#garfield Logs

Jan 22 2021

12:06 AM rue_mohr: and I have no idea which way to go
06:13 AM spline: on feeding your cat: we do wet (can) and dry. Have an autofeeder https://www.amazon.com/WOPET-Automatic-Dispenser-Distribution-Recorder/dp/B089GG57QL/ - really nice and I've got a litter robot v3-wifi (also nice)
06:13 AM spline: solves some of the work
07:08 AM polprog: https://polprog.net/rozne1/ircjunk/various/cs_lens.png
07:08 AM * polprog is drawing lenses in inkscape
07:08 AM polprog: https://polprog.net/rozne1/ircjunk/various/lensmounts.svg
07:09 AM polprog: this will probably misalign the text because you dont have the font installed
07:09 AM polprog: but i think im getting better at inkscape
07:09 AM polprog: :)
07:10 AM polprog: i think the circuit board should be green
07:18 AM spline: what are the lenses for?
08:30 AM aandrew: rue_mohr: get yourself converted to what?
08:31 AM aandrew: spline: my dad had an auto-feeder when he went on his first real vacation of his life a couple years ago
08:32 AM spline: it works really well with our dry food and lasts about 2 weeks w/hopper size
08:34 AM aandrew: his was a "rotating pie slice" kind of design for the wet food, it did a decent job keeping the wet food wet which is a challenge
08:45 AM spline: oh yeah seen those and.. just thought it'd be a mess. they're fine with the dry but like both. we have our neighbor come-by
08:45 AM spline: since he's got two cats, too
08:48 AM rue_mohr: I have an auto feeder thing, not set it up
10:15 AM polprog: spline: a CCTV camera, i modified one to take C mount lenses instead of CS mount
10:15 AM polprog: ive got a couple of them and they are all CS mount, so i thought ill use one with the lens i have
10:16 AM polprog: C and CS use the same thread but the flange distance is shorter on CS
10:16 AM polprog: soi just had to move the image sensor PCB back a little
10:19 AM spline: couldn't make a mount? reminds me of canon ef vs r
10:20 AM spline: think canon is 44.5mm
10:29 AM polprog: its easier for me to put some spacers behind the sensor PCB than to machine out an adapter :)
10:29 AM polprog: i dont have any machining skills or tools
11:15 AM spline: ahh gotcha and that's very precise
11:15 AM spline: I've had to deal with that in the past when I had a modified Canon 6D. The "hot" filter was removed so the CCD could see UV/IR
11:16 AM spline: I had to learn a bit about this because you need to put some type of filter (infront of lens, behind before the mirror-box or sensor)
11:16 AM spline: but that's cool - I didn't know what C or CS mount are
11:16 AM spline: used to Canon with EF, now R (mirrorless where they moved to I think 24mm)
11:17 AM spline: and I've got some old M43 lenses from the 60s (Cooke Triplets) I shoot flowers with
11:49 AM polprog: from what i read C mount is what they used on 8mm film cameras
11:49 AM polprog: and CS is a shorter version for CCTV and machine vision (cool stuff)
11:50 AM polprog: a nice thing is that the flange distance (between the lens mount plane and the film/sensor plane) is exactly 5mm longer on C
11:50 AM polprog: so i just had to put some 5mm spacers :)
11:51 AM polprog: and the tolerances are pretty lax to, im sure these CCTV cameras arent very strictly manufactured, the focus is set individually anyway
11:51 AM polprog: and they make great cheap toys
11:51 AM polprog: ill be using this one for laser stuff
11:51 AM polprog: not a problem if i burn a hole in the sensor
11:52 AM polprog: i can get a new one, unlike a pair of eyes...
11:52 AM polprog: they also see IR very well (CCTV kind)
11:52 AM polprog: they need to see in the night time
04:29 PM Tom_itx is now known as Tom_L
07:12 PM rue_mohr: cool
07:14 PM rue_mohr: ok, my demimation filter is dividing its result by 128
07:14 PM rue_mohr: its over-attenuating by -8db
07:15 PM rue_mohr: is that a double negative
07:15 PM rue_mohr: hmm
08:26 PM rue_mohr: http://ruemohr.org/~ircjunk/tutorials/elex/1bitanalog/slide.htm
08:26 PM rue_mohr: I'm playing with a software version
08:26 PM rue_mohr: I'm crashing on the math
08:27 PM rue_mohr: I could do better if I used a dartbaord
08:27 PM rue_mohr: dartboard
08:27 PM rue_mohr: I wouldn't hit the board, but I might hit a nearby book with the right content
08:30 PM aandrew: oh ffs
08:30 PM aandrew: I think my garage door openers/etc are all 315MHz not 433
08:31 PM aandrew: hm, 433 is the only one in canada I thought, 315 and 915 being elsewhere
08:32 PM rue_mohr: damnit, I'm no good at this
08:32 PM rue_mohr: I was dividing the signal by 51, and it was +48db
08:33 PM rue_mohr: so I'm dividing it by 12000 and now its -19db
08:33 PM aandrew: I'm really keen to understand what you're up to, it's jsut ever time I've got myself here with time my brain's fogged
08:33 PM rue_mohr: :)
08:33 PM rue_mohr: so, check out that link ^
08:33 PM aandrew: I'm gonna have to fire up another rtl-sdr for 315mhz, ugh
08:33 PM rue_mohr: slideshow
08:34 PM rue_mohr: I'm doing all that in software, and comparing the rc output with the output of a "decimation filter" hooked up to the comparator output
08:34 PM rue_mohr: at /256, I'm +34db
08:34 PM rue_mohr: I want to be 0db
08:35 PM aandrew: is there a way to get your slideshow to NOT autoload the next pic?
08:35 PM rue_mohr: 34db, /20 and 10^x = 50
08:35 PM aandrew: oh there's a fucking pause button right in the middle
08:35 PM aandrew: how blind am I
08:35 PM rue_mohr: so, my divider needs to be 50x higher than 256
08:36 PM rue_mohr: goes back to saying I should /12000, NO!
08:36 PM rue_mohr: hmm
08:37 PM rue_mohr: last week I was making digital 0 and 1 pole filters, high and low pass
08:39 PM aandrew: yeah I remember seeing stuff about that
08:40 PM rue_mohr: there was a post that caught my eye
08:40 PM aandrew: oh - my son (the one you met) has been making good use of your care package. He's taken a simple mech kit (https://www.amazon.ca/gp/product/B01D37OV2Y), built a crane, and has added a motor to automatically wind and unwind it
08:41 PM rue_mohr: https://twitter.com/twi_kingyo/status/1352288017197588480
08:41 PM rue_mohr: :)
08:41 PM aandrew: now he's struggling to get more torque out of the motor so he can lift things heavier than those rectangular white erasers so I hinted at him about using some gearing or a belt with different sized pulleys
08:41 PM aandrew: poor guy, he's not got a chance
08:41 PM aandrew: he's going to be an engineer before he's 16
08:41 PM rue_mohr: hah
08:42 PM aandrew: I need to hustle and get my garage workshop set up, he's already complaining that he can't use a saw or cut metal in the basement because of mom
08:42 PM aandrew: just wait until I turn him loose with the stick welder, she's gonna have a heart attack when she sees him strike it for the first time
08:43 PM rue_mohr: ok, so I think thats the best I can get the decimation filter to behave
08:43 PM rue_mohr: http://paste.debian.net/1182347/
08:43 PM rue_mohr: this is the code
08:43 PM rue_mohr: it sets up a sine wave, at a freq, then samples it at a rate, and feeds the results to -whatever-
08:44 PM aandrew: my younger one destroyed my nice handy 4lb sledge hammer by taking it outside to smash rocks
08:44 PM rue_mohr: in this case, a 1 bit ADC
08:44 PM rue_mohr: heh, eyewear I hope
08:44 PM aandrew: no, when I discovered him doing that I gave him a lecture on eyewear and gave him some
08:44 PM aandrew: but mom is a killjoy
08:44 PM rue_mohr: it takes the output of the virtual comparator and runs it thru a 'decimation filter'
08:45 PM rue_mohr: but the decimation filters output is awefull
08:45 PM aandrew: interesting. I have to dig into the code a bit to understand but it looks like you're sweeping frequencies
08:46 PM aandrew: it won't matter but why did you vo= before computing comp?
08:49 PM aandrew: this would be easier to follow if you labelled your vars. :-)
08:51 PM aandrew: dammit I'm sure I have another cheap RTLSDR here
08:51 PM aandrew: but where
09:02 PM aandrew: balls, I can't find it
09:05 PM aandrew: yeah - i've had them a couple weeks now. Mpow H19 "hybrid echo cancellation" BT headphones are pretty legit
09:06 PM aandrew: way way way way way way way cheaper ($80) than anything bose, and at least for non-airplane stuff (house noise, fans, even a decent job of TV) noise cancellation is pretty good and doesn't give me that head cold feeling
09:09 PM rue_mohr: yup
09:09 PM rue_mohr: sweep and sample
09:10 PM rue_mohr: Vi is the source sinewave, the "real signal"
09:10 PM rue_mohr: Vo is the output of the RC filter
09:10 PM rue_mohr: because its compounded, its next
09:12 PM aandrew: right, but what is Ph/Pl and Ph1/Pl1. decif[] looks like an array of comparator "decisions" and DFResult the sum of those decisions
09:17 PM rue_mohr: aha
09:17 PM rue_mohr: so I measure the output signal peak-to-peak with a peak detector
09:17 PM rue_mohr: one param for the high peak and one for the low
09:17 PM rue_mohr: the p-p is the difference
09:18 PM rue_mohr: then I do the log thing to get it in db
09:23 PM rue_mohr: but it shows that the decimation filter is awefull compared to a simulated RC
09:23 PM rue_mohr: the cool thing is that there can be an external RC and compartator to the chip,
09:24 PM rue_mohr: you can use a simulated matching RC on the inside of the chip and regenerate the analog value without an ADC
09:24 PM rue_mohr: and that japaneese person used a differential signalling line as the comparator
09:25 PM rue_mohr: aka an fpga diff input
09:34 PM Tom_L: rue_mohr has been stuck on this filter thing for quite a while
09:43 PM rue_mohr: its not stuck
09:43 PM rue_mohr: its moving
10:07 PM rue_mohr: ok, the goal was to find a way to turn a 1 bit audio steam into values
10:08 PM rue_mohr: and I can do that with my RC code
10:08 PM rue_mohr: the decimator isn't a good method
10:09 PM rue_mohr: which is neat cause of how it walk all over the analog/digital line
10:12 PM rue_mohr: the -3db of the decimation filter is bad
11:58 PM rue_mohr: whats that fpga we got