My servomotor controller for hobby CNC machines

[ajv2803959]

Thanks dubble.
I think that tou have a good and innovating product.
Could you made any recommendation about good servos and the technical characteristics in order to get the maximum performance from the system

Thanks in advanced

[dubble]

Could you made any recommendation about good servos

Thanks for your appritiation!
What do you mean under "good servos"? Servomotors or good tuning method?

[ajv2803959]

Dubble I mean Servomotors.

Thanks in Advanced

[tyutyi]

-

[dubble]

Hi,

Good quality Brushed DC servomotors has the following:

- long motor armature
- lower coil inductance is better
- high stall torque
- linear Current vs Torque curve diagram.
- linear Voltage vs Speed curve diagram.
- replecable brushes (longer lifetime, about 1000 - 2000hours continious work/brush life)

These things makes a servomotor good performance.
The industrial servos with good parameters are much more expensive, than standard DC motors, but a medium quality DC motor is usable with the controller.
You can find small DC motors in newer printers, etc. These don't have much torque, only usable with gearheads to get the torque required for a CNC machine.

A good performance high torque servomotor looks like this:

[kmcwhq]

Dubble,
How is the drive mounted?
Should the aluminum plate be a bit bigger with screw holes for mounting?

How do you accept payment on an order?

dk

[dubble]

Hi,

How is the drive mounted?
Should the aluminum plate be a bit bigger with screw holes for mounting?

There are 4 holes at the bottom of the 5mm Alu-case with M5 female threads in them. You can mount it to a plate or whatever from behind.

How do you accept payment on an order?

We accept T/T direct bank transfer only at the moment. Please ask info on shipping and payment in private: dubblecnc(at)gmail.com

[FPV_GTp]

Hi dubble

in reference to reply #24

Good quality Brushed DC servomotors has the following:
- long motor armature
- lower coil inductance is better
- high stall torque
- linear Current vs Torque curve diagram.
- linear Voltage vs Speed curve diagram.
- replecable brushes (longer lifetime, about 1000 - 2000hours continious work/brush life)
These things makes a servomotor good performance.
The industrial servos with good parameters are much more expensive, than standard DC motors, but a medium quality DC motor is usable with the controller.
You can find small DC motors in newer printers, etc. These don't have much torque, only usable with gearheads to get the torque required for a CNC machine.

I'm aware there are difference in quality in industrial purpose built servomotors as apposed to permanent magnet DC motors that are commonly found in and around the house in toys , photocopiers , electrical appliances , automobiles ie electric window motors , wiper motors , ABS-motor and so on. With a bit of modification to some of these motors they can be used for hobby-cnc purposes. Like fit proper roller bearings , make new ends for the motors for the purpose of mount things and so on .


My question is , can your driver control 12Volt permanent magnet dc motors ??


I am under the impression these motors run higher currents through them .

These motors are cheap and easily obtained , and the amount of fabricating and machining we do to make a cnc machine a little modification of a windscreen wiper motor or a ABS-motor will not be a issue considering the amount of work that goes into a hobby project.

Since this is a hobby , people have limited funds to make a cnc machine and can not afford to buy commercial servo motors .

I am sure many of you guys have priced commercial small servo motors from manufactures and fallen backwards at the prices there asking for them.

I know one can be searching on ebay for the rest of there life for cheaper alternatives , or source through surplus stock shops for bargains , but that makes one hobby very time consuming in wasted effort in searching. And i am sure there have been many people burnt buying secondhand servomotors from half way around the world to find that the servomotor is faulty or is too small or large for the intention they had in mind to use the servomotor.

So my thoughts are why not just modify a available motors that are sitting around us everyday , very cheap , easy to modify , plentiful and so on.

We have wreckers ( auto dismantlers ) around us , which one can buy a windscreen wiper motor for under $10.

Here is a link to one of threads i started
http://www.cnczone.com/forums/showthread.php?t=34602

what is the link to your website ?

It is a pitty http://www.google.com/language_tools?hl=en doesn't have a Hungirian to English translation website feature.

Let see what else is out there for website translations.

cheers

[FPV_GTp]

Hi dubble

I found you website http://dubble.extra.hu/ , i just followered your link http://dubble.extra.hu/hun/whale/seroconfig.zip

now i have to learn how to read Hungirian


cheers

[FPV_GTp]

hi dubble

have found a txt translation program that will do Hungarian to English

It is time consuming , as you can only cut and paste a certian number of txt

seems like some of the resolution in the sentance is lost but one can make out what is being siad on you website

if anyone else wish to use this program here is a link http://www.foreignword.com/Tools/transnow.htm

If anyone finds a website that can convert a complete website pleae let me know , otherwise im still searching

cheers

[dubble]

My question is , can your driver control 12Volt permanent magnet dc motors ??

Yes, you can use a cheap PM DC motor. There are many PMDC motor manifacturer on the web, which makes motors with nearly linear P-I and v-I diagramms, for only some*10$s . Motors, which can be found in printers are a good choice, because they used for nearly the same way, as in a CNC machine, so they are good for a hobbyCNC project. The only problem with them is that they have low torque, so you must use speed reducers on it to get the required torque to drive a CNC machine.

The video, I put on the thread before is also with a PMDC motor, and as you can see, it runs well. The encoder hacked up to the back of the motor.
We tested some other cheap motors, also, like motors from a waterpump with a very high torque, and it runs good aswell. Those high torque, wide armature motors seems to be the most tolerant to PID values change, but they are 'lazy' can't accelerate fast...

The difference will be, that an industrial high price servomotor is able to speed up from 0 to maximum RPM in nearly notime. A $10 PMDC motor will need some milliseconds to speed up. Because it's reaction time is not as good as an industrial servo's. So you have to pull down the acceleration in your controller software a bit, but I think it's not a problem for the 99% of the users. An other difference is that an industrial servomotor can give about 10X it's rated torque when it starting to move.
As you can see on my video an old PMDC motor can be used for hobby purpose, the controller is able to drive PMDC motors.

I found you website http://dubble.extra.hu/

Yes, it's our website The english translation is in progress, but our time is limited, so it will take some time to complete... the only interesting on our site for non hungarian people are the videos and pictures, I think, because you are not able to read the text.

[rkonnen]

I have been told that the PID autotune on other drives do not do a very good job. Does yours tune preatty cose or is just a good start point, then fine tune manually? Also found your English site www.cncdrive.com
Thanks,
Richard

[dubble]

Hi Richard,

You are right, PID tuning is the hardest part of the controlling mechanism. There's no 100% tuning method, which can support all kind of motors. Let's say, that the autotuning software can tune the system to 80% performance. Ofcourse this 80% is for industrial servomotors, which has linear changing parameters, which curves I listed above.
For non linear plants, you must tune the system manually, the automatic tuning is only a good starting point, as you said. I have heard that in industry, the systems running PID algorithms are tuned about to 70% performance in avarage, witch is enough to get a good enough result, so the worker says, it's well tuned

Also found your English site www.cncdrive.com

Yes, it's our domain. It will be our website, but it's under construction at the time, so the contents of the site may change in these days, so sometimes you will not be able to reach the site. That's why I didn't give a link yet to that URL. Until we complete the website you can download the documents, from the previous links above... from our hungarian language website.

[rkonnen]

These are some features used to compare other servo drives. Would you mind suppling information to these catagories so that everyone has a better understanding of your drives.

Control loop type i.e...(32 bit PID ,8 Bit PID ,Analog Gain Dampening)
Error lock range i.e...(Programable 64-2000,1000 counts fixed,127 counts fixed)
Over-current warning output (non fatal)
Position error warning output (non fatal)
Over current shutdown output (fatal)
Output for remote error LED
Multiple configurations stored
Programable sample loop time
On board Fuse
Replaceable drivers and MosFets
Differential Encoder input
Opto-Isolated power and control sections
Maximum recomended encoder cnts
Separate voltage regulator for Encoder
Main onboard capacitor size

Thank you,
Richard

[dubble]

Hi,

Control loop type i.e... (8 Bit PID)
Error lock range i.e...(Programable 1 to 100)
Over-current warning output = Over current shutdown output (yes, with shutdown LED and error signal output shuts down other controllers connected onto the error line)
Position error warning output (yes, with shutdown and error signal output to shutdown other controllers connected onto the error line)
Output for remote error LED (yes, 2 error LEDs + error in/out line...)
Multiple configurations stored (1 config can be stored at a time in onboard EEPROM)
Programable sample loop time (yes, (1 to 255)*60usec)
On board Fuse (No, motorpower fuse must be in powersupply, digital power supply range is 8..30Volts)
Replaceable drivers and MosFets(The power stage is isolated from the digital part of the controller with replaceable optocouplers, the USB and the step/dir inputs are also opto-coupled, mosfets are in to-220 package, so they can be easily replaced)
Differential Encoder input(no diff. encoder input, only connect "+" lines of differencial encoder)
Opto-Isolated power and control sections (yes)
Maximum recomended encoder cnts (it depends on the motor's type, reaction time.)
Separate voltage regulator for Encoder(yes, and built-in noise rejection with 3 samples per encoder period to avoid false counts. with Ts=1/2.5Mhz sampling time)
Main onboard capacitor size( I don't know what does it mean)

BRS,
Balazs

[ESjaavik]

Quote: "Patch cable = UTP cable with RJ45 connectors on it! We're using UTP cable and yes, it's flexible enough It's a CAT5 UTP cable, if you think in it deeper, it is used in 100Mbit ethernet connections, it's up to 100MHz data transfer speed without error. If you use Mach3, the frequency is maximum 45kHz, it's much slower, so it's safe.

Yes, I meant 10 meters/axis. If you're using e.g. a 2m printer cable from PC to breakout board, then 8meter UTP/axis. I think you understand it from the example."

Your user manual state that you use white/orange for step and orange for direction. Those are two of a twisted pair in a standard UTP patch cable. Not a lucky choice IMO. I could give you the long version, but will try the short one first:
The signal should have the 2 wires of a differential pair twisted.
If using single ended, the signal and common should be twisted. In this case connect at both ends so no current loops can be created through the multiple commons.

You might consider using a slower (cheaper) optocoupler on the step/dir inputs. Max input frequency*f, where 2
What do you mean by "modified PID"? Modified in what respect?

What microcontroller is used?

I assume the capacitor asked for is the PWM reservoir. There must be a capacitor close to the driver transistors that can deliver the current for the fast rising current pulses in the short timespan before the external capacitor can get it's current in through the inductance of the PCB traces and external wires. This capacitor is one that is specified for high ripple current operation.

[dubble]

The signal should have the 2 wires of a differential pair twisted.
If using single ended, the signal and common should be twisted. In this case connect at both ends so no current loops can be created through the multiple commons.

It's not a 100Mbit/s connection, so as I said before for short distances this kind of wiring seems to be ok. But a patch cable is easy to make, so if you have doubts, you can put RJ45 male connector onto the UTP cable as you like I mean you can change wiring.
If I like, I could make a long version, with RS485 line drivers, but then the circuit would be more expensive and I think most of the people don't need a 100m connection.

You might consider using a slower (cheaper) optocoupler on the step/dir inputs. Max input frequency*f, where 2<10 is usually a good choice. Choose f closer to 10 if sampling input.

It's an interrupt input, not a smpling... I know the theory . But there are controller programs, which generates only short step pulses, even if the frequency is low. ,like 1usec wide step pulse and then the opto should handle a minimum of 1Mbit/s frequency (1/1usec)! In other words the rising time of the opto should be<1uSec. You are right, that's an expensive opto 2in1 package, DIP8, which is about $4.

What do you mean by "modified PID"? Modified in what respect?

If you learnet controlling theory, you might know, that PID is only a general algorithm. There are several modified PID algorithms like P^2 alg., proportional to I alg. (I have learnet controlling theory in my language "hungarian" so maybe the english translation of these names are not correct, sorry). The algorithm is a bit optimalisated to control motors, I don't want to go into details...

What microcontroller is used?

One of the PIC18f series, I don't want to go into details, because maybe you know what happend to gecko, it had been ripped off by another company.

I assume the capacitor asked for is the PWM reservoir. There must be a capacitor close to the driver transistors that can deliver the current for the fast rising current pulses in the short timespan before the external capacitor can get it's current in through the inductance of the PCB traces and external wires. This capacitor is one that is specified for high ripple current operation.

Oh, I see know, what do Mr.Konnen asked about, there's no cap. but the traces are wide enough on the PCB, so it has a low inductance and resistance, so it's enough to connect a capacitor into the screw terminals. But the circuit can work without it, as the tests shown...

[ESjaavik]

It's not a 100Mbit/s connection, so as I said before for short distances this kind of wiring seems to be ok. But a patch cable is easy to make, so if you have doubts, you can put RJ45 male connector onto the UTP cable as you like I mean you can change wiring.
If I like, I could make a long version, with RS485 line drivers, but then the circuit would be more expensive and I think most of the people don't need a 100m connection.

I'm not concerned about bitrate. Most drivers can output max 300KHz anyway. The concern is that the cabinet is a hostile EMC environment. Check the theory on why a twisted pair is used in almost every (copper segment) data communications line in use.

You are right, that's an expensive opto 2in1 package, DIP8, which is about $4.

And if not needed, those $4 x number of units comes right off your bottom line.

Re: Control algoritms and MCU, I was just curious. Of course you have the right not to tell us.

Oh, I see know, what do Mr.Konnen asked about, there's no cap. but the traces are wide enough on the PCB, so it has a low inductance and resistance, so it's enough to connect a capacitor into the screw terminals. But the circuit can work without it, as the tests shown...

In theory (and my experience) even a short wire/trace of high cross section should be considered an inductance at these currents and short risetimes.
The video shows a motor jogging with a sector disk as load. Did you try it out with a real machine at accelerations demanding 90V 20A (or close to it)?
In my next conversion project that is about the least I will need (peak load).

Lest you think I'm one of your competitors: no, I earn my food and toys by solving problems in industrial data communications.

Einar

[dubble]

Hi,

I'm not concerned about bitrate. Most drivers can output max 300KHz anyway. The concern is that the cabinet is a hostile EMC environment. Check the theory on why a twisted pair is used in almost every (copper segment) data communications line in use.

Yes, I know the theory, and I learnet the theory at the Univ. some years ago
All I can say, that the distance is short, the frequency is low and there was no problem even in noisy enviroment, but I'll think about it and maybe change the pinout in the next REV.

Quote:

You are right, that's an expensive opto 2in1 package, DIP8, which is about $4.

And if not needed, those $4 x number of units comes right off your bottom line.

What kind of opto would you advice? An 1 Mbit/s opto would work with upto 1usec pulse widths. There's an 1Mbit/s opto with the same pinout as which is in use now and it's $2 only. Thanks for the advice, i'll try it out and measure throughput waveforms... maybe you saved us $2/controller.

In theory (and my experience) even a short wire/trace of high cross section should be considered an inductance at these currents and short risetimes.
The video shows a motor jogging with a sector disk as load. Did you try it out with a real machine at accelerations demanding 90V 20A (or close to it)?
In my next conversion project that is about the least I will need (peak load).

We made the PCB with the minimum of cross-sections as you may see on the PCB's picture. As I said before you can still use an external capacitor, placed into the screw terminals. My advice is to always use a well buffered power supply, minimum 1000uF/Amper.

The video shows a motor jogging with a sector disk as load. Did you try it out with a real machine at accelerations demanding 90V 20A (or close to it)?
In my next conversion project that is about the least I will need (peak load).

Yes, on the video it's jogging only, and it's a small 3Amp(with total load) motor. We tested bigger ones ofcourse up to 20Amp and yes on a real machine Our machine is a DIY hobby only one, it works well and accurate, but it looks dirty it's not a beautiful and shiny machine, like which you buy in a shop, that's why I didn't make a video from the running machine.
I agree, that you need buffer, when there are short peak loads, when you use a 90V power supply and the peak load is 20Amps, its 1800W=1.8kWatts!!! So your supply should be very well buffered and high power!!
In our tests there was 3x10000uF caps. in the power supply and we used about 2meters of supply cable. In this case the voltage didn't fall down with high current peak load, MosFETs switched corretly.
You should use a buffer capacitor close to the device, if you use long powersupply cables and the requirement is high pulse current.

Lest you think I'm one of your competitors: no, I earn my food and toys by solving problems in industrial data communications.

I don't know anything about you, except what in your post, but I hope, that I'll know much more and you will continue posting in this thread.

[happymarius]

Double, check your forum at CNC drive.
I need some answers.

Marius