guy is cool though]
02:03 PM solarwind: so for anything leadshine, I'm going through him
02:03 PM solarwind: sensille very high chance to miss steps at that RPM if you run open loop
02:03 PM solarwind: like almost guaranteed unless you drive at like 200V or something
02:03 PM solarwind: if you run closed loop, the drive errors out if it can't do it with the supplied voltage
02:04 PM CaptHindsight[m]: solarwind: yeah he is local, he tries to help, some of his machines are not so great
02:04 PM solarwind: CaptHindsight[m] as far as leadshine USA, I'm continuing to send reminder emails with more company contacts CCd
02:05 PM CaptHindsight[m]: you have to figure they are like HF tools that need to be rebuilt properly
02:05 PM solarwind: yeah I don't plan on getting any machinery from him. Maybe ball screws but the 20mm ones are out of stock
02:05 PM solarwind: it's just the servo drives and motors I want
02:06 PM CaptHindsight[m]: be aware of bearings as well, I am just getting rid of some now that have slight detents
02:07 PM CaptHindsight[m]: for motors, break out boards and drives they aren't too bad
02:07 PM solarwind: oh yeah for sure, I repalced most of the bearings on my 6040 router. I got it a long time ago but aliexpress bearings are so bad, you can feel the grittiness
02:07 PM CaptHindsight[m]: CaptHindsight: linear bearings were all the crunchy type
02:07 PM CaptHindsight[m]: I used them only for non automated applications like drawers, never for a machine axis
02:08 PM solarwind: they are priced very cheap
02:08 PM solarwind: but LOL drawers that's actually a good idea
02:08 PM CaptHindsight[m]: he stores his parts fine, unfortunately he gets them in like they were stored on a beach before packing
02:09 PM CaptHindsight[m]: drawers, machine models that operate by hand etc
02:10 PM sensille: solarwind: what do you do with the ALM output?
02:11 PM CaptHindsight[m]: solarwind: who did you find for closed loop steppers with integrated drives besides clearpath?
02:11 PM solarwind: sensille if it alarms, that means you pushed it so hard that it lost a step that it couldn't corect
02:11 PM solarwind: that means abort your operation
02:11 PM solarwind: CaptHindsight[m] you mean integrated servos?
02:11 PM solarwind: leadshine ISV and ISV2 series
02:12 PM solarwind: same performance, waaaaayyyyyyy cheaper
02:12 PM CaptHindsight[m]: https://www.leadshine.com/product_three/pro-4-19-26.html
02:12 PM sensille: i mean, how do you wire it. emergency stop doesn't sound like a good idea to me. bringing the tool to safety would be good
02:12 PM CaptHindsight[m]: https://www.leadshine.com/products/servo/integrated-servo-motor/isv2-rs-series.html
02:12 PM solarwind: sensille it's an open collector output
02:12 PM solarwind: an optoisolator output
02:13 PM solarwind: so stick a pull up resistor on the collector, tie the other end to GND
02:13 PM solarwind: and read the voltage at the terminal
02:13 PM sensille: sorry, i mean, which kind of operation do you trigger with it
02:13 PM sensille: or just manual
02:13 PM solarwind: no it's not an input
02:13 PM solarwind: it's an output
02:13 PM sensille: i got that
02:14 PM solarwind: Oh, well just abort operation on linuxcnc
02:14 PM CaptHindsight[m]: I am getting rid of the last of my open loop steppers and drives
02:14 PM solarwind: CaptHindsight[m] I'm repurposing them to throttle control on my forklifts, things like that
02:14 PM CaptHindsight[m]: new old stock that was sometimes used as test motors or for a model
02:14 PM solarwind: sensille if you have a mesa card, just read it from there and configure LinuxCNC. I don't remember exactly how I configured mine
02:15 PM sensille: solarwind: can you have a short look at this? https://imgur.com/nxNU4Gb.png
02:15 PM CaptHindsight[m]: unless someone wants a bargain machine I don't use them anymore
02:16 PM solarwind: I have plenty of one off uses like opening the tank drain valve on my compressor tank
02:16 PM CaptHindsight[m]: looks like a private label leadshine
02:16 PM solarwind: slowly, not like the solenoid valve that scares my cat every single time
02:16 PM solarwind: CaptHindsight[m] they are leadshine, confirmed by the guy
02:16 PM sensille: so the current setting is new, the brk output and the 5/24V output selector for PWR/ALM
02:16 PM solarwind: it's obvious if you check the manual
02:16 PM solarwind: manual is identical, just different logo in header
02:17 PM solarwind: sensille yeah I don't think there's any difference in control algorithm
02:17 PM CaptHindsight[m]: solarwind: Automation technology also private labels Leadshine as Keiling
02:17 PM solarwind: it's nothing that I personally would care about
02:17 PM solarwind: CaptHindsight[m] yup, I check the photos to see if they match up
02:17 PM solarwind: and the manual
02:17 PM solarwind: I don't know why they bother to do that
02:17 PM CaptHindsight[m]: just FYI, no judgements
02:18 PM solarwind: as long as they don't change the model number internally so that the software doesn't refuse it, I don't care
02:19 PM solarwind: because then it'll be a pain in my ass to open up IDA Pro and reverse engineer that shit
02:19 PM CaptHindsight[m]: solarwind: I'd say he does it to get a discount price from Leadshine
02:19 PM sensille: solarwind: so just to double check, this one https://www.omc-stepperonline.com/ts-series-2-0-nm-283-28oz-in-1-axis-closed-loop-stepper-cnc-kit-nema-23-motor-driver-w-2m-cable-1-cl57t-s20
02:19 PM CaptHindsight[m]: he orders in bulk and the relabelmeans that HE supports them and not Leadshine
02:20 PM solarwind: sensille yeah that one is solid
02:21 PM solarwind: 1N•m at 900 RPM, my measurements are better
02:21 PM solarwind: But I run them at higher than rated voltages, which I don't recommend
02:23 PM solarwind: sensille also the "microstep" setting on the driver is more like "electronic gear ratio" for servo drives
02:23 PM solarwind: if you pick something like 40,000, it's not actually doing 40,000 like an analog drive would
02:24 PM solarwind: it has a CPU inside that ultimately decides the actual microstep resolution to what is optimal
02:24 PM solarwind: so more is not better, but also doesn't reduce torque like it does for an open loop drive
02:25 PM solarwind: I measured and characterised in detail with my oscilloscope and logic analyzer and all that
02:25 PM Tom_L: have you looked at gecko's patterns?
02:25 PM solarwind: Tom_L I would but I haven't purchased a Gecko drive
02:25 PM solarwind: They're all open loop and expensive
02:26 PM solarwind: Other people have though, I stumbled upon some analyses on the Internet
02:28 PM solarwind: Gecko makes a big stink about midband resonance and why their drives are superior because they compensate. Great, but _all_ the closed loop drives I tested also do that
02:28 PM CaptHindsight[m]: a properly designed and tuned closed loop stepper is going to perform better than the geckos
02:29 PM solarwind: yeah
02:29 PM solarwind: 100%
02:29 PM CaptHindsight[m]: geckos could be part of a closed loop controller
02:30 PM solarwind: yup, but for the price of the imported closed loop controllers, they're thoroughly impressive
02:30 PM CaptHindsight[m]: i tend to use AC servos but many applications are fine with closed loop steppers
02:31 PM solarwind: agreed
02:31 PM solarwind: and avoiding tuning is a big part of it. They have very predictable behaviour as long as you run them within their capabilites
02:33 PM CaptHindsight[m]: https://www.ebay.com/itm/Parker-CM232AE-00083-Compumotor-Servo-Motor-3-8-Shaft-92oz-in-/185642237420
02:34 PM CaptHindsight[m]: there are lots of parker and similar servo on ebay, many are custom models but once you decode the versions many are real bargains since most people can't use a custom version
02:34 PM solarwind: CaptHindsight[m] seems to be pretty much the same as this https://www.omc-stepperonline.com/nema-23-integrated-easy-servo-motor-180w-3000rpm-0-6nm-84-98oz-in-20-50vdc-servo-motor-short-shaft-isv57t-180s
02:34 PM solarwind: same price too
02:35 PM solarwind: this one is the leadshine ISV
02:35 PM CaptHindsight[m]: custom might mean a different shaft length or encoder count
02:35 PM solarwind: but yeah that one has a nice high input voltage
02:36 PM Tom_L: CaptHindsight[m], you catch 3.4.6 mesaflash is in mainstream debian now?
02:36 PM Tom_L: or soon to be
02:36 PM CaptHindsight[m]: ok
02:37 PM CaptHindsight[m]: 👍️
02:37 PM Tom_L: requied for the 7i92T
02:37 PM Tom_L: i think that's the one
02:42 PM CaptHindsight[m]: solarwind: 7500 rpm https://www.ebay.com/itm/202736014073
02:43 PM solarwind: wait till you see what I did to the transformer coolilng fan from my old microwave
02:44 PM solarwind: 8,000 RPM with a VFD
02:44 PM solarwind: shitty little shaded pole induction motor
02:44 PM CaptHindsight[m]: solarwind: nice find for the motor comparison test videos
02:45 PM sensille: solarwind: can make a higher microstepping than the encoder resolution make any sense?
02:45 PM CaptHindsight[m]: I've wondered about some of those
02:46 PM solarwind: sensille yeah that's the question I'm working on now
02:46 PM solarwind: I'm seeing how they behave when I bypass the encoder and feed them a custom signal from the microcontroller
02:47 PM solarwind: but again, the setting is more of an electronic divider, not really "microstep resolution"
02:47 PM solarwind: the DSP inside ultimately controls how they're driven and they're not driven exactly the same way an open loop driver does
02:47 PM solarwind: they do some level of field oriented control
02:47 PM CaptHindsight[m]: sensille: they don't actually step at what they print on the label
02:47 PM CaptHindsight[m]: yeah ^^
02:48 PM sensille: yeah, but that's the resolution of the desired position i feed
02:48 PM solarwind: yes, so if you pick an absurdly high setting, it may choose to do nothing at all if you feed it a single step
02:48 PM solarwind: I'm testing that sort of thing over the next couple of days in my free time
02:49 PM CaptHindsight[m]: sensille: how much do you oversize your motors when using really small microsteps?
02:49 PM CaptHindsight[m]: I tested one at a few thousand microsteps
02:50 PM sensille: why oversize?
02:50 PM solarwind: torque at microstep is significantly lower than full step
02:50 PM solarwind: torque is highest when the teeth align due to maximum reluctance
02:50 PM sensille: really? at the same load angle?
02:50 PM CaptHindsight[m]: was pretty dead on, it was a 5mm lead screw with pulley that theoretically gave 1um Z resolution
02:51 PM solarwind: sensille yes
02:51 PM solarwind: anything less than perfect step alignment gives less than full torque
02:51 PM solarwind: usually
02:51 PM sensille: i can only see the detention torque as a difference
02:51 PM solarwind: but again, the closed loop drivers do some special control that significantly improves that
02:52 PM roguish[m]: it's usually a good idea to have some reduction from motor motion to linear motion.......ultra small linear motion is tricky
02:52 PM CaptHindsight[m]: so for most torque and high res use a real servo
02:52 PM solarwind: yeah
02:52 PM sensille: the torque from the coils at a 90deg load angle is independent of the microstep position afaik
02:52 PM solarwind: On a 6040 router for something like PCBs, it's a non-issue
02:52 PM CaptHindsight[m]: what linear servos are for
02:53 PM CaptHindsight[m]: 10 G's accel
02:53 PM solarwind: sensille nope, they're reluctance motors
02:53 PM solarwind: well "hybrid", but for all intents and purposes, reluctance
02:54 PM CaptHindsight[m]: and <5um accuracy when temp controlled pretty easily
02:57 PM CaptHindsight[m]: https://www.ebay.com/itm/194227672692
02:58 PM CaptHindsight[m]: https://www.ebay.com/itm/401541401656
02:58 PM sensille: solarwind: okay, so you thoroughly talked my into it. just ordered 3 of them.
02:58 PM sensille: this is going to be fun.
02:59 PM CaptHindsight[m]: Travel: 29.75"
02:59 PM CaptHindsight[m]: sensille: leadshines?
02:59 PM sensille: this kit https://www.omc-stepperonline.com/ts-series-2-0-nm-283-28oz-in-1-axis-closed-loop-stepper-cnc-kit-nema-23-motor-driver-w-2m-cable-1-cl57t-s20
02:59 PM sensille: leashine/omc
03:00 PM solarwind: Nice. I have no complaints about the performance of mine, it's perfect for the 6040 class machine
03:03 PM sensille: roughly calculated 20W per motor, so my 240W 48V LED brick should be fine
03:03 PM solarwind: yeah, I haven't seen it draw more than 2A per axis in practice
03:03 PM solarwind: usually around 1A
03:04 PM sensille: 2A sustained would be a lot
03:04 PM solarwind: not sustained, peak
03:04 PM solarwind: you'll be fine with the 240W
03:05 PM solarwind: at no load and while holding position, it does not get warm at all
03:05 PM sensille: that is already a big advantage
03:05 PM sensille: running cooler
03:06 PM roguish[m]: sensille: try one of these. way better than a switcher.... https://www.antekinc.com/ps-15n50-1500w-50v-power-supply/
03:07 PM solarwind: roguish[m] so just a transformer and full bridge rectifier?
03:07 PM solarwind: can you say RIPPLE?
03:08 PM solarwind: wait what, it has linear regulators?
03:08 PM sensille: fuuull bridge rectifier!
03:08 PM roguish[m]: they have caps in e'm
03:08 PM solarwind: yeah, at 30A, those caps won't do shit
03:09 PM sensille: switching power supplies are a great invention
03:10 PM solarwind: LOL 41V p-p ripple at 50A
03:11 PM roguish[m]: what ever floats your boat........
03:11 PM solarwind: don't get me wrong, that's a nice transformer
03:11 PM sensille: i want that for a lab power supply
03:13 PM solarwind: I wonder what 41V ripple will do to those cheap high ESR wet electrolytic caps
03:13 PM solarwind: how long before they blow up
03:13 PM CaptHindsight[m]: automation tech ships their torroid supplies out of the east cost
03:13 PM CaptHindsight[m]: i once ordered one and then called a few minutes later to cancel and it had already shipped
03:14 PM CaptHindsight[m]: https://www.automationtechnologiesinc.com/products-page/torroidal-power-supplies
03:18 PM sensille: for my understanding ... when i run linuxcnc with a mesa card, at least for my case the only feedback from the machine are endstops and probing. i assume those are handled in the fpga (stopping the axis). why does it still have realtime requirements? can't it queue motion commands for seconds in advance?
03:21 PM solarwind: sensille it doesn't do that
03:21 PM CaptHindsight[m]: sensille: it doesn't buffer commands like repraps and mach
03:21 PM solarwind: it's not a G code interpreter
03:22 PM solarwind: the MESA card is not much more than a step generator
03:22 PM CaptHindsight[m]: it refactors position commands every cycle
03:23 PM solarwind: "anything I/O card" is a very accurate description
03:23 PM solarwind: it really is an I/O interface
03:23 PM solarwind: and not much more
03:23 PM sensille: but wouldn't it be possible to do that? or am i missing some kind of feedback?
03:23 PM solarwind: it would be possible
03:23 PM solarwind: but that's the point of LinuxCNC
03:23 PM solarwind: all that logic happens in LinuxCNC
03:23 PM CaptHindsight[m]: people coming from reprap and mach always wonder why this
03:24 PM solarwind: or then you just end up building GRBL
03:24 PM JT-Shop2: less and less parts on the floor...
03:24 PM solarwind: CaptHindsight[m] why mach? Mach does the same thing as LinuxCNC
03:24 PM CaptHindsight[m]: and then they learn about closed loop and things like rigid tapping
03:25 PM CaptHindsight[m]: did mach change? They used to buffer
03:25 PM solarwind: all the I/O cards that people use with Mach do the same thing as MESA cards
03:25 PM CaptHindsight[m]: over USB
03:25 PM solarwind: like the ethercat cards just give you a plugin for mach
03:25 PM solarwind: and mach tells them to run in position mode
03:26 PM CaptHindsight[m]: is this Mach4?
03:26 PM solarwind: yes
03:26 PM solarwind: but also mach 3 right? there's very little buffering
03:26 PM CaptHindsight[m]: oh so they changed
03:27 PM solarwind: those USB interface cards provide a mach3 plugin so mach3 can provide position instructions
03:27 PM CaptHindsight[m]: like a liittle pregnant?
03:27 PM CaptHindsight[m]: mach3 was LPT and USB
03:27 PM solarwind: yeah
03:27 PM solarwind: well its LPT was a whole different thing
03:28 PM solarwind: but anything other than that, as far as I know, it didn't need to do any insane buffering on the USB I/O card
03:28 PM solarwind: I don't know, I might be wrong
03:28 PM CaptHindsight[m]: only used it on one machine, it came with mach3
03:29 PM CaptHindsight[m]: why i have hits on the mill bed
03:29 PM Loetmichel: i am really interested how someone does passive IO-cards (no onboard tracking/servo) over usb... considering that the usb standard allows for up to 1 second packet delay
03:29 PM CaptHindsight[m]: would lose steps
03:29 PM sensille: ... and you can't go below 1ms
03:30 PM CaptHindsight[m]: guesstimation
03:30 PM solarwind: no they're definitely not passive
03:30 PM CaptHindsight[m]: like open loop it's fine as long as everything goes as planned
03:30 PM Loetmichel: or are they more or less a full "grbl" implementation on that usb card and the PC is demoted to supplying Gcode over a serial connection?
03:30 PM solarwind: Loetmichel definitely not that
03:31 PM solarwind: no way in hell
03:31 PM solarwind: they must be buffering a little bit
03:31 PM CaptHindsight[m]: otherwise it's maybe 100mS until the PID gets feedback over USB
03:31 PM solarwind: and they are indeed open loop step/dir outputs
03:32 PM Loetmichel: i know that some of the 6040 machines come with an "USB cable" that is basically an 8 bit cpu in the plug that does the GRBL/marlin thing and the PC is just the GUI
03:32 PM CaptHindsight[m]: there is a robot co in Japan that used LCNC over USB
03:32 PM solarwind: Loetmichel I have one of those
03:32 PM solarwind: I can't imagine there's a G code interpreter on there
03:32 PM Loetmichel: there is
03:32 PM CaptHindsight[m]: machinekit looked at their model
03:32 PM Loetmichel: google for GRBL
03:32 PM solarwind: I know about GRBL
03:32 PM CaptHindsight[m]: I'll look for the old link to their source
03:33 PM Loetmichel: the adapter itself is actually announcing itself as such over the serial interface when plugged in
03:33 PM Loetmichel: at least mine does
03:33 PM solarwind: damn
03:33 PM solarwind: well then
03:33 PM solarwind: Time to hijack the link and see what's going on if I can find where I put it
03:34 PM sensille: and while we're at stupid noob questions ... has anyone tried to use an old rpi as a mesa replacement, receiving commands via ethernet and generating steps, running bare metal?
03:35 PM solarwind: don't, it has to be insanely tight timing
03:35 PM solarwind: You _can_, just like how you can generate steps out your parallel port
03:35 PM solarwind: but don't
03:35 PM sensille: bare metal, no OS to interfere
03:36 PM Loetmichel: sensille: wouldnt work because network needs interrupts to work properly
03:36 PM solarwind: I'm sure it can be done with something like FreeRTOS
03:36 PM sensille: you should be able to have insanely tight timing
03:36 PM solarwind: but an RPi isn't a microcontroller
03:36 PM Loetmichel: thats why the 8 bit grbl units do with an usb->serial chip to get that timing loard off the small cpu
03:36 PM solarwind: it has very limited facilities for that kind of low level control
03:36 PM sensille: you don't _need_ interrupts, you can poll the network, too. but you also have more than one core
03:37 PM solarwind: I don't think it's worth it. Way better to use an STM32 if you really wanted to go that route
03:37 PM Loetmichel: yes, AND unimpeded access to the bus the i/os are on
03:37 PM solarwind: which a microcontroller provides
03:37 PM Loetmichel: there IS a version of LCNC for raspi4 though
03:38 PM sensille: not sure why an rpi isn't a glorified microcontroller
03:38 PM CaptHindsight[m]: solarwind: IF you can find available stm32's :) remora
03:38 PM solarwind: sensille very different purpose
03:38 PM solarwind: microcontrollers have lots of "macrocells" for dedicated functionality
03:38 PM CaptHindsight[m]: https://github.com/scottalford75/Remora
03:38 PM Loetmichel: sensille: because a rp4 is basically a multitasking computer like any phone or tablet
03:38 PM sensille: the rpi has, too
03:38 PM solarwind: an RPi is a general purpose CPU with very few facilities for that
03:38 PM Loetmichel: its MADE for an OS, not for bare metal access
03:38 PM solarwind: not nearly to the same extent that any mcu does
03:39 PM solarwind: it has a full MMU, and a whole bunch of other things
03:39 PM sensille: you don't need to use the MMU
03:39 PM CaptHindsight[m]: well you can't buy rpi4's now anyway but IF you can turn of the L2 cache on one of the cores you could use it like a micro for RT threads
03:39 PM solarwind: I know, it's a bit hard to explain
03:39 PM sensille: but RAM access is slower
03:39 PM solarwind: but it's very different in the way it works
03:39 PM sensille: rpi4 would be overkill
03:39 PM solarwind: STM32 has things like tightly coupled SRAM with zero wait states for memory
03:39 PM sensille: more like the opi zero
03:39 PM solarwind: and so many other random things
03:40 PM CaptHindsight[m]: skip the latency of the L2 cache
03:40 PM solarwind: a general purpose CPU doesn't
03:40 PM Loetmichel: sensille: ARM CPUs and the RP archictecture are meant for "doing things in quasi-parallel", not for hard timing critical access. as is a PC
03:40 PM Loetmichel: thats why you need that biiiiiig overhead in cpu power to get LCNC to work with LPT
03:40 PM sensille: _with_ an OS
03:40 PM solarwind: and there are cycle by cycle timing guarantees that the RPi chip doesn't
03:40 PM Loetmichel: exactly.
03:40 PM CaptHindsight[m]: AMD made some APU's with GPIO, they could toggle IO's to about 2MHz
03:41 PM solarwind: damn, didn't know that
03:41 PM sensille: well, i haven't read the rpi datasheet closely, but it has enough peripherals, even with DMA support
03:41 PM CaptHindsight[m]: only had 6-8 GPIO :(
03:41 PM solarwind: sensille like I said, MCUs have lots of properties like zero wait for RAM, etc.
03:41 PM sensille: but i'm not sure it hasn't got something like a bios to interfere
03:41 PM solarwind: and cycle by cycle timing guarantees
03:42 PM solarwind: a general purpose CPU like a desktop CPU and RPi don't have those timing guarantees
03:42 PM CaptHindsight[m]: Seb did some RT testing with the RP4 and got it down to ~5uS jitter
03:42 PM Loetmichel: CaptainHindsight: i have a bunch of industrial PC boards at the company that have 8 to 20 GPIO with their own 8bit MCU attached to it
03:42 PM solarwind: the CPU has to wait for memory access
03:42 PM solarwind: and the bus
03:42 PM sensille: i don't think that's needed in a 600mhz cpu
03:42 PM Loetmichel: that would probably work
03:42 PM solarwind: and other things
03:42 PM CaptHindsight[m]: I forget which timer he used
03:42 PM solarwind: and microcontrollers have lots of dedicated timers with interrupts
03:42 PM solarwind: RPi does not
03:42 PM Loetmichel: funny thing is that said GPIOs are accessed via SMbus ;)
03:43 PM solarwind: microcontrollers have lots of registers for clock/pll manipulation
03:43 PM solarwind: general purpose CPU do not
03:43 PM sensille: https://www.raspberrypi.org/app/uploads/2012/02/BCM2835-ARM-Peripherals.pdf
03:43 PM solarwind: all that is very important
03:43 PM sensille: timers
03:43 PM Loetmichel: what solarwind says
03:44 PM Loetmichel: its not about speed per se, its about repeatability timing wise
03:44 PM solarwind: I've been working on this to get torque control over EtherCAT on an STM32F7 this past month. I've been reading very detailed datasheets
03:44 PM solarwind: it's so much to explain
03:44 PM sensille: i'm not convinced :_)
03:45 PM sensille: the only real argument is RAM latency. but 70ns isn't bad
03:45 PM solarwind: even things like DMA access, memory sharing/buffering/caching options
03:45 PM solarwind: it's very highly controlled on an MCU
03:45 PM solarwind: a general purpose CPU doesn't give you that because you don't need it for what it's designed to do
03:45 PM sensille: you can do all that if you run bare metal
03:45 PM solarwind: no
03:45 PM solarwind: you most certainly cannot
03:45 PM Loetmichel: i have successfully run CNC machines with a 100Mhz 486dx4 with a dos program and LPT. it CAN be done without any timers, but you need to be "uninterupted"
03:46 PM sensille: it is not a general purpose cpu
03:46 PM sensille: you have timers
03:46 PM sensille: look at the datasheet
03:46 PM solarwind: I know, like I said, I've been doing this in depth for like the last 10 years, it's a lot to explain
03:46 PM solarwind: there are _so many_ details that affect timing
03:47 PM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=4591&g2_imageViewsIndex=1
03:47 PM solarwind: and timing is critical for real time signal control
03:47 PM solarwind: that's why microcontrollers exist
03:47 PM sensille: well, it can even be done on a PC running linux
03:47 PM solarwind: at 1kHz, sure
03:47 PM Loetmichel: maan those were the times. That gray box on the styrofoam down there is the 90s version of a "Portable PC"... 100MHz
03:47 PM sensille: shouldn't be hard to be _much_ better on the rpi without OS
03:47 PM solarwind: yeah at 1kHz servo update rate, no problem
03:48 PM CaptHindsight[m]: current Ryzen are under 1uS jitter IF you don't use a Nic
03:48 PM sensille: 1kHz? more like 40kHz?
03:48 PM solarwind: but there's a very real reason all industrial motion controllers offload their real time control to dedicated hardware
03:48 PM sensille: i know
03:49 PM solarwind: If you get 40kHz deterministic control, I will eat a bowl of shit
03:49 PM CaptHindsight[m]: I don't think I have seen any tests under 5uS or so when a Nic is used with x86
03:49 PM sensille: and an stm32 is much cheaper than an rpi
03:49 PM Loetmichel: solarwind: bon apetit ;)
03:49 PM solarwind: modern CPUs are insanely complicated too
03:49 PM solarwind: then there's DRAM latency and all that
03:50 PM sensille: ... which is deterministic
03:50 PM solarwind: not exactly
03:50 PM solarwind: not the way you think
03:50 PM sensille: quite exactly
03:50 PM sensille: at least according to the sdram datasheets
03:50 PM solarwind: the memory controller makes it a non issue for most use cases
03:51 PM solarwind: but for precision timing, FPGAs and µC are used
03:51 PM solarwind: you're talking about positioning and synchronizing things on the micrometer level
03:51 PM solarwind: yes you can do it to some extent like you said, it's been done with the parallel port since forever
03:51 PM sensille: micrometer? last i looked my 6040 has trouble with 1/100 mm
03:52 PM sensille: i'm not talking about the LHC here
03:52 PM Loetmichel: solarwind: its relative. the Two 6040 i have at the company have 200 steps/mm and can run at 40mm/sec... and no step loss. BTW: the PCs are core2duo 3ghz
03:52 PM sensille: just about lcnc
03:52 PM solarwind: my lathe can do 1µm tolerance
03:52 PM solarwind: I've done it multiple times by hand
03:53 PM Loetmichel: sensille: 1/100mm is achievable with a 6040, but is very slow going.
03:53 PM Loetmichel: uM? no chance
03:53 PM solarwind: Loetmichel yeah for sure, I'm just saying in general that CPUs designed for throughput and not latency are good for exactly that
03:53 PM solarwind: µC are designed with other tradeoffs
03:53 PM solarwind: Loetmichel 100%
03:53 PM Loetmichel: indeed
03:53 PM solarwind: I've verified
03:54 PM sensille: i'm still not convinced. i don't argue that fpgas are better than general purpose cpu. just that a bare metal rpi can be much better than a PC running rt-linux
03:54 PM solarwind: my Mitutoyo micrometer along 15cm long shaft, 20mm diameter
03:54 PM * Loetmichel has to check his two machines soon for wear and tear though
03:54 PM solarwind: vertical shear tool
03:54 PM solarwind: I could not find a spot more than 1µm out
03:54 PM Loetmichel: have made 10 PC doors last week. each with 2100 hexagonal 4mm air holes.
03:54 PM Loetmichel: that was quite a bit of "moving in place"
03:54 PM solarwind: sensille on that I agree, of course
03:55 PM solarwind: so can any computer
03:55 PM solarwind: I would say run FreeRTOS even
03:56 PM solarwind: that gives you the same level of control. You never even have to yield to the scheduler
03:56 PM Loetmichel: solarwind: "can not find a spot more than 1um out" and "doing that repeatably and on pourpose every time" is a difference though
03:56 PM solarwind: Loetmichel yes you're right, I'm just saying that I _have_ done it a few times
03:58 PM Loetmichel: yeah, and then there are the guys that stand on a lathe thats SHRIEKING with chatter and asked why they dont do anything they say "Isnt that normal?"
03:58 PM CaptHindsight[m]: sensille: help us find a rpi4 replacement
03:58 PM CaptHindsight[m]: https://forum.linuxcnc.org/18-computer/46077-potential-all-in-one-single-board-computers
03:59 PM Loetmichel: <- "well, Lothar, i am an electrician with 2 weeks of lathe training but i know about half a dozen ways to stop that chatter. Arent you the one who did a 3.5 year lathe apprenticeship?"
04:00 PM Loetmichel: sometimes i am SO disappointed in the motivation of friends and co-workers.
04:02 PM CaptHindsight[m]: there are some new boards with either an ARM or RISC-V SOC to run Linux and an stm32 on board for RT
04:02 PM solarwind: CaptHindsight[m] there's also the Artix(?) Kintex(?) FPGAS with an ARM core onboard
04:02 PM CaptHindsight[m]: https://biqu.equipment/products/manta-m4p-m8p
04:02 PM sensille: zynq
04:03 PM solarwind: yeah that's the one
04:03 PM CaptHindsight[m]: solarwind: but expensive or no GPU for GUI or latency is actually sorta slow with the arm core inside
04:04 PM sensille: why is realtime so hard on a PC even when you take a core completely away from linux? CaptHindsight[m], you said something about BIOS?
04:05 PM solarwind: sensille "realtime" to what extent?
04:05 PM solarwind: cycle by cycle timing? yeah
04:05 PM sensille: linuxcnc-grade
04:05 PM solarwind: 1kHz servo update rate? not a problem
04:05 PM solarwind: linuxcnc grade is a non issue
04:05 PM sensille: 10000ns base thread
04:06 PM solarwind: it does it already just fine with rt linux
04:06 PM sensille: or 5000
04:06 PM solarwind: just because you have 10 CPU instructions, doesn't mean they will all execute in the period of 10 clock cycles
04:06 PM solarwind: OS or not
04:07 PM CaptHindsight[m]: sensille: PC's have very shitty firmware
04:07 PM sensille: solarwind: i'm _still_ not talking cycle-by-cycle timing
04:07 PM solarwind: 10µs is insanely fast
04:07 PM solarwind: like sending an ethernet frame alone takes that long lol
04:08 PM solarwind: I know becuase I had to calculate it for EtherCAT stuff I'm working on
04:08 PM sensille: but toggling a line on lpt does not
04:08 PM CaptHindsight[m]: RDMA is just under 2uS latency in a PC
04:08 PM solarwind: no, but parallel port chip doesn't go that fast
04:09 PM solarwind: it's connected through the chipset and 10 other layers
04:09 PM solarwind: PCIe is about as close as you can get, and that's what's used for real time control
04:09 PM solarwind: everything else has orders more latency along the stack
04:11 PM sensille: well, toggling a GPIO on rpi has nothing in between ;)
04:11 PM solarwind: yeah, I'm curious now
04:11 PM solarwind: try it, write some inline assembly in a loop
04:11 PM solarwind: measure it on a scope and let us know how fast you got it
04:12 PM solarwind: I wonder how fast its peripheral bus is clocked
04:12 PM sensille: not how fast, but how low the jitter can go
04:12 PM solarwind: it's not going to be anywhere close to core clock speed
04:12 PM sensille: we don't need insane toggle speeds, but good timing
04:13 PM solarwind: Well try it and see I guess. People on forums are saying they got 80MHz toggle speed
04:13 PM solarwind: Which is pretty impressive
04:14 PM solarwind: I think that was with DMA though
04:15 PM solarwind: and DMA necessarily has some extra latency
04:16 PM CaptHindsight[m]: solarwind: have a link to the forums and source?
04:16 PM solarwind: https://forums.raspberrypi.com/viewtopic.php?t=283800#:~:text=Re%3A%20Raspberry%20PI%204%20GPIO%20lines%20maximum%20speed,-Thu%20Aug%2027&text=But%20anyway%2C%20we%20are%20in,MHz%20is%20the%20upper%20limit.
04:16 PM CaptHindsight[m]: or is it well hidden?
04:17 PM solarwind: that link format looks to be damn well hidden
04:17 PM CaptHindsight[m]: on a Pi Zero using SMI is 80 megasamples per second;
04:17 PM CaptHindsight[m]: or sorry i though it was GPIO on an RPI4
04:18 PM solarwind: https://raspberrypi.stackexchange.com/questions/87846/how-fast-can-gpio-pins-toggle
04:18 PM solarwind: "Pi 3 up to 65.8MHz"
04:19 PM CaptHindsight[m]: The speed limit for the Pi 4 is determined by the tendency of the SMI interface to go completely crazy (driving the data bus during read cycles) at speeds above 45 Mbyte/s - this behaviour is so problematic that you'd need to choose a rate significantly lower in order to stay out of trouble, maybe 40 Mbyte/s maximum.
04:19 PM solarwind: and something about dropping samples
04:19 PM CaptHindsight[m]: just curious but we can't buy RPI4's anyway
04:19 PM CaptHindsight[m]: now try this on an RK3588 or similar available board
04:19 PM xxcodery: yeah I had hit bus hammering too many times lately
04:20 PM solarwind: I wonder how much of this is documented. Probably next to none
04:20 PM CaptHindsight[m]: https://pine64.com/product/quartz64-model-b-4gb-single-board-computer/
04:20 PM xxcodery: yes im still on rpi4 lol
04:20 PM CaptHindsight[m]: why all this energy on a not available CPU board?
04:21 PM solarwind: It'll be back soon
04:21 PM solarwind: I got one a while ago
04:23 PM solarwind: I'm loving the availability of the Pi Pico W though
04:25 PM solarwind: I would LOVE to see them make an upgraded Pico with dual Cortex M7
04:25 PM solarwind: give us hardware double precision, increase the instruction space for the programmable I/O
04:35 PM CaptHindsight[m]: rockchip boards will give us u-boot
04:35 PM CaptHindsight[m]: so we can control the cpu clock throttling, unlike the rpi4's
04:36 PM CaptHindsight[m]: https://de.aliexpress.com/item/1005004266637486.html
04:37 PM CaptHindsight[m]: I'm surprised that you guys aren't all over these boards
04:37 PM CaptHindsight[m]: vs tinkering with the rpi4's
04:38 PM CaptHindsight[m]: https://www.aliexpress.com/item/1005004359533685.html
04:39 PM CaptHindsight[m]: https://jlcpcb.com/partdetail/Stmicroelectronics-STM32F407VET6/C28730 ~11K in stock (09/22) , now 1
04:40 PM CaptHindsight[m]: https://jlcpcb.com/parts/2nd/Single_Chip_Microcomputer_Microcontroller/Microcontroller_Units_(MCUs_MPUs_SOCs)_79443
04:41 PM CaptHindsight[m]: https://linuxgizmos.com/mangopi-allwinner-based-sbc-available-for-27us/
04:43 PM solarwind: das boot?
04:45 PM CloudEvil: CaptHindsight[m]: CPU chip clock throttling only matters to a limited extent, and doesn't actually give decent power saving without more things.
04:45 PM CloudEvil: For example, I have a phone which wants about 5mWish to maintain the linux system in a mostly alive state, whereas simialar dev-board with identical CPU wants 400mW
04:47 PM CaptHindsight[m]: CloudEvil: it's about the added latency jitter
04:47 PM CaptHindsight[m]: hell we max the cpu cores out sometime to get lowest latency
04:47 PM CaptHindsight[m]: power be damned
04:48 PM CaptHindsight[m]: most cpu's take about 100 clock cycles to stabilize after changing speeds
04:48 PM CaptHindsight[m]: time the OS gets paused
04:49 PM CloudEvil: ah
04:49 PM CloudEvil: I was in the wrong channel
04:57 PM CaptHindsight[m]: Alec is adding RTAI's i-pipe to the kernel to try and see if we can have this working so that RTAI doesn't have to be built against a given LCNC and vice versa
05:15 PM acer: I'm liking this shell that I'm rebuilding the motor controller for the new cnc machine into. It also houses the freq drive for the spindle
05:38 PM * JT-Shop calls it a day
05:39 PM Tom_L: warmed up a bit here
05:41 PM roycroft: i just got back from our board retreat
05:41 PM roycroft: we held it on our land, and about a half hour before the retreat ended it started snowing
05:41 PM xxcodery: snowing this early?
05:42 PM Tom_L: we had a sprinkling of snow but it melted right away couple days ago
05:42 PM roycroft: it did not snow in town
05:42 PM roycroft: but it did out there
05:42 PM xxcodery: tom on west coast it hardly snows but when it does it tend to be later not nov
05:43 PM roycroft: and it was so funny - people started freaking out and wanting to end the retreat early so they "wouldn't get killed" on the drive home
05:43 PM roycroft: they all dashed out as quickly as they could after the retreat
05:43 PM roycroft: but i realised that it was a very rare opportunity to walk around our fair site and the meadows in the snow
05:43 PM Tom_L: if they smell it in the air in the south they shut everything down and declare a state of emergency
05:44 PM roycroft: it snows so rarely around here, and usually when it does, it's well into flood season, and our fair site is under water
05:44 PM roycroft: so i spent an hour walking around in the snow before i came home
05:44 PM roycroft: it was really nice
05:45 PM xxcodery: roy nice
05:47 PM Tom_L: i've seen snow on halloween and in may
05:47 PM Tom_L: not usually the same year
05:48 PM roycroft: we have never had snow in may
05:48 PM roycroft: we have had it in october, but that's exceptionally rare
05:49 PM Tom_L: same here but it's happened
05:51 PM roycroft: https://roycroft.us/Dragon_Gate_Snow.jpeg
05:51 PM xxcodery: heh I remember that april snow
05:52 PM roycroft: our worst snows have been in april over the past few years
05:52 PM xxcodery: it was only snow of the "winter" so i guess nature wanted to get some in.
05:52 PM xxcodery: or was it may?
05:52 PM roycroft: not here it wasn't
05:53 PM roycroft: the snow in april 2010 was pretty bad
05:53 PM roycroft: i'll never forget that one, because it took down several trees at my place
05:53 PM roycroft: it took me months to get things all cleaned up
05:54 PM xxcodery: 2010 I was still at vancouver
06:08 PM roycroft: vancouver washington?
06:09 PM xxcodery: yeah
06:09 PM roycroft: does vancouver get washboard ice when it's snowy like portland does?
06:09 PM xxcodery: i always wanted a trip from vancouver to vancouver but never got around to that
06:09 PM xxcodery: washboard ice?
06:10 PM roycroft: yeah, where the roads are covered with ice that resembles a washboard
06:10 PM roycroft: i.e. wavy ice
06:10 PM roycroft: sinusoidal waves
06:10 PM xxcodery: ahh thats a rarity in vancouver but it does happen
06:10 PM roycroft: it happens all the time in portland when it snows
06:10 PM xxcodery: never saw any here at tacoma
06:10 PM roycroft: i think it's the way they plow the roads
06:11 PM roycroft: i had a friend in minneapolis who was in lincoln city one december, and i went to visit him and drive him back to the airport
06:11 PM roycroft: lincoln city is on the north-central oregon coast
06:12 PM roycroft: the day he was to return home we had a pretty bad snowstorm
06:12 PM roycroft: i drove up there anyway
06:12 PM roycroft: and had to drive him over the coast range and all the way to pdx in the heavy snow
06:12 PM roycroft: there was washboard ice on the roads as soon as we came out of the mountains
06:13 PM roycroft: iirc it took us 6 hours to get to the airport, and it's a 2 hour drive normally
06:13 PM roycroft: and when we finally got there the airport closed 10 minutes later, and he was stuck in portland overnight
06:14 PM roycroft: i still had to drive back down to eugene that night, so i ende up leaving him there and making the long drive home
06:14 PM roycroft: which again is normally 2 hours, but was more like 4 that day
06:14 PM xxcodery: ow
06:14 PM roycroft: yeah
06:15 PM roycroft: but since i grew up back east i had no realy problems
06:15 PM roycroft: real
06:15 PM roycroft: other than i had to drive slowly
06:15 PM roycroft: but i never felt in danger
06:16 PM roycroft: when it started snowing at the retreat today, before there was even anything sticking on the ground, one of our board members literally said "we need to ajourn this meeting right now - i don't want to die driving home"
06:16 PM roycroft: he lives in eugene
06:16 PM roycroft: it's a 20 minute drive in dry conditions
06:16 PM roycroft: probably 40 minute drive for him today
06:17 PM xxcodery: i still dont get why people is so bad in driving in rain here
06:17 PM xxcodery: hellp, its west wa
06:17 PM roycroft: one of our new board members said "i'm originally from la. i dont know how to drive in this stuff"
06:17 PM xxcodery: *hello
06:17 PM roycroft: and if there was actual snow on the ground i could kind of understand
06:17 PM roycroft: but there wasn't
06:17 PM roycroft: the roads were just wet
06:18 PM roycroft: you know, like they are almost continuously from september through may most years
06:18 PM xxcodery: usually I would make a joke of it raining when its not raining but its very little rain here so far. bit concerning
06:18 PM xxcodery: yeah normally i forget what dry roads look like for months
06:19 PM xxcodery: i remember one year I was driving home from work, and roads looked weird to me
06:19 PM xxcodery: then I finally realized it, its DRY
06:19 PM roycroft: the first rain in september actually is somewhat hazardous
06:19 PM roycroft: but not because people aren't used to the rain yet
06:20 PM xxcodery: usually power issues. we had it here late at OCT. normally earlier
06:20 PM xxcodery: *too
06:20 PM roycroft: it's because that first rain raises a lot of oils that are on the roads, and they do get very slick in a way that makes it hard to handle a vehicle
06:20 PM xxcodery: yep
06:20 PM xxcodery: fresh rain I always drive like its snowy or icy.
06:21 PM roycroft: but after that gets washed away, as long as you have a decent amount of tread on your tires, and they are inflated properly, it's not hard to drive in the rain
06:22 PM xxcodery: that reminds me, need to get new tires for my car
06:22 PM xxcodery: its getting quite bald
06:23 PM xxcodery: it still has some but yeah
06:23 PM roycroft: i almost drove my pickup today, because it was suppose to rain really really hard, and i have to drive on some gravel roads that are getting muddy out at the fair site
06:23 PM roycroft: but i ended up driving the prius instead
06:23 PM roycroft: with its small, fairly smooth tires
06:23 PM roycroft: it handled fine though
06:25 PM xxcodery: thats good
07:24 PM roycroft: it's so dark already
07:27 PM xxcodery: yep
07:28 PM xxcodery: we time traveled a hour so it's going to be dark early for quite a while.
07:28 PM xxcodery: I know we pst states all voted for stopping time changes but dunno when if ever it will be implemented.
07:28 PM roycroft: were that we would be able to stay with this
07:28 PM roycroft: it looks like next year it may actually happen
07:28 PM roycroft: but in a stupid way
07:28 PM roycroft: we'll end up effectively in year-round mst
07:29 PM roycroft: instead of year-round pst
07:29 PM xxcodery: honestly I dont care as long as time stops changing
07:29 PM roycroft: so the high sun will not occur at noon ever again
07:29 PM xxcodery: pst forever? fine. mst forever? sure
07:29 PM roycroft: i care somewhat because it's being done arbitrarily
07:29 PM roycroft: and without regard to nature
07:30 PM roycroft: but i agree that it's more important to not switch back and forth any more
07:31 PM xxcodery: im fine with pst also
07:32 PM xxcodery: it stopped matering as soon as gas and candles was dropped as light sources
07:40 PM cradek: if we switch to permanent summer time (solar noon at 1pm) we'll switch back shortly as soon as people see how terrible winter mornings are, just like the last two times we did the same stupid thing
07:42 PM roycroft: yes, especially north of the 45th parallel
07:43 PM cradek: (we did the same stupid thing in 1942 and 1973, each time lasted just a few years)
07:43 PM roycroft: we're just south of the 45th, but even yesterday, before we switched back to pst, the sun did not rise until almost 8am
07:43 PM cradek: senator joe biden voted for it in 73, and then the reversion in 74
07:43 PM roycroft: if we stay on dst year-round the sun won't come up until 9am at the solstice
07:43 PM cradek: he *might* remember that
07:45 PM xxcodery: dunno, 50 years is hard to remember I bet
07:46 PM xxcodery: I barely remember anything 10 years ago
07:48 PM cradek: we should call sunrise 6am and sunset 6pm as god intended, and do away with these socialist time zones. they're a relic of before cell phones, when we had ... a useful railroad system
08:39 PM firephoto__ is now known as firephoto_
09:20 PM roycroft: you're in tukwila now, xxcoder?
09:21 PM roycroft: or have qwest just relocated you?
09:24 PM xxcodery: weird
09:24 PM xxcodery: I just rebooted my rpi
09:49 PM roycroft: *chuckle*
09:49 PM roycroft: abom just realised that cnc tapping really works
09:49 PM xxcodery: lol
09:50 PM xxcodery: I hoped it would work, expecially since i used cnc tapping thousands of times
09:50 PM roycroft: he did his first ever tapped holes on the new cnc mill
09:50 PM roycroft: and they were blind holes
09:50 PM roycroft: he was really nervous
09:50 PM roycroft: then it worked
09:52 PM roycroft: i should not laugh, because when i finally get my cnc conversion done on my mill, i'll probably be pretty nervous the first few operations
09:52 PM xxcodery: I bet. first times I did it it was quite nervious
09:52 PM roycroft: but i'm not a journeyman machinist
09:54 PM roycroft: i am hoping this board work slows down a bit over the winter, so i can get some significant progress done on my shop projects
09:55 PM roycroft: my plan this summer was to get into the cnc router table work hot and heavy as soon as i was done building those new cabinets
09:55 PM roycroft: but what happened was my elections work started right after that, and has consumed almost all my spare time for the past three months
09:56 PM roycroft: i'm optimistic that most of that work is over until late spring
09:56 PM xxcodery: nice
09:56 PM roycroft: and my standing desk for my cad workstation arrives this week
09:56 PM roycroft: i'm sure i'll want to spend some time working with it, as i've not had a standing desk before
09:57 PM roycroft: i'm quite anxious to see how well it works out
09:57 PM roycroft: and that will require doing a lot of work at the cad workstation, so i should make a lot of progress on the router drawings
09:57 PM xxcodery: hope it works out
09:57 PM roycroft: everyone i know who is using a standing desk loves it
09:58 PM roycroft: they all say it's best to get one that is easily adjustable from standing to sitting, and this one is not
09:58 PM roycroft: but if i build one for my main desk, it will have an electronic lift, and pre-sets for the heights
09:58 PM xxcodery: theres system to automate switches
09:58 PM roycroft: so i can just push a button to adjust the height
09:58 PM xxcodery: say cycle of 2 hours sit and 2 hours stand
09:59 PM roycroft: https://www.amazon.com/dp/B07WFR5WB8/
09:59 PM roycroft: that's what i'm looking at for my main desk
10:00 PM roycroft: it has four presets for the height
10:00 PM roycroft: i probably only need two
10:01 PM xxcodery: no top hm
10:01 PM xxcodery: you will make the top?
10:01 PM roycroft: no, i'll build that
10:01 PM xxcodery: nice
10:01 PM roycroft: one of my hardwood suppliers is looking for a slab that will work for me
10:01 PM xxcodery: I picture fancy inlay patterb or something lol
10:02 PM xxcodery: or complex pattern using 2x2 wood
10:02 PM roycroft: since it's a corner desk, i need a slab that curves 90 degrees
10:02 PM roycroft: and if i can find a nice crotchwood slab, that could be perfect
10:02 PM xxcodery: cant just put one together?
10:02 PM roycroft: it would be nice if i found a single piece that naturally curves
10:02 PM roycroft: and i don't mind paying $1500 for a slab, if it's the right one
10:03 PM roycroft: my supplier already has some walnut slabs that are almost big enough
10:03 PM roycroft: i'd prefer something lighter, though
10:03 PM roycroft: cherry would be ideal
10:05 PM xxcodery: lol ok
10:47 PM xxcodery: sinned6915 yo
11:19 PM solarwind: I have these servo motors I got for like $5. They were industrial surplus, made by Sanyo I think in Japan
11:19 PM solarwind: can't find a datasheet
11:19 PM solarwind: But the encoder has 4 differential pairs
11:20 PM solarwind: 2 of the pairs look like quadrature A/B signals on the scope
11:20 PM solarwind: The other two are also differential pairs but I have no idea what the hell I'm looking at
11:20 PM -!- #linuxcnc mode set to +v by ChanServ
11:21 PM solarwind: The quadrature signals look like they're about 1000 pulses per revolution just from turning the shaft with my finger at ~1 revolution/s and reading the measured frequency on the scope
11:21 PM solarwind: The other two pairs are significantly faster and seem to output a steady pattern even when the shaft is not moving
11:21 PM solarwind: any idea as to what they can be?
11:22 PM solarwind: the pulses in the signal are around 1µs