#garfield Logs
May 09 2021
#garfield Calendar
12:01 AM rue_mohr: ok I cant work out that servo code I did
06:09 AM Tom_itx is now known as Tom_L
10:50 AM rue_mohr: polprog, iirc the 74123 is good for pulses upwards of 20Mhz eh?
10:50 AM rue_mohr: er, tesla chips!?
11:05 AM polprog: Tesla was a czech[oslovakian] semiconductor factory
11:07 AM rue_mohr: oh cool!
11:08 AM rue_mohr: ebay them as replcement firmware for the model S :)
11:09 AM polprog: haha
11:10 AM polprog: i dont know whats the S logo (like the TBA120 in the corner) or the t-in-circle like the A232D
11:17 AM rue_mohr: no, I'd ask twitter tho
11:18 AM rue_mohr: oh wait
11:19 AM rue_mohr: the TBA120S is sig....
11:19 AM rue_mohr: sigenetics?
11:27 AM rue_mohr: https://www.mqp.com/images/g_ident.gif
11:27 AM rue_mohr: siemens
11:28 AM rue_mohr: haha I have a sheet for the tba120
11:29 AM rue_mohr: FM IF amp & demodulatr
11:29 AM rue_mohr: oh damn, I just closed the wrong window
11:30 AM rue_mohr: that'll be hard to put back
11:32 AM rue_mohr: found info on the mca660
11:32 AM rue_mohr: hows your russian?
11:32 AM rue_mohr: its something to do with colour video
11:33 AM rue_mohr: oh hah
11:33 AM rue_mohr: they have some english in there
11:33 AM rue_mohr: circuit for contrast, colour satuationand brightness control in colour tv sets
11:34 AM rue_mohr: so there is a lot of tv/radio stuff in there
02:26 PM Tom_itx is now known as Tom_L
03:19 PM polprog: rue_mohr: my russian is usable
03:19 PM polprog: I saw that tba120s be an IF related part but i was in the forest, and only got my phone and 2g
03:19 PM polprog: let me read up now
03:20 PM polprog: tba120 looks like a TV sound demod, so most likely UHF
03:20 PM * polprog makes a radiomodem from old TV parts
03:21 PM aandrew: heh
03:21 PM aandrew: speaking of russians
03:21 PM aandrew: saw a funny meme a few weeks ago talking about how the grandchildren of Rasputin were alive in the late 70s when that Boney M song became famous and how awkward it must have been to hear people singing about how much of a stud your grandpa was
03:22 PM polprog: looks like it's for IF, so 5MHz
03:22 PM polprog: FM-ZF-Verstarker
03:22 PM polprog: mit Demodulator!
03:22 PM polprog: ermahgerd
03:23 PM polprog: aandrew: hahaha
03:24 PM aandrew: "ermahgerd" is my second favourite ancient meme. My #1 is "whargarblgrbl": https://i.kym-cdn.com/entries/icons/original/000/001/625/135891.jpg
03:24 PM polprog: 0-12 Mhz
03:24 PM polprog: huh, i never heard about that meme
03:24 PM polprog: lol
03:25 PM aandrew: actually ermahgerd is probably #2. Flamenwerfer is above it: https://i.kym-cdn.com/entries/icons/original/000/007/168/flamenwerfer.jpg
03:25 PM aandrew: that to me is just about the most German thing I can think of
03:25 PM aandrew: such a logical name :-)
03:26 PM polprog: ah yes
03:27 PM aandrew: yeah, whargarblgrbl #1, flammenwerfer #2 and ermagherd #3
03:27 PM aandrew: and now I've got to step outside, there are some children on my lawn
03:27 PM aandrew: and a cloud I need to shake my fist at
03:36 PM polprog: clint eastwood intensifies
03:40 PM polprog: TBA920: Television horizontal oscillators
03:40 PM polprog: holy crap. That one is a power house.
03:41 PM polprog: Sync separator, which has it's output brok out on a pin, so you can use it as a sync separator, two phase comparators and a horizontal oscillator
03:41 PM polprog: DAMN
03:44 PM polprog: TBA530 - three 6MHz amplifiers - RGB video amplifier
08:13 PM rue_mohr: :)
08:13 PM rue_mohr: I wonder if you can use the phase comparators for PLLs
08:48 PM aandrew: wtf is with everybody's shipping
08:48 PM aandrew: even amazon prime is 7 days now
08:59 PM rue_mohr: during this whole thing nobody will admit that local manufacturing is a big problem solder
09:00 PM aandrew: well given that local mfg would still have the prices way above what I'm willing to pay for my junk electronics, that'd be right, but I agree with you on a general note
09:06 PM Tom_L: ok i did a test cut with the new cutters: https://www.youtube.com/watch?v=A7PcbEKrtkI
09:06 PM Tom_L: surface finish after that cut: http://tom-itx.no-ip.biz:81/~webpage/cnc/cutters/60Deg_3Flute_3500R_F30_Finish.jpg
09:07 PM Tom_L: rue_mohr, how can i get the spindle to respond a little quicker without overshooting?
09:22 PM rue_mohr: wow
09:22 PM rue_mohr: nice
09:22 PM rue_mohr: well, what control method are you using right now
09:22 PM Tom_L: pid for the analog in linuxcnc and whatever pid method the vhd has
09:22 PM Tom_L: it's not that bad but i wish it was a tiny bit better
09:22 PM rue_mohr: no like 4-20mA -10-10V?
09:23 PM rue_mohr: i2c
09:23 PM rue_mohr: msp
09:23 PM Tom_L: 0-10v
09:23 PM rue_mohr: ok
09:23 PM rue_mohr: and the feedback is an enocder, right?
09:23 PM Tom_L: i kinda hate to mess with it really
09:23 PM Tom_L: for the linuxcnc loop yes
09:24 PM rue_mohr: k, the vfd loop uses its own?
09:24 PM Tom_L: yes
09:25 PM Tom_L: inductive feedback i presume
09:25 PM rue_mohr: ok, so you want to set up the loop without linux in there first
09:25 PM rue_mohr: but you didn't find PID settings
09:25 PM Tom_L: umm iirc just P and I
09:25 PM Tom_L: i think
09:25 PM aandrew: Tom_L: AC or DC motor?
09:25 PM Tom_L: i wrote the defaults down
09:25 PM aandrew: VFD suggests the former
09:25 PM Tom_L: bldc
09:26 PM rue_mohr: ok, PI is fine
09:26 PM aandrew: sensorless or tach feedback?
09:26 PM rue_mohr: yea
09:26 PM Tom_L: Tom_L> inductive feedback i presume
09:26 PM aandrew: ah sensorless
09:27 PM Tom_L: setp pid.0.Igain 1.8
09:27 PM Tom_L: setp pid.0.FF0 .75
09:27 PM Tom_L: setp pid.0.FF1 .75
09:27 PM rue_mohr: so if you remove the linux feeback
09:27 PM aandrew: there's lag on that and the internal motor model for speed/position estimation could also be lagging
09:27 PM Tom_L: that's for the analog signal
09:27 PM rue_mohr: and set it, fron 0 to go to some speed
09:27 PM Tom_L: it doesn't use P I or D
09:27 PM rue_mohr: is it slow or does it overshoot
09:28 PM Tom_L: it comes up fairly quick and overshoots a tiny bit then settles
09:28 PM Tom_L: it was about as good as i could get it in linuxcnc
09:28 PM rue_mohr: so thats its responcew ithout linux feedbac
09:28 PM Tom_L: which?
09:29 PM Tom_L: the video is how i'm using it with both PID loops
09:29 PM Tom_L: linuxcnc and the vfd
09:29 PM aandrew: man I hate those scabs that are "deep" and they kind of dry out and feel like a little stone on the skin
09:30 PM rue_mohr: step a) DISCONNECT the linux feeback
09:30 PM rue_mohr: step b wtih the spindle stopped, set a speed
09:31 PM rue_mohr: step c monitor the sleep goes it overshoot
09:31 PM rue_mohr: how fast does it come up
09:31 PM Tom_L: should i not use the PID on linux for the analog at all?
09:32 PM Tom_L: just assume the analog out is what it's supposed to be...
09:32 PM rue_mohr: its a test, you do this before connecting the linux feedback
09:32 PM rue_mohr: after testing the performace without it, you reconnect
09:32 PM Tom_L: i'm not sure how much software rewiring i'll need to do to try that
09:33 PM Tom_L: so run linuxcnc analog with no pid
09:33 PM Tom_L: to the vfd
09:33 PM Tom_L: i'll have to look over the config to see what all to disconnect
09:34 PM Tom_L: it's not a matter of unhooking a wire
09:34 PM Tom_L: but i must say, i was impressed with the cut and finish on that
09:36 PM rue_mohr: just do a step test without the linux feeback
09:36 PM rue_mohr: <rue_mohr> step a) DISCONNECT the linux feeback
09:36 PM rue_mohr: <rue_mohr> step b wtih the spindle stopped, set a speed
09:36 PM rue_mohr: <rue_mohr> step c monitor the sleep goes it overshoot
09:36 PM rue_mohr: <rue_mohr> how fast does it come up
09:36 PM Tom_L: yeah i get what you're saying
09:37 PM rue_mohr: after thats tuned you add the linux feedback again
09:37 PM rue_mohr: ok
09:37 PM Tom_L: not a tonight project
09:37 PM rue_mohr: I'm being terribly distracted right now too
09:37 PM rue_mohr: someone brought up a youtube window and now there are all sorts of interesting shorts
09:38 PM Tom_L: :)
09:41 PM * aandrew is reading about photodiode amplifiers
09:41 PM rue_mohr: oo thats trickey
09:41 PM aandrew: 150,000,000 gain type stuff
09:41 PM rue_mohr: oh, not the high speed ones?
09:41 PM aandrew: not so much high speed here, but I might change that as I learn more
09:42 PM aandrew: I might investigate silicon photomultipliers too
09:43 PM rue_mohr: I did the rotate on the feedback register of the interrupt callback thing and its working great
09:44 PM aandrew: oh? the z80 stuff?
09:44 PM rue_mohr: yes
09:44 PM rue_mohr: oh and on the other front
09:44 PM rue_mohr: I didn't take apart the DC servo breadboard
09:45 PM rue_mohr: I'm going to assemble an atmega8 with a DAC and see if I can compare a digital filter against an analog one
09:45 PM aandrew: yeah this weekend was pretty much a loss for any work
09:46 PM rue_mohr: yesterday me and ma got lots of plants from the woods
09:46 PM rue_mohr: "woods"
09:46 PM rue_mohr: its all logged, whatever
09:47 PM rue_mohr: trees, berry bushes, and some flowers
09:47 PM rue_mohr: see what lives
09:48 PM rue_mohr: the next challange for the z80 is to totally re-arrange the memory map
09:48 PM rue_mohr: bit
09:49 PM rue_mohr: the decoding is going to be trickey
09:49 PM rue_mohr: so I'm not sure how to do it yet
09:49 PM rue_mohr: I think I need to use a GAL for the chip select decoder
09:49 PM rue_mohr: I'm going to end up with a mix of 16k, 32k, and 64k chips
09:50 PM rue_mohr: maybe even a 256k
09:50 PM rue_mohr: I should probably upgrade the ram to a 256k if it'll fit in the socket
09:51 PM rue_mohr: The rom needs to go to 32k
09:51 PM rue_mohr: not for space, for chip-programmability
09:51 PM rue_mohr: er 64k
09:51 PM rue_mohr: @64k I can use an eeprom
09:51 PM rue_mohr: they dont make 32k eeproms
09:51 PM aandrew: interesting
09:52 PM rue_mohr: so I'll end up with a 64k rom, 256k ram, 32k of nvram, and 32k of space for a paged LUT rom
09:52 PM rue_mohr: which is super cool
09:53 PM rue_mohr: not sure what I do with that last 128k
09:54 PM rue_mohr: the Paged LUT rom is cool
09:54 PM aandrew: ore RAM of course
09:54 PM rue_mohr: the page sets the value to the function you want
09:54 PM rue_mohr: you read the address for your function
09:55 PM rue_mohr: so, for the value you write, you have 4096 answers available
09:55 PM rue_mohr: if you want to know the sine and cosine of a value, you only have to write the value once
09:55 PM rue_mohr: then read from one address for the sine answer, and another for the cosine answer
09:56 PM rue_mohr: answers can be grouped
09:56 PM rue_mohr: for example, hex to ascii
09:56 PM rue_mohr: there can be one answer for the first character, next for the second, next for the 3rd
09:57 PM rue_mohr: two more locations can be the first and second characters for the hex string
09:58 PM rue_mohr: I suppose it would only chew up 512 addresses to have 8x8 multiply and divide tables
09:59 PM rue_mohr: it takes 8 cycles (machine cycles, not clocks) to do a lookup
10:00 PM rue_mohr: if I implement the whole thing, it takes an 8Mbit rom
10:03 PM rue_mohr: ok, I need an atmega8 with a DAC
10:04 PM rue_mohr: oh, and I want a keyboard clicker
10:04 PM rue_mohr: hmm
10:04 PM rue_mohr: 8255->transistor->resistor->speaker?
10:22 PM rue_mohr: should I to an RC lowpass?
10:23 PM rue_mohr: and some volume controle?
10:23 PM Tom_L: i thought you did already
10:23 PM Tom_L: the lowpass
10:23 PM rue_mohr: no
10:23 PM rue_mohr: I did software emulation
10:24 PM rue_mohr: and worked out all the formulas
10:24 PM Tom_L: oh
10:24 PM rue_mohr: I didn't actually build the digital filter
10:25 PM Tom_L: ok, my eyes are done for today
10:25 PM Tom_L: gnite
10:25 PM rue_mohr: alrady?
10:26 PM Tom_L: i did the numbers on those cutters in fswizzard and it told me that the difference between a 30 deg helix cutter and a 60 deg helix cutter was an additional 10 ipm
10:26 PM rue_mohr: sounds good
10:26 PM Tom_L: that's pretty impressive geometry
10:27 PM rue_mohr: so, the chips come off more straight up and not at an angle?
10:27 PM Tom_L: think of them like an auger
10:28 PM Tom_L: it seems to clear them better yes
10:31 PM Tom_L: what impressed me was the finish it left at those feeds
10:31 PM Tom_L: for a home built mill
10:31 PM rue_shop1: :)
10:32 PM Tom_L: now go back to your electronics. i'm goin to bed
10:36 PM rue_shop1: damnit, cant they just put one port in a contigious row!
10:37 PM rue_shop1: oh I wonder if I should put an RTC on my z80