Voice Coil Rail
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
The first thing I thought of, years ago, to shove a rail, was a radio-control model's SERVO.
They have 12 bit resolution (claimed) all the torque and speed you need, and are a doddle to drive.
there's some descriptive chat here but other sources may be better:
http://www.helifreak.com/showthread.php?t=263175
You CAN use resistors and capacitors to set the pulse width which determines the position, even with a (horrible but useful) 555 timer, but an astable and counter is a much better plan.
They're also cheap, small and run off small batteries.
There are linear-actuator versions, but I know no more. A rotating one could pull a cord and slide a slider.
Analogue voltages and coils of wire? No thanks
You can also easily build in things like flash detector - something which is bugging me at the moment. You could do it from eg an Arduino, but the electronics is simple enough to get going on a breadboard.
I think they were 8 bit when I looked last, perhaps I'll have a go at school
They have 12 bit resolution (claimed) all the torque and speed you need, and are a doddle to drive.
there's some descriptive chat here but other sources may be better:
http://www.helifreak.com/showthread.php?t=263175
You CAN use resistors and capacitors to set the pulse width which determines the position, even with a (horrible but useful) 555 timer, but an astable and counter is a much better plan.
They're also cheap, small and run off small batteries.
There are linear-actuator versions, but I know no more. A rotating one could pull a cord and slide a slider.
Analogue voltages and coils of wire? No thanks
You can also easily build in things like flash detector - something which is bugging me at the moment. You could do it from eg an Arduino, but the electronics is simple enough to get going on a breadboard.
I think they were 8 bit when I looked last, perhaps I'll have a go at school
Chris R
Peter,
Even though you are using dc you will find accurately controlling the coil current will be very difficult in voltage mode. First off the coil temperature will case an error since copper has a tc of over 3000 ppm.
Since you are using a dac and voltage buffer, a current mode driver is almost as simple, probably just one extra resistor.
Best,
Mike
Even though you are using dc you will find accurately controlling the coil current will be very difficult in voltage mode. First off the coil temperature will case an error since copper has a tc of over 3000 ppm.
Since you are using a dac and voltage buffer, a current mode driver is almost as simple, probably just one extra resistor.
Best,
Mike
mjkzz wrote:Mike, thanks! And here is what I think.
Lets put the word speaker aside, so our goal is to move something up and down (my initial goal is vertical setup) electromagnetically by 2mm max at very fine steps. So yes, as far as I know, this displacement (or rather force) will be proportional to current inside the voice coil if a permanent magnetic is used. But we are not making it oscillate, just moving it to displace some subject linearly, so this sounds like almost DC because it does not oscillate. Of course, since we are changing the current over time, there bound to have some oscillation if you analyze it in frequency domain, but I somehow think we can treat it as DC overall.
Since it is DC (or almost DC) operation, we really do not have to worry about frequency response characteristic of the setup and only consider the DC side of it. Having said that, I think since the resistance (rather than impedance) of the coil is the major factor here, we can turn a current mode device into a voltage mode one -- by adjusting the voltage across the coil, we can alter the current.
Base on the above, my circuit is basically a 12bit DAC and a voltage follower as a buffer to supply enough current with the coil as resistive load. I have not submitted the circuit for PCB fabrication yet, but will in a couple of days.
I am sure there will be something to be aware of from the frequency domain, such as oscillation when changing current (voltage) applied to the coil, or maybe even PCB design, and if they are causing enough problem, I think I will go for piezo route with simple voltage control (with less displacement)
http://www.makro-forum.de/viewtopic.php?f=2&t=94401
in a german Forum somebody showed measurements on using a Speaker as actuator.
edit:
another stacking solution
http://www.mikroskopie-bonn.de/_downloa ... ermann.pdf
in a german Forum somebody showed measurements on using a Speaker as actuator.
edit:
another stacking solution
http://www.mikroskopie-bonn.de/_downloa ... ermann.pdf
Mike,
Thanks for advice, yeah, the thermal part is kinda troublesome, this is why I have to consider the weight of subject so that I can drive the thing at lower current (ie, voltage) so that the coil will not over heat and remain stable. Yes, so at low current, a good power transistor should be enough.
Regardless which transistor will be used, the output of my PCB is the buffered voltage from the DAC so I can play with different options to drive the coil.
Regards
Peter
Thanks for advice, yeah, the thermal part is kinda troublesome, this is why I have to consider the weight of subject so that I can drive the thing at lower current (ie, voltage) so that the coil will not over heat and remain stable. Yes, so at low current, a good power transistor should be enough.
Regardless which transistor will be used, the output of my PCB is the buffered voltage from the DAC so I can play with different options to drive the coil.
Regards
Peter
Not sure if you are addressing me, if so, my name is Peter.Brilliant idea William! I will be very interested to see the result. I may even give it a try myself using a coin as a subject. Are you patenting the idea?
In terms of patent, another intention of this post is to prevent people from patenting it as this post will serve as public domain knowledge. So at least for USA, since the "new" patent law (passed back in 2013?) will count this post as prior art, any patent issued (after this post) based on this idea will be invalidated.
Regards
Peter
I sketched out a simple circuit to use a counter for (my preferred ) servo.
This is most of it. It would need some simple extras
You need to set a mark/space ratio.
Simplified -
In other words if your total period is 10, and you have the Mark (high period) from 1 to 6 and Space (low) from 6-10, you're set to 60% of the servo movement range.
If you change that to 7 Mark and 3 Space, you have the servo set to 70% of its range.
So, if you make a circuit where you input a number and a counter counts to that number, output being high while it does it, then low for the rest of the number range, you have control.
Here's a Noddy circuit where you set a number with up/down buttons. (top display),
the middle counter starts from that number (here it's 8 ) and counts down to 1 and repeats, and
the bottom one starts from 1 and counts up to 8 then repeats, if you need that.
You get your output when they get to 0.
Here the counter clock is running at 2 Hz - make it what you need.
It's only working for one digit 0-9 , you'd cascade/duplicate the counter chips to make it 0-99, or 0-999, or with binary counters, 0-F, 0-FF etc.
You don't need the displays 2 and 3, so you only need a few chips for one digit
Three decimal digits would let you choose from one of 999 positions, about 3 more chips per digit.
if you can program a controller it would be quicker, but the hardware route costs under $20.
I can upload a .cwz file if someone wants to play.
This is most of it. It would need some simple extras
You need to set a mark/space ratio.
Simplified -
In other words if your total period is 10, and you have the Mark (high period) from 1 to 6 and Space (low) from 6-10, you're set to 60% of the servo movement range.
If you change that to 7 Mark and 3 Space, you have the servo set to 70% of its range.
So, if you make a circuit where you input a number and a counter counts to that number, output being high while it does it, then low for the rest of the number range, you have control.
Here's a Noddy circuit where you set a number with up/down buttons. (top display),
the middle counter starts from that number (here it's 8 ) and counts down to 1 and repeats, and
the bottom one starts from 1 and counts up to 8 then repeats, if you need that.
You get your output when they get to 0.
Here the counter clock is running at 2 Hz - make it what you need.
It's only working for one digit 0-9 , you'd cascade/duplicate the counter chips to make it 0-99, or 0-999, or with binary counters, 0-F, 0-FF etc.
You don't need the displays 2 and 3, so you only need a few chips for one digit
Three decimal digits would let you choose from one of 999 positions, about 3 more chips per digit.
if you can program a controller it would be quicker, but the hardware route costs under $20.
I can upload a .cwz file if someone wants to play.
Last edited by ChrisR on Thu Mar 02, 2017 8:17 pm, edited 1 time in total.
Chris R
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
I'm sorry Peter, I got your name wrong.mjkzz wrote:Not sure if you are addressing me, if so, my name is Peter.Brilliant idea William! I will be very interested to see the result. I may even give it a try myself using a coin as a subject. Are you patenting the idea?
In terms of patent, another intention of this post is to prevent people from patenting it as this post will serve as public domain knowledge. So at least for USA, since the "new" patent law (passed back in 2013?) will count this post as prior art, any patent issued (after this post) based on this idea will be invalidated.
Regards
Peter
Thanks for putting this into public domain as I think it is truly brilliant and wish I had thought of it myself. Not too many things come along that I can say that about.
Peter,
TI has Darlington transistors in the TIP series devices, these are very inexpensive. IR has the HexFet power NMos devices. Use the "snubber" to attenuate any glitches on the waveform edges. With the proper selection of the DAC time-constant RC you can creat an almost linear ramp as mentioned by Rik.
Mike
TI has Darlington transistors in the TIP series devices, these are very inexpensive. IR has the HexFet power NMos devices. Use the "snubber" to attenuate any glitches on the waveform edges. With the proper selection of the DAC time-constant RC you can creat an almost linear ramp as mentioned by Rik.
Mike
Thanks Chris and Mike, wow you guys are super fast designing things!
I went to market today and checked out some magnets, maybe it is worth it to wind my own coils.
by the way, Mike, my design is similar to yours, I use LM358 for op-amp, I think it is a little more stable with single power supply, and it is configured as (identical) voltage follower, but yours seems to more advanced.
I also plan to use ULN2803 darlington, this is the part I have to experiment as I have little knowledge about analog circuit.
Regards
Peter
I went to market today and checked out some magnets, maybe it is worth it to wind my own coils.
by the way, Mike, my design is similar to yours, I use LM358 for op-amp, I think it is a little more stable with single power supply, and it is configured as (identical) voltage follower, but yours seems to more advanced.
I also plan to use ULN2803 darlington, this is the part I have to experiment as I have little knowledge about analog circuit.
Regards
Peter