#robotics Logs

Jul 26 2020

02:02 AM mbrumlow: So one of the problems I have in robotics is it takes so long to get everything around.
02:08 AM mbrumlow: I am coming from a software world, were if I just bang out the code, i can see results.
02:08 AM mbrumlow: I am working on my first self balancing robot. And part of my issue is I need to do design a and 3d print a chassis, I need to (at least for this project) learn some kicad so i can make a small board to plug everything into.
02:08 AM mbrumlow: I guess my fear is fear of failure, that I will spend hours and hours, and send this thing off to be fabricated and it comes back all broken.
02:08 AM mbrumlow: Is there any advise on moving forward in these situations?
02:18 AM mrdata: mbrumlow, rapid prototyping is your friend
02:18 AM mrdata: have a version of your circuit you can make rapidly, or a subcircuit
02:18 AM mrdata: that you can test and debug
02:19 AM mrdata: with breadboards or manhattan construction
02:19 AM mrdata: manhattan construction is advisable for high speed
02:21 AM mrdata: anywhere you need to control parasitic properties of the circuit
02:22 AM mrdata: prototypes of mind go from breadboard to perf board
02:22 AM mrdata: and usually stop there because it either works or i abandon it
02:22 AM mrdata: and i only needed 1
02:23 AM mrdata: DIY etching is possible as well; i have done very little of that
02:25 AM mrdata: s/mind/mine/
02:26 AM mrdata: by the way manhattan construction can do surface mount
02:28 AM mrdata: http://www.unixnut.net/files/manart.pdf
02:34 AM mrdata: the effort of doing layout with kicad or eagle pays back when you have to make more than about 3 of what ever thing you designed
02:34 AM mrdata: and you would do that after the first prototype is debugged
03:40 AM Jak_o_Shadows: Why am I getting a 40Hz update rate?
03:40 AM Jak_o_Shadows: I asked for 50
03:43 AM mrdata: what clock source
03:58 AM Jak_o_Shadows: I am using a STM32F0's Timer . So I think that's 48 MHz, prescaler of 4, period of 240000 clocks
03:58 AM Jak_o_Shadows: TIM2*
04:07 AM Jak_o_Shadows: Yeah, it reckons it's 48 MHz.
04:22 AM Jak_o_Shadows: Ok. So if I assume it's 38 MHz, and just deal with the calibration error for now
04:22 AM Jak_o_Shadows: that means I have about 9500 clock ticks per ms
04:23 AM Jak_o_Shadows: So over the range of 1ms -> 2.5 ms, I have 14250 values
04:23 AM Jak_o_Shadows: that's nearly 14 bits worth
04:32 AM Jak_o_Shadows: But if we ignore that for now
04:32 AM Jak_o_Shadows: I just need to port the serial control over to this board
04:32 AM Jak_o_Shadows: then once I respin the boards, I have teensy tiny little PCBs that can control 3 servos.
04:32 AM Jak_o_Shadows: They are small enough to almost fit on a 9g servo, and are chainable too
04:33 AM Jak_o_Shadows: https://github.com/Jak-o-Shadows/stm32SerialServo
04:52 AM mumptai: you might use this for your serial communication: https://github.com/modm-io/modm/blob/develop/examples/stm32f4_discovery/sab2/main.cpp https://github.com/modm-io/modm/blob/develop/examples/stm32f4_discovery/sab2/main.cpp
05:02 AM Jak_o_Shadows: Oh yeah, i'm not concerned about it
05:04 AM Jak_o_Shadows: and also, I'm only using C here
05:04 AM Jak_o_Shadows: tbh, I think I would skip C++ and jump to rust or something
05:32 AM mumptai: just a suggestion
05:33 AM mumptai: its also a multi-drop uart-based "bus"
05:34 AM mumptai: a simple thing to add sensors and actuators to some more central component, what ever that might be, a PC, some MCU or FPGA
12:27 PM rue_bed2: getting replies from devices on a rs232 bus is trickey, even if its just at ttl levels,
12:27 PM rue_bed2: collisions
12:46 PM mrdata: rs232 isnt a bus?
12:47 PM Tom_L: isn't a car either
12:47 PM mrdata: i thought there are just two devices on it, the terminal and the modem
12:48 PM Tom_L: afik that is so
12:48 PM Tom_L: spi can have multiple
12:48 PM Tom_L: as can i2c
12:49 PM mrdata: yes