[ivancasten]

Sorry about the attachment on my previous post.

Ivan

[krazatchu]

Ivan,

The 75176 is a differential level shifter... A cheaper version of the max 487...
It's used for multi-drop, long distance data transmission...
As in RS485, or DMX ....

The signal gets level shifted so that each of the two transmission wires is doing the opposite...
IE one wire goes +, while the other goes - ... This cancels out interference, allowing for some long distance transmission...
IIRC, it's something like 1 km... Also good for noisy environments..

In this configuration, it would be point to point, not multidrop...
The disadvantage would be extra wires and extra parts...
It's prolly not necessary unless noise and distance are a concern...

Ciao

[krazatchu]

Ivan,

I used this part a little while back to build a DMX512 8ch soft PWM lighting system...
Looking at the schematic, I pretty sure it's wired incorrectly ...

On the encoder side, the encoder should be tied to input "D"...
Also, "DE" and "RE" should be pulled high not low...
The Tiny2313 side is ok...

Ciao

[Rony.sn]

Hello,

Now the hardware is working.
But tuning is not that simpel .
Is there a guide line to handel that?
One time i have a smooth motor , without any torque .
So useless , changing parameters result in a violent drive .
But also useless, both ways result in an error , and no movement
as it should be.
Maybe someone has a working set of parameters ?
I use a 110w motor , that should be able to do better than the present
steppers .
The motor shaft is coupled with a belt , with 4/1 reduction and a
screw pitch of 5mm.

thx Rony

[epineh]

Hi Rony, I did a walkthrough somewhere regarding tuning, I will try find it, rather than typing it all out again, just have to find it...

Russell.

[ivancasten]

G'Day Rony,
See Russell's post #326 for a walk through re. tuning.
He may have produced another somewhere.
Cheers Ivan

[Rony.sn]

Ivan,

Thanks for the reminder .
Wil try this as soon as possible.
In Belgium it is 10 past 10 pm now , so it will be tomorrow.



grts , Rony

[ivancasten]

G'Day All,

I have two questions if I may.
1) See attachment.
Michael I cut the attachment from your design.
What function does the four diodes serve?
I noticed Emil removed them fom his design.
2) As I'm an inveterate fiddler is the black part of a codewheel
printed on a laser printer opague to infrared eg a H21A1?

I must admit I'm enjoying the tooing and frowing, I'm learning a lot from the various questions and answers.
Thanks to all.

Cheers Ivan

[krazatchu]

Ivan,

1) Mosfets, as a byproduct of their design, already have diodes...
In the mosfet, we call them them body diodes....

So depending on the mosfet used, the body diode might be up to the task of dissipating the voltage spike that occurs when we stop feeding an inductive load and it's magnetic field collapses...

In this case, the motor is the inductive load and it's field is collapsing at the rate of the PWM base frequency...

If the body diodes are not be fast enough to react ... then some fast schotkey diodes would be ideal...
I've seen the BYV27 used frequently, however they were very expensive here so I went with the MUR420...

The other option is using an RC snubber to "slow down" the inductive spike and give the body diodes more time to react....

2) Not so long ago, I tried making an encoder by printing (laser printing) a radial pattern onto a piece of inkjet transparency plastic...

I used a couple of paired IR beam break sensors...
And to answer your question, yes the black toner printed onto a piece of clear plastic was sufficient to trigger a LM339 comparator and give a nice square wave output...

Attached is a pic... this had no specific purpose, just for fun...

Ciao~~

[ivancasten]

Michael.
Thanks for the informative reply.
I also do lots of things without purpose but to add to knowledge base.
Ivan

[Rony.sn]

Hello,

The body diodes in the IRF 540 or 530 are not fast enough to kill the spike's.
It realy makes a difference ading 4 diodes byv 27.
By trying an learning i discoverd that.
So in my aproach of the controller i use extra diode's and the R C filter.
Good luck,

Rony

[youthreewire]

Hey,

Good to hear there is improvement...
Next thing I would try is to split your ground plane with a single connection point between the signal side and the power side...
The continuous ground plane can carry interference...
In audio designs you will often see star grounding where all grounds terminate at a common point...

Opto-isolate the drivers from the Tiny2313? Interesting idea, I think with proper layout, it shouldn't be necessary...
What voltage are you running on the power side?

The solution to the eeprom problem is prolly somewhere in the thread...
There is modified code that moves the eeprom storage address further down...
If you can't find it, let me know...

In total, there are 4 versions, I think....

1) Original Source courtesy of Chan
2) Modified to start in position mode
3) EEprom address change/fix
4) Added "D" functionality courtesy of CrazyIvan via Russell

I really didn't want to post my version, not because I don't want to help, but because it's far from perfect....
So... use it as a reference and hopefully someone more experienced can provide some pointers how to make it better....

Anyhow, attached is my version... Eagle format...
Package changes will likely be required depending on what parts you have...
The power connector is too big, I made a new one but can't find the library now...
At some point in future, when I get a lathe to convert, I will likely do an all SMD version...
But that may not happen for a while....

EDIT: Posts 313 and 319 have the most recent versions...
Althou, I'm not sure if the EEprom fix was carried forward to Ivan's version...

EDIT: zip file was without sch, fixed

Hey,Could you please post the corresponding hex file (and even source code).I want to build my own servo controller for a school project.

[youthreewire]

Hi, I want to build a servo controller and the entire servo mechanical setup.I have experience in AVR programming and embedded circuit making.Could someone please suggest something.(In terms of the encoder and possibley a wrking hex file for an AVR like ELM chan's SMC3a,or atmega8 or atmega128 version of servos).Please help me out.

[Rony.sn]

Hello all,

Already tried the opto way .
But that is not a perfect solution.
Looking at the short guideline to tune the drive ,
i was able to get some movement in a behaved way.
So that is progres .
When i try to use more than 35v as motor supply , the controller still
goes crazy.
So those fet drivers act like a radio station i think.
There are still other ways to achieve victory.
Don't intend to gif up .
I would like to use between 60 and 70 v , so still a way to go .
As for youthreewire , if you go down this thread, you will find the code for the atiny .

Rony

[youthreewire]

Hi,

I have seen ELM chans link and the code as well which is in .asm (I am new to .asm programming and cannot make anything out of it).I want to build a servo using a quadrature encoder (mechanical resistive and not an optical encoder).So what are the changes do I need to make in the code.I know that in RC servos any potentiometer can be used for position feedback.But in case of quadrature encoder there is a unit "ppr" or "cpr".So does this value effect the code and design of the servo?

Ron I have downloaded many hex and code files but exactly don't know which one is working properly.And setting motor parameters and stuff.Please help me.

[Rony.sn]

youthreewire,

The code's found here are indeed working.
Changing the source code (asm file ) is also beyond my capability's.
The source code need to be compiled into an hex and een eeprom file.
Already compiled files can also be find in this comunity.
You have to load the file's in your controller .
For that purpose i use ponyprog , and a self build programmer .
Quadrature encoders produce minimum 2 square wave signals with a 90° phase shift .
If you inted to use a different kind of encoder , than an optical , if you got the same output signals , it should work .
Encoders are specified by the number of counts per revolution ( cpr )
The more pulses per rovolution , the more precise you can position .
On the other hand , more pulses need to counted by the controller .
So eventualy slowing down the drive .
I think most used are a few hundred cpr .
Can you share your alternative encoder approach ?
Could be interesting .

Rony

[youthreewire]

Rony,

Quadrature encoders need not be optical,they can even be mechanical resistive.As shown here :
http://www.robotroom.com/Counter5.html.I am looking for a good source online to buy these.I think they are some common electronic components found in radios.They look like potentiometers but have 4 terminals.There is even one more alternative to use these contactless magnetic encoders:

But mechanical resistive would be a cheaper option.

[krazatchu]

I would avoid the mechanical ones, there will be numerous issues...

1) they will wear quickly...
2) they will be noisy, necessitating at least a comparator...
3) they are generally very low count (20 ppr) ...

There is also the option of capacitive encoders....
I have never used them but, they are cheaper and offer similar performance to optical encoders...
They also have the added benefit of of being impervious to dust...

Perhaps Russell can relay his experience with the capacitive type...

[epineh]

Perhaps Russell can relay his experience with the capacitive type...

Yup, no problems, the ones I have cost about $25.00 bucks US and have selectable resolution from 96 to 2048 CPR and work very well.

They also have different sized sleeves for mounting on different sized shafts, very handy indeed.

I have had them working on my router for about 10 months without issue.

I am not sure if youthreewire is trying to make a CNC servo or RC servo here, if it is an RC servo the ELM obviously wont work.

Russell.

[hesham morsy]

Rony,

Quadrature encoders need not be optical,they can even be mechanical resistive.As shown here :
http://www.robotroom.com/Counter5.html.I am looking for a good source online to buy these.I think they are some common electronic components found in radios.They look like potentiometers but have 4 terminals.There is even one more alternative to use these contactless magnetic encoders:

But mechanical resistive would be a cheaper option.

http://en.wikipedia.org/wiki/Rotary_encoder (nuts)

http://zone.ni.com/devzone/cda/tut/p/id/4763

http://www.seattlerobotics.org/encod.../Mousebot.html