how to remote control a rotary switch based attenuator

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Thanks ! It is good to know there is a motor with resolution high enough and power is high enough. So far I have tried with a small motor which position it well without the detent ball, but can not turn with detent. The big motor I have tried (200 step) was stopped sometimes between two positions.

Regards,

JG
 
Giordano said:
gootee, I do not thrust LDR-s. After the experiment with the realy network, I decided I do not go again away from the known good way of solving this problem. I just want to add remote control to the rotary switch.

Regards,

JG

Hi JG,

Please, I am seriously interested in why you say you do not trust LDRs, because I have used them in a non-audio commercial product, and am considering designing them into another commercial product. So, if there are any issues or even negative opinions of which I might not be aware, I would definitely very-much-appreciate hearing about any related experience or even just thoughts about them that you might be kind-enough to share. (Please, feel free to email me, if you do not want to put the off-topic material into this thread.)

Back on topic (sort of):

I really don't want to try to 'convince you to use LDRs'. I have not even used them in an audio attenuator, myself. And I, too, am very interested in your eventual solution for a remote-controlled rotary switch. But, I also wouldn't casually dismiss LDRs. (Please, note that I am not saying that that is what you have done. i.e. 'casually dismissed' them. I can't even know if that was the case unless you expand on your comment about them. I'm SORRY for possibly *seeming* to have made that assumption. I really just want to 'get my thoughts down' about them while I'm here again, maybe even just for the benefit of others' future 'search results' if nothing else.).

Although I, personally, have not used LDRs for an audio attenuator, the LED/LDR optoisolators ARE, apparently, a "known good way" to implement such an attenuator, and have been used that way for at least 20 years, even in some high-end commercial audio products. And the circuit is elegantly simple. (Also, the concept and design seem sound, to me. And the 'pure resistance' of the LDRs' photocells seems quite attractive, as does the 'no moving parts' aspect.)

The Lightspeed Attenuator example, which has a thread here on diyaudio.com (which I have recently read in its entirety), has received many 'rave reviews', and basically no substantive complaints that I know of. The guy who posted the Lightspeed design on diyaudio, georgeHiFi, had already built and sold them commercially for about 15 years, before sharing it here, and has reported no optoisolator/LDR problems, that I know of, even with older units. (And it SEEMS like almost everyone who has heard those 'optical attenuators' has been amazed by how much better they sounded than anything containing movable contacts. That's their story, anyway. ;-)

However, one probably-important caveat: As I mentioned earlier, since you are using a vacuum-tube-based system, you would probably want to first make sure that the attenuator's input and output impedances would be compatible, if you wanted to use the Lightspeed design 'verbatim'. (And if the Lightspeed design was NOT compatible with your impedance requirements, there should be slightly-different alternatives that would be fine.)

Thanks, JG. I will try not to pull this thread off-topic, any more!
 
Giordano said:
Eva, linearity and noise is not my primary target, but I do not really want to open this tubes vs. solid stage etc. discussion here. Bottom line is, I do non wnat to see them on my audio chain nor most people I know who loves to listen to music.
I do respect them, they are fine for most application.
A year ago I was happy with my black alps, but recently I have tried a stepper with my new amp (2A3 PSE).

Regards,

JG

Mechanical contacts have been used for ages and over time they are well known to become dirty, oxidized, noisy, coated by a layer of every substance volatilized in the ambient, etc... In the end they produce great amounts of distortion when small signals are involved. For example, relays have a minimum current rating and special ones are required in order to switch small signals. After some aging, the contacts in standard relays and switches have trouble to start conducting when current/voltage is too low. This becomes worse when some DC component is involved due to electrolysis.

Then again, we invented solid state devices to overcome mechanical contacts and get ever lasting signal integrity. I've had quite bad experiences both with rotary switches (as frequency selectors in crossovers) and relays (for audio signal routing). Now I prefer solid state without doubts.
 
gootee, thanks a lot for your reply.

I have used a wrong word about LDRs, or rather closed out with too short explanation.
My reason is, I have "burnt" myself with the relay network attenuator. I did so much time into it, changed relays for different version etc. but it did not win over a black alps. That time I have looked at the Lightspeed topic (and now I will look again), but I did not really found someone who used it in a tube based chain. That is why I decided I go back to the well known rotary switch. The shame is, I have a 15 years old attenuator not used for 10 years, what I just tried and kicked both relay network and black alps. (That time I did it for a M. Leach amp, but when I finalised my preamp I have used the black alps, I heard no significant difference) The switch itself was a used one from an instrument, I did not even cleaned it. The resistors are various types from the local shop. It is a series configuration.
On the new amplifier it takes about 3 seconds to hear it is far better. I have tried it in a friends system with the same result.

It is crazy, I was a great believer in solid states, M. Leach amp gave us a lot. I did not believed in cables, tubes etc. at all. I have tried Mission Cyrus, Audiolab 8000A, Audio Innovations Alto (this I really liked) but liked the Leach better.

Now I have changed to this triode SE amp, the jump is huge. Now the bloody cables are making so significant difference, this attenuator thing is just like that. I'm sick of this cable, attenuator, equipment stand ... story, but I have to accept these are respected with a reason.
I still hate tubes (lifetime, power consumption etc.) , but love the music they provide and that argument is stronger, so... I have decided I follow the known way but try to extend with a few thing which does not effect sound.

What I'm working on now, because of the page what Peter Daniel linked in, see again the stepper motor control, without the feedback disc.
My problem was, when I de energized the motor colis, the motor control IC lost it's phase info. With half step mode, (0,9degrees) my control could step some backwards before it steps the right direction, depending on what phase the motor is left in. The result was, most of the time it worked and the detent pulled in position accurately when I de energized the motor coils, but not always. Sometime it was "hasitating" to start and than stopped half way between two positions. I will try to tune it with adding some steps because the detent can also pull it back if I step over some. Also, I got some L9935 ICs which can be controlled to de energize the coils without loosing the phase info.

Also, I look at the Lightspeed thread again, looking for a 100k attenuator.

Thanks,

JG
 
Hi JG,

Thank you for the good reply.

For the stepper-motor problem, if you cannot get them to always do it right, maybe you could get them to always do something consistently, even if it's 'wrong', so you could just fix it in the controller software. It would be similar to the idea used for some DIY CNC machines with steppers, where backlash (and even leadscrew error) is compensated-out in the software, instead of trying to make the hardware perfect, with, for example, expensive anti-backlash nuts or ball-screw drives. But, to do that, you'd need for it to have a consistently repeatable 'error'. Maybe there's a way. But that's 'just an idea', so far.

OK. Forget all of the above, I think. I just re-read your message and now see (I hope I understand, now, anyway) that the problem is the motor 'losing its phase info', and basically not re-starting from where the system thinks it stopped.

For that: If you could assume that it was starting in a good detent-locked switch position, couldn't you just re-initialize the motor-position data, from there? i.e. Each time you get to a new switch position, throw away all of the previous step and position information and re-initialize it with the new (assumed 'known') switch-position. Otherwise, any errors might accumulate. Of course, this way, if it does ever fail to correctly get to the next detent position, it might never correct the error, unless you had some way for it to detect at least one known position (maybe a limit switch for the minimum-volume position, at the least), which you could maybe use as a manual 'reset' [hopefully not needed very often, if at all]. (And maybe the mid-volume position would be more convenient, for that. Or use both, etc.) At any rate, with a scheme like that, you would only need to concentrate on being able to get from any one position to any one adjacent position.

Even if the above type of scheme worked OK, I think the ideal way would be to have some better feedback, so you could at least know that each switch-position was properly reached, even if you didn't know the exact motor-step position (from feedback) at all other times.

Obviously, an extra switch pole would be able to provide that coarse position information. But that might not be practical. I suppose that if a signal could be assumed to always be present, then a very-high-input-impedance (FET) circuit, i.e. a very, very simple one for each switch position, could give the needed information. That might not work well during the times when the source was 'silent'. But that might not matter, too much. Anyway, that's just something 'off the top of my head', and might be ridiculous to even consider. I will try to think about it, some more. (Another way might be to inject 'pilot tones', that were way out of the audio frequency range, and detect those to detect the switch position. That might be easier than it sounds.)
 
Hi JG,

I saw your posts in the Lightspeed Attenuator thread. A 100K attenuator should not be a problem. It takes LESS current to get a higher LDR resistance. So the max current needed for a given ratio between the series and shunt LDRs should be even less than the Lightspeed uses. You can find the datasheets for some alternative Perkin-Elmer optos through alliedelec.com, e.g. VTL5C2, VTL5C3, et al. They also have 'dual' units, e.g. VTL5C2/2 and VTL5C3/2, which might be easier to use, for two channels, i.e. maybe some advantage for 'matching'. For the Silonex NSL-32SR2 that the Lightspeed uses, you apparently have to go to the silonex.com site and find the document with the typical R vs I curve and grab the bitmap from within the webpage, since the actual 'datasheet' doesn't seem to provide that.

The NSL-32SR2 apparently CAN get up to about 300k Ohms. But it looks like that is for 1 microamp LED current (graph says 1m current but current scale seems to be already in mA), and 100k Ohms would require 5 uA. So, I'm guessing that it might be better to use a different opto, such that your resistance range is provided by somewhat larger LED currents, so that they would be easier to control accurately, and so current-source noise would not be as significant, relative to the required control levels. Also, on that graph, Silonex did not give variations of the typical response, that are due to temperature differences, and 'light-memory/history', which ARE provided for the Perkin-Elmer parts, and might be significant, especially if someone wants to use either extreme of the resistance range of a particular part. Perkin Elmer also gives additional data plots, such as response speeds.

Coincidentally, I have been working on (analog) circuits that can completely (and basically perfectly) linearize any LED/LDRs' R vs I response, over almost their entire range, with also other (non-audio) applications in mind, for them. (That work is getting interesting [as well as productive], and might even be generalizable for wide-range linearization of almost any non-linear property of almost any component or device. I guess the underlying theory must be well-known, but is probably often not easy to implement.)
 
How about putting 2 ratchet gears on the shaft, with a sawtooth profile. Like what stops a winch from running backward.

Use 2 solenoids, on the end of each solenoid is a roller on a short arm with a pivot where it connects to the solenoid by a couple of side plates, think of a single link of a bike chain.
The rollers rest on each gear at the top, the solenoid shafts are horizontal.

The solenoid pushes the roller into the steep side of a tooth to turn the shaft for roughly one detent then releases. It only has to be approximate because the detent of the switch does the rest.
the other one just slides up the gradual slope of the backward running gear.

Pulsing one solenoid pushes the shaft one way, click by click.
the other solenoid pushes it the other way.
 
How about coupling the motor to the switch more loosely?
By doing so, you can introduce a controlled degree of backlash, enabling the switch to 'find' its detent position without turning the motor to a position between steps. Subsequent operation of the motor will, of course, 'take up the slack', then continue as before.
 
gootee, you are right in the second part, it is not a constant lag what I can compensate in firmware. Unfortunately there is such stepper motor where you can step exact 15 degrees. Most steppers are 200 steps / rev or 400 steps. Half step mode is easy, that looks to be very accurate, but the problem is, because of this slight error, when the detent pulles it into position, you do not know the phase any more.

After Daniel's mail I did some more experiment.
I have used the small stepper in half step mode, AND I modified the detent spring a bit, so I can turn it now with the floppy motor easy. I have to make a better frame for the motor and the switch, but it seems it will be fine. The phase mismatch at the start is so small > the error at the end of the 15degree (actually 13.84 because I use a 26 position switch for trial) the detent always pull it to position. The frame is a piece of wood, a spacer, and lot of insulation tape :)

I also found, that the detent of my rotary switch is not the best by far. By hand, I have tried the blue closed switch what you can get on ebay and also an ELMA in an equipment and I found these far far better. On my swith from one center to the other only like 1/5 ends are the portions where the detent pulles it in position. On the blue one and much more on the elma it is just the middle 1/3 - 1/4 shere the detent does not pull it into position. This means, this all is probably meaningless for those switch, even the 200 step motors error is small enough for the detent mechanism. :)))))

I will see the LDR now anyway, I'm interested in how it sounds, but the fine motor trial looks very promising and I plan to buy an ELMA anyway. I believe it will work fine. Unfortunately on the ELMA switch I can not try my motor and control, because that is built in, but soon I will try on the blue one. The blue one is very easy to turn, so the floppy motor will handle it fine. About the elma, I feel it easier than my old switch, but that is 100% not sure.

Regards,

JG
 
dnsey, you touched a very important point, I forgot to mention. Even for the alps pot turning it is very important to keep the current switched on for a bit after turning. Even the 220mm long aluminium rod I use have that much flexibility (with the junctions) that you step 2-4, if you release the current immediately, the motor is turned back, because the pot is not turned yet !

An easy experiment shows it clearly. Years ago when I have tried my first program, for every IR receive it stepped 2. I have tried it on a bench, nothing was connected to the motor. When I pushed the button on the remote controller, the motor did crazy things sometime. Even making noise but rod not rotating. That was because after the step I released the current immediately. First I added a bit of delay and it solved the problem immediately. Crazy, but I have spent hours to find out what a hell is going on. I did not tried the motor and the control IC before, I thought one is faulty or I did not wired it correctly.

So, coming back to your idea, it might help, but yet I found the more flexibility I have in the connection rod, the more difficult to tune the control. For the pot, I have tried to make it as solid as possible. For small steps, you need to wait more for the pot to follow it if the rod is more flexible.

Regards,

JG
 
gmphadte, to be honest I did not found a good angle sensor under 100EUR. the one I found is 230EUR and quite big.

Anyway now I'm quite confident it is not necessary. The fine motor will be enough (400 step). The error will be about 2.5 degrees only, worst case. I'm about to finish a dac now. In ~2 weeks I will get back to the vol control.

Regards,

JG
 
You could easily make an optical position sensor yourself.
Prepare the artwork using drawing software, then print it onto clear acetate.
Once you have this fixed to the attenuator shaft, you don't need a stepper motor at all! Just use an ordinary geared motor, and let the position sensor control it directly.
 
"Starting over, fresh", on this, in my mind, and thinking about it for just a few seconds: Just 'off the top of my head', I kind of like the idea of using a rotary switch and a stepper motor that both have the same number of steps per revolution. Solenoids also seem like they might be simple-enough. But, with a small-enough DC motor, probably with some sort of gearing reduction, with a small-enough clutch, it could VERY-easily be made to work, with position feedback available, if size could somehow not be a problem. If it weren't for the possible size problem, the last method seems easiest to implement, for me at least.
 
Giordano wrote
gmphadte, to be honest I did not found a good angle sensor under 100EUR. the one I found is 230EUR and quite big.

After I went home and was at a highly relaxing place(!), I got another idea.

Issue is to detect the motion without drive signal.

A slot sensor can be used along with a slotted wheel to detect uncontrolled motion. For manual rotation, u will have to use quadrature detector to subtract cw/ccw counts in case of manual turning.

Now this may not be of interest but is an easy and cheap solution.

Gajanan Phadte
 
I am successfully using a bit larger stepper motor which has very small breaking force when de-energized (not sure my English is clear, with other words, very easy to rotate when not drived), so when I stop with the motor and de enrgize it, it jumps into the middle of the step. The starting problem/phase problem is still here, it stats sometime a wrong direction with 1 step. This motor I found (old 5.25" floppy head motor) is a 400 step motor, so with half step it takes
22 half steps (as I remember) for a step at the switch. 4 half step in the wrong direction worst case is 1.8 degrees.
Not perfect, needs matching of the switch and the motor, but works and there is no additional gears, cluthes etc. That would probably be better, but I'm not good in mechanics.

Regards,

JG
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.