#garfield Logs
May 01 2020
#garfield Calendar
12:30 AM zhanx: rue_mohr, crap photo https://imgur.com/a/QC4sURw
12:36 AM Tom_L: what board is that?
12:36 AM zhanx: nano arduino
01:33 AM Tom_L: yay i got it
02:05 AM rue_mohr: ah yea
02:06 AM rue_mohr: mg90
02:06 AM rue_mohr: mg90?
02:06 AM * rue_mohr quints... nmnmnmmnmnmnmnmnm
02:09 AM zhanx: sg90 metal gears
02:12 AM rue_mohr: yea
02:12 AM rue_mohr: mg
02:13 AM rue_mohr: so, I realized that my motor driver could save 18 parts by using a uln2003
02:16 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150878.jpg
02:16 AM rue_mohr: I'll use the resistors for 4.4us deadtime
02:16 AM rue_mohr: otherwise I'm pushing a bit more power than I want from my dc-dc
02:18 AM rue_mohr: and the uln version, as the quick & dirty zener regulator is fine for the 74ls06, but wastes more power again
02:18 AM rue_mohr: I'm gradually working out where I dont need to have precision
02:19 AM rue_mohr: a supply for an inverter that can take from 4.75 to 7, is just fine on a 5.6V zener regulator
02:19 AM rue_mohr: pulls 35mA tho, so you end up dissipating .28W
02:20 AM rue_mohr: zippo:/usr/src$ ohm -v 7 -i 40m
02:20 AM rue_mohr: Power is: 280.000mW
02:20 AM rue_mohr: Current is: 40.000mA
02:20 AM rue_mohr: Voltage is: 7.000V
02:20 AM rue_mohr: Resistance is: 175.000Ohms
02:20 AM rue_mohr: (bit of headroom)
02:49 AM rue_mohr: 42 holes reduced to 16
02:50 AM rue_mohr: if only I could trim down the power supply
05:55 AM polprog: dead time is when it keeps both transistors active?
07:02 AM polprog: rue_mohr, new version is in the ircjunk
07:02 AM polprog: same place
07:03 AM polprog: it has make install now, managed to fix the e24 bugs
07:03 AM polprog: i think it could calculate attenuation and phase shift if R, C, F is given
07:03 AM polprog: you can add that if you want :D ill be doing uni stuff today
09:33 AM aandrew: rue_mohr: why use the ULN for one side?
09:34 AM aandrew: use two inverters on the one leg or an inverter and buffer to try to balance the +ve and -ve delays
12:16 PM rue_bed: polprog, both odd
12:16 PM rue_bed: off
12:17 PM rue_bed: its all uln
12:17 PM rue_bed: the uln does not require 5V
12:17 PM rue_bed: the 7406 is an alternate version
12:18 PM rue_bed: I can do the whole H bridge in 1 uln2003
12:18 PM rue_bed: all the logic and fet driving
12:20 PM rue_bed: so, the whole thing is, a 12->24V converter, the 4 main fets, a uln2003, and like, 6 resistors
12:21 PM rue_bed: the fets dont generate "shoot thru" (positive to negative short) due to the deadtime, there is a ~4us period between everyone being off and someone turning on
12:21 PM rue_bed: I could make it less than 4us, but the power dissipated by the 24V pullup goes too high for the dc-dc converter
12:21 PM Tom_L: might be a toaster oven if they crossed
12:21 PM polprog: rue_bed: what's both odd?
12:22 PM rue_bed: oooh I designed a new dc-dc too!
12:22 PM rue_bed: both off
12:22 PM rue_bed: I just woke up, I'm still seeing double
12:22 PM polprog: ah
12:22 PM Tom_L: no, there's really 2 of you
12:22 PM polprog: why is that a problem?
12:22 PM rue_bed: I cant tell which key is really what
12:22 PM rue_bed: and reading is really hard
12:23 PM rue_bed: Tom_L, yea, the cool part was when I realized that I could create the deadtime by making bad fet drivers that turned off fast, but on slow
12:23 PM rue_bed: so their carefully tuned bad drivers
12:23 PM rue_bed: "_
12:24 PM rue_bed: ... :)
12:24 PM rue_bed: ... I have 4 hands....
12:24 PM * rue_bed stuff his head under his pillow
12:48 PM polprog: haha
12:48 PM polprog: i got up at 1130 today
12:48 PM polprog: and made a nice plot in gnuplot
12:48 PM polprog: https://polprog.net/rozne1/ircjunk/1b.svg
01:00 PM rue_mohr: whats pomiarowe?
01:02 PM rue_mohr: something is odd about those values...
01:05 PM polprog: measurement data
01:05 PM polprog: with linear regression
01:05 PM polprog: the data is made up
01:05 PM rue_mohr: ah!
01:05 PM polprog: the way remote labs work is that we run a simulator (on next experiments we can also book a time to remotely control a lab setup)
01:05 PM polprog: and we take the data from the sim
01:06 PM polprog: well, in this case we just got the data
01:06 PM polprog: what do you find odd about it ?
01:06 PM polprog: :P
01:08 PM rue_mohr: the sudden change in the horizontal deviation between v=10 and v=10.5
01:09 PM polprog: range change on the analog vom
01:09 PM polprog: 10 -> 3
01:09 PM polprog: sorry, 30 -> 10
01:10 PM polprog: the current was measured with a 2000 counts dmm
01:10 PM rue_mohr: oooh, its the uncertainty of the vom reading!
01:10 PM polprog: it doesnt make sense :D
01:10 PM polprog: yep!
01:10 PM polprog: its all practice
01:10 PM rue_mohr: ok
01:10 PM rue_mohr: you gonna chart some stuff from the RC library?
01:11 PM polprog: hmm, i could
01:11 PM rue_mohr: I'v never used gnuplot (directly) I use gnumeric
01:11 PM rue_mohr: ( which I'm sure uses gnuplot)
01:11 PM polprog: i should install gnumeric on the bench PC, its got some features i coul duse
01:11 PM polprog: could use
01:11 PM rue_mohr: I need a shower, breakfast, and to build this circuit
01:11 PM polprog: and i couldnt find a non shady source of gnumeric for windows
01:12 PM rue_mohr: heh
01:12 PM polprog: the gnuplot process is a bit overcomplicated, i do the data processing and writing in libreoffice on windows
01:12 PM polprog: i export a CSV file with the plot data, the sftp it to this server and do gnuplot here
01:12 PM polprog: into an SVG :D
01:12 PM rue_mohr: gnumeric?
01:13 PM rue_mohr: can do all that?
01:13 PM polprog: i think
01:13 PM polprog: i could plot stuff in libreoffice but the plots arent that nice
01:13 PM polprog: i dont think they are that nice in gnumeric either
01:14 PM polprog: and that linear reg in gnuplot is like 2 lines of script.. in gnumeric it takes a key combo i forgot
01:15 PM polprog: gnuplot has tektronix graphics output, so i could draw the curves in terminal ;)
01:16 PM aandrew: I think that rue_mohr will appreciate this: https://hackaday.com/2020/04/30/ultrasonic-sound-gun-precisely-aims-your-music/
01:16 PM aandrew: it uses a 555
01:27 PM rue_mohr: heh
01:27 PM rue_mohr: interesting
01:27 PM rue_mohr: it looks like its just one driving set tho
01:29 PM aandrew: yeah .. I thought this worked by constructive interference
01:29 PM rue_mohr: I want to make one like that
01:29 PM rue_mohr: let it naturally hetrodyne into what you wanted
02:14 PM rue_mohr: ARG, BREAKFAST
02:27 PM Tom_L: odd the latency was better on my old D525 than an asrock Q1900
02:28 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/screenshots/ASRock_Q1900_MB_Buster_RTAI.png
02:28 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/screenshots/D525_Latency_Wheezy.png
02:28 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/screenshots/D525_Buster_Rtai_2.png
02:29 PM Tom_L: and my gigabyte i5
02:29 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/screenshots/Gigabyte_i5_MB_Buster_RTAI.png
02:38 PM rue_mohr: not really
02:39 PM rue_mohr: latency is something that new designers completely ignore, they want speed
02:39 PM rue_mohr: and they latency the hell out of things to get it
02:40 PM Tom_L: i was concerned buster wouldn't even start up on the old board
02:47 PM ShopMoon: A function call taht looks wierd.
02:48 PM ShopMoon: MyReturn = ((FUCKINGTHING1)(MaybeFunctionName)) (Arg1, arg2, arg3);
02:49 PM ShopMoon: my initial thoght was it was a typecast to the return type, but maybe not with those extra parenthasese
02:54 PM ShopMoon: ahhhh, I'v come to the conclusion that it is a typecast on the return type
02:54 PM ShopMoon: never mind :P
03:22 PM rue_shop1: hi
03:22 PM rue_shop1: ok
03:22 PM rue_shop1: that makes sense
04:11 PM zhanx: ok finally got the hip joint right
04:42 PM rue_shop1: hows the scrap bin looking?
05:01 PM zhanx: bare
05:01 PM zhanx: TrainingCycle: 400 Error = 0.00604
05:01 PM zhanx: its getting there
05:19 PM aandrew: well, I asked for a nice sliding mitre saw for my birthday: https://www.canadiantire.ca/en/pdp/mastercraft-dual-bevel-sliding-mitre-saw-10-in-0556745p.html
05:19 PM aandrew: 50% off is hard to say no to
05:39 PM rue_mohr: ah! happy birthday
05:40 PM zhanx: who's birthday is it?
05:41 PM rue_mohr: aandrew, maybe
05:41 PM zhanx: ok one leg on the robot 3 more hips to print
06:56 PM Tom_L: make it hop
06:57 PM zhanx: not this version tom
06:57 PM zhanx: next one maybe
06:57 PM zhanx: really need to change them threaded rods out on the printer
07:02 PM rue_shop1: what are you using now?
07:02 PM rue_shop1: M10?
07:03 PM zhanx: 8mm
07:03 PM rue_shop1: ouch
07:03 PM rue_shop1: 10-24
07:03 PM zhanx: i know i know
07:03 PM rue_shop1: you want flimsy
07:03 PM zhanx: you said to change them out a long time ago
07:03 PM rue_shop1: did you get the stl for the "nutgrade" that adapts a 10-24 to the nut size for the #8 so you can just drop it in
07:04 PM zhanx: you talked about it but, don't think so
07:05 PM Tom_L: how do you circumsize a hillbilly?
07:06 PM zhanx: same way you do a pakistani
07:06 PM Tom_L: you kick his sister in the jaw
07:06 PM Tom_L: some of the shit my kids bring home from work....
07:07 PM zhanx: its an old joke
07:35 PM zhanx: 2 legs on
07:49 PM rue_mohr: pff
07:49 PM rue_mohr: k, well, I have a descent charge pump going