Motorised fader seek algorithm

[ruffrecords]

See, when you drive a stepper motor, the only control you have is the time between steps. As you speed up, you decrease the time between steps until it is some minimum and then you're slewing. At some point you have to start decelerating, which you do by gradually increasing the time between steps.

I may be wrong but I thought these motorised faders used simple permanent magnet dc motors?

I honestly don't know, either, but it seems to me that a stepper might be simplest. Or not!

-a

Indeed, but also more expensive. Looks like the ALPS ones use dc motors:

http://www.conrad-uk.com/ce/en/product/442094/Motor-fader

heers

Ian

 

[JohnRoberts]

Yup, the cheap alps I recall were simple PM motor. This is consistent with his H switch driver.

I have thought about this some more. these are the kind of problems that micro's handle well... as robot makers have figured out long ago.

If i was doing this I would drive the motor with a short fixed on time step for convenience that represents something like say 10% of total travel. For simpler micro math maybe 12.5% (easier to divide).  So bang the motor hard on for one of these predicted (12.5%) on time interval, then when it finishes measure how far it actually traveled and adjust the time to move the next 12.5%, etc, After several steps your predication will be more accurate.

When you get to within one large step from the final target, change to using the smallest step size, to nudge it to the final target.

While this code will seem clunky and stepwise, in practice these steps are occurring with such short time delays between them that the movement will appear continuous.

You will need to do one long seek every time the fader is first powered up to learn the precise step size relationship, unless you use a micro with flash memory that can save what you learned from prior sessions.

Of course look at the robot maker literature for simple canned solutions. 

JR

 

[sahib]

See, when you drive a stepper motor, the only control you have is the time between steps. As you speed up, you decrease the time between steps until it is some minimum and then you're slewing. At some point you have to start decelerating, which you do by gradually increasing the time between steps.

I may be wrong but I thought these motorised faders used simple permanent magnet dc motors?

Cheers

Ian

Yes but there is a way to control  PM like a stepper. Those RC servomotors used in model planes etc. Check the IC on this link I have used it many times and I still have about 500 in stock.

http://pdf1.alldatasheet.com/datasheet-pdf/view/876/MITSUBISHI/M51660.html

If you send the command in absolute then the motor moves at full speed. If you send it incrementally it moves in steps. By controlling the off time you control the speed, hence ramp up/down.

The resolution will be 0.39mm in 8 bits over a fader travel of 100mm. It might be o.k. but moving up to 10 bits (by using an appropriate micro) will nail it. The ramp up/down will be extremely smooth and better controlled.

Normally the RC system is open loop externally but closed loop internally. However, on this application the same feedback will also be used for external loop to track the position of the fader when moved manually.

 

[ruffrecords]

See, when you drive a stepper motor, the only control you have is the time between steps. As you speed up, you decrease the time between steps until it is some minimum and then you're slewing. At some point you have to start decelerating, which you do by gradually increasing the time between steps.

I may be wrong but I thought these motorised faders used simple permanent magnet dc motors?

Cheers

Ian

Yes but there is a way to control  PM like a stepper. Those RC servomotors used in model planes etc. Check the IC on this link I have used it many times and I still have about 500 in stock.

I am only too well aware of that. Over 20 years ago, when I was gainfully employed!!, we developed a label printer for Esselte Dymo that was based on a stepper motor as were all the label printer designs of that time.

For the next generation, cheaper, faster product, we used a dc motor with an opto-interrupter and closed the loop in software making it act just like a stepper. They made a million of these in the first year of production. I am pretty sure that most label printers use this technique now. And here is a link to the original patent:

http://www.google.com/patents?id=4igdAAAAEBAJ&printsec=frontcover&dq=ian+thompson-bell&hl=en&sa=X&ei=KewKUOzkB8Gm0QXN2YnTCg&ved=0CDYQ6AEwAg

Cheers

Ian

 

[sahib]

Hi Ian,

I did not mean to try to teach you how to suck egg. I meant in the context of driving the fader motor in a stepper fashion.

I have checked the link and your system and to be perfectly honest I don't know how and why  you would make savings over a stepper. All things being equal a stepper will obviously be more expensive than a PM but you have the additional drive belt and the opto feedback system.
However,  your accountants were oviously happy that it went into production.

I am not particularly into repairing printers etc but all the printers I came across over the past twenty years used steppers.

 

[ruffrecords]

Hi Ian,

I did not mean to try to teach you how to suck egg. I meant in the context of driving the fader motor in a stepper fashion.

That's OK. I did not think you were trying to teach granddad to suck eggs!! LOL

I have checked the link and your system and to be perfectly honest I don't know how and why  you would make savings over a stepper. All things being equal a stepper will obviously be more expensive than a PM but you have the additional drive belt and the opto feedback system.
However,  your accountants were oviously happy that it went into production.

The savings were considerable and label printing is an extremely cost sensitive market. The drive requirements for a stepper and a dc motor in this application are almost identical. There's no drive belt, just some plastic reduction gears - very cheap. The opto too is very cheap - a few cents. Overall I think we save about $1 which over 1 million units is well worth while. This hand held battery powered label printer sold in the early 1990s for 70GBP and I think the BOM cost was about $25.

I am not particularly into repairing printers etc but all the printers I came across over the past twenty years used steppers.

For regular printers you are quite right and we later developed a franking machine that did use a stepper as for that application it was the most cost effective.

Label printers, especially battery powered hand held ones, are a whole different ball game. The latest version of that printer we developed 20 years ago now sells for just over 20GBP.

http://global.dymo.com/enGB/Products/LetraTag_LT-100H.html

Cheers

Ian

 

[sahib]

The savings were considerable and label printing is an extremely cost sensitive market. The drive requirements for a stepper and a dc motor in this application are almost identical. There's no drive belt, just some plastic reduction gears - very cheap. The opto too is very cheap - a few cents. Overall I think we save about $1 which over 1 million units is well worth while. This hand held battery powered label printer sold in the early 1990s for 70GBP and I think the BOM cost was about $25.

The patent application showed timing belts so I assumed. I actually kept a few of those gear drives. They are unbeliavebly fine for injection moulded parts. Literally zero backlash. One of my hobbies since childhood is to take things apart to study designs both eletronics and mechanical.

I have done quite a few consumer electronics stuff in the past so I know how even the pennies count. And  certainly  for $1 over a million run you would even dismantle a government.

Great stuff.

 

[ruffrecords]

The patent application showed timing belts so I assumed. I actually kept a few of those gear drives. They are unbeliavebly fine for injection moulded parts. Literally zero backlash. One of my hobbies since childhood is to take things apart to study designs both eletronics and mechanical.

Until you mentioned it I had forgotten the diagram showed a belt drive rather than gears. I am not sure why but it was probably on the advice of the patent agent to make the concept easier to grasp. In the text it does say that the belt is just a diagrammatic representation of a gear train.

I used to take things apart as a kif too. Used to drive my mum crazy. Have you seen the Dilbert video about the kid engineer??

http://www.youtube.com/watch?v=FlJsPa6UwcM

Cheers

Ian

 

[sahib]

;D ;D ;D ;D ;D ;D ;D ;D

That was superb. I love when the doctor says "he'll be an engineer".

I know this post will belong to the Brewery but my father was an electrician and we, the four brothers obvioulsy grew up in "the shop" and all studied electrical and electronics engineering. Did not really have much choice. Anyhow, when my eldest brother was 8 years old (he is 60 now) my father brings this state of the art, brand spanking new valve radio home. Next day my mum reaches out to turn it on the thing is dead. No lights no sound. When my dad comes home my mum asks him to look at it. He turns the radi over all gutted out. Just an empty shell. My bro takes the whole thing apart and makes it headlights etc for his cart. Imagine this.

 

[ruffrecords]

Nice one, Sahib!!!

Cheers

Ian

 

[JohnRoberts]

I had two older brothers so most stuff was taken apart by them before I had a shot...  :'(

My dad was an engineer but died while I was still young so the main thing i remember him bringing home from work was candy and some comics he'd tear out of the daily newspaper for us. (note: there were some old western electric 19" rack gear in the basement from one of his earlier day jobs.)

Regarding that patent and pulleys, by law the patent is supposed to show the preferred embodiment or best way to execute the invention. It is not unusual for patents to show other than best or most economic  solutions as long as they effectively define the invention. The claims are the real nuts and bolts (enforceable part) of the patent, but to break a tie in court cases they sometimes can dig down past the claims into embodiment drawings and text to parse out uncertain meanings.

=======

I don't have any stories about taking things apart that I couldn't get back together, but I did fix a dishwasher that a repairman declared unfixable. FWIW I was only allowed to mess with it only after it was considered a basket case, since I was just a kid.  8)

JR

 

[glorymorris]

I designed an automation system for the Rupert Neve Designs 5088 console, which was never completed due to some bad business. If you need help, I can answer your questions. Generally, the way this has to be approached is as an actuator. PI (no D) loops, and read up on some Control Theory. You need Proportional and Integral Drive, and for the Integrator to work well (and especially if you use D, which will be minimal if at all) you'll need some serious low pass filtering, which can be implemented on a DSP level post A/D.

 

[glorymorris]

To keep the fader from stalling out within a certain range of it's target, you need to manually tune a PI loop. The integrator will destabilize the control math unless it's tuned right, but it will land you right in the target within a certain (very small) tolerance based on how clean your servo reading is. This is how it has always been done. It can't be done with procedural or conditional methods.

 

[jdbakker]

One of the better starting points for PID design is the article PID without a PhD by Tim Wescott:

http://www.eetimes.com/ContentEETimes/Documents/Embedded.com/2000/f-wescot.pdf

(or, in HTML: http://www.eetimes.com/design/embedded/4211211/PID-without-a-PhD).

JDB.

 

[Twenty Log]

When I did a PID for an environmental chamber (to prevent over/under shoot from target temperatures and humidity to test our 5kW solar inverters), I found the coefficients for each of the I and D portions to be extremely fractional, which may have some caveats in an integer based processor without some scaling

Granted temperature slow moving stuff versus a fader is a different ball game, but then again my sample time was 1 second which simplified the D math in code (trivially)...

 

[glorymorris]

The Differential coefficient will almost always be the smallest unit, and definitely fractional. For parking a fader at a target, the D will usually be significant, but for tracking a stream of targets pouring out of HUI from a DAW, almost any amount of D will cause a stutter. Clean A/D and good DSP LP Filtering with only Proportional and Integral will work very, very well. If the fader has to reach a target at a high distance, there will be a minor amount of overshoot, but this is considered completely acceptable industry wide.

If you don't implement good filtering, the PI Loop will not work well.

 

[glorymorris]

Here's a video of my system a few months before what was supposed to be the initial production run. Apologies on the video quality and the state of my shop! This video was shot for the staff at RND, and the A/D Voltage Reference had yet to be cleaned up, so overshoot wasn't yet minimised. Some stuttering was still present at this state, also due the the vref circuit. Had a bit of noise that was challenging to deal with.

Obviously, we had some issues that left me not wishing to ever work with a corporate entity again, and the design is mine free and clear. I've been debating open sourcing it so that people can sell kits or fully assembled units and modify or improve the design if they wish. The only issue I have is that I was never paid anything and so I've lost quite a lot of money on the design and I'd kind of like to make that money back.

http://www.youtube.com/watch?v=0UspHkmytjo&feature=youtu.be