Linear electrolyte charging?

[~arph]

Hi,

I'm wondering if it is possible to make an elektrolyte charge in a more linear fashion. What I'm trying to do is to slowly fade a LED (in a vactrol) in or out by charging a cap. It all works fine, but I wonder if there is a way of getting a more linear response out of the LDR by having a more linear charging electrolyte?
Any thoughts on this?

Thanks,

Arnoud

[oskar]

Yes...

http://bertrik.sikken.nl/bat/pix/funcgen.gif

If you charge a capacitor over a resistance you get the classic sharkfin look on the curve. This is because the charge-current
decreases when the closer the driving voltage the cap gets.

http://www.tpub.com/neets/book2/3d.htm

oskar

[~arph]

Hi thanks,

I know about the sharkfin. Maybe I should be a little more clear. The LED I'm charging is not used in the signal path. I want to mechanically switch the LED on/off after which it (de)charges at a speed determined by a potentiometer in front of the cap (in say five seconds). I want this to happen in a more linear fashion.
Let me see if I can draw something up here. (pic coming) to make it clearer.

EDIT: here's the crude image



..Strangely the image is not showing.. here's the link

http://www.friedair.com/images/charge.png

S1 determines whether S2 charges or decharges the cap..

Regards,

Arnoud

[oskar]

Not without OP-amps anyway. But what is the circuit supposed to do? This is only an approximation right?

[~arph]

It's driving the speed section of an LFO.. this way you can either speed it up or slow it down to a certain setting after pressing S2 (which is the bypass stompswitch). Using op-amps will not be a problem. So I could in theory also use an opamp like in the triangle wave generator to integrate the switches "squarewave signal" to an approximated linear triwave? that would be fine for me.
What about the ramp up/down method used in the Neovibe adapted LFO on geofex? that one uses a 2N3904 and not op -amps (the p90 version does)

[oskar]

Ah... ramp up/down is what you're looking for. And off course you can go discrete too... I haven't tried it though.
Also you could check out the LM13600 datasheet. It has ramp up/down in it.

[~arph]

Thanks,

The LM13600 looks promising. I'll have to get one and test it out.

[~arph]

Hey, I have an idea... would it be possible to use a FET as a charging resistor for an electrolyte and have the voltage over the electrolyte control the resistance of the FET so the voltage on the cap determines how slow/fast the cap will charge. Well it's just a thought, I haven't had time to really think it over so I could be horribly wrong.

[R.G.]

An OTA like the LM13700 or now obsolete 3080/3094 are the only simple way to linearly charge and discharge a resistor to ground. You can get linear charge/discharge on a floating cap by using it as the feedback element on an opamp.

As always, there are issues. Unless your vactrol can run on no more than 2ma, the OTA is out. This is because the 13700 burns up if you get over 2ma in the Iabc, which is the same as the current in/out of the output pin. Maybe that's OK, maybe not. Looks like what you really want to do is ramp the current in an LED. You could put the LED in the feedback path of an opamp, set so the LED always sees a current in the correct direction, and then control its current directly with a voltage to the input resistor of the opamp. That converts the problem to getting a voltage ramp up/down, which is easier to solve.

The light from an LED is almost but not quite linear. The resistance per unit of brightness on an LDR is NOT linear. So if linearity is what you're looking for, you have some more problems.

The reason this all comes easily to me is that I went through that same path. I found an alternate way. The ramp up/down LFOs for the Easyvibe and P90 at GEO are my answers to this problem. This LFO produces an output frequency that is a linear function of an input voltage. I used a pot and a cap for ramping, but you can replace that with any voltage signal and it still works the same. If you replace the speed pot with an OTA which is set up to follow a input voltage with a limited slew rate, you're there. I didn't want to shoot over everyone's heads, so I left it a cap to ground.

[~arph]

Thanks R.G.

So in fact it boils down to the fact that the LDR light/resistance response is non-linear as well and both curves result in a rather useless resistance change curve. I think I should drop the whole vactrol speed control and switch to an op-amp based one.
So (on a sidetrack now) purely theoretically will this work for flattening out the charge curve on a cap?


http://www.friedair.com/stupid-idea.jpg
Excuse my drawing skills...

Regards,

Arnoud

[Sir H C]

If you want to get a real triangle wave out, you need to use current sources to charge/discharge the cap.  That would be perfectly linear (in the ideal case, which you can come close to).

You have to be able to switch in/out a current source to charge the cap then one to discharge it.  These sources should be the same value so that you get equal rise/fall times.

To build this circuit requires a bunch of bits, and a comparator to switch from one current source to the other.

[~arph]

All I was after was getting a linear ramp up/down on a LFO using a LED/LDR combination for speed control and using a capacitor charge/decharge to drive the LED. This however does not appear to be feasable due to the nature of the cap charging and the LDR's response to light. So I have to find another solution