LDRs as pots?

[Gregg]

This also looks like a good idea :

http://www.elecdesign.com/Files/29/9944/Figure_01.gif

I liked the idea for a kit or why not a presoldered board /no way I'm soldering TSSOP packages!/. I was thinking more of a /relatively/ universal tube preamp - the ten or twenty most famous guitar preamps in one. Imagine a tube amp with six knobs only but capable of much more than that.

[Don Symes]

This also looks like a good idea :

http://www.elecdesign.com/Files/29/9944/Figure_01.gif

I liked the idea for a kit or why not a presoldered board /no way I'm soldering TSSOP packages!/. I was thinking more of a /relatively/ universal tube preamp - the ten or twenty most famous guitar preamps in one. Imagine a tube amp with six knobs only but capable of much more than that.

The difficulty is that the LED current/brightness relationship isn't real linear and neither is the CdS cell's illumination/resistance (AFAIK). What you need is a simple, predictable way to replicate a known rotation of a pot of known value and taper - that way you can keep the user interface's 'feel'. Using an op-amp to 'servo' the resistor value in a photocoupler is the closest you're going to get.

The universal preamp and/or amplifier - I called my project ToneLego and was gonna house it in an old Hot-Swap Drive Array chassis - each module a preamp, with a couple of low-power power amps in the back. All I can say is: Don't start down that path until you have a couple of solid builds under your belt - the complexity is factorial, not exponential.

Hope this helps!

[Steve Conner]

Gregg, all I can say on the subject of SMT homebrewing is http://scopeboy.com/elec/misc/prototype.jpg so there

I think Don's circuit is probably the best basic building block to use for a system like this. It could probably be simplified somewhat (especially if the op-amp has enough beans to drive the LED directly) and adapted to take a voltage input from a DAC instead of a resistance from a digital pot. Driving the LED directly would make the thing a lot more stable too, since the MOSFET adds gain in the loop and the R-C network on its gate adds more phase lag. That was probably part of the reason for the low-frequency oscillations.

Another cool method for computer gain control is the multiplying DAC, which acts somewhat like a pot. If you only needed 16 steps, say, you could make a high voltage multiplying DAC using four LDRs and a R-2R ladder network. The LDRs are just used as on/off switches, so they don't need to be specified so tightly.

If you needed a variable resistor, you could use four binary weighted resistors and four LDRs to short them in the right combinations.

P.S. I love the ToneLego thing! A Seymour Duncan Convertible for the internet generation

[Ray Ivers]

Steve,

all I can say on the subject of SMT homebrewing is http://scopeboy.com/elec/misc/prototype.jpg so there

I couldn't have said it better myself.

Ray

[Gregg]

It could probably be simplified somewhat (especially if the op-amp has enough beans to drive the LED directly)

Are you talking about an opamp like this:

http://www.analog.com/en/prod/0%2C28...8532%2C00.html

[Steve Conner]

No, that's overkill. LEDs only need about 20mA max, so most op-amps can probably drive one.

[Gregg]

Then maybe NE5532 would be a good contender?

[Don Symes]

Gregg, all I can say on the subject of SMT homebrewing is http://scopeboy.com/elec/misc/prototype.jpg so there

I think Don's circuit is probably the best basic building block to use for a system like this. It could probably be simplified somewhat (especially if the op-amp has enough beans to drive the LED directly) and adapted to take a voltage input from a DAC instead of a resistance from a digital pot. Driving the LED directly would make the thing a lot more stable too, since the MOSFET adds gain in the loop and the R-C network on its gate adds more phase lag. That was probably part of the reason for the low-frequency oscillations.

Hey, thanks! I hadn't really thought about an op-amp with an output that would be happy at 20mA. Slowing the response adequately was the problem that was eating me up.

Anybody up to designing a stable discrete diff amp that doesn't need a ton of trimming?

If you only needed 16 steps, say, you could make a high voltage multiplying DAC using four LDRs and a R-2R ladder network. The LDRs are just used as on/off switches, so they don't need to be specified so tightly.

Two points here - the Mesa patent uses what amounts to an audio-taper resistor ladder, so that implementation might infringe enough to cost a mint in defending a civil suit. Also, the R-2R ladder really only works for linear pots, and at least half the pots we need are audio tapers of various slopes.

P.S. I love the ToneLego thing! A Seymour Duncan Convertible for the internet generation

Viso drawings are available - as of abandonment a couple of years ago.

[Tiago]

Hi all!
Some time ago I did some research on how to digitally control a potentiometer.
LDR's (optocouplers) look like a perfect approach, but IMHO they are not.
Mesa Boogie, ENGL, and even ADA 20 years ago used them one their programmable preamps.
Mesa Boogie use lot's and lots of circuitry to control them (as you can see on the triaxis schematics), maybe the aren't aware that there is new technology out there which would simplify it. I don't have access to ENGL's preamp schematics, but by reading their manual, it seems that they use a lot less circuitry by using micro-controllers.
What I've seen about them is that their specs will change a LOT with temperature changes. That's why these manufacturers use some king of feedback to control them.

I think that where they place a Optocoupler to control something (gain, bass, volume, etc...) they will place another identical optocoupler close to this one so that the two are subject of the same climatic conditions and the second one will just be there for feedback purposes. As ENGL's says in the manual, for temperatures above 50ºC theres no reliability on these controls.
And we might think that a Optocoupler is "transparent", but they are not. As you can see in some datasheets, their produce some distortion, and this dirstortion percentage increases as signal leves increases. Altough, guitar preamps are mostly distortion generators, so we might get away with a tiny bit of it from the LDR's circuitry.
So, in order to control them we could use microcontrollers(I would't imagine my life without them....). PIC's for example have very cool built-in features, like a built in USART, very usefull for MIDI communication, several built in ADC's, very usefull for the feedback, a built in PWN, usefull to control the Led fo the Optocouplers if you dont want to use a DAC. Not to mention Flash, ROM and RAM memory, and the option to use a bootloader! Very usefull for debugging! By the way, features like PWM, are not restricted to the hardware capabilities. If we need more, it can be implemented by software.

Altough, in the end, I prefered the Soldano approach, using motorized potenciometers...in better words, normal potentiometers attached to stepper using heatshrink tubing.
Signal integrity is the best. Maybe it has some mechanical stress...but as far as my experience go with stepper motors..I never saw one going bad.

But if I had to chose between non-mechanical approaches, I would chose the Marshall JMP1 approach, simply controlling the signal using digital pots. I had one JMP1 and I liked its sound (except the OD2 channel).

Just my 2 cents....My heater is on, so its 5 cents

[Ray Ivers]

Tiago,

Do you (or anyone else) know of a good, easy-to-read schematic of a motorized-pot setup?

Ray

[Tiago]

Yes I know, but it's in my head and my breadboard...
But I will show some of the resources that helped me to figure it out.
This application note (http://www.st.com/stonline/products/...re/an/1679.pdf) from ST Microelectronics, shows what you need to know to drive stepper motors and it includes a schematic of the worldwide famous L298 and L297 combo. I also have to mention that I did not used the L297/L298 combo, but it would be much more simpler if you use it. I used only the ucontroller (PIC 16F876) and the L298, which is the motor driver chip. By the way, I used a unipolar motor.
Using only the L298, you will have to deal with the connecting / disconnecting order of the windings of the motors. But using the L297 you just have to say were you want it to rotate, if would be half or full step...etc(check the app note).
To know the resistance of the potentiometer, we have to use a dual ganged potentiometer. One for the part we want (Gain, presence, etc), another for the position feedback. For feedback, we just have to connect the potentiometer to an A/D converter pin of the PIC. (I connected one of the pot "legs" to Vcc,another to ground, and center one to the PIC). I also have to mention that for the higher resistances found in most tube amps, the center pin of the potentiometer should be buffered (just put an opamp there).
For the hardware, I think that's all. Now it just needs to be programmed so that the motor rotates until the feedback value is the one we want (I also advice for some tolerance there). The pot then, could be also used as an encoder.

To me, the electronic part isn't the worst, the worst is to tie the motor to the potentiometer while having the shaft available to the front panel of a amp or preamp...

To motors don't need to be high-speed high-torque million dolar ones. Size 17 (I dont know what this value really means or which units is it) 200 steps to start are great, ebay is full of them. For my tests I used a small motor from a hold printer (a 48step) and it was ok, but just "ok", not great.
The potentiometer I used was an ALPHA dual gang (you can rotate the shaft with an allen key from the opposite side of the shaft).

Sorry for my broken English, if you can't figure it out by this tips, tell me of a good and free schematic capture program and I would draw it to you.



EDIT:
I've just remembered that some time ago, I founded "Diy motorized volume control" project on the net. Here it is: http://home1.stofanet.dk/hvaba/stepp...e/stepper.html
In the middle of the page there is a very good schematic, just like you want. The L293 controls the motor, and the Rservo is the feedback source.

[Don Symes]

A (probably) overlooked advantage of the servo-pot idea is that the pot connected to the 'setting' knob and the pot that reads the servo/stepper position are the two that have to match. The 'controlled' pot (that actually operates the amp) can be whatever value/taper is required by the circuit. All you really need is a fraction-of-rotation to sense and apply.

A multi-channel ADC to read the knobs and the feedback pots and a multi-channel stepper controller tied to something like this gets you cooking.

[Gregg]

For various reasons I definately don't like the idea of motorized pots and stuff. If this is the alternative I'd prefer to skip it altogether or better wait for a 100V digital pot or something.

[loudthud]

The circuit posted by Don is like one that was posted on the old Ampage as a pdf file. I don't remember who posted it but I have a copy. The date on the copy I have is Jan6, 2005. It's actually a brilliant circuit. To make it into a pot I would remove the ground from the lower end of the programmable pot and run it to an identical circuit. Leave the wiper of the programmable pot grounded. Now the second circuit controls it's LDR to mimic the resistance of the bottom portion of the programmable pot. Hook the LDR's from the two circuits to form a voltage divider and you have a pot.

While I can't speak to the distortion of LDR's in current production, I know that the ones used in the Tektronix AA501 distortion analyzer had very low distortion. I'm pretty sure they were Vactrols.

[Steve Conner]

I bet money there will never be a 100V digital pot chip. The IC maker would need to move his digital pot design to a whole other semiconductor process, and he wouldn't invest that time and effort unless he thought there was a big market.

How about redesigning a tube circuit to work with regular digital pots that run off +/-15V? I'm sure with some thought you could come up with a circuit that covered tones similar to classic amps, but never applied more than 30V p-p of signal across any one pot. It may involve throwing away a few dB of headroom here and there, but does that really matter? A case in point is a passive tone stack which may have 15-20dB of loss anyway: you could make the tone stack out of digital pots and put a potential divider ahead of it (or ahead of the cathode follower driving it a la Marshall, unless you think the heavily driven CF adds to the tone) and just make up the extra 6dB or 9dB or whatever, in the following stages.

I've managed a few reasonable hybrid designs, my favourite is the "Toaster" 2-band parametric EQ, a cathode-follower-driven tone stack that went on a very long trip and ended up with 10 op-amps bolted onto it. The way I designed it, the tubes are guaranteed to always clip before the op-amps do, therefore all of the knobs in this circuit could be replaced with a mixture of digital pots and multiplying DACs, with no other mods needed.