#robotics Logs
Jul 15 2018
#robotics Calendar
12:00 AM rue_shop4: 24 counts from a 3 bit encoder
12:03 AM rue_shop4: 15 degrees
12:05 AM rue_shop4: actaully, thats 25 positions
12:05 AM rue_shop4: so 14.4 degrees
12:05 AM mrdata: rue_shop4, quartz oscillators are good to parts per million, typically
12:06 AM rue_shop4: yes, but not variable from 1M to 10M
12:06 AM mrdata: right
12:06 AM rue_shop4: SOME SAW filters are tho
12:06 AM mrdata: the discussion on the vackar oscillator page is interesting
12:06 AM rue_shop4: but dunno about stability
12:07 AM mrdata: https://en.wikipedia.org/wiki/Vack%C3%A1%C5%99_oscillator
12:07 AM rue_shop4: did you hear my pll took it from a drift of about 300hz to a jitter of 4Khz?
12:07 AM mrdata: it mentions the things that go into that
12:08 AM rue_shop4: gaddamn jfets!
12:08 AM rue_shop4: you cant just go out and buy a jfet!
12:08 AM mrdata: you probably dont have to
12:08 AM mrdata: should be possible to taylor a mosfet
12:09 AM mrdata: read the discussion tho
12:09 AM mrdata: and the talk page of the article
12:10 AM mrdata: probably go find the patent too and read that
12:11 AM mrdata: a friend gave me a bag of crystals, an assortment from 4 to 75 MHz
12:11 AM mrdata: this week
12:11 AM mrdata: he asked, "can you use these?"
12:11 AM rue_shop4: :) I have a bucket of them
12:11 AM mrdata: sure!
12:11 AM rue_shop4: small, but a bucket
12:11 AM mrdata: so he wants to know if i can figure out what the original collector of these did
12:11 AM rue_shop4: I think mine go from 500Khz to 100-and-somethingMhz
12:11 AM rue_shop4: heh
12:12 AM rue_shop4: 4? 75?
12:12 AM rue_shop4: hmm, computer?
12:12 AM mrdata: i reckon the answer lies in the precise frequencies
12:12 AM mrdata: eg: 27.125 MHz
12:12 AM mrdata: is one of them
12:13 AM mrdata: 26.670 is another
12:13 AM rue_shop4: CB isn't it?
12:13 AM mrdata: idk, i didnt look on the spectrum chart yet for those
12:14 AM rue_shop4: https://en.wikipedia.org/wiki/Citizens_band_radio
12:14 AM rue_shop4: table most of the way down
12:16 AM rue_shop4: the old CBs used to swtich crystals, so there would be banks of them in there
12:16 AM mrdata: and 17.734
12:17 AM mrdata: 14.31818
12:17 AM mrdata: 6.144
12:18 AM rue_shop3: iirc 14.31818 is a multiple of the colour crystal for tv
12:18 AM mrdata: so these just came out of salvaged derelict consumer tech, i guess, then
12:19 AM rue_shop4: thats what you would see going thru my bin
12:19 AM rue_shop4: everything from 32Khz resonators to 100-and-somethingMhz computer
12:31 AM rue_shop4: hey, if I put a ceramic in parallel with a poly cap, do I get one thats stable?
12:35 AM mrdata: uh
12:35 AM mrdata: how unstable is each
12:35 AM mrdata: how do they vary under what conditions
12:35 AM rue_shop4: apparently inversly so from the chart on that page
12:37 AM mrdata: you'll have funny 2nd order effects maybe
12:37 AM rue_shop4: I wonder if this program will find anything better than 24 transitions by morning
12:37 AM rue_shop4: all the high freq stuff I can find uses smt
12:42 AM Tom_L: ok cad file modified a bit, we'll see what that does
12:51 AM rue_shop4: mrdata, I'd like to know what makes the current osc unstable
12:51 AM rue_shop4: I realize it could be a lot of things
12:52 AM mrdata: could be
12:52 AM mrdata: can i see the schematic
12:53 AM rue_shop4: :( my webserver is still gone
12:53 AM rue_shop4: oh
12:53 AM rue_shop4: !
12:53 AM rue_shop4: its back!
12:53 AM rue_shop4: sweet gee
12:56 AM mrdata: plus, a picture of the board, both sides?
12:59 AM * Jak_o_Shadows sighs
12:59 AM Jak_o_Shadows: searching for animaronic tail comes up with too much furry stuff
12:59 AM rue_shop4: drawing! :)
01:01 AM mrdata: animatronic?
01:01 AM mrdata: try mechanical snakes
01:02 AM Jak_o_Shadows: I sorta do want it as a tail - but like, a lot smaller, for the robot
01:03 AM mrdata: something snakes and tails have in common: a tail
01:04 AM rue_shop4: oh, camera, I'll try that again
01:08 AM mrdata: so imagine each vertebra is a cylinder with a cone on its end, and each attaches to the next; then you have 3 wires running the length of it, which you can tension or slack
01:08 AM rue_shop4: mrdata, the 74hc14 decoupling cap is under the chip in amongst the socket
01:08 AM rue_shop4: oh Jak_o_Shadows I got just the image for you
01:08 AM rue_mohr: https://a.pololu-files.com/picture/0J7327.1200.png?246c9d81e84fa2ea89cd6c6482211dde
01:08 AM rue_mohr: there ya go
01:08 AM rue_mohr: yes?
01:08 AM Jak_o_Shadows: yes!
01:08 AM rue_mohr: https://i.ytimg.com/vi/e-AClwxINBU/maxresdefault.jpg <-- this one is cool when you realize what your looking at
01:11 AM Jak_o_Shadows: oh, that's a cool way of controlling it
01:15 AM rue_mohr: http://ruemohr.org/~ircjunk/projects/siggen/
01:15 AM rue_mohr: ^ mrdata
01:15 AM rue_mohr: tell me if thats too small
01:16 AM mrdata: checking...
01:16 AM rue_mohr: it wasn't meant to be particularily pretty, I had to get off the solderless breadboard to make it work
01:18 AM rue_mohr: low component count tho
01:18 AM rue_mohr: for the record, the drift it about the same with a 2 gang pot instead of the CDS system
01:19 AM rue_mohr: for a while, it was fun trying to figure out the larger part of the drift problem, till I realized the cds were picking up stray light
01:20 AM rue_mohr: I'v been thinking about making a transistor version like your test
01:21 AM Jak_o_Shadows: 3/4 through the print, I realised I designed it completely wrong
01:21 AM Jak_o_Shadows: But I fuged i may as well let it finish, it wan'st a big print
01:21 AM rue_mohr: heh
01:23 AM mrdata: ok
01:23 AM mrdata: you need decoupling caps
01:23 AM mrdata: as close as possible to the chip
01:23 AM rue_mohr: have one per chip in there
01:23 AM rue_mohr: the op-amp is on the bottom
01:24 AM rue_mohr: the 7414 is in the socket
01:24 AM rue_mohr: 0.1uF ea
01:24 AM mrdata: 0.1uF from each chip's + to its -
01:24 AM rue_mohr: yup
01:24 AM rue_mohr: thats whats there
01:25 AM rue_mohr: http://ruemohr.org/~ircjunk/projects/siggen/p1100328.jpg <-- you can see the op-amp one there
01:25 AM mrdata: that mustard coloured thing? okay
01:25 AM rue_mohr: yep
01:25 AM mrdata: what about the other chip
01:26 AM rue_mohr: you know that gap in the socket?
01:26 AM mrdata: ??
01:26 AM rue_mohr: its in there
01:26 AM rue_mohr: .......
01:26 AM rue_mohr: -cap here-
01:26 AM rue_mohr: .....
01:27 AM rue_mohr: http://ruemohr.org/~ircjunk/projects/siggen/p1100328.jpg the rightmost dip is the 74hc14, the bottom right corner is vcc
01:27 AM rue_mohr: there is a leg of the cap stretched over to the vcc pin
01:27 AM rue_mohr: I couldn't get the litttle *&^&^$ thru the other hole
01:28 AM mrdata: weird way to install it; i'm lagging here
01:28 AM rue_mohr: maybe i should see how sensitive the whole thing is to power supply level, run it from 4.75 to 5.25 and see what the difference is
01:29 AM mrdata: and if it is, then put a voltage regulator on the board?
01:30 AM rue_mohr: sockets are handy space savers, in DIP40 days, you could tuck a narrow DIP18 under the dip40, in the socket, which worked great for 808x chips that needed the address latch
01:31 AM rue_mohr: maybe I can find some better 20pf caps too
01:31 AM rue_mohr: from what i'v read today, I'm better off putting leads on a smt cap and using that
01:31 AM rue_mohr: less just using smt caps for it
01:32 AM rue_mohr: I needed to be able to play with values
01:32 AM Jak_o_Shadows: This is the good youtunbe series I think
01:32 AM Jak_o_Shadows: https://www.youtube.com/watch?v=H4v6TTRIW-I&t=13s
01:32 AM Jak_o_Shadows: for the animatronic tails
01:33 AM mrdata: your connection wires are long and flop around each other
01:33 AM mrdata: each is a parasitic inductance
01:34 AM mrdata: rules of radio: signal wires should be straight and as short as possible
01:35 AM mrdata: manhattan construction would help control parasitics
01:35 AM mrdata: because everything stands off from a continuous ground plane
01:36 AM mrdata: you could replace those signal wires with coax maybe and ground the shield
01:37 AM Jak_o_Shadows: https://www.youtube.com/watch?v=rukhEogdWPM&list=PLQdu_G7xyFIQTkVES0KELgHnIPn9v5AxK is the playlist
01:44 AM rue_shop4: mrdata, I can clean it up a lot, it was mostly built that way so I could change parts to work out values
01:52 AM rue_shop4: 4.75V, 9.20Mhz and dropping
01:53 AM rue_shop4: 5.25V, 10.0Mhz and dropping
01:53 AM rue_shop4: wow
01:53 AM mrdata: yeah quite sensitive
01:53 AM rue_shop4: you can see it shift like crazy on the scope
01:53 AM mrdata: which makes sense from the datasheet fro the 74HC14
01:55 AM mrdata: so 0.1V will be like 80 kHz
01:56 AM mrdata: so get those power rails stable to 0.001V or better
01:56 AM rue_shop4: so, the pll should tweek the 7414 voltage :)
01:58 AM mrdata: probably add a 10uF decoupling cap to the board
01:59 AM mrdata: (but i didnt calculate that value)
01:59 AM rue_shop4: that must be related to the schmitt levels
02:03 AM mrdata: ugh. i misread 4.75V as 4.25; so 0.1 V will be like 160 kHz rather
02:04 AM mrdata: so get the rails stable to 0.0001V
02:04 AM rue_shop4: I can really see it swing when I varry the voltage
02:06 AM mrdata: q = CV; 0.0001V * 10 uF = 1 nC
02:07 AM mrdata: are you using cmos or ttl load currents
02:07 AM rue_shop4: just the scope so far
02:07 AM rue_shop4: it'd be cmos probably
02:08 AM mrdata: ok. +/- 0.02 mA for high vs low
02:08 AM rue_shop4: I have that output buffer too
02:10 AM mrdata: 10 uF decoupling isnt enough
02:12 AM mrdata: also change your power rails to big fat wires
02:18 AM rue_shop4: hmm
02:18 AM rue_shop4: I could etch a board
02:18 AM rue_shop4: I know the values now
02:18 AM rue_shop4: well, I could etch a baord if the equipment was up and going
02:22 AM mrdata: imagine two or three 0.04 mA p2p chirps at the oscillator's frequency (say 1 MHz) rippling back into the supply; each cycle is then 0.12 nC of ripple. q = CV, so at 0.1uF that's 0.0012 V; to keep the supply stable to 0.0001 V under this condition, you need 12x the protection
02:23 AM mrdata: replace 0.1uF decoupling caps with 1.5uF SMT
02:24 AM rue_shop4: can I use 10uF?
02:24 AM rue_shop4: my selection kinda goes from 1 to 10
02:24 AM mrdata: ok
02:24 AM mrdata: what is ESR of these
02:24 AM mrdata: it has to be very low
02:24 AM rue_shop4: heh, ... china
02:25 AM mrdata: for sure need < 0.01 ohm; maybe lots lower
02:25 AM rue_shop4: just to have done it tho, I'm gonna change the ceramic 20pf for smt
02:25 AM rue_shop4: I'm watching the power supply, its a steady 5.10
02:25 AM rue_shop4: mmm lets see on scope tho
02:27 AM rue_shop4: 60mV noise
02:27 AM mrdata: at 0.01 ohm and 10 uF, tau = 10^-7 s which becomes a 1.6 MHz low-pass filter
02:28 AM mrdata: 60mV noise on the supply is 600x more than you want
02:28 AM rue_shop4: hahaha I dont have 20pF
02:29 AM * rue_shop4 thinks
02:29 AM rue_shop4: the osc can go low, so, if I use smaller I'm ok
02:29 AM rue_shop4: 18pf
02:32 AM mrdata: okay, 10 uF SMT 0.001 ohm ESR is preferred for decoupling
02:36 AM mrdata: but i might be calculating that wrong
02:40 AM mrdata: do you have 1 uF SMT caps?
02:41 AM rue_shop4: yes
02:41 AM rue_shop4: I can give it a go
02:41 AM mrdata: add a pair of those to further decouple the 74HC14, and test supply noise
02:41 AM rue_shop4: you know, I'm suspect of the electrolytic caps, I think they can add noise
02:41 AM rue_shop4: ok
02:42 AM rue_shop4: the results aren't much different with the smt 20pF (18pF) freq is a big higher on average
02:43 AM mrdata: electrolytics are too-high ESR
02:43 AM mrdata: you'd put one on the board to decouple it from the supply tho
02:44 AM mrdata: 100uF maybe
02:44 AM mrdata: but 1uF's first, and test i think
02:46 AM mrdata: i'm guessing those will bring supply noise down to 3mV but who can say
02:49 AM rue_shop4: uh, when I said the noise was 60mV p-p...
02:49 AM rue_shop4: I think I had it on x10
02:49 AM mrdata: it's 6mV?
02:49 AM mrdata: so only 60x your goal, then
02:50 AM rue_shop4: with the 1uF caps, were at 400mV p-p
02:50 AM rue_shop4: other way :(
02:50 AM mrdata: fuck
02:50 AM rue_shop4: so, it must have been 600mv p-p
02:50 AM rue_shop4: hehehehehe
02:50 AM rue_shop4: just a bit of noise
02:50 AM mrdata: a fuck ton of noise
02:50 AM rue_shop4: there is something else
02:51 AM rue_shop4: its a cmos chip
02:51 AM rue_shop4: I didn't tie down the unused inputs
02:51 AM mrdata: ok, better do that
02:51 AM mrdata: they'll flap wildly
02:51 AM rue_shop4: so, there could be stuff switching all over the place
02:51 AM rue_shop4: yea
02:51 AM rue_shop4: esp with that much power noise
02:52 AM mrdata: the parasitic inductances are coupling with the supply, too, i expect
02:52 AM mrdata: can you get a ground plane in between those
02:53 AM rue_shop4: ok, down to 350mV
02:54 AM rue_shop4: I'm getting an idea how I might build the next rev of this
02:56 AM mrdata: try stacking another pair of 1 uF in parallel?
02:57 AM mrdata: fatten up the supply rails
02:57 AM mrdata: add 100uF electrolytic to the board
02:57 AM mrdata: where are you measuring the supply noise?
02:57 AM mrdata: near the 74HC14 or near the pin headers
02:58 AM rue_shop4: this could in part be my scope
02:58 AM rue_shop4: I'm measuring at the 7414 pins
02:58 AM mrdata: ok
02:59 AM rue_shop4: but if I short the probe and touch the ground to the board...
02:59 AM rue_shop4: yea, like I thought
02:59 AM mrdata: ?
03:00 AM rue_shop4: ok, ground clip *ON* probe, clipped right to it, signal is ~20mv p-p
03:00 AM mrdata: cheesy
03:00 AM rue_shop4: now I keep it clipped on and touch the ground to the osc power supply ground, bam, 350mv p-p
03:00 AM rue_shop4: I remember this from something else I was doing
03:01 AM rue_shop4: maybe if I dual probe and diff it..
03:01 AM mrdata: the probe is coax, isnt it?
03:02 AM mrdata: dual probe and diff is a possible strategy
03:03 AM mrdata: with phase adjust
03:03 AM mrdata: how to measure a known reference
03:06 AM mrdata: Gah! i was looking at the wrong datasheet
03:06 AM rue_shop4: I dont know if its real
03:07 AM rue_shop4: I cant see a difference nomatter what I do from a shorted probe and the power rails
03:07 AM rue_shop4: its interesting tho, the drift is always down...
03:10 AM rue_shop4: so many cool projects, so little time
03:10 AM rue_shop4: its only 1am, and my eyes are snapping shut
03:10 AM rue_shop4: arg
03:11 AM rue_shop4: I'd like a 1M to 10M linear detector I could use for feedback to that op-amp instead of the led current
03:12 AM rue_shop4: filters wont work, not over that range
03:15 AM rue_shop4: I might be able to do it by making an edge detector just like the delay uses, set to 1/10Mhz, use it to create a pwm signal that is filtered and fed to the amp as feedback
03:15 AM rue_shop4: or I get a pll that can handle it
03:16 AM rue_shop4: the problem seemed to be that it naturally drifts too fast
03:18 AM rue_shop4: I also want to make up a 2 to 4 gang cds array to play with
03:20 AM mrdata: ugh. lagged to oblivion
03:20 AM mrdata: crashed
03:31 AM rue_mohr: zlog
03:42 AM rue_mohr: 0,1,2,0,3,1,4,0,5,1,6,2,3,4,5,2,7,3,6,4,7,5,6,7
03:43 AM rue_mohr: ok, so, I need some kinda binary wheel generator
03:44 AM rue_mohr: there must be a longer sequence, every digit should come up 7 times
03:44 AM rue_mohr: anyhow
03:45 AM rue_mohr: hmm, cnc? 3d print? etch?
04:05 AM -!- #robotics mode set to +o by ChanServ
04:06 AM rue_mohr changed topic of #robotics to: If you ask a question, you must wait for the answer. | http://tom-itx.no-ip.biz:81/~webpage/index.php | N is for nanometer.
04:06 AM -!- #robotics mode set to -o by rue_mohr
09:27 AM robotustra: M
09:28 AM theBear: that sound a bit like a challenge ! i counter your emotive-sound with a "MMMMmmmmmmmmmmmmmm"
09:29 AM robotustra: M is for morning
09:33 AM robotustra: hm, acorn controller + pro verstion of software is a bit expensive
09:34 AM robotustra: + if to buy encoder for it
09:34 AM robotustra: but it can do a rigid tapping
09:37 AM robotustra: does it matter if the encoder NPN or PNP?
09:37 AM theBear: heh, probly cos acorn's are almost older than me, and no doubt getting a bit "thin on the ground" this late in the game <grin>
09:40 AM robotustra: you mean that china propose better for less money?
09:41 AM theBear: aha, there 'tis ! https://www.youtube.com/watch?v=j2gu3aetaOs&feature=youtu.be&t=1399 tphat almost 20 minutes was 5 totally worth spending to find the maaaaybe 6 or 8 second clip that mighta seemed relevant if it had arrived about19.5 minutes sooner :-)
09:42 AM theBear: but since "MMMMmmmmmmmm" is kinda his catchphrase i figured if there was even a chance of a contest, that he shoulda been involved :-D
09:42 AM robotustra: ok
09:43 AM robotustra: Mmmmm - is just a bug in your comprehension of reality
09:44 AM robotustra: I know people who start talking with "Mmmm" but I'm not :)
09:51 AM robotustra: < theBear> heh, probly cos acorn's are almost older than me
09:52 AM robotustra: it means it has no bugs
09:59 AM Tom_L: https://imgur.com/gallery/cKQA0Lh
10:03 AM robotustra: not clear what is it for
10:12 AM theBear: robotustra, indeed... i got nothing against the things, actually more than a few fond memories of 'em, it was purely a comment re: electronics with plastic casings very rarely seem to last for 30-35+ years, and if they do, they probably aren't exactly easily found in some huge "stuff you might wanna buy" list like more current things
10:21 AM robotustra: btw, chinese lathe controllers look suspisious
10:21 AM robotustra: I don't know why
10:22 AM robotustra: usually because people do not give much info on ebay
10:57 AM theBear: meh, we're robot builders in the age of cheap teeny weeny microcontrollers... we can build/program/make our own controller/power-stagey thingers quicker than the slow boat can even get from china to here
10:57 AM theBear: on reflection tho, some people,maybe even yourself, proably live a bit closer to china than myself
11:02 AM robotustra: theBear, yes we can, but it's a matter of time
11:03 AM robotustra: it's better to spend time on the REAL things like new robot design rather than to waste time on tools which already had been done millions of times by others and costs relatively cheap
11:04 AM robotustra: I would spend 2-3-4 month of making my own cnc lathe controller
11:04 AM robotustra: and my time will cost N times more than even acorn controller
11:05 AM robotustra: I prefer to build tools only if I'm sure that I need this tool and I can't get it anywhere for reasonable price
11:08 AM theBear: average 33-40 business days default quoted shipping time is what pretty much everywhere in china says for this capital city right here these days, some of them recently seem to have found an even slower slow boat and quote up to about 60days, which is enough time for at least a ok-ish controller r+d + implement <grin> maybe it not so bad there
11:09 AM weyland|yutani: theBear, last stuff i ordered was here in 1 to 2 weeks
11:09 AM robotustra: acorn is shipping from usa
11:09 AM weyland|yutani: theBear, maximum to this da took 4 weeks
11:09 AM robotustra: I usually get stuff in 21 days
11:09 AM robotustra: average
11:09 AM robotustra: still 21 days is not enouph to get things done because it's only 3 weekends
11:11 AM robotustra: if I build lathe during 1 year, ETA 33-40 days is still good for me
11:12 AM robotustra: hm, gonna go to shop and buy some wood to make a hutch on my table and thow out ikea shelvs
11:12 AM rue_mohr: in BC, canada post can take up to 6 months to 'get to' sorting a shipment
11:13 AM robotustra: is there any reason to live in a such hole like BC?
11:14 AM robotustra: I lived all my life in 1-2M people cities
11:14 AM robotustra: no problem with post what so ever
11:16 AM robotustra: wanna make tall hutch to free some space in the room
11:17 AM robotustra: for newly bulding lathe
11:24 AM rue_bed: how about having a half acre of space?
11:25 AM weyland|yutani: still need a shed on that half acre and plumbing electric etc
11:31 AM rue_bed: yea, me too
08:08 PM robotustra: made a hutch today
08:08 PM robotustra: now thinking what to throw out from the room
08:15 PM z64555: the old hutch, obviously
08:25 PM robotustra: new one
08:25 PM robotustra: ah
08:25 PM robotustra: there are an old chest of drawers
08:26 PM robotustra: but it keeps valuable things
08:26 PM theBear: heh, say hutch one more time and someone will have to shout 3 seemingly random numbers out then throw the heck outta a football
08:32 PM z64555: I have 5 numbers, tho :(
08:38 PM theBear: hmmm, and i have no football
08:39 PM z64555: what about a honey ham
08:39 PM durrf: put the hutch on the chest
09:33 PM robotustra: chest is rather high itself
09:33 PM robotustra: I put 3d printer on the chest