Univibe/Neovibe circuit insights

[R.G.]

I got to thinking about the LFO in the 'vibe.

Other guys dream about boats, cars, women, money; I daydream about circuits. 

Anyway, I had a great time messing in the simulator with it and found out a couple of things that I then verified on a breadboard.

1. The vibe LFO is critically dependent on some unsuspected parts - the 2.2M biasing resistor and the back-to-back diodes on the middle capacitor. If either of these is wrong, it will not start.
2. "Wrongness" means "too low an impedance". You can't, for instance, use a 220K for the 2.2M - it won't start. A bigger surprise is that power diodes like the 1N400x series won't let it start. I know that these can be used in real life, but that indicates that there are some values of 1N400x that will not let the LFO start. 1N4148 seems to be really very good there in all conditions.
3. You can extend the range of the LFO by subbing in an N-channel JFET for the first transistor. I used the 2N5485 model, verified on the breadboard with a 2SK30a.
4.This runs the impedance up so high that you can then raise the biasing resistor to 22M. Now the frequency range is wider. I hadn't thought about it, but the impedance of a 1uF cap at 1/2Hz is Z = 1/(2*pi*1/2*1E-6) = 320K. So a 2.2M is marginal for "not mattering" compared to the 1uF conducting the feedback signal into the two transistors. Likewise, it gives you room to lower the 1uF caps down to perhaps 0.22 or 0.1 and keep the same low speeds with bigger pot resistances.
5. A JFET opamp set up as a unity gain buffer works there and can directly replace both transistors. This makes the 10K to ground from the emitter superfluous. Should start more reliably in the real world.
6. The unloaded voltage output of the LFO rises with frequency. This probably has something to do with being a first-order compensation for the poorer response of the lamp at higher frequencies.
7. I've messed a bit with replacing the input preamp with an opamp. No good results yet.

[Nitefly182]

Hey RG - do you know which caps around Q11/12 are the specific caps for the LFO range? I thought they were c19 and c22 on the GGG board but lowering those two didn't up the LFO rate.                                       

[ohmelter]

Hey R.G. Interesting stuff.
In part three of your "insights" you mention substituting an N-channel JFET for the first transistor...this would be the first transistor of the darlington pair in series with the 2.2m resistor, correct?

[slideman82]

Other guys dream about boats, cars, women, money; I daydream about circuits. 

That's not a matter... it's worst if you dream about eating MN3007 and other 2x4 IC's... I don't remember them, but there's was a MN3007, I'm sure!

[R.G.]

Hey RG - do you know which caps around Q11/12 are the specific caps for the LFO range? I thought they were c19 and c22 on the GGG board but lowering those two didn't up the LFO rate.

It is C19, C20, and C21. those all need to be the same value. But because of the impedance things I mentioned, you may not get much change if you don't also increase R40 proportionately.

In part three of your "insights" you mention substituting an N-channel JFET for the first transistor...this would be the first transistor of the darlington pair in series with the 2.2m resistor, correct?

Yes - Q11. However, this does change the DC bias point of Q12 upwards by Vth of the JFET minus the old transistor's Vbe. If you use a 1V - 4V JFET like the common 2N5292, 2N5485, or 2SK30, you might get a JFET with any voltage in that range; let's say it's 3V. The base and emitter of Q12 would be raised by 3V-0.5V = 2.5V. This may make no difference at all, but it's worth checking to be sure. A J201 could be used as well, probably.

[Nitefly182]

Hey RG - do you know which caps around Q11/12 are the specific caps for the LFO range? I thought they were c19 and c22 on the GGG board but lowering those two didn't up the LFO rate.

It is C19, C20, and C21. those all need to be the same value. But because of the impedance things I mentioned, you may not get much change if you don't also increase R40 proportionately.

In part three of your "insights" you mention substituting an N-channel JFET for the first transistor...this would be the first transistor of the darlington pair in series with the 2.2m resistor, correct?

Yes - Q11. However, this does change the DC bias point of Q12 upwards by Vth of the JFET minus the old transistor's Vbe. If you use a 1V - 4V JFET like the common 2N5292, 2N5485, or 2SK30, you might get a JFET with any voltage in that range; let's say it's 3V. The base and emitter of Q12 would be raised by 3V-0.5V = 2.5V. This may make no difference at all, but it's worth checking to be sure. A J201 could be used as well, probably.

So if I halved the values of those three caps I would have to double the value of R40 to get everything going?

[R.G.]

That's what I would do. At some point the impedance into the base of the darlington will limit you, but that's relatively unpredictable. When that happens, you'll have to go over to a JFET.

[mac]

A bigger surprise is that power diodes like the 1N400x series won't let it start. I know that these can be used in real life, but that indicates that there are some values of 1N400x that will not let the LFO start. 1N4148 seems to be really very good there in all conditions.

I had that problem last year. I used 4002 and couldn't get oscillations. I tweaked everything except diodes!!! Too complicated, finally I gave up.
But a few weeks later a friend who build commercial pedals gave me a nice sounding univibe clone as a gift.

Other guys dream about boats, cars, women, money

mac

[R.G.]

'Nother insight.

The 330pF compensation cap in the 'vibe is probably not the best value or position in the circuit. It makes for a big almost-oscillating peak out in the 1MHz region. Better to put a 30pF from collector of Q2 to base of Q2. This is much less marginal on stability, and much closer to the concepts of the diffamp-input opamp with voltage-gain middle stage and a buffer output. Q1, Q2, and Q3 perform those functions almost exactly, but modern opamp practice is to compensate in the voltage-gain stage.

I wonder why Mieda-san didn't do that?

[Eb7+9]

I got to thinking about the LFO in the 'vibe.
Anyway, I had a great time messing in the simulator with it and found out a couple of things that I then verified on a breadboard.

great insights fer sure ...

http://www.diystompboxes.com/smfforum/index.php?topic=38977.0

for the last couple of years I keep the bipolar darlington but replace the 2M2 resistor by something between 4M7 and 10M
to give nice extension on the lower end range ... if you keep playing with the simulator you might stumble on how to increase LFO strength at the lower speeds ...

[bluesdevil]

Just thought I would add this: I've been messing around with my ol' Neovibe build trying to get a less choppy bass heavy throb and discovered I needed a lot more resistance (around 400 ohms) on the emitter of the bulb driver transistor and played around with the value of the resistor going from ground to the same transistor's base and found that reducing it to 33k did the trick. Basically, I couldn't bias the bulb dim enough with the stock parts values.... maybe using a 2n5089 instead of 2n3904 for the driver had something to do with it?
   While playing around with it I loosened something up and the LFO freezes once in a while until I wiggle the board around a bit..... maybe a good excuse to build another with the new insights mentioned here in consideration. Also this time with a bulb socket!!!!!!!!!!!!

[mac]

Is the bulb/LFO better than a LED/Phase90 LFO?

mac

[mike_a]

Why not use an opamp LFO on the univibe?   
It can handle the high voltage and you only use a single pot.
It also will not exhibit the same artifacts that the transistor LFO exhibits...
It shouldn't have any effect on the sound, as long as the bulb is driven properly, which can be done with the same old bulb driver.

[R.G.]

Largely because the LFO/lamp artifacts are a big part of the 'vibe sound. The 'vibe is the only widely available phaser with a sine wave LFO. The others are by and large triangle. 

You CAN replace the darlington transistor with a JFET opamp; that's obvious, and does the same job as the transistors. You still get the sine wave. You could use an XR 2206 sine/sqare/triangle chip instead of the LFO, but that's a 16 pin chip. There are no reliable two-opamp sine wave LFOs I know of, but there are a plethora of ways to do it with more than two opamps or dedicated chips.

Getting a sine wave LFO is not a simple task. The 'vibe's precursor, the RT-18 (Resly-Tone...) had a multivibrator LFO.

[mac]

I guess that the LED turn on/off is not as good as the lamp.

mac

[R.G.]

It's the other way round - the LED is much, much faster.  The bulb has a lag time because the electricity has to heat up the filament to glowing; then the filament has to get rid of heat by radiation to cool off when the electricity gets less. Both processes are heavily nonlinear, so the bulb's response to electrical drive is at best, lumpy, and the lumpiness varies with speed. If the electrical variation is too fast, the bulb simply can't follow it. That's why light bulbs do not flicker, even though the electrical power through them reverses at power line frequency.

[mike_a]

But there are sine oscillators using opamps.
They have a bit more parts than their triangle counterparts, but not that much more.

It is also possible to approximate a sine wave using a triangle wave with clipping diodes.
It won't give a perfect sine, but I highly doubt that anyone can hear the difference.

[Nitefly182]

But there are sine oscillators using opamps.
They have a bit more parts than their triangle counterparts, but not that much more.

It is also possible to approximate a sine wave using a triangle wave with clipping diodes.
It won't give a perfect sine, but I highly doubt that anyone can hear the difference.

I dont see the point of this argument though. The design works well and people want vibes because they sound like vibes. Everything about them is what makes them sound unique and why they are just more organic than a vanilla phase 90.

[R.G.]

But there are sine oscillators using opamps. They have a bit more parts than their triangle counterparts, but not that much more.
It is also possible to approximate a sine wave using a triangle wave with clipping diodes. It won't give a perfect sine, but I highly doubt that anyone can hear the difference.

You are correct on all counts.

I should have said that "getting a good sine wave oscillator with opamps is not all that simple".

There is presumably some reason that one goes from a two transistor circuit to an opamp circuit. Usually that is more accurate performance, wider range, etc. It is hard to replicate the oddities of the transistor LFO/bulb setup in the vibe with opamps in a simple circuit that cures some of the vibe oscillator's ills and at the same time preserves some of the quirks people like... and at the same time is a simple enough circuit to be easily buildable and won't need a day or two a week handholding builders.

My point was "Largely because the LFO/lamp artifacts are a big part of the 'vibe sound." which I think is accurate.

I dont see the point of this argument though. The design works well and people want vibes because they sound like vibes.

I'm with you.

[mac]

Sorry RG, I was not clear.
I am aware of the effect of temp as a function of current. Anyway, thanks.

What I was trying to say is that the LED has a narrow voltage range in which it goes on and off.  What about the light radiated as a function of current, voltage? Maybe LEDs are less or more linear than a bulb, I do not know, but IMHO the bulb is more realistic at low speeds. The LED/LDR-P90 LFO or FET/P90 LFO seem to stay in the "sine wave" tops for a while, that is, like taking a rest without an audible change of phase. At least, this is the difference between my P90 and my vibe clone; could be just my "experience"

I dont see the point of this argument though. The design works well and people want vibes because they sound like vibes.

... or peolple like the vibe because Jimi liked it?

IIRC russians say: if something works ok then do not change it.
I agree that the vibe should be left as is.
But for educational purposes talking about the flaws of the vibe, and how to improve it, although in theory, is a good thing.

mac