#robotics Logs

Jul 12 2018

12:00 AM mrdata: the voltage source can be told to kick a small pulse at the start
12:04 AM mrdata: which helps, too
12:04 AM rue_: did you hear that I got a coupled CDS/LED to control the freq?
12:04 AM rue_: I put it on an op-amp for current control
12:04 AM rue_: hmm transconductance or trans... uh... geez, I get them mixed up even if I remember the names... impedence
12:04 AM rue_: prolly v/I or I/V would make the names make sense
12:04 AM rue_: k V/A or k A/V
12:04 AM rue_: I'm doing A/V, so, my guess is transconductance amp...
12:04 AM rue_: 4.3mA/4.3V
12:04 AM mrdata: give me a few minutes, i'll reconstruct the schematic here and show a screenshot
12:04 AM rue_: sure
12:20 AM rue_: ooo if I get some PMMA side light fiber, I can have one led run several CDS cells at once
12:20 AM rue_: get multichannel with linear feedback
12:21 AM rue_: its $5 and I'v already spent my $5 today
12:30 AM mrdata: https://imgur.com/a/bPvZxAH
12:30 AM rue_: capacitor to ground, resistor directly yo base
12:30 AM rue_: so, right side of R1a to base, C1 from base to ground
12:30 AM rue_: to act as rc delay
12:30 AM mrdata: this still yields an RC delay; the caps also isolate output voltages from bias between the stages
12:30 AM rue_: oooh, right ac couple it
12:30 AM rue_: yea, that would be a dc bias mess :)
12:30 AM rue_: interesting amplitude chart...
12:31 AM rue_: definitly goes to hell not long after 10Mhz
12:31 AM rue_: hows 3.3V / 5V -50db?
12:33 AM mrdata: eh?
12:33 AM rue_: it averages -50db from 1M to 10M
12:33 AM rue_: but its 3.3V
12:33 AM mrdata: 0 dB is 1V p2p i guess
12:33 AM rue_: but its almost 3v p-p
12:34 AM rue_: that should be +1db
12:34 AM rue_: er, something
12:34 AM mrdata: i'm not sure
12:34 AM rue_: er, more than 1
12:34 AM rue_: 1.5ish?
12:34 AM rue_: no, more than... 3
12:34 AM rue_: so like 4
12:34 AM rue_: final answer, honest
12:35 AM rue_: :)
12:35 AM rue_: les than 6db
12:35 AM rue_: more than 3db
12:35 AM mrdata: part of what youre seeing is a sample over a short interval just ass it starts up
12:35 AM rue_: oh
12:35 AM mrdata: i'm running as sample for a bit longer...
12:35 AM rue_: the right chart is a resistor sweep tho, isn't it?
12:36 AM mrdata: no it's an FFT of the output terminal
12:36 AM mrdata: while the pot is set for 16 MHz
12:37 AM mrdata: 0.1 k
12:39 AM mrdata: what matters is the sharpness of the first spectral peak
12:45 AM rue_: oh
12:45 AM rue_: but its for a squarewave output
12:45 AM mrdata: yes
12:45 AM rue_: dont quite get why the peaks are where they are then, maybe its cause its not that sqare
12:45 AM mrdata: it was my understanding thats what you wanted
12:46 AM rue_: its looking good
12:46 AM mrdata: rise and fall times both have RC time contant
12:46 AM mrdata: *constant
12:47 AM rue_: actually, the wave looks quite a bit like on the scope too, there is that funny shadow of a sawtooth in there
12:47 AM mrdata: this affects the slope
12:48 AM rue_: so, I'm going to try to pll it
12:49 AM rue_: for now, against my current signal gen
12:50 AM rue_: mrdata, so, take an xor to two of the stages, (works better with a 2stage ring osc, but whatever)
12:50 AM rue_: get a higher freq, as each stage toggles the xor
12:51 AM mrdata: yeah each stage is the same, just phase shifted
12:51 AM rue_: yea
12:51 AM rue_: so, 2 stacked xors, should get you 3x freq
12:52 AM mrdata: i did odd number of stages to encourage each transistor to switch
12:52 AM mrdata: since this is open-collector output, each npn is a not gate
12:56 AM rue_: y
01:01 AM rue_ is now known as rue_mohr
01:04 AM mrdata: anyhow, 2n3904 is 9 cents each in qty 100 from Newark right now
01:04 AM rue_mohr: well, I'm kinda concerned about my website dissapearing
01:05 AM rue_mohr: 9c ea?
01:05 AM mrdata: yeah
01:05 AM mrdata: https://octopart.com/search?q=2n3904&start=0
01:05 AM mrdata: not quite as cheap from digikey
01:05 AM rue_mohr: https://www.aliexpress.com/item/Free-shipping-100pcs-2N3904-in-line-triode-transistor-NPN-General-Purpose-Amplifier-TO-92-0-2A/32823910408.html
01:05 AM rue_mohr: I dont know what currency you use
01:06 AM mrdata: i guess octopart quotes USD
01:06 AM mrdata: i'm in canada
01:06 AM rue_mohr: .13c in canadian currency
01:06 AM rue_mohr: oh, me too :)
01:06 AM mrdata: k
01:06 AM rue_mohr: er, hmm
01:06 AM mrdata: oh wait
01:07 AM mrdata: no
01:07 AM rue_mohr: 1.31c ea
01:07 AM mrdata: octopart knows i'm in canada
01:07 AM mrdata: C$0.059 in qty 1000 it says
01:08 AM rue_mohr: 0.0131 from china qty 100
01:08 AM mrdata: but here's something odd
01:09 AM mrdata: every aliexpress link you give me redirects to ru.aliexpress.com
01:09 AM mrdata: idk why
01:09 AM rue_mohr: oh, it searches everyone
01:10 AM rue_mohr: oh, yea, they give me russian sometimes too
01:10 AM mrdata: so it tells me in russian; i was going to ask you about that
01:10 AM rue_mohr: click on the thing ont eh top right
01:10 AM mrdata: i maybe should set my preferences
01:10 AM mrdata: k
01:10 AM rue_mohr: huh, so the china prices are even cheaper than the 10k prices anywhere
01:11 AM rue_mohr: aha, way down, diotec
01:11 AM rue_mohr: starts to compare
01:16 AM rue_mohr: ok, so, tommorow I can finish putting the led driver on the osc board, and start building the phase comparator
01:16 AM rue_mohr: I'm protoboarding things so far to get around breadboard capacitance
01:20 AM rue_mohr: want to make more cds based opto units
01:20 AM rue_mohr: I suppose if I mix toner with epoxy I'll get something pretty opaque
01:22 AM mrdata: ok, it knows i'm in canada, finally
01:22 AM mrdata: yeah those are cheap
01:22 AM rue_bed: never had any faults
01:22 AM rue_bed: not with those
01:23 AM rue_bed: trying to work out a way of simplifying this up-down voltage servo
01:23 AM rue_bed: on the phase comparitor
01:24 AM rue_bed: it looks like 2 transistors is the bottom of it
01:24 AM rue_bed: http://images.books24x7.com/bookimages/id_15239/p30001d62g770003vpp_thm.jpg
01:24 AM rue_bed: not really much I can do thats smaller
01:26 AM rue_bed: I did re-arrange the resistors so it cant crowbar
01:27 AM rue_bed: https://m.eet.com/media/1059461/avago2b.gif
01:28 AM rue_bed: neat trick I can do with cds cells, to linearalize the output
01:32 AM rue_bed: http://www.next.gr/uploads/135-8691.png
01:33 AM rue_bed: lots of things latley have me thinking in new ways about the + and - inputs on an op-amp
01:33 AM rue_bed: negitive feedback is just a perspective
10:04 AM z64555: erwhat
11:05 AM mrdata: what
11:19 AM z64555: "negative feedback is just a perspective"
06:45 PM rue_mohr: negitive feedback only exists because there is a single ended output,
06:46 PM rue_mohr: when you have a systemw with complimentrary outputs, suddenly you cant call it negitive and positive feedback
06:51 PM mrdata-: you can call it push-pull maybe; or (a) and (b)
06:51 PM mrdata-: or +- and -+
07:23 PM rue_shop3: I need to make a system that does curve prediction of RC curves
07:24 PM rue_shop3: at a 'part way' mark I need to know the settle time and value
07:24 PM rue_shop3: so I'm thinking I can watch the slope,
07:24 PM rue_shop3: have two known tou values for slopes on an RC curve, know those and do a curve match
07:24 PM mrdata-: tau
07:25 PM mrdata-: these are exponential / logarithmic curves
07:25 PM rue_shop3: thanks, always just used the symbol
07:25 PM rue_shop3: yes
07:25 PM rue_shop3: I need to make a simulator (I think I ahve one) and find some key slopes to use
07:26 PM rue_shop3: hey I never did come up with a good aluminum etchant
07:27 PM rue_shop3: (I'm building a list for week-after-next I'll be on vaccation)
07:28 PM rue_shop3: so far I ahve: restock shop, rebuild pcb etching system, build 12' robot, put solar on house, redo shop audio system, build cnc machine, make force feedback demo, make new tea machine, make reflow oven, and make plastic injection molder
07:29 PM rue_shop3: gonna be a heavy week
09:52 PM rue_shop3: hmm, I think I'll pick on someone
09:52 PM rue_shop3: bolt,
09:52 PM rue_shop3: bolt
09:52 PM rue_shop3: hey bolt
09:54 PM rue_shop3: bolt is a spambot...
09:54 PM rue_shop3: collecting data to use for fake ads