Phaser rate ramp help!

[armdnrdy]

I'm working on an eight stage vactrol phaser design that I found in "Electronic Music Circuits" by Barry Klein.

It has a few bells and whistles....and I thought I'd add some more.

One of the "bells" is a "Vibrato" switch which cuts the dry signal going to the mixing node.

This is where I had an idea...why not add the vibrato switch as a 3PDT footswitch, switch a "vibrato" mode indicator light with the second switch section, and switch between the phaser rate pot and an added vibrato rate pot with the third switch section.

That way the vibrato can have it's own rate independent of the phaser and can be switched on and off on the fly.

This was all good....until I had the notion to add a ramping effect when switching from phaser to vibrato. (different speeds)

Now I've studied R.G.s LFO ramp documents as well as the Roland AP5 and AP7 ramp circuits but.....the LFO circuit I'm working with is a bit different.
In all of the aforementioned circuits the Rate control is of the voltage divider type. The rate control I'm dealing with is a variable resistor.

Here is the redrawn LFO circuit without the vibrato switch/pot:



And here is the same LFO with the "vibrato" switch/indicator/rate pot: (Switch section A which cuts the dry signal for vibrato is not depicted)



Does anyone have suggestions for a good starting point to add a ramp mod to the second LFO circuit?

Thanks

[duck_arse]

use the diodes across the rate pot method, as shown here. the oscillator rate will increase when ramp is selected as the diode shorts the resistance for half a cycle.

[armdnrdy]

Thanks for the reply.

I've been looking over that schematic as well as this one using the same approach:



I'm wondering if the diodes change the character of the sweep due to the diodes voltage drop....and if so how to compensate for it.

[duck_arse]

the diodes convert the triangle wave to a sawtooth. when they go forward bias, they "short" their resistor, vastly reducing the charge/switch time. when you switch to the other diode, it shorts on the opposite half cycle. I've bredded that circuit plenny times, works pretty well.

not sure what you mean by the character of the sweep, changing from tri to saw is a fair change.

[armdnrdy]

"not sure what you mean by the character of the sweep"

"the diodes convert the triangle wave to a sawtooth."

I think you answered my question.

I'll give the diode method a try.

Thanks

[R.G.]

At some point in making oscillators more complex, it becomes worthwhile to change over to voltage control so you can generate the control path you want by making a voltage emulate it, and then having the oscillator follow suit. Generally this offers wider range, and independent control of the speed, path, etc. for the control.

That was the motivation for this: http://www.geofex.com/FX_images/ramp-lfo.pdf
and this:
http://www.geofex.com/FX_images/p90ramp.pdf

I don't know if that's one of the ones you looked at.

The nice thing here is that within wide limits, the speed of the LFO is linearly controlled by the voltage fed to its control input. That can be switched semi-instantly from one value to another, or smoothed out by caps, integrators, etc., or stepped. This is an adaptation of an even more flexible circuit I did for an eight stage version of the P90 at one point. It's very flexible, and only needs one dual opamp and a couple of transistors, one of which is purely a buffer to feed it a low impedance control voltage.

[armdnrdy]

Thanks for the info R.G.

I've printed out and have been studying both the modified Easyvibe and P90 LFO drawings.

I guess it would be easier to drop in the modified P-90 LFO since it does exactly what I want it to do.

I believe for some people, it's human nature to over think things.....and then the other swarm of people just don't think. 

[R.G.]

I've breadboarded these, and the VCO does work as expected. A couple of things are critical, though. The opamp used simply must have an input common mode range that includes the negative supply. The LM324 and its dual cousins the LM2902 and 2904 do this, as does the LM358 and other single-supply opamps, and all rail-to-rail input amps. The NPN used for pulling the input low must saturate to very low voltages at low currents. The 2N5088 works well, but probably many others would too.

You may have to watch the hysteresis and bias voltage for the Schmitt trigger to get it working right. I several times made it not work by asking the Schmitt to trip at voltages the integrator could not reach, but I think both of these work as intended.

You can do real oddities with this. One is to combine a pseudorandom voltage source and this LFO to have the LFO frequency shift around mysteriously. The R and C on the input voltage determine how fast it shifts from frequency to frequency. This is another place to play with it. With little low passing on the control voltage it jumps, with some slowing down on the control voltage it ramps from frequency to frequency.

[armdnrdy]

Thanks again for taking the time to explain things.

Now I have to do the opposite of what is usually needed......I have to convert the LFO to +15/0/-15 bipolar supply.

Is there anything that I might need to know about this circuit to accomplish this?

[R.G.]

Maybe. What is the actual modulation that's happening? Is it currents to LEDs? Or voltages to FETs?


This circuit LIKES single power supplies. But if you're sweeping LEDs, the conversion is simple.  The LED/opto isolation lets you do whatever for the LEDs that give them the right current, and they run the other stuff for you.

[armdnrdy]

Aaah!!!

Brilliant! 

The LFO is running Vactrol LEDs.

So the LFO can stay "as is" fed by +V and GND.

[R.G.]

Even better is the issue of current reuse. By far the biggest current drain this thing makes is the LED currents. You're using four or eight LEDs.

If you contrive the output to drive a voltage-to-current converter and stack the LEDs in series from one of your power supplies, all the LEDs use the same current, not N times an LED current.

Consider the creative use of current mirrors to drive your LEDs. It's really handy, and much more accurate than using a resistor from your LFO voltage.

[armdnrdy]

I'm using 4 LEDs. (dual vactrols)

The image in my first post (right side) shows a voltage follower and a 100K trimmer. The trimmer (TR2) has an arrow coming out of  lug 3 labeled "to vactrol LEDs" which connects to the anode of the first LED.

The LEDs are then ran in series, (cathode to anode) terminating at the last LED, cathode to a 10K resistor to -V.

I'll check out voltage to current converters and current mirrors for use in this application.

Thanks for the info.

[R.G.]

An issue is the LED forward voltage. You get no current, and no light, when the LED forward voltage is below - um, 1.2V for the Vactrol? Can't remember the spec exactly, but its in the low 1's.

The other lurking issues are the standard ones for LFOs - idle point, and increasing width up from idle, down from idle, or balanced around idle.

Your first circuit shows the LFO being fed to the LED driver from a width pot tied to ground. Since the circuit is fed from a bipolar supply and biased at ground, the LFO waveform will be balanced around ground. What current the LEDs get will depend on what voltage their return current is tied to. If the LEDs are tied to ground for return current, the balance above/below ground LFO drive voltage will put current in to them when the voltage on the LFO exceeds four LED drops above ground. That's about 4.8V or more. They'll hardly ever turn on.

If you think about it a little bit, the output of an LFO is not the voltage waveform. It's whatever currency the actual modulated thing needs. That may be a voltage or a current, and it may have a positive or negative offset from the "baseline" or "idle" current for the device. JFETs in a phaser need a volt or two of negative bias, and an LFO that wiggles around that. Some tremolos using bipolar transistors for modulation may use a zero current for idle, and only run current into the base (at 0.4-0.7V!) to make the transistor conduct and gain to go down; or for a maximum current all the time and reduction in current for the gain to increase.

Your vactrol LEDs need an idle voltage of either four times one LED voltage, two times an LED (with two each in parallel) or four times (with all in series). What you want to change is the LED current, and let the voltage fall where it may. There will be a min, max, and idle current to be set for them, and light output will be set by that.

To do that, you have to have a setup so you can let the LEDs have enough forward voltage to go from just-barely-zero current to full current.

One way I might approach this is to use a current mirror run from the +15 supply. This would be two bipolar transistors, with the input transistor emitter grounded, base and collector tied together. The output transistor is emitter to ground, base to the input transistor base/collector, and collector the pull-down output. When you feed a current into the input transistor's collector/base, the base voltage sets to whatever base voltage makes the collector conduct that current, and this will be something like 0.45-0.6V most of the time. The output trannie has the same base voltage, so its collector current matches (at least to the degree that it is identical to the first one; it happens that under these conditions, transistor match is easier than in most setups).

So I might set up the mirror, put all four LEDs in series from the power supply to the collector of the mirror output. Then I'd work on feeding the input the current I want, this time having only a diode drop offset and a fixed voltage to ground to work on. From there, you have better control of the current into the transistor by voltage.

You could then set up your LFO to be offset from ground by a bias voltage - which is a pain. Or you could feed a constant current into the mirror's input via a resistor and use your LFO output voltage through a resistor to subtract current from the constant current in the resistor.

In any case, you need to know the min, idle, and max currents the LEDs need, and provide for that, including what voltage feeds them and what voltage that current returns to. Ground is a very nice place for return currents if you can get it there. 

[armdnrdy]

Once again thank you for the direction R.G.


I spent yesterday going through your last post and researching current mirror/LFOs to try to find a similar application.

I didn't find anything close.

I put together this drawing of how I understand your description.



There were parts mentioned that I wasn't exactly sure of.

I was wondering if a trimmer should be placed between Q2 emitter and ground or between Q2 collector and the first vactrol LED for adjustment.

[R.G.]

I spent yesterday going through your last post and researching current mirror/LFOs to try to find a similar application.
I didn't find anything close.

I'm not terribly surprised about nothing close. I make these things up as I go...   

I put together this drawing of how I understand your description.

And that's pretty much correct.

I was wondering if a trimmer should be placed between Q2 emitter and ground or between Q2 collector and the first vactrol LED for adjustment.

Both are bad ideas. Do any trimming on the input side, where you convert incoming LFO voltage to input current.

The resistor in the emitter of Q2 imbalances the current mirror, unless you do both Q1 and Q2 equally. That's OK, and good in some situations. A resistor of 0 to 22 ohms or so is common in precision current mirrors.

The resistor in series with the LEDs does nothing but limit the voltage available to the LEDs.

The resistor from the LFO to the first transistor is what scales the current.

[armdnrdy]

Thanks for the reply,

I'm surprised I got that much correct!

So...I will remove R3 and replace R1 with a trimmer?

I tried to find an example of how to calculate the value for resistor R2.

I'm a bit in the dark as to what I'm looking for.

[R.G.]

Let's put the LFO aside for a moment and just look at the current mirror, LEDs R2 and R3, as well as what we're trying to do.

For the LEDs, we have the idea that the resistance on the Vactrols is what we're trying to change. I suspect that somewhere there is a chart of resistance on the LDR side of the Vactrol as a function of LED current. If the LED current is zero, it's really dark in there and the LDR resistances go as high as they can. If the LED current is some large value, then it's as bright as the LDRs can take for changing resistance, so they are at their minimum resistance.  We change resistance by changing LED current, and that's pretty much a single line on a chart. 0ma = 2Meg (for example) and 5ma = 4K maybe, just to pull some numbers out of the air. So we can make the resistance be anything between those two by changing LED current.

The phaser sections have a sensitive range, generally about 10:1 to 100:1, over which resistance change makes a phase shift change. Resistances outside that range don't move the phase much. Most phaser cells I've seen are about 10:1 max, and the resistances tend to be between a few K and several hundred K. So if we knew the curve of resistance versus LED current, we'd have an answer. Or four of them, because the Vactrols are probably not identical; but close.

Those min and max currents tell you the min and max currents you have to get the current mirror to pull through the LEDs, and also the min and max currents you have to put INTO the current mirror input.

Assuming for the moment that R1 = three inches of dry air     we could change R2 to get the min and max currents we got from above. The current is always (15V - Vbe)/R2, and we can set the current to any value by selecting R2. But that's a fixed current and does not change.

Now we decrease R1 and mess with the LFO. Let's assume for the moment that the LFO is not oscillating, but is instead a variable voltage DC power supply.

Since the voltage at the base of the current mirror input is always 0.6v +/- a bit, then the current that the variable DC level subbing in for the LFO adds and subtracts from the current mirror input is always (Vlfo-Vbe)/R1, independent of whatever R2 is doing. If Vlfo-Vbe is positive, it adds a current of (Vlfo-Vbe)/R1 into the current mirror and increases the current the LEDs see by that amount; it's added to the R2 current. If the voltage from the ersatz LFO DC power supply goes negative, then it SUBTRACTS a current of (-Vbe-Vlfo)/R1 from the current R2 is already adding in.

So R2 sets a DC "bias current" that remains fixed for the LEDs. The LFO voltage adds to this current if it's above Vbe for the current mirror, and subtracts from it to the extent that it's below Vbe.

Now we go back to that current versus voltage thing for the Vactrol LEDs. You adjust ( or calculate) R2 to be the DC bias current you want, and the LFO swing min to max to make the LED current change up and down as you like it. In practice, you can either change the LFO peak to peak voltage to change "width" or change the R1 value to scale it down. Either way works, and you may want to do both for tinkering.

[R.G.]

Forgot - R3 does do one thing for you - it soaks up some of the voltage across Q2, and might be useful for making Q2's power dissipation be smaller if Q2 was dissipating a lot of power. The LED voltages will not change - it will always be 4x Vled, about 5-7V maybe. If R3 is zero, then 15V - (4xVled) is dropped across Q2. This will be about 8-10V, and Q2 really should not dissipate more than 100mW, which means the max current you'd want to use is 0.1/10 = 10ma. That's probably OK. Another thing you could do is to put the blinky-rate LED in series with the Vactrol LEDs and a resistor in parallel with it so it ramped on and off with the Vactrol current and ate up some of the excess voltage/dissipation away from Q2. You might still want an R3 to eat up more of it, just making sure that the value of R3 is small enough not to rob the LEDs of enough voltage at max current.

[armdnrdy]

R.G.

I really want to thank you for all of the information you've been kind enough to provide!

Your explanation has given me a much better understanding of this section!

I'll make some revisions to the full schematic and post it.

Thanks again!