#robotics Logs
Jul 07 2018
#robotics Calendar
12:10 AM rue_shop4: ooo ALS14 is 39Mhz
12:20 AM rue_shop4: haha!
12:20 AM rue_shop4: ha!
12:20 AM rue_shop4: a 74als14, with the input current offset, with a LP using a 56pF cap, 470ohms feedback = 11Mhz, 10K = 0.7Mhz
12:21 AM rue_shop4: "I can use a fet-output opto to control it"
12:30 AM rue_shop4: now how about a 74hct1G14...
12:32 AM rue_shop4: hah, they arguably just fit on my dip8 adapter boards
12:53 AM rue_shop4: ok, HCT tends to stop oscillating, so that wont work
12:53 AM rue_shop4: stability of these is in question
01:21 AM rue_shop3: huh, the limit on the 4060 is 30Mhz
01:21 AM rue_shop3: but..
01:23 AM zhanx: but...?
01:25 AM rue_shop3: but its datasheet dosnt mention rc osc
01:34 AM rue_shop4: nope by about 5Mhz, the cd4060 is done
01:34 AM rue_shop4: er, 74hct4060
01:41 AM rue_shop4: *blink* oh
01:41 AM rue_shop4: in a fully differential amp, there is no - and + input
01:54 AM mrdata: i thought i might have a use for the insertable ferrite version for radio, so i embellished the drawing a bit, https://imgur.com/a/qNfJiBz
01:57 AM mrdata: stability probably depends on Q-factor
01:58 AM mrdata: getting the capacitor pair down to 10 to 20 pF range might be a good thing, because that brings the Q factor up
01:59 AM mrdata: stability probably also depends on parasitics and interference
01:59 AM mrdata: and with too heavy a load, it will stop oscillating
02:03 AM mrdata: oh, temperature too, is a concern
02:04 AM mrdata: this is why many oscillators are operated in a temperature-controlled oven
02:26 AM Jak_o_Shadows: Rue, once you have your slip ring/light joint thing together, would you mind sharing a photo?
02:45 AM rue_shop3: np
02:45 AM rue_shop3: alignment will be a snot
02:45 AM rue_shop3: I'm not sure what datarate I will get
02:45 AM rue_shop3: but come to think of it, whats the datarate on irda?
02:52 AM Jak_o_Shadows: TBH, i think you'd trade raw datarate for reliability
02:53 AM Jak_o_Shadows: I mean, say you want 50Hz, and are sending like 128 bytes per packet.
02:53 AM Jak_o_Shadows: that's only 6.4 kbs
02:53 AM Jak_o_Shadows: But if every other packet fails...
03:02 AM Jak_o_Shadows: irda is pretty damn fast. Even the lowest is 9.6 kbits
03:02 AM Jak_o_Shadows: also, when i said 6.4kbs, i meant 6.4 kbits
03:15 AM rue_shop4: hmm
03:15 AM rue_shop4: I wonder how fast I can switch a laser
03:16 AM rue_shop4: NE592 and wien was a fail, 30Mhz, regardless of the components used
03:22 AM Jak_o_Shadows: hmmm. how would you meausre the switching speed? Do you have something that you already have semi-calibrated for receiving?
03:23 AM Jak_o_Shadows: Or just connect a LeD to a scope
03:26 AM Jak_o_Shadows: Apparently irDA you only need to line it up to within 15 deg..
03:28 AM rue_shop4: I need to get the rings made and see how the alignment is
03:28 AM rue_shop4: it'll varry too
03:29 AM Jak_o_Shadows: Is it going to be full duplex, or half duplex?
03:29 AM rue_shop3: each side is half
03:29 AM rue_shop3: two channels on each ring
03:39 AM Jak_o_Shadows: Do the channels to go to different places? What makes it not full duplex if you have two half duplex channels?
03:40 AM rue_shop3: no, each pin is just unidirectional
03:40 AM rue_shop3: so, as a whole, the comms are duplex
03:40 AM rue_shop3: full
03:40 AM Jak_o_Shadows: ah
03:40 AM Jak_o_Shadows: fair
03:42 AM Jak_o_Shadows: https://www.st.com/content/ccc/resource/technical/document/application_note/c7/d1/63/f7/80/06/41/a4/CD00267896.pdf/files/CD00267896.pdf/jcr:content/translations/en.CD00267896.pdf is helpful (for me, as I would stick to using stm32's)
03:45 AM rue_shop3: the AGC in those makes them.... fun
05:12 AM rue_shop3: OK, the GOOD transistors, come out of CRT video amplifier baords
05:12 AM rue_shop3: 500Mhz min.
05:12 AM rue_shop3: a few power 2Ghz ones
05:13 AM Jak_o_Shadows: New leg idea; http://pasteall.org/pic/show.php?id=90cfac3107320f4d0c42e913e56274e8
05:13 AM Jak_o_Shadows: this is the lower leg. The bolted on bit is a force sensor bar
05:14 AM Jak_o_Shadows: https://www.ebay.com.au/itm/YZC-133-Mini-Scale-Electronic-Load-Cell-Weighing-Sensor-1-2-3-5-10-50Kg-100g/191982568921?hash=item2cb30d85d9%3Am%3Ame7xv0T4jSTtY2WgJ3S3HlQ&var=491535453335 like that
05:14 AM Jak_o_Shadows: One of the bar type variants
05:15 AM Jak_o_Shadows: Driven remotely via belt.
05:19 AM rue_shop4: hwo did you manage to switch colours int eh middle of a print?
05:22 AM Jak_o_Shadows: I coloured the filament in with a permanent texta
05:23 AM Jak_o_Shadows: one of your arms, the joint was remtote driven wasn't it?
05:36 AM rue_shop4: oof, 7.5mV in, 1.6V out, happy at 1Mhz
05:37 AM rue_shop4: joint?
05:37 AM rue_shop4: I have a controller for the arms, realtime style
05:37 AM rue_shop4: via serial stream
05:37 AM rue_shop4: Av of 213
05:39 AM Jak_o_Shadows: nah, sorry, i mean the motor - one of the joints was driven remotely.
05:43 AM rue_shop4: I'm working on a force feedback servo if thats what your reffering to
05:43 AM rue_shop4: oh you mean the cable drive?
05:46 AM Jak_o_Shadows: probably the cable drive?
05:55 AM Jak_o_Shadows: Through, looks like this project is getting abandoned for a month, I don't have any belt
07:01 AM rue_shop3: dollar store, beading wire
07:01 AM rue_shop3: http://ruemohr.org/~ircjunk/robots/arm10/p1080879.jpg
07:01 AM rue_shop3: see folder for more
07:01 AM rue_shop3: http://ruemohr.org/~ircjunk/robots/arm10/p1080873.jpg
07:03 AM Jak_o_Shadows: Thanks!
07:07 AM Jak_o_Shadows: The nice thing about hte wire is that you can use bowden cables.
07:22 AM rue_bed: not too much tho, the drag adds up
07:22 AM rue_bed: but I can totally understand why movie people do it, its awesome
07:24 AM Jak_o_Shadows: Just cuts down on your acceleration & increases your max torque right?
07:25 AM Jak_o_Shadows: Doesn't change the max load capability of your thing?
05:59 PM rue_: mrdata,
05:59 PM rue_: two VCO, countermodulated, mixed, amplified
05:59 PM rue_: each modules 5Mhz
06:00 PM rue_: modulates
06:00 PM rue_: total difference 10Mhz,
06:06 PM Jak_o_Shadows: They beat at 10Mhz?
06:06 PM rue_shop4: oh, and there are SAW filters, that can be made into oscillators, that can have 10Mhz bandwidth
06:06 PM rue_shop4: that would be an idea
06:06 PM rue_shop4: I need 1-10Mhz
06:06 PM rue_shop4: its being a pain
06:07 PM rue_shop4: NO the chineese modules WONT do it
06:07 PM rue_shop4: not in a continious variation EVEN THO THEY SAY IT
06:07 PM Jak_o_Shadows: Does getting a stabe pulse out of beating depend on the phase difference betweent the two?
06:07 PM rue_shop4: the next step down from 10Mhz is 8.3Mhz
06:07 PM rue_shop4: Jak_o_Shadows, yea, there is a stability factor
06:07 PM rue_shop4: I still have a few RC based options to try
06:08 PM rue_shop4: using 2Ghz transistors to push audio oscillators to places they were never meant to go
06:59 PM Jak_o_Shadows: ah, piss, i forgot to put the hole for the axle in
08:33 PM Jak_o_Shadows: Hmm, the only problem with te belt driven knee is that you can't go and add any compliance or elastic elements very easily
08:35 PM Jak_o_Shadows: Hmm, and I probably want to sensor the knee joint angle. So need to put a lock screw on the shaft.
08:42 PM Jak_o_Shadows: I suppose I could put an elastic elemetn in line with the belt.
08:42 PM Jak_o_Shadows: I'd 100% need to sensor the knee actual position.
09:28 PM mrdata_: i hammered on oscillator stability for a bit
09:29 PM mrdata_: finding that the output impedance was a bit high,
09:29 PM mrdata_: and the frequency not very stable
09:29 PM mrdata_: so i fixed those issues
09:33 PM mrdata_: https://imgur.com/a/fTJK498
09:33 PM mrdata_: rue_ rue_shop4 : ^
09:35 PM mrdata_: supply reduced to 3V, amp stage added, 50 ohm output
09:36 PM rue_shop4: hmm
09:36 PM rue_shop4: will it actually work tho
09:38 PM rue_shop4: L1 and C1 are the tank parts, right?
09:39 PM mrdata_: L1 and C1,C2 yes
09:40 PM rue_shop4: c2 isn't emitter stabalization?
09:40 PM mrdata_: a major step toward stability was getting Q1 base bias to a nice value
09:40 PM mrdata_: um, well,...
09:40 PM mrdata_: i'm no sure
09:40 PM rue_shop4: why not self-bias it?
09:41 PM rue_shop4: (one resistor off collector)
09:41 PM mrdata_: colpitts topology uses the split capacitors
09:41 PM rue_shop4: y
09:41 PM rue_shop4: 3p for coupling??
09:42 PM mrdata_: ys
09:42 PM rue_shop4: why so small?
09:42 PM rue_shop4: oh, its the only input impedence
09:42 PM rue_shop4: er... ?
09:42 PM mrdata_: i found the first stage output impedance to be about 3k ohm; so, Xc = 1/(2 pi f C ) ---> C ~ 3 pF
09:43 PM mrdata_: for the mid range of 19 MHz
09:44 PM rue_shop4: with no emitter resistor, whats the gain of stage 2?
09:44 PM mrdata_: i'm not sure but it brings the 3k ohm impedance to 50 ohm
09:45 PM mrdata_: which is nicer for mixing
09:45 PM rue_shop4: and doesn't R6<->load cause an average 30mA flow?
09:45 PM mrdata_: more like 20 mA
09:45 PM mrdata_: RMS
09:46 PM rue_shop4: if I make two, maybe the core can shuttle from one to the other
09:46 PM rue_shop4: use that differential mixer trick
09:46 PM rue_shop4: then its got way more than enough range
09:47 PM rue_shop4: suppose you wouldn't be into doing a heartly version of that :)
09:48 PM mrdata_: you would use a product mixer and a 14 MHz fixed oscillator; then you get your 1 .. 10 MHz
09:48 PM mrdata_: after low-pass filter
09:48 PM rue_shop4: or two of those, opposing adjust, mixed with each other
09:48 PM rue_shop4: but yea
09:48 PM rue_shop4: 14Mhz...
09:49 PM rue_shop4: LF drawr...
09:50 PM rue_shop4: 4, 4, 1.84, 1,84,1,84....3.68...6...6.. 8... 10...2.4......hm
09:50 PM rue_shop4: 15....
09:51 PM rue_shop4: suppose we could bump it up a bit to mix with a more common freq... 16Mhz
09:52 PM mrdata_: if you like
09:52 PM mrdata_: so then it would be 17 to 26 MHz
09:53 PM rue_shop4: I have lots of 16Mhz modules
09:53 PM rue_shop4: 16.0000
09:53 PM mrdata_: ok, so then min inductance becomes 2.38 uH
09:54 PM mrdata_: testing...
09:54 PM mrdata_: looks good in sim
09:57 PM mrdata_: so the inductor range becomes 2.38 to 5.5 uH
10:12 PM justanotheruser is now known as swampmonster
10:36 PM mrdata_: this seems quite doable with the gapped coil i linked here, https://imgur.com/a/qNfJiBz with 18 turns and min gap 5.7 mm, max 15.7 mm