#robotics Logs
Mar 01 2018
#robotics Calendar
01:04 AM rue_bed: did j'[t you say that yesterday?
01:07 AM rue_bed: still too many voices in my head
01:07 AM rue_bed: where are my headphones, I wonder if this laptp goes up to 'cant hear myself think'
01:08 AM z64555: probably does
01:08 AM z64555: largely dpends on the quality of the headphones, really
01:08 AM rue_bed: the problem is that if I damage my hearing i'm stuck listing to me
01:09 AM z64555: yes, there is that. How about your food situation, have you ate recently?
01:11 AM rue_bed: damnit, Its not even close
01:13 AM rue_bed: where am I gonna find a booster amp at 11pm
01:13 AM rue_bed: FFFFFFFFF
01:14 AM rue_bed: I'll have to make a usb powered thing with one of those class D modules
01:14 AM rue_bed: wait, maybe I can just use a usb headphone thing
01:46 AM z64556 is now known as z64555
01:55 AM testbed_: Evening all
01:55 AM testbed_: This isn't strictly regarding robotics but, I'm about to power on a bunch of closely packed LEDs on a board and I'm wondering if anyone knows WHICH part of the an LED typically fails due to heat
01:56 AM testbed_: like, does the lens melt before the diode or does the diode burn up like a fuse
01:57 AM z64555: diode burns up
01:57 AM testbed_: also, this isn't over heating due to over current but rather ambient temp because the close proximity
01:58 AM testbed_: got it
01:58 AM z64555: Diodes put off any heat at all, compared to "normal" light bulbs
01:58 AM z64555: They do put off *some* amount of heat in the form of IR radiation
01:59 AM z64555: More likely the heat that you're feeling is from the resistances in the soldering and traces from the board
01:59 AM z64555: *don't put off
01:59 AM testbed_: gotcha
02:00 AM z64555: hm.
02:00 AM z64555: what color LED's?
02:00 AM testbed_: I think you're right -- the whole board is warming up, I think it's the solder
02:00 AM testbed_: white
02:00 AM z64555: ok.
02:01 AM z64555: So that actually is heat from the LED themselves
02:01 AM testbed_: You're thinking that the broader spectrum white LEDs are emitting more IR?
02:01 AM z64555: White and blue LED's run hotter than other colors, due to their nature. White LED's actually emit in the x-ray spectrum and then use a phospher coating inside the lens to generate white light
02:02 AM z64555: The leads to the LEDs act as its heat sink
02:03 AM z64555: Eh, no, they actually are producing amount of heat, theory is that the emitted x-rays bounce around inside the metals instead of out into the lens
02:03 AM testbed_: hmm, well the leads are snipped short, and soldered in real close rows with one another. Any heat piping going on is probably poor
02:03 AM z64555: They can't escape so they dump their energy into the metal again, in the form of heat
02:04 AM testbed_: gotcha. so, the HIGHER frequency light is absorbed by the phospher, warming it all up?
02:04 AM z64555: no
02:04 AM z64555: It's absorbed by the metal
02:05 AM testbed_: hmm, now I'm really confused. The metal of the leads?
02:05 AM testbed_: or solder, in this case?
02:05 AM z64555: the metal of the diode, which is also the leads
02:06 AM z64555: hm, IIRC, that is
02:06 AM z64555: Solder of course has a bit of resistance in it, which is a cause of power loss and heat generation
02:07 AM z64555: now about the heat sink, you'll want some sort of tape or glue that's thermally conductive, but electrically insulated
02:07 AM testbed_: The amperage was so low, I didn't think it'
02:07 AM testbed_: d be an issue
02:07 AM testbed_: but the high freq light.. ..I hadn't considered that as a factor
02:07 AM z64555: Low freq light, like red, is easier to produce
02:08 AM testbed_: Sorry to treat the room like google but do you know if that typical thermal CPU paste is non-conductive?
02:08 AM z64555: sadly, I do not know
02:08 AM testbed_: No prob
02:08 AM z64555: to clarify about the low frequency light. Low frequency = larger wavelength
02:09 AM z64555: Large wavelength means a larger antenna, so to speak
02:09 AM z64555: Silicon also has an innate red/IR spectrum
02:09 AM testbed_: yeah, I thought Infra-red light could jiggle things up but high freq (like X-rays, you mentioned) wouldn't hardly interact with most substances, and wouldn't generate any heat
02:10 AM z64555: eh, you'd think that
02:10 AM testbed_: I'm having a eye opening night looks like
02:11 AM z64555: x-rays bounce off, or are absorbed, by conductive surfaces
02:12 AM z64555: IF they're large enough
02:12 AM z64555: The duality of light, having both particle and electromagnetic wave properties, makes it difficult to understand at times
02:15 AM testbed_: Well, here was my thinking before I asked the question; you can help me untangle it: LEDs generate light in a narrow spectrum (ie, only visable light, no IR or UV) and that all the heat generated by the LED was due to it's innate resistance as an electrical load. I thought that if and when the LED fails due to heat, it would be a mechanical failing of the diode just like a burnt fuse.
02:16 AM testbed_: I hadn't considered color, or solder or any of that
02:16 AM z64555: I don't think an LED fails actually to heat, just over-amping the junction
02:16 AM testbed_: let alone IR or X-ray
02:16 AM testbed_: hmm
02:17 AM z64555: Heat *does* have an effecton the junction, however, making a bit weaker
02:17 AM z64555: effect on
02:17 AM z64555: Makes it a bit weaker, and makes the light production less efficient
02:18 AM testbed_: well, I'm going to fail two LEDs right now. One I'll over-amp with my bench PSU and the other I'll melt with my heat gun while it's on
02:18 AM z64555: heat gun will just melt the lens
02:18 AM z64555: no
02:18 AM z64555: To do a heat test you need to head the leads
02:18 AM z64555: *heat the leads
02:19 AM testbed_: Power on an LED, then heat a lead with a solder iron or something?
02:19 AM z64555: yes
02:19 AM z64555: uh
02:19 AM z64555: Make sure your iron is grounded
02:19 AM testbed_: :D
02:21 AM z64555: hm, might also try unplugging the iron. Just to make sure that there's no electrical interferance from it
02:22 AM testbed_: Shouldn't be an issue at all. I'll report back in a few minutes
02:31 AM testbed_: Suprisingly hard to make them fail from heat alone, both with the heating of the leads and through just the ambiant heatgun on the lens
02:32 AM testbed_: Just for fun I failed one to overcurrent
02:32 AM testbed_: I'm checking them out under a scope now because they're interesting but, I'm pretty satisfied that I won't have the heat issues I thought I'd have
02:33 AM z64555: Yeah, more likely something else will fail than the LED itself
02:33 AM testbed_: (microscope not oscilascope, sadly)
02:33 AM testbed_: Yep
02:33 AM testbed_: Thanks for being my wingman
02:33 AM z64555: you're welcome
02:34 AM z64555: btw, I do recommend getting an oscilloscope when you can
02:34 AM z64555: They even come in pocket sizes now
02:34 AM testbed_: Dude. I want one so bad.
02:34 AM testbed_: I know, I saw a kit on ebay the other day, I wanted to checkout some reviews but havn't yet
02:34 AM testbed_: do you have one? a pocket one I mean?
02:35 AM z64555: sadly no. I'm still limited to a "beginner hobbyist" pay
02:35 AM z64555: and, uh...
02:35 AM z64555: From what I've heard, the "make your own" o-scope kits shouldn't be used for serious lab work
02:36 AM z64555: they're good for understanding how they work, and some light-duty testing
02:36 AM z64555: BUt when you start needing precision measurements is when you have to bring in the professional grade stuff
02:36 AM testbed_: I figured--due to my own budget constraints, I have to pick between one of those pocket kits, or a scope from the former soviet union
02:36 AM testbed_: yeah, the dream is to buy a keysight, or a house
02:37 AM z64555: hm
02:37 AM z64555: Coulda sworn there was also a PC based o-scope
02:38 AM z64555: It uses the PC as the display, and also uses the CPU for some computations
02:38 AM testbed_: that'd be nice. I'm not sure how much of the cost is just an LCD and some logic. If I could do that with my laptop then..
02:38 AM testbed_: yeah, exactly what you just said
02:38 AM z64555: display electronics isn't the bulk of the pricing, I'm afraid. That's all in the sampling circuitry
02:39 AM z64555: If you happen to have a microcontroller, say, an arduino
02:39 AM z64555: You could theoritically make a crude o-scope with it
02:40 AM z64555: you'd be limited to the clock frequency of the uC, however
02:41 AM z64555: so if you had like, a 2MHz uC, you could do reliable testing up to, hm, maybe 1Mhz
02:42 AM testbed_: At the moment, the only use for a scope I'd have is for troubleshooting some motors
02:42 AM testbed_: I doubt I'd need to see frequencies in the Mega Hz
02:43 AM testbed_: but, I bet if I did get a scope, I'd find a need for it real quick
02:43 AM testbed_: It'd enable me to break things in a more sophisticated way
02:43 AM z64555: the o-scope kits should be good to use, then
01:18 PM veverak: ok
01:18 PM veverak: I don't understand prescale in PWM driver
01:18 PM veverak: prescale = ( CLOCK_FREQ / pow( 2, DRIVER_BITS ) ) / freq - 1;
01:19 PM veverak: I understand why it is based on clock_freq, and why the value is divided by 'freq' (because the bigger the freq, the more frequently it has to send the pulses, the smaller the prescale)
01:19 PM veverak: but
01:19 PM veverak: why divided by 2^DRIVER_BITS ?
01:21 PM veverak: yeah, this fits with what is in the manual
01:21 PM veverak: althoug the manual specifies / 4096
01:22 PM veverak: ( DRIVER_BITS = 12 )
01:25 PM veverak: I don't understand why 'power' of it
02:14 PM SpeedEvil: read the datasheet for the microcontroller in question
02:14 PM SpeedEvil: It is reflective of the actual hardware design
02:14 PM SpeedEvil: usually
02:35 PM veverak: manual == datasheet
02:48 PM polprog: what micro is that by the way?
02:55 PM veverak: pca9685
06:49 PM AchiestDragon: well plan b , since mg996r servos just are not up to the task , have to say not had one where its melted the servo housing as yet , had it bend the internal drive pins , strip the brass gears , break the servo horns and burn out the internal drivers but not melt the servo housings
06:53 PM AchiestDragon: plan b should of been plan a in the first place but i opted to go for servos ,, as rs-775 dc motors and 5mm threaded bar + breaings + 15A (30 peek) h bridge's x18 of is not cheap
06:55 PM AchiestDragon: not to mention the extra batteries for 24 to 36v motor power , although sould give about 40x the torque of the servos , so it should actualy be able to lift itself and walk
06:56 PM * orlock did some grinding and JB-welding last night
06:56 PM orlock: now theres just some minor electronics work
06:56 PM orlock: a bit of cad and printing
06:57 PM orlock: and my secondhand equatorial mount will be in working order with a brain transplant and PC interface
06:57 PM orlock: its already working much better than the old mount
06:59 PM AchiestDragon: kool , what software kstars and lib indi ? or other goto software ?
07:00 PM orlock: the defacto "standard" is the LX200 protocol, which is well documented
07:00 PM orlock: Indi and Ascom both have drivers for the LX200 protocol
07:00 PM AchiestDragon: yea
07:00 PM orlock: so it should work with anything that works with those
07:01 PM orlock: i wrote a partial LX200 emulator
07:01 PM orlock: it doesnt include co-ordinate tracking
07:01 PM orlock: purely movement controls
07:02 PM orlock: so using the arduino removes the need for a handset too
07:02 PM orlock: sort of
07:02 PM orlock: lets me control it via a small python script
07:03 PM AchiestDragon: i had a mead lx10 drive for my scope but all plastic gears and backlash was cronic , i got as far as working out a antibacklash gear system but never got the gear cutting kit for the mill to make them
07:03 PM AchiestDragon: mead / meade
07:04 PM orlock: fork mounted 8" SCT?
07:04 PM orlock: Probably uses the same control protocol as the one i'm emulating - its pretty simple
07:05 PM orlock: :Ms# to move south, for example
07:05 PM orlock: or was it #Ms: ?
07:05 PM AchiestDragon: but after evenings of loding up the car , driving to the top of the mountain here and setting up the kit in -5 temps for most of the night , its just too cold , remote would be the only way realy and not got a perminent site i could use
07:06 PM AchiestDragon: yea fork mounted 8"
07:07 PM AchiestDragon: walk within 2 foot of it and it wabbles
07:07 PM orlock: I'm using a telephoto ense from the 70's
07:07 PM orlock: lense
07:07 PM orlock: 200mm focal length, F4 iirc
07:08 PM orlock: one day i'll splurge and buy a little RC - they have a 6" RC Astrograph for under $500 now
07:10 PM orlock: i'm not happy with the crispness of the stars
07:10 PM orlock: and the chromatic abberation
07:11 PM AchiestDragon: i was using 35mm film asa 1200 and 35 min exposures , so 2 hrs +1 hr setup and 2 hrs for the scope to aclimatise at -5c for 4 pics , sort of decided therers better things to do on an evening
07:12 PM orlock: Damn, so you had a wedge?
07:13 PM AchiestDragon: and its -10c outside with snow here tonight , prob a lot colder on the mountan with wind chill
07:13 PM orlock: See, i just fire up a bash script on the laptop
07:13 PM orlock: and get away from the mount
07:13 PM orlock: but i still have to drift align it to get the equatorial mount in line
07:14 PM orlock: but hopefully that's a once-off
07:14 PM orlock: i just set up in my backyard
07:14 PM orlock: pavers with holes in them to get the position the same each time
07:17 PM orlock: The new mount is a Heq5, which has actual bearings, and motors prorperly attached to the mount body
07:18 PM orlock: the old mount was an EQ3, which had a single motor from the factory, mounted poorly, and no braings, and a lot of slip/wobble/gear noise
07:18 PM AchiestDragon: i got prity good at getting it alligned but the drift sometimes was bad , theres a spot on the mountain about 6 miles away from here sort of a sheltered dip that is good , no light polution
07:18 PM orlock: but it let me test the LX200 emulator code with two motors, and perform tracking/autoguiding
07:20 PM orlock: It seems slightly silly to be complaining about using optics from the 70's on a system with a reasonably modern microcontroller for a brain
07:20 PM orlock: especially considering my employer
07:20 PM orlock: (Leica)
07:21 PM orlock: But we dont actually make that sort of gear, especially not here.
07:23 PM AchiestDragon: anyway i came to the conclusion i needed a better antibacklash worm drive gear , did the cad and calulations so i could stepper drive it but like the £600 for a custom gear or make your own , and have not managed to get the needed gear cutting attacments for the mill
07:24 PM orlock: And it's still a fork mount :)
07:26 PM AchiestDragon: no , , actualy the fork and equ mounts are in bits and the scope is just without one at the moment
07:27 PM orlock: http://somepicturesof.space/TrackingGRRR/crappyPOS01.JPG - example of some of the tracking issues i saw on my old mount
07:27 PM orlock: the stars appearing as mountain ranges is due to the mechanical noise, etc
07:32 PM orlock: work i did last night was replacing the "locating nub" on the tripod
07:32 PM orlock: that's used to adjust the location of the EQ head
07:32 PM orlock: previous owner accidentally snapped it off
07:32 PM orlock: so i'd drilled+tapped a hole, then i grund down a square nut to add some reinforcement
07:32 PM orlock: and JB welded that and a bolt in place
07:33 PM orlock: and some square rack nuts, etc
07:40 PM AchiestDragon: https://drive.google.com/file/d/140vzzR4IusV8Wt8c477HaXQ5n_Dr7q3M/view?usp=sharing
07:40 PM orlock: hahah, awesome
07:40 PM orlock: i feel the pain
07:41 PM orlock: manually tracked? or trails?
07:42 PM AchiestDragon: thats with the meade goto controler tracking
07:42 PM AchiestDragon: andromida
07:43 PM AchiestDragon: at least it was sort of on target
07:43 PM orlock: So the basic "rotation compensation" or "clock motor" as it's sometimes called?
07:43 PM orlock: no active/live input?
07:43 PM orlock: i heard that people who used to use film cameras would try to keep a star in the crosshairs over the whole exposure by using the N-S-E-W buttons on the handset
07:45 PM orlock: http://somepicturesof.space/ - my pics
07:46 PM AchiestDragon: yea onto 35mm film and thats a 30 min exposure ,, if you look though the eyepice with the cam attached to try and aim it you have to pick 3 stars just out of the filed of view and try to get the point in the middle of them wher all you see is black and take the pic
07:48 PM orlock: The only thing i'm unsure about now
07:48 PM orlock: and it's no blockers
07:48 PM orlock: but i'm going to have an arduino nano and a pair of A4988 stepper drivers controlling the mount
07:48 PM orlock: I'd like to have it all socketed onto the one PCB
07:50 PM AchiestDragon: then wait till the next day , use up the remaning 32 exposures taking pics of random crap and get the film developed to see if 1 you actualy got a pic , 2 if it was actualy in focus as focusing on something you cant actualy see is not easy , and 3 if the tracking was ok ...
07:51 PM orlock: Yeah
07:51 PM orlock: Yay for CCD's
07:51 PM orlock: and instant feedback
07:53 PM AchiestDragon: i got a cannon 5d digital ,good cam but bad move , mecanical shutter , extra vibrations to the scope tube , more noise , and no live view while taking frames
07:54 PM AchiestDragon: i did start to make a ccd but gave up basicaly dew to the cold when using it , not good for my athritus , ache for days after sometimes
07:57 PM AchiestDragon: and get cramps in fingers and neck , so not realy a suitable hobby to continue with for me
08:17 PM AchiestDragon: https://drive.google.com/file/d/1P6Ncn0i93I26TUujyrVw7h164Yx8Cz2S/view?usp=sharing cloud/mist from the mountain , the morning after spending all night in the car ontop of it , with telescope set up trying to get pics
08:27 PM orlock: nice pic!
08:49 PM AchiestDragon: but on the rare occasions when you get it right , you get a pic worth the effort https://drive.google.com/file/d/1DkKEHeJZHaVxAw3xLW4DHBe_UfybqmAn/view?usp=sharing
08:49 PM AchiestDragon: or https://drive.google.com/file/d/1-tNBWbkwspNhozrd5IlgDTM2aOGcJh6P/view?usp=sharing
09:02 PM AchiestDragon: found one ,,, pic of scope with goto (fork) mount fitted ,
09:02 PM AchiestDragon: https://drive.google.com/file/d/1xY_uCaCZNVVOOpzXy7fD7xTxD5khtkta/view?usp=sharing
10:53 PM rue_shop3: I spent 3 hours looking for a peice of aluminu
10:53 PM rue_shop3: before I gave up and went with a peice of lexan
10:53 PM rue_shop3: which took me 2 hours to machine,
10:56 PM rue_shop3: and made me realize I should have used a different design, and done it in steel