[mcm_xyz]

Hello again!

I've made some additional optimizations in code and got the PID to work at 9.12 kHz! very good...
Also there is a new procedure in place allowing a 5 % decrease in power when motor is in stopped position making this way the vibrations induced by the integral windup ZERO!!! so there are no more unwanted vibrations when stopped.
Also there is no decrease in actuation performance, the servo response curve is near perfect, only 13% overshoot and 52 ms settling time for a 100 steps jump.

I will come back with pictures and videos of my new router when a new encoder for the third axis will arrive. I accidentally connected it wrong and smoked it... shame on me...

For now the new firmware...

Mihai

PS. I see a growing interest in this project. Rest assured, it will newer transform in a commercial one. But I am interested for a better design in the schematic and board. If someone could improve it, this could be a nice addition. Also it would be nice to separate the inputs for step / dir / enable with optocouplers and maybe even the encoder, given the advantages and the low price for such circuits.

[grabik]

do you test in code something like this to prevent anti windup


If(PWM_out {
process integral calculation
}

[mcm_xyz]

do you test in code something like this to prevent anti windup


If(PWM_out {
process integral calculation
}

Yes, simple protection like this is also present. However if the motor refuses to stay in a certain position when power is not applied, because of the magnets inside, then we have a problem. Something had to be done to be able to stop.
Check the code.

Mihai

[mcm_xyz]

3 servo drives working on my new router. Max speed you see is 10 meters / minute.
System seems to be ok. After a hour long program zero position is kept, so no steps lost.
Enjoy

Mihai

https://www.youtube.com/watch?v=T7vj81t6yPk

PS. Video is public now.

[jdholbrook]

Hello Mihai,

That's great news. I can't wait to get mine working.
I tried to watch the video but it says "Private".
Would really like to see it.

Thanks
James

[H500]

Looks good. It would be interesting to see the maximum acceleration /deceleration capabilities once the parameters are optimized.

[jdholbrook]

That does look good. Nice and smooth.
Totally the reason for getting away from stepper drives.
Can't wait!

James

[boldive]

Excelent work! Next step is test it under a load.

Mihai, what kind of encoders you are using, model number?
Thanks

[moghul]

hi
i want to make this but here atmega 644 is not available. but i can found atmega 2560.
help me for this.
also available atmega 88 n 162.
shabbir

[mcm_xyz]

Excelent work! Next step is test it under a load.

Mihai, what kind of encoders you are using, model number?
Thanks

http://www.protehno.ro/rotary-encoders/module/1161169/

[mcm_xyz]

hi
i want to make this but here atmega 644 is not available. but i can found atmega 2560.
help me for this.
also available atmega 88 n 162.
shabbir

I just looked over atmega 2560 datasheet. It has all the features needed. Lots of pwm outputs, adc, uart, external interrupts. You just have to change the pin out, maybe some additional initializations.
You can use 6 pwm outputs in 16 bit mode so in theory you can make the movement smoother.

Mihai

[moghul]

Hi Mihia
Thanks for reply. Now I start study of Ur last firmware.
One question more.
All 3 phase a.c. servo drives use commutation feedback must. But in here it is not used.
I m feeling after watching Ur video there is no problem in this.
Thanx
Shabbir

[mcm_xyz]

Somebody asked me about acceleration... Check this out. Movement on X axis (left to right) in the middle of the movie is done from limit to limit so practically deceleration is enormous.I also have some other changes in firmware and servo tuning program. I will post asap.

Mihai

https://www.youtube.com/watch?v=AuhWhJ6WaOs

[mcm_xyz]

Hi Mihia
Thanks for reply. Now I start study of Ur last firmware.
One question more.
All 3 phase a.c. servo drives use commutation feedback must. But in here it is not used.
I m feeling after watching Ur video there is no problem in this.
Thanx
Shabbir

No. No back EMF or Hall Sensors are used. As I explained before, the motor is initialized first by applying power to 2 phases, then it is kept in phase by the encoder. So very important to know is: Encoder PPR and motor Poles.

Mihai

[mcm_xyz]

New firmware in place. Ver 1.5
Major changes: Enable / Disable is done by interrupts now.
Stop position stabilization is done by decreasing ki and kp values.

Servo Drive Tuner
You can see the encoder register now if you want.

Mihai

[H500]

Here is a very low cost 400 ppr encoder that might be suitable for this project.
Encoder 400P R Incremental Rotary Encoder AB 2 Phase 6mm Shaft 5V 24V W Coupling | eBay

[Larken]

Hi Mihai
One simple thing you could add is a voltage divider to one (or 2) of the A-D pins to be able to measure the power supply voltages for "Low Voltage detection".
This lets you only enable the output if the voltage is over 15 volts , because if the voltage dips lower, then you will blow the mosfets if they are under biased.
Also you can measure the voltage rise during braking and fault if it is over a limit.
It also tells you if main power is on.

Larry

[boldive]

Here is a very low cost 400 ppr encoder that might be suitable for this project.
Encoder 400P R Incremental Rotary Encoder AB 2 Phase 6mm Shaft 5V 24V W Coupling | eBay

It looks like the shipping cost can change the story.

There are few available encoders for reasonable price:
non-contact, magnetic AMS AS5040-ASSU
or something like CUI INC AMT312Q-V

Magnetic is very cheap and has high resolution.

[H500]

It looks like the shipping cost can change the story.

That vendor sold out, possibly because I linked to their site. Shipping was free when they had stock. Several others are offering the part at the same cost with free shipping.

The magnetic chip looks interesting, but I'm not sure how accurate it is. I think they are using an estimator algorithm to increase the reported resolution, like some of the CUI's. Any lag would making them unsuitable for some applications.

The low cost Chinese units might be doing the same also. If I don't already have 10 HEDS 5500's lying around, I would purchase one just to test.

[mcm_xyz]

Hi Mihai
One simple thing you could add is a voltage divider to one (or 2) of the A-D pins to be able to measure the power supply voltages for "Low Voltage detection".
This lets you only enable the output if the voltage is over 15 volts , because if the voltage dips lower, then you will blow the mosfets if they are under biased.
Also you can measure the voltage rise during braking and fault if it is over a limit.
It also tells you if main power is on.

Larry

Interesting ideas. Noted. it's easy to implement in future versions.

Mihai