#robotics Logs
Mar 01 2020
#robotics Calendar
01:09 AM rue_mohr: <rue_mohr> http://ruemohr.org/~ircjunk/projects/cnc6/p1150377.jpg
01:14 AM rue_mohr: <rue_mohr> http://ruemohr.org/~ircjunk/projects/cnc6/p1150378.jpg
01:14 AM rue_mohr: <rue_mohr> http://ruemohr.org/~ircjunk/projects/cnc6/p1150379.jpg
01:14 AM rue_mohr: motor driver!
01:27 AM mumptai: http://cjt.users.sourceforge.net/misc/motor_driver/MG_2781_cut.jpg
01:27 AM mumptai: motor driver!
01:34 AM rue_mohr: heh, not gonna take 20A at 30V
01:34 AM rue_mohr: that sure it a lot of discretes tho
01:35 AM rue_mohr: I dont even like the pin size on the to-220s
01:35 AM rue_mohr: oh well
01:36 AM mumptai: its more for around 5-8A max, even though there are way better so-8 mosfet
01:36 AM rue_mohr: are the discretes the fet drivers?
01:36 AM mumptai: yes they are
01:37 AM rue_mohr: and level shifters?
01:37 AM rue_mohr: must be level shifters too, or is it low voltage (12 or less)
01:37 AM mumptai: http://cjt.users.sourceforge.net/misc/motor_driver/gate_drv.pdf
01:38 AM rue_mohr: pnp bipolar!!!
01:38 AM mumptai: iirc, the gate-drv voltage is externally generated (assumption is that there will be many drifers)
01:38 AM rue_mohr: huh, I'd like to time that circuit
01:38 AM mumptai: its not that fast
01:39 AM rue_mohr: I'm winding a transformer with two isolated outputs for floating gate drives for high side
01:40 AM mumptai: both high sides of the have full-bridge?
01:41 AM rue_mohr: both high sides of the bridge have a transformer isolated high side drive supply
01:41 AM rue_mohr: good for more than 25mA
01:41 AM rue_mohr: which, is apparently ok (!?!)
01:42 AM rue_mohr: that pnp arrangement is really hurting my brain
01:42 AM mumptai: average maybe, but certainly not peak for a high-power mosfet
01:43 AM rue_mohr: yea, its feeding a 10uF capacitor
01:43 AM rue_mohr: at about 40kHz
01:43 AM mumptai: any high dV/dt across drain/source will cause parasitic turn-on/offs
01:43 AM mumptai: okay, that micht be okay
01:44 AM mumptai: s/micht/might/
01:44 AM rue_mohr: the pnps are installed backwards thats what it is
01:44 AM rue_mohr: the emitter should be high side
01:44 AM mumptai: yes
01:44 AM rue_mohr: I wonder how those bipolars perform reversed
01:44 AM mumptai: not well
01:44 AM mumptai: at all
01:44 AM rue_mohr: ah
01:45 AM rue_mohr: sometimes you get higher gain, sometimes you get them acting like a current source
01:45 AM mumptai: there is a version of the pcb wit sot23 flipped
01:45 AM rue_mohr: oooh
01:45 AM rue_mohr: hah
01:45 AM rue_mohr: evil
01:46 AM mumptai: yes, was some symbol swap mistake
01:46 AM rue_mohr: hmm, the 2.2nF is confusing, to slow down the siwtching?
01:47 AM rue_mohr: that 10mohm fet probably already has like 15nF itself
01:47 AM mumptai: yes, and it can be used to fix parasitic switching
01:47 AM rue_mohr: it looks like its set up to switch quite softly
01:47 AM rue_mohr: low noise
01:48 AM rue_mohr: its almost that circuit I keep using
01:48 AM rue_mohr: but in my case its known to be good up to 1Mhz
01:49 AM mumptai: the value 2n2 can be changed in the BoM
01:49 AM mumptai: its a standard totem tole driver, just mirrored to use pnp
01:50 AM mumptai: it makes it easier to reference its input to the supply of the gatedriver, not ground
01:50 AM rue_mohr: well there is a circuit that uses a fallback type drive
01:51 AM mumptai: whats that?
01:51 AM rue_mohr: how does google drive just make my webbrowser instantly halt?
01:54 AM rue_mohr: its on my google drive
01:55 AM rue_mohr: here it is on something not google driven
01:56 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150101.jpg
01:57 AM mumptai: ahh, like http://elm-chan.org/works/smc/rs/smc.png
01:57 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150102.jpg
01:57 AM rue_mohr: ^^ 10nF load
01:57 AM rue_mohr: its a popular circuit
01:57 AM rue_mohr: yup
01:57 AM rue_mohr: and its fast
01:58 AM mumptai: for my application "no audible whining" is good enough
01:58 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150099.jpg
01:58 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150100.jpg
01:58 AM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/p1150102.jpg
01:58 AM rue_mohr: the voltage follower is better
01:58 AM rue_mohr: but ~level shifting~
01:59 AM mumptai: also the mosfet is an improvement
02:00 AM rue_mohr: yea, it reacts faster when first stage is a fet
02:00 AM rue_mohr: AND you dont need a base resistor
02:00 AM rue_mohr: score 1 on parts count
02:01 AM rue_mohr: I have a PC817 based opto thats good to 500kHz
02:01 AM mumptai: isnt it obvious, i don't care much about parts count
02:01 AM rue_mohr: I have an ethernet transformer based isolator, but I havn'tbeen able to time it yet
02:02 AM rue_mohr: heh, ok
02:02 AM rue_mohr: a 6N137 is good to 1.5Mhz
02:02 AM mumptai: ir2184 would be a much shorter list
02:02 AM rue_mohr: On this new one I'm building, I'm rapidly just going for HIP8041
02:03 AM rue_mohr: its expensive, but it radically simplifies things
02:04 AM mumptai: simple? https://www.infineon.com/cms/en/product/power/motor-control-ics/intelligent-motor-control-ics/single-half-bridge-ics/ifx007t/
02:05 AM rue_mohr: two L6384 is an option too
02:06 AM rue_mohr: hmm, yea, I'm not confident in that chip
02:07 AM rue_mohr: I cant see it taking 50A continious
02:08 AM mumptai: who claims 50A?
02:08 AM rue_mohr: its in the datasheet
02:09 AM rue_mohr: looking for the "* package can only take 30A"
02:09 AM mumptai: steady-state thermal limit will be lower
02:10 AM mumptai: its 6.5mOhm at 25°C
02:10 AM rue_mohr: maybe its a large package than it looks, but I dont think the pins can take 50A
02:13 AM mumptai: actually the die might just handle 50A, but its running hot, and need unreasonable good cooling
02:13 AM mumptai: i've seen wore claims in datasheet
02:14 AM rue_mohr: I dont see the *
02:14 AM rue_mohr: Current limitation detection
02:14 AM rue_mohr: level HS/LS
02:14 AM rue_mohr: ICLH0
02:14 AM rue_mohr: ICLL0
02:14 AM rue_mohr: 55 77 98 A
02:14 AM rue_mohr: thats a funny one too
02:14 AM mumptai: (50A**2) * 0.008Ohm * 5 K/W = 100 °C
02:15 AM mumptai: so to stay under 130°C we need an environment below 30°C, which is "okay"
02:15 AM rue_mohr: another good point would be "how the hell did you find a trace to take 50A?"
02:16 AM rue_mohr: 8 layer board with vias straight thru to fat traces on each layer?
02:16 AM mumptai: wire-laid pcb?
02:17 AM rue_mohr: a photo of the test setup might be priceless
02:18 AM mumptai: no need to get hung up on that one. it will likely do fine with 10-25A
02:18 AM mumptai: and i've seen way worse datasheet
02:18 AM rue_mohr: my target is 4m ohms
02:19 AM rue_mohr: how are the pins 1.27mm wide, with 0.25mm spacing?
02:19 AM mumptai: just parallel more mosfet, and use bigger packages
02:19 AM mumptai: there are also nice smd ones for high-currents
02:19 AM rue_mohr: its the size of a regular to-220 tho
02:20 AM rue_mohr: but with 7 pins
02:20 AM mumptai: it looks comparable to D²-Pak
02:20 AM rue_mohr: ground and supply are each 1 pin
02:21 AM rue_mohr: so one pin, smaller than a pin of a regular to-220, is supposed to carry 50A
02:21 AM rue_mohr: ....
02:22 AM rue_mohr: Specified RthJA value is according to Jedec JESD51-2,-5,-7 at natural convection on FR4 2s2p board; The Product
02:22 AM rue_mohr: (Chip+Package) was simulated on a 76.2 x 114.3 x 1.5 mm board with 2 inner copper layers (2 x 70μm Cu, 2 x 35μm Cu).
02:22 AM rue_mohr: Where applicable a thermal via array under the exposed pad contacted the first inner copper layer.
01:25 PM -!- #robotics mode set to +o by wolfe.freenode.net