|
Post by tronicgr on Nov 24, 2016 13:58:47 GMT
I 3D printed mounting axle for the magnet as it has to be aligned with certain distance from the chip inside the sensor. I used ninjaflex for the mount to be a little flexible and provide some friction for the testing. The axle that holds the magnet is made from harder PLA filament. First tests with quick code shows it works giving noise free absolute position, but for some reason it spits out the values in increments of 17... have to check things with my logic analyser... Thanks, Thanos
|
|
|
Post by paulg100 on Nov 24, 2016 15:22:11 GMT
Brilliant look forward to seeing more, gotta love these 3D printers Just to be clear the value is being sent to the AMC digitally rather than analog? Also i did more reading online and there seems to be quite allot of posts about using arduino to filter analog pots via software/code, Is this something youve looked into previously? ie software filters for pot signal? or would that cause to much lag or inaccuracy in the position reading?
|
|
|
Post by tronicgr on Nov 24, 2016 15:46:20 GMT
Brilliant look forward to seeing more, gotta love these 3D printers Just to be clear the value is being sent to the AMC digitally rather than analog? Also i did more reading online and there seems to be quite allot of posts about using arduino to filter analog pots via software/code, Is this something youve looked into previously? ie software filters for pot signal? or would that cause to much lag or inaccuracy in the position reading? 3D printers is the best thing ever happened for DIY people. I can design something that can be build by anyone with same exact dimensions. Especially for custom parts, saves on using cnc and other tools. This sensor is totally digital, using Synchronous Serial Interface, which is much faster than normal serial uart. I already implemented software filters for the analog sensors on the amc1280usb, but it cannot fight the strong consistent noise that motors emmit sometimes, thus the need for better shielded wiring. Hopefully using SSI sensors will make analog sensors thing of the past.
|
|
|
Post by tronicgr on Nov 30, 2016 11:45:38 GMT
|
|
|
Post by riton39 on Nov 30, 2016 16:57:50 GMT
The quality on the movements is even better? What is the best solution? 180 ° pot? 360 ° pot with reduction to 180 °? 10 turns POT with reduction to 180 °?
Someone tested the difference in games? The sensations are different?
|
|
|
Post by tronicgr on Nov 30, 2016 17:34:05 GMT
The quality on the movements is even better? What is the best solution? 180 ° pot? 360 ° pot with reduction to 180 °? 10 turns POT with reduction to 180 °? Someone tested the difference in games? The sensations are different? Its not matter of degrees of rotation. The new digital SSI sensors I'm adding in next firmware are noise free. So far it looks like they are faster to read too than the analog sensors where you need time to settle the reading. I would say the digital sensors I show above should make the motion much more smooth if you eliminate mechanical issues like backlash of the gearboxes of the motors. Did you see my post about the string potentiometer using magnetic hall effect sensor? That would be even smoother to use but it's more fitted for linear actuators. Thanks Thanos
|
|
|
Post by paulg100 on Nov 30, 2016 20:42:45 GMT
great work Thanos,
Sorry if this was already clarified but are these sensors compatible with the existing AMC or it needs a new board?
Also were the gears taken from a resource somewhere? or did you design yourself? if so what software did you use?
i read that the nylon materials are good for making gears as its more hard wearing then pla or abs for this use.
thanks
|
|
|
Post by riton39 on Nov 30, 2016 23:23:32 GMT
I was referring to the resolution. What interests me is to have the most details in the sensations in games.
I thought that more track, more details in the transcription of the sensation in games no?
Attenuate or smooth out the defects, also smoothes the feelings in game?
I prefer to find a solution to the noise rather than to smooth the sensor and make the sensations in games too smooth / attenuate.
sorry for my English, it's not easy to explain .
A lot of resolution on the sensor = a lot of sensation in games.
Much analog track = a lot of sensations in games.
Analog sensor = infinite resolution = a lot of sensations and details in games
Digital sensor limits resolution compared to analog potentiometer, true?
|
|
|
Post by hexpod on Dec 1, 2016 7:15:58 GMT
The quality on the movements is even better? What is the best solution? 180 ° pot? 360 ° pot with reduction to 180 °? 10 turns POT with reduction to 180 °? Someone tested the difference in games? The sensations are different? Its not matter of degrees of rotation. The new digital SSI sensors I'm adding in next firmware are noise free. So far it looks like they are faster to read too than the analog sensors where you need time to settle the reading. I would say the digital sensors I show above should make the motion much more smooth if you eliminate mechanical issues like backlash of the gearboxes of the motors. Did you see my post about the string potentiometer using magnetic hall effect sensor? That would be even smoother to use but it's more fitted for linear actuators. Thanks Thanos Hi Thanos,
are you able to get full 4096 resolution with steps of 1 instead of 17?
cheers
|
|
|
Post by hexpod on Dec 1, 2016 7:43:41 GMT
I tested on my small DC motor (12v) and didn't see any noise or unstable position. The PWM signal was clean as well for 31khz and 4khz settings. It could be the Clearpath failing to capture certain pulse ratio. I'll have to test on my actual Clearpath motor to see if I can replicate this. On other testing of the i2c 12bit pwm chip and the ssi sensor, I will be receiving the ssi sensor today, but I'll be testing these by Monday or Tuesday as I'm moving my lab this weekend. Thanks Thanos "It could be the Clearpath failing to capture certain pulse ratio"
could you please help by performing a test with clearpath motor and analog sensor? it's hard to know if the hunting issue (only in some arm positions) comes from the pot noise or the unability to capture properly the PWM signals.
- put the clearpath accel around 16000 rpms - put the P around 26 and move the Platform starting position (with Fabi's tool) from 0 to 100.
You should notice a strong hunting issue on several positions.
Thanks alot!
|
|
|
Post by tronicgr on Dec 1, 2016 11:52:13 GMT
Its not matter of degrees of rotation. The new digital SSI sensors I'm adding in next firmware are noise free. So far it looks like they are faster to read too than the analog sensors where you need time to settle the reading. I would say the digital sensors I show above should make the motion much more smooth if you eliminate mechanical issues like backlash of the gearboxes of the motors. Did you see my post about the string potentiometer using magnetic hall effect sensor? That would be even smoother to use but it's more fitted for linear actuators. Thanks Thanos Hi Thanos,
are you able to get full 4096 resolution with steps of 1 instead of 17?
cheers
Not yet, I was busy the last days submitting a project before deadline. I have to attach my logic analyser to find out if I'm missing bytes during the communication. Its not standard 8-bit packets, it consists of 12-bit values that I have to read at once and wonder if I miss a few bits on start (MSB).
|
|
|
Post by tronicgr on Dec 1, 2016 11:56:20 GMT
I tested on my small DC motor (12v) and didn't see any noise or unstable position. The PWM signal was clean as well for 31khz and 4khz settings. It could be the Clearpath failing to capture certain pulse ratio. I'll have to test on my actual Clearpath motor to see if I can replicate this. On other testing of the i2c 12bit pwm chip and the ssi sensor, I will be receiving the ssi sensor today, but I'll be testing these by Monday or Tuesday as I'm moving my lab this weekend. Thanks Thanos "It could be the Clearpath failing to capture certain pulse ratio"
could you please help by performing a test with clearpath motor and analog sensor? it's hard to know if the hunting issue (only in some arm positions) comes from the pot noise or the unability to capture properly the PWM signals.
- put the clearpath accel around 16000 rpms - put the P around 26 and move the Platform starting position (with Fabi's tool) from 0 to 100.
You should notice a strong hunting issue on several positions.
Thanks alot!
I'll check, thanks for the conditions of failure info. What is the gearbox ratio you have? I'll have to adjust the acceleration accordingly as I don't have gearbox on my clearpath motor.
|
|
|
Post by hexpod on Dec 1, 2016 12:03:58 GMT
"It could be the Clearpath failing to capture certain pulse ratio"
could you please help by performing a test with clearpath motor and analog sensor? it's hard to know if the hunting issue (only in some arm positions) comes from the pot noise or the unability to capture properly the PWM signals.
- put the clearpath accel around 16000 rpms - put the P around 26 and move the Platform starting position (with Fabi's tool) from 0 to 100.
You should notice a strong hunting issue on several positions.
Thanks alot!
I'll check, thanks for the conditions of failure info. What is the gearbox ratio you have? I'll have to adjust the acceleration accordingly as I don't have gearbox on my clearpath motor. it's a 50:1. motor max speed limited to 800rpm
thanks
|
|
|
Post by tronicgr on Dec 1, 2016 21:25:59 GMT
Its not matter of degrees of rotation. The new digital SSI sensors I'm adding in next firmware are noise free. So far it looks like they are faster to read too than the analog sensors where you need time to settle the reading. I would say the digital sensors I show above should make the motion much more smooth if you eliminate mechanical issues like backlash of the gearboxes of the motors. Did you see my post about the string potentiometer using magnetic hall effect sensor? That would be even smoother to use but it's more fitted for linear actuators. Thanks Thanos Hi Thanos,
are you able to get full 4096 resolution with steps of 1 instead of 17?
cheers
OK, got the ssi sensor to read correct the values with steps of 1. The total time to read all six digital ssi sensors is 794us which is about 1200 times per second. I'll be publish the new firmware as soon I have the 12bit PWM going as well, to take advantage of the increased resolution. I have a couple of videos about it, I'll upload them soon. Thanks Thanos
|
|
|
Post by tronicgr on Dec 2, 2016 0:09:18 GMT
|
|