#garfield Logs
Nov 27 2020
#garfield Calendar
12:21 AM rue_mohr: polprog, :) of course, wildcards work in the pdf search
12:27 AM rue_mohr: 340?3
03:02 AM polprog: rue_shop1: :)
06:26 AM rue_mohr: http://machineryequipmentonline.com/electric-equipment/wp-content/uploads/2016/02/introduction-to-basic-electricity-and-electronics-technology-0450_thumb.jpg
06:26 AM rue_mohr: its a relaxation osc? but for an inductor?
06:33 AM rue_mohr: https://www.industrial-electronics.com/measurement-testing-com/images/aole_10-22.jpg
06:34 AM rue_mohr: https://www.industrial-electronics.com/measurement-testing-com/images/aole_10-24.jpg
06:37 AM rue_mohr: heh polprog : https://www.industrial-electronics.com/measurement-testing-com/images/aole_10-37.jpg
06:37 AM rue_mohr: funny, I'd never come across an scr osc before
06:38 AM rue_mohr: less the ones we did
06:38 AM rue_mohr: wait that a PUT, their common
06:40 AM rue_mohr: ...
06:40 AM rue_mohr: if a capacitor can be used in series across an op-amp to get a trangle wave, why cant an inductor be used in parallel
06:48 AM rue_mohr: https://www.learningaboutelectronics.com/images/Triangle-wave-generator-circuit-with-an-LM741-op-amp.png
06:48 AM rue_mohr: maybe I can ditch the 555
06:49 AM rue_mohr: funny, I was wondering if I could do just that last night
06:49 AM rue_mohr: its hard to find the circuits that are real triangles and not made of RC curves
07:03 AM rue_mohr: https://lh3.googleusercontent.com/proxy/e1valUPNwSvFtUarLyr940Y9iCixQwULN4pZ7Qjx84ONM-YlXiiBExbtJXd2kvh_0wtPZaSj7NsL9cjZho3RCUTSiAVfwpDI
07:04 AM rue_mohr: thats a variant (a complex one) of what one of the ones I was playing with
07:05 AM rue_mohr: LOTS of the circuits that are comming up are fakes
07:06 AM rue_mohr: you can stick a square wave into an integrator without feedback, the output bias drifts
07:06 AM rue_mohr: tiny changes in duty slam the output from clipping one or the other rail
07:23 AM rue_mohr: http://all-electric.com/schematic/eticircuits/555-triangle-with-independent-slopes.htm
07:23 AM rue_mohr: there is another one
07:24 AM rue_mohr: :) still, nobody does a current source like I do
07:27 AM Tom_L: mornning. you're up early
07:27 AM rue_mohr: yea i woke up
07:27 AM rue_mohr: in the "ugh damnit" way
07:28 AM rue_mohr: brain is on, sleep is hopeless
07:28 AM Tom_L: i could but can't
07:28 AM Tom_L: gotta run..
08:38 AM rue_mohr: ok, the RC one with the RC output filter, nope
08:38 AM rue_mohr: the one with the seemingly reversed integrator feedback, nope (does not even oscillate)
08:39 AM rue_mohr: so I'm left with (didn't try the transformer based one, as its discharge is RC)
08:39 AM rue_mohr: is the integrator/schitt combo
08:39 AM rue_mohr: which might get rid of the need for the 555
08:39 AM rue_mohr: I got it stuck in my head that I needed the latch, but a schmitt is ok
08:40 AM rue_mohr: but I need two things
08:40 AM rue_mohr: a dc-dc to make the op-amp supply
08:41 AM rue_mohr: and a osc that can clock the 4017 faster than the 5kHz coil oscillator
08:41 AM rue_mohr: and I think I can combine them
08:41 AM rue_mohr: if I use a power converter like the joule theif
09:27 AM rue_mohr: 4 minutes till the exam starts
09:27 AM rue_mohr: arg
09:27 AM rue_mohr: after which, I'm trying my power converter
09:27 AM rue_mohr: }:|
09:27 AM aandrew: good god man you never sleep
09:27 AM aandrew: how do you do it
09:28 AM rue_mohr: hah
09:28 AM rue_mohr: I woke up early and was just awake
09:28 AM aandrew: too much coffee gives me jitters. 5hr energy works REALLY nicely but will only get me to about 4am and then I am truly ready for bed
09:28 AM rue_mohr: so I found a bunch of circuits and tried them out
09:28 AM aandrew: yeah I've been lazily following your progress
09:29 AM rue_mohr: I think I can do the whole lvdt with two dual op-amp chips* and a 4017
09:29 AM aandrew: 4017? for the "rising edge to sample time" delay?
09:29 AM rue_mohr: I might be using 2 duals instead of a quad due to a particular ali purchase...
09:29 AM rue_mohr: yea, I need to have a sample timer
09:29 AM aandrew: nifty
09:30 AM aandrew: this is all for cpu-less operation right?
09:30 AM rue_mohr: if I run the dc-dc about 5x to 10x the clock of the lvdt, I can pick some outputs off the 4017 to give me the sample pulse
09:30 AM rue_mohr: the unit will just output a 0-5V analog signal from the lvdt
09:31 AM rue_mohr: I can pour a bunch of them into a uC with an ADC
09:31 AM rue_mohr: ok, gtg do exam, bbl!
09:41 AM polprog: oh, i just missed him
09:41 AM polprog: i think the scr osc you posted works on a similar basis
09:41 AM polprog: im so tired
09:41 AM polprog: but i have some free time
09:41 AM polprog: i got the VNA software working
09:55 AM aandrew: nice, which VNA do you have?
09:55 AM aandrew: I bought an ali SA+TG which is really more of an SNA but good enough for now
10:25 AM aandrew: https://www.sainsmart.com/products/sainsmart-genmitsu-cnc-router-3018-mx3-diy-kit
10:25 AM aandrew: just picked that up, CAD$300 (kit, no tooling or software)
10:25 AM aandrew: it's usually almost $500
10:47 AM rue_mohr: hey
10:47 AM rue_mohr: hey there, now you can do lamacoids :)
10:48 AM rue_mohr: aandrew, I wonder what kinda speed controller they used with it
10:48 AM rue_mohr: I dont think the motor on the one I made is much larger, but I have an optical interrupter regulating a pwm controller
10:53 AM rue_mohr: https://www.aliexpress.com/item/4000016863432.html
10:53 AM rue_mohr: 300x180
10:53 AM rue_mohr: is that the same?
10:55 AM rue_mohr: 230...hmm
10:55 AM rue_mohr: +- I'd adjust things...
10:56 AM rue_mohr: https://www.aliexpress.com/item/33006255751.html
10:56 AM rue_mohr: hahaha look at the shipping!!?
10:56 AM rue_mohr: hahah
10:56 AM rue_mohr: ...
10:56 AM rue_mohr: ECHO...
11:08 AM aandrew: yep ali shipping is all over the map
11:08 AM aandrew: and the controller is some proprietary shit, another $70 for a grbl-aware one
11:08 AM aandrew: I'll skip that for the time being
11:20 AM rue_mohr: pff. grbl would probably work find
11:20 AM rue_mohr: e on that board
11:27 AM Tom_itx is now known as Tom_L
12:12 PM aandrew: https://old.reddit.com/r/nextfuckinglevel/comments/k1wmek/this_man_made_a_flying_bathtub_using_drones_and/?ref=share&ref_source=link
12:12 PM aandrew: rue, is that you?
12:42 PM aandrew: sigh, looks like both mach3 and drufelcnc are windows apps. balls
12:47 PM aandrew: ooh it looks like linuxcnc might speak the language
12:49 PM aandrew: er what, linuxcnc doesn't know how to talk via usb?
12:51 PM Tom_L: no, usb is not realtime
12:52 PM Tom_L: it's smarter than that
01:00 PM aandrew: Tom_L: USB can be realtime
01:00 PM aandrew: stop using bulk transfers for everything
01:01 PM aandrew: although I figured it just spat out gcode like the 3d printer, isn't that also a realtime requirement?
01:06 PM rue_shop1: geez
01:06 PM rue_shop1: not much battery bank is it?
01:07 PM aandrew: rue_mohr: just take a little gas inverter up there with you :-)
01:08 PM rue_shop1: the dc-dc and clock is gonna work
01:08 PM rue_shop1: if I can just dial it down about 600khz
01:08 PM rue_shop1: aandrew, thats halarious
01:08 PM rue_shop1: I wonder what the flight time is
01:09 PM aandrew: do you use ball nose or flat nose bits for the initial cuts?
01:12 PM rue_shop1: hmm, it likes inductors that saturate
01:12 PM rue_shop1: aandrew, me? you mean tom, right?
01:15 PM rue_shop1: if L~~n^2 (inductance, turns) then if I want to divide the L by two
01:16 PM rue_shop1: lets say I had 10uH and 49 turns
01:16 PM rue_shop1: L = k*n^2
01:50 PM Tom_L: aandrew, but it's not
01:53 PM aandrew: Tom_L: forgive me, but why does g-code have to be realtime? "move here" "now move here" "now here" has no feedback that requires real time conditions
01:53 PM aandrew: if the PC is counting encoder steps or something I understand
01:54 PM Tom_L: most of the control in linuxcnc is on the pc not on an adapter box that just gets gcode
01:54 PM Tom_L: it passes realtime information out
02:04 PM aandrew: ok, that makes a lot of sense then, but I still argue that 125us guaranteed latency is absolutely realtime :-)
03:18 PM polprog: ive cleaned the bench
05:29 PM rue_mohr: aandrew, with GRBL, when you stop feeding commands to it, it finishes spooling out the ~20 it might have cached
07:11 PM aandrew: rue_mohr: whicih is exactly what I'd expect it to do
08:01 PM rue_mohr: it would be good if it had better status reporting on its buffer tho
08:01 PM rue_mohr: it can be hard to tell just what the machine has and hasn't done
08:01 PM rue_mohr: my spooler waits for "ok" before sending commands
08:02 PM rue_mohr: but that doesn't always work
08:02 PM rue_mohr: there are a few cases where lines return more than one ok
08:38 PM aandrew: rue_mohr: sure, if I were to design it you could either implement a progressive ACK system or something
08:38 PM aandrew: but I still think something that just drives isochronous traffic and lets LinuxCNC do what it has always done would work as well with hard realtime
08:45 PM rue_bed: yea, grbl sure is easy and cheap
08:46 PM rue_bed: I know a few people who do step/dir interfaces via paarallal port and can't stand how slow it is
08:56 PM Tom_L: on linuxcnc? yes parport is slow
08:56 PM Tom_L: on a budget it does work
09:00 PM rue_bed: its the parallel port, the other guy here was saying that the rate his machine would move under mach3 was really frustrating
09:00 PM rue_bed: thats the machine I have here
09:00 PM rue_bed: using grbl, and speed is no issue
09:00 PM rue_bed: gecko 10microstep drives
09:00 PM rue_bed: _10_ not 8 or 16, _10_
09:01 PM rue_bed: like, wtf
09:01 PM rue_bed: WHY are NE5532 op-amps so STUPID CHEAP from china?
09:01 PM rue_bed: I'm scared to pay only $6 for 100
09:03 PM rue_bed: that said, TL081 is $8 per 100
09:04 PM rue_bed: https://www.aliexpress.com/item/32971071879.html
09:04 PM rue_bed: I think those ones have worked out well for me before
09:06 PM rue_bed: so.... who ya talkin to there Rue...
09:06 PM rue_bed: hey
09:06 PM rue_bed: shall we go build a circuit
09:06 PM rue_bed: well, I aint gettin much sleep am I
09:06 PM rue_bed: cmon then!
09:07 PM rue_bed: we gonna wind an inductor?
09:07 PM rue_bed: yes! lets do it!
09:10 PM Tom_L: yeah geckos use 10
09:24 PM rue_mohr: like to make my own step/dir driver for dc motors
09:24 PM rue_mohr: keep getting sidetracked
09:28 PM aandrew: interesting, so this thing has a USB connection and I guess all the cool kids have custom DLLs that let mach3 drive them using whatever proprietary protocolst hey use
09:31 PM rue_mohr: makes sense
10:04 PM rue_mohr: hey I'v just learned that hunting for things on aliepxress and adding them to a cart is all the fun of shopping, without all the cost!