[moraru]

arisov,

I have a big problem.

First time I tried to power my DC motor with 16v, and all was OK (with this parameters #1-1500, #2-35, #3-6, #4-0,1016, #5-200, #6-1400),
but when we increased the supply voltage to 34v, nothing has worked, whenever I tried to change parameters to find suitable parameters to work well. With other voltages I have not tried (ex. 48v or 120v).
Once I found the motor parameters are appropriate, (#1-80, #2-35, #3-6, #4-0,1016 and 0,3075, #5-150, #6-2500) but only one move in the opposite sense gave servo error.

Another issue that came all the parameters or erased from memory (bank1 to bank4 all is 0), wen I tried to find parameters with 34v.

How to calculate these parameters?
I tried with ELM site information but don't work.

Tanks.

Moraru.

[arisov77]

Visky, if I have understood it is correct translation (beside You there is AtMega88 and You did the scheme from message 13 or 22), that You need the code from 1 messages "Servo_AtMega88_TimerInt".

Programme Dead-Time in scheme with L298 does not need, I earlier mistakenned, since I did not do servocontroller with this microcircuit. And not has attentively studied functioning(working) the scheme . So with code smc2 (ElmChan) on base which works this servocontroller forms on output PWM + Direct. So programme Dead-Time (which I earlier entered) is formed in pause between reverse (the direction) of the motion and accordingly he does not need.

---------
Moraru, "ServoError" appears only when governing from ServoTuning or else when governing from Mach3 (under "correct" adjustment in Config / Motor Tuning)?
If "ServoError" only in ServoTuning, but in Mach (on slow velocity) all orderly - i'm try to explain why this occurs. In servocontroller there is two counters: counter given positions (Pos_ "Purpose" from Step/Dir or ServoTuning) and counter current (real) to positions - (Pos_"Current" from encoder). When control from Step/Dir given in "Pos_ "Purpose" enter gradually, step by step (slowly) and big difference between "Pos_ "Purpose" and "Pos_"Current" does not approach. When control through ServoTuning given in "Pos_ "Purpose" enter immediately (fast). And if beside You PID adjustment are installed thereby that velocity of the rotation of the motor small and acceleration time big, that approaches the moment, when operates protection from unbalance of the positions.
Beside me also was this, but beside I was got this avoid more careful adjustment PID parameter. Possible with Your engine this is not got and then come to settle for only work through Step/Dir.



Also want to note that real on CNC machine beside I well work, only 2 servocontroller on AtMega88 with code Servo_AtMega88_CompInt_Modif.


I much it is difficult to express their own thoughts on english , so if who I have understood, can that my have wrote above to describe and add something its on "clean" english.

[moraru]

Hi arisov,

I quit the project because, in my case, nothing was working.

Sorry for disturbance.

Tanks ,

Moraru.

[krazatchu]

Moraru, don't give up~~

I've got the Mega88 comp-int working for me...

With higher voltages (tension?) you have to be careful with the PCB layout....
You should have a split ground plane for the power side and the logic side, connecting at only one place...
Also, you should keep the gate drive components (the components between the ir2184 and the FET) as close as possible to the FET....

If you take a look a the "HP UHU" thread you will see that designing a high power servo drive is no easy task...
You might even want to copy some, if not all of the "HP UHU" layout features...

A long time ago, I tried building the original ELMCHAN servo drive...
I had to make 2 prototypes before I had one that worked allright at 50vdc...

So don't give up, redesign your PCB, keep in mind the HP UHU, post the PCB before you make it, and someone here will help check your board...
You will learn alot in the process, I know I sure did....

Michael

[moraru]

Hi Michael,

Tanks for you advices.
Probably I have something wrong in my PCB.

Can you take a look for my PCB?
I make with ExpressPCB.

Tanks,

Moraru.

[krazatchu]

I just downloaded ExpressPCB and now I have an urge to format my HDD....
I think you should download the free version of Eagle and use it instead...
Eagle can be a pain sometimes but it's much better than ExpressPCB....

Looking at your board I see a number of things that need fixing...
Here are a few things for starters...

1) 7805 needs some 0.1uF (104) caps as close as possible to the in and out to prevent oscillation...
Personally I try to never use linear regs, they just get too hot for me...
Check the datasheet for notes on capacitor use and placement...

2) Gate components need to be closer to the Mosfets to prevent ringing...

3) You should really have a ground plane for both the power side and the logic side... Attached at only one spot...
This means you really need to make it a double sided board...
Double sided isn't that hard to do at home...

4) You should have 0.1uF decoupling caps as close as possible to every ICs power pins...

5) Is that a crystal unit or a resonator? You really need a proper crystal on there,if it's a crystal, it might need caps (22pf ??) to gnd as well...
Should check the datasheet on the part...

6) Where are the H-bridge diodes? Don't rely on the intrinsic Mosfet diodes...
Use some ultrafast or schottky diodes there...

7) There is not much else I can make sense of without the schematic...
Only the layout is in the zip file (.pcb) ....

As it is, I would expect there is too much noise and oscillation to run it at anymore than 24volts...
I would think the AVR is prolly going into reset and even with the EEPROM precautions in the software, it's likely the stored data is getting corrupted...

Get yourself the free version of Eagle from http://www.cadsoft.de/ ...
Then take a look at the .brd and .sch files I posted a few pages back...
You might also find some other pcb layouts in the ELMChan thread...

Michael~~

[moraru]

Hi Michael,

To many wrongs in my PCB.

1. OK I will install Eagle and I rebuild.

Schematic is arisov without power supply.

2. I will remove 7805

5. I use a resonator

Now I see what my servo don't work more than 14v.

Thanks Michael,


Moraru.

[arisov77]

Moraru, don't give up~~

Moraru, I can for You to do the code, where will are unplugged protection from big unbalance.
If beside You problems with PCB or join then with this code - an engine (motor) will by leaps and bounds revolve, without stop. If will all orderly, that deal in code. So, in many servocontroller, for instance UHU Servo, value of the unbalance - controlled. To possible was optimum to select for each concrete engine.

Join to wordses Michael (krazatchu) - servocontroller with engine on 110V much harder to force it is correct to work. Beside me with 24V too no problems was not until gave 110V...

For increasing of noise-immunity possible wire from encoder do shielded. The shield to connect only on the one hand.
Also possible try to connect the body of the motor with the general wire (-110V) of the feeding. Beside me in this case hindrances from PWM on chain encoder sharply decreased.

[moraru]

arisov,

Tanks, I will redesign my PCB and try again.

Moraru.

[moraru]

arisov,

We have 10 pages,

[moraru]

arisov,

I modify my PCB, please take a look.

Tanks,

Moraru.

[arisov77]

Excuse me, but here (see PCB) I shall not be able You to help, on this no time and I have no desires to install for viewing special programs.

[krazatchu]

Moraru, resonators are terrible for accuracy...
That alone could cause alot of problems with the RS232 interface...

Please use Eagle, it's a great piece of software once you figure it out...
I took a look at the ground plane.... It's not right....
It needs two ground planes... One for the power side... One for the logic side...
They need to be joined at only on place...
As it is now, you will have ground transients and noise from the FETs making problems for the MCU....
Get eagle and take a look at the layout I posted previously... You will see what I mean...

What Arisov77 is saying is that every servo motor will have different PID tuning values...
If the PID tuning values are wrong, then the motor may vibrate or jump around...
When it vibrates it can stress the FETS alot....

He also says, lower voltage is easy compared to higher voltage...
24v is pretty easy compared to 110v, at higher voltage small problems become big problems....

Also, the encoder cable should be shielded, so a 4 wire cable with shield...
The shield should be connected to GND only at the servo controller...
The shield should not be connected at the encoder side....
If the shield is connected at both end, a ground loop will form and that's bad...

Shortly I will post a pic of a partially populated professionally made PCB...

Michael

[krazatchu]

Hey...

Here is the most recent layout, v16...
And a partly populated board...

I just got the boards today from the PCB manufacturer...
I'm too tired to add anymore parts today.... (busy day)...

Mararu, you can see the split ground plane...
The two ground planes are joined at the top....
The logic section (5v) is separated from the high voltage section (12v & 30vdc) ...
Opposite the FETs are the schottky diodes...
Also note the gate drive components are as close to the FET as I could put them....

The power for this board will come from the control board, the power supply is a pair of tps5420's, one for 5v and one for 12 to 15v...
The control board also sports a Mega128 and an AT86RF230 RF transceiver chip, among other things...
The control board will receive distance and velocity commands by RF and produce Step pulses based on a trapezoidal velocity profile ...

The intended use is for an art project, eventually the servo boards may replace the steppers on my SX3 mill as well...

This version is a bit updated from the last eagle files I posted...
If anyone wants the new version in Eagle format, I will upload them...

Michael...

[arisov77]

Michael, good PCB!:banana:
I think that Your variant PCB & Scheme many peopl's needs.

With the code for AtMega8 - you have not earned?
With AtMega88 well works?
You did not try the code from message #88?

[krazatchu]

Thanks~~~

I didn't get it to work with the Mega8...
I think there is an EEprom problem with it...
But I'm not very good with ASM yet....
So I'm not sure....

The Mega88 (comp_int_modif) works perfect...
I haven't tried the code from post #88....

Michael~~

[moraru]

Hi Michael,

Wen I see you PCB I left without words.

Great works, my PCB is garbage, compared to you PCB.

I'm beginner in electronics, but I want to learn.

Tanks,

Moraru.

[krazatchu]

It wasn't so long ago I was a beginner too....
Though, compared to some people, I still am a beginner...
Here are a couple of good links on PCB layout....

Perhaps other ppl can suggest some reading material as well...
http://www.alternatezone.com/electro...torialRevA.pdf
http://www.smps.us/pcb-design.html

Have fun and ask questions if you have any....

Michael

[krazatchu]

Hey...

I just found something that at first look seemed counter-intuitive...

I was testing the driver with a couple of motors....
One motor worked perfect... It has a standard Ailigent encoder...

Another motor, an E543 with a Renco encoder was only giving about 2.7 volts on the encoder output after the HC4093D...
The four inputs on the Mega88 were static as I rotated the motor..
At first it seemed really strange as the Renco has it's own LM339 ...

It turns out the two 100 Ohm resisters on the encoder power line were current limiting sufficiently such that output drive current was too weak ...

So, I suggest anyone with an encoder that has it's own drive output to take note...
You may need to reduce or bypass the two 100 ohms resisters in order to get sufficient current to drive the input on the Servo_CompInt schematic....

Michael...

Edit: I previously tested without the HC4093 but the Renco with LM339 wouldn't drive it either..
At that time I didn't think about the resistors limiting the current...
So had I bypassed them then, it might have worked without the HC4093...
I prolly won't test that theory as it's easier to just inline the HC4093 ...

Michael

[arisov77]

Michael, two 100 Ohm resisters - this R26 on your scheme (I am a review messages #84)?
For firm switching the input node of the scheme of the encoder, possible change supporting voltage LM339 - the resistor R15 & R17, R25 & R28 or change the nominal values a resistor divisor of the voltage (R14 & R7, R23 & R24).
In general input divisors of the voltage (R14 & R7, R23 & R24) possible to exclude and supporting voltage LM339 to install around 2,5V. Originally these divisors of the voltage were planned by me for reduction noise on chain of the encoder. But they have not justified my waiting .

…The control board also sports a Mega128 and an AT86RF230 RF transceiver chip, among other things...
The control board will receive distance and velocity commands by RF and produce Step pulses based on a trapezoidal velocity profile ...Michael...

Possible, in detail about this?