DIY "Dual Magnetic" Draw Knob

[Coenraads]

Anyone putting together a console has probably lusted after Syndyne's SDK Solenoid Draw Knob. There's nothing like moving draw knobs IMHO. However, these units are prohibitively expensive and I've often thought, "Is there not an inexpensive way to make a draw knob work?" So, I decided to do a crude mockup where I hooked up a small servo motor to a draw knob.


The draw knob has to be loosely linked to the servo since the servo cannot be forced to run in reverse through the gear train.
The servo normally rests in the 90 degree (neutral) position so that the draw knob can manually be pulled in and out. But when commanded by the Arduino it goes to the 180 degree position to turn the stop ON after which it returns to the neutral position. Similarly, by going to 0 degrees followed by 90 degrees, the stop turns OFF. And indeed it functions as hoped and the current draw is minimal.

Two drawbacks:
1. The servo is "whiny" and I don't think I'd care to hear 30 of them run at the same time.
2. It needs to be driven using Pulse Width Modulation (PWM) circuitry, not a problem for an Arduino Mega which has 12 pins configured this way.

So my thoughts are, using a higher quality servo, could this actually be engineered as a viable, inexpensive option?
Your thoughts?

[circa1949]

How much is prohibitively expensive?

[Coenraads]

$68.55 USD = $97.30 CAD + 5% Federal tax + 7% Provincial Tax + shipping.
And on March 23 possibly 25% Canadian retaliatory Tariff????
Yes! Prohibitive.

[gtc]

How much is prohibitively expensive?

LMGTFY: https://us.midiworks.ca/index.php?op...&product_id=85

[gtc]

So my thoughts are, using a higher quality servo, could this actually be engineered as a viable, inexpensive option?
Your thoughts?

It's a very worthwhile project, however I'd be inclined to investigate b-directional/bistable linear solenoids, e.g. such as are used in electric locks for cupboards. These days such components can be sourced/made for quite low cost in China and India. The trick there would be to allow manual operation.

[Coenraads]

So, something like this?

https://www.aliexpress.com/item/1005...8aAs93EALw_wcB

But the 6mm stroke makes it not useful.

[gtc]

Yes, like that. The factory may make, or be able to make, a longer stroke version. This sort of component is often made in India.

[circa1949]

At the end of the day, precisely winding great lengths of copper wire is costly and time consuming. (There are incredibly fascinating Youtube videos of toroid winding machines in action!)

Although there are no Chinese pipe organ parts suppliers we know about, I would be rather surprised if there wasn't at least one occidental organ supply company having the actual winding done somewhere in Asia.

[Coenraads]

The solenoid in #4.1 still uses two coils. I'd like to reduce this to one coil. So I stacked about 8 tiny disc magnets inside the shaft of a discarded pen.
This pen slides smoothly inside the coil removed from a solenoid. Applying one polarity pulls the little stack of magnets into the coil while the opposite polarity pushes them out. The action is prompt and remarkably forceful with a stroke of 3/4 inch or more.



To apply the reverse polarity, an H-Bridge is needed. This also acts as a 12V driver and is Arduino compatible.

https://www.amazon.com/Motor-Shield-...%2C155&sr=8-16

These are now on order and we'll see how that goes.

[Larason2]

I don't think it's just winding the copper, it's making any circuit compatible with 15 amps! So pretty much every component has to be power rated. Solenoids act a bit like resistors, but all the proven designs you see here have other components to prolong the solenoid's life and reduce the risk of fire. To design circuits like that, you have to incorporate back EMF and other phenomena that are less fun to play with at 15 amps. Then, you need someone who's never done it before to wire up the 15 amp circuit! For that reason, these are one of the few things I wouldn't be comfortable with DIY. Maybe as a fun exercise, but not to recommend someone hook up for themself.

[Coenraads]

15A is not that scary but 75A or 100A is! These are currents I encountered when replacing a power supply that operates the slider motors on a tracker organ. It is then I discovered that the original wiring did not follow the manufacturer's specification. It is at that point that one starts paying close attention to wire gauges and lengths and the solidity of all terminal connections. It is also wise to insulate or cover up any exposed connections so there is no risk of an errant tool or even a ring on one's finger creating a short circuit.

When dealing with 15A or even 30A the trick for a safe installation is to connect the power supply directly to a fuse block so that the current can be subdivided into branch circuits carrying at most 5A. But even then connections need to be secure and when crimping terminals I usually like to solder them too. The fuse blocks I use are the ones usually found in marine applications.

https://www.amazon.com/Deyooxi-Circu...1zcF9hdGY&th=1

These have nice solid screw terminals and use automotive (blade) fuses which are readily available. They even include LEDs to indicate when a fuse has failed.

[Coenraads]

BTW, as I continue to perfect the single coil design, I discovered that I can get a nice toggle feel through the simple expedient of mounting a steel washer at the end of the coil where the little stack of magnets go in and out. It's the two ends of the magnet stack that are naturally attracted to the washer thus holding the draw knob securely in the ON or OFF position.
The length of the stroke can be adjusted by changing the number of magnets in the magnet stack.
A third use of the magnets is that they can trigger a reed switch.
Now to design a simple bracket and bushings for mounting the parts.

[Coenraads]

The L293D dual motor control/H-switch has arrived and I just finished testing it on the bench using it to drive the single coil draw knob I built.
This chip drives the coil at 12V although higher voltages are possible plus it has built in diode spark suppression. The H-switch (polarity reversing switch) is controlled by two pins on the Arduino. 10 = positive polarity on, 01 = negative polarity on and 00 = coil off. The draw knob "bops" back and forth without hesitation as commanded by the test program. The red LED indicates that I caught it in the middle of its half-second off transition.



The question then is, does the cost saving of going from two to one coils warrant taking this route? The downside is an increased chip count. Whereas the usual ULN2003 driver chip has 7 drivers, typically driving 3 stops (on and off coil) the L203D drives only 2 stops.

And of course, to my knowledge, the single coil draw knob is incompatible with all currently available combinations actions making unlikely that it will adopted. It's been an interesting exercise!

[OpusIII]

I have found recently your old discussion in 2021 and now this open and more recent one and was reading them both with interest.

Single solenoid coil works even without additional H-Bridge to reverse polarity - a lady enthuast in Germany has built a nice prototype here (unfortunately in German, but Google can transcript and subtitle easily). At about 11:30 to 14:00 she explains the single coil approach, that of course has some disadvantages, as the software needs to consider the current position, but on the other hand, it spares winding many solenoids... BTW, winding - she has also built a machine for that here, but this is another story.

Still, the outcome is relatively big with not much possibilities to miniaturize any further - for my project commercially available moving drawknobs or Sabine's solution would not work, as I have very limited space available for the stops. My main design constraints were: a stop should fit in 50mm (around 2") cube including all electronics and cabling, needs a nice bistable toggle feeling for manual operation, thrust somewhere between 3/4 to 1" (decided to go with 20mm in the middle), color coded lights as I intend to use the stops with multiple assignments in banks, touch sensitive for displaying the currently assigned register and finally affordable (I am currently at roughly 15-20 USD in total per drawknob including all materials and parts).

I have recently finished prototyping with a similar approach like yours: using cheap servo motors and instead of trying to decouple, rather not couple at all. I cannot recognize in your photos the moving parts, but likely you have solved the problem in a similar way.



The main advantage of the servos is indeed the current consumption - with about 200-250 mAmps in operation I do not need thick wiring and/or a built in fire extinguisher in the console and the stops can operate on 5-6V.

I use a Teensy 4.1 as controller as it is very fast and has plenty of I/O where I could realize also touch sensitivity for my knobs - implementing MIDI was also a breeze as the thing can be set up as class compliant USB MIDI. Additionally I used cheap and readily available 16 channel PWM drivers. The servos are a slightly better quality with metal gears, but still very commonly available and affordable. The lights are programmable Neopixel rings. Finally for the clean bistability with well feelable endpoints I use small neodym magnets with opposite poling. For the switch I went for IR optical bridges - too many magnetic fields in a small place for reliable reed or hall detection.

Most parts are 3D printed, this makes it more easy to produce them with a reliable tolerance in small batches.



Here I have a sketch and a short video of the prototype available "in action" - I suppose I cannot upload video files to the forum..

The only thing I am not yet sure about is the noise of the servos (as you have also mentioned - in the open prototypes the zzzip-zzzap is kind of nerving, I will see if the housing dampens sufficiently, possibly needs some foam or felt, we will see...

Currently I plan to do some decent "stress-testing" to see how long the stops would last - if the basic design works out as expected, likely I will start with the "production" of my 32 stops this winter.

[OpusIII]

Somehow it looks like I did not manage to upload some pics - possibly a limitation for “moderated” newcomers or just me being hopelessly silly.

Also a pity I could not edit my post anymore but now it shall remain as it is :-) forgive me the typos.

[myorgan]

The photos you uploaded should appear soon. The OrganForum server was just migrated, and the photos that have been uploaded will follow soon.

Michael

[Coenraads]

What you have accomplished is nothing short of amazing! I'm impressed. Instead of just toying with an idea, like I do, you've gone ahead and designed and built a working prototype. Thanks for sharing what you have done on the Forum.
I've played around with Neopixels but didn't realize they are available in ring format. They could be used as expression pedal or crescendo indicators. Hmmm...

[Larason2]

Using servos is a clever idea, and certainly the decreased amperage required is very attractive. However, I would expect a servo to have reduced life compared to a similar solenoid, especially if you're pulling and pushing on it when it's running in and out on a regular basis. The weakest link is the brushes on the electric motor. But for the price, why not try it? If they start giving out after a few months, you can always try something else! I believe Michael often judges robotics competitions, and from what I've seen of videos of these competitions, it's not unusual for servos to go AWOL over the course of a match. You can always overbuild it, but with that goes up the price and amperage, and its advantage over solenoids decreases!

[OpusIII]

instead of just toying with an idea, like I do, you've gone ahead

He, who starts earlier, will get ready sooner 😁 I have been toying with these ideas for many years by now, I am kind of "planning" my dream organ console since nearly a decade by now - a future project for my upcoming retirement. Before losing track of all the ideas and not finding some sketches made years ago, just picked up the stops as a "feasibility study".

They could be used as ... Hmmm ...

Indeed, tasty little things that make lot of fun, especially when it comes to animated effects. For someone who grew up with James T. Kirk, Uhura and Spock, like me, the big screen in the middle and many christmas lights around make a serious impression. You get Neopixels in many different form factors, see Adafruit - the small rings with 8 LEDs are getting rare, so likely I will need to do some serious shopping soon.

expect a servo to have reduced life compared to a similar solenoid

This is exactly my fear also - soleonids with 20mm stroke are not easily available, winding my own like Sabine does, is something I would like to avoid. My main issue with coils (besides the amperage) is the place required - for something with 1" stroke you need at least a depth of 3-4" that is too much for my plans.

I have chosen MG90S which are supposed to be a bit stronger and hopefully also more robust with metal gears at the same form factor like the SG90, possibly some other brushless make would last longer, but the prices are going up fast for 32 pieces...

My first test round did not last very long - one of the three prototypes was gone after 2-3 hours of continuous operation, it did not manage more than 2-3000 cycles. As it turned out later this was likely a software issue - the servos are rather dumb and if the PWM signal goes over their endpoint, they keep on trying. For a single cycle with half a second you do not recognize that, but if running continously, it sums up and fries the beast at the end.

I have ordered now a proper laboratory power supply to have better control of the current drawn and will try to setup the next round more carefully.

...after a few months, you can always try something else!

This I would like to avoid at any price. Basically the servo can be replaced easily, but I want an organ console for many years to come and not a repair practice shop with open guts all the time. So if the things do not work reliably at least something like 20-25000 cycles or more than I will not start building 32 of them - I need roughly two days for a single one, excluding the printing time, my old eyes will fall out of my head with all the soldering with magnifying glass, so I won't give it a go before I am sure.