Transistor Phase Shift Oscillator and Phase Splitter - Recommended Schematic?

[frequencycentral]

As part of my seemingly endless research I'm now looking at transistor phase shift oscillators and phase splitters.

I'm looking for a workable schematic for control duties in modulation circuits. Preferably with one transistor as the oscillator and another as the splitter.

I also have a question: usually transistor phase shift oscillators have three capacitors performing the phase shift - can they be polarised capacitors or should they be non-polarised?

Thanks!

[R.G.]

http://geofex.com/FX_images/FakeFenderVib3.pdf

LFO section is bipolar PSO with emitter follower buffer, driving phase splitter.

[frequencycentral]

Thanks R.G. just what i was looking for.

What about caps? Can they be polarised?

[R.G.]

What about caps? Can they be polarised?

If you can guarantee that at no time is the cap reverse biased by even a fraction of a volt, they can be polarized. Otherwise, use NP or film.

[frequencycentral]

R.G., sorry to bother you on this again.

I breadboarded the PSO/buffer last night without success.

I didn't build the Vsrc part - D1, D2, D3, C2, C3, R5, as I assume that's the Vref for Q6.

The transistors I tried it with were BC547 and BC107. I also have a couple of 2N2369. Should it work with any of these, or should I get hold of the transistors detailed in your schematic?

I'm using NP caps.

Also, can you confirm that R27 is 4K7 not 47K, and R29 is 100R not 100K?

Thanks

[R.G.]

I breadboarded the PSO/buffer last night without success.

What exactly does it do? And what are the DC voltages on the pins of the active devices?

I didn't build the Vsrc part - D1, D2, D3, C2, C3, R5, as I assume that's the Vref for Q6.

That's correct.

The transistors I tried it with were BC547 and BC107. I also have a couple of 2N2369. Should it work with any of these, or should I get hold of the transistors detailed in your schematic?

I doubt that the transistors are your problem. It's possible, but unlikely. More likely are the usual suspects - pinout problems, bad connections, unwanted shorts, etc. It may be that you will need to adjust the values of a couple of resistors, but the phase shift oscillator is pretty simple to get running. Let's look at the pin voltages.

I'm using NP caps.

That's fine.

Also, can you confirm that R27 is 4K7 not 47K, and R29 is 100R not 100K?

Yes, that is correct.

[frequencycentral]

Thanks R.G.

I took it apart and breadboarded something completely different right after, so I'll have to build it again!

I''ll post voltages tomorrow.

Rick

[frequencycentral]

Ok, built it again. Voltages:

Both transistors are BC107. Power supply is 9.10v.

Q7
E - 0.017v
B - 0.606v
C - 7.14v

Q8
E - 6.50v
B - 7.13v
C - 9.10v

At the moment I am using a 4k7 resistor in place of the 5K depth pot to simulate maximum depth. The voltage at the conjuction of R27 and this point is 3.24v.

Both speed pots are in the mid position.

I have only built up to the 1uf cap between the depth control and Q9's Base. The voltage at the output of that cap is 0.00v. Should that cap too be NP? I'm using two 0.47 NP in parallel.

[frequencycentral]

I should mention that I'm ultimately trying to make a low voltage version of the schematic in this AX84 thread http://www.ax84.com/bbs/index.php?id=292675

I'll be using 6111 submini tubes with a 30 volt plate voltage. So I'm eventully hoping to run the transistors at 30 volts too.

[R.G.]

Ok, built it again. Voltages:

The transistor voltages look OK-ish. I would prefer if the collector of Q7 sat lower, maybe at about 3V. That could happen by making the value of the 3.3M collector-to-base resistor less, maybe 2.2M. But this should be working if the transistors are the only problem. You could try shorting the 100 ohm emitter resistor on Q7 to see if it starts. If so, it's suffering from a lack of gain on Q7 and you need a higher gain transistor. Beyond that, it's likely the phase shift network, the three caps and speed pot; or a combination of low gain and the network.

Try momentarily shorting the base of Q7 to ground, then releasing it. If it's a low-gain problem, it will ring a couple of times, then fade out. If it's the network, it won't ring at all.

I should mention that I'm ultimately trying to make a low voltage version of the schematic in this AX84 thread http://www.ax84.com/bbs/index.php?id=292675
I'll be using 6111 submini tubes with a 30 volt plate voltage. So I'm eventully hoping to run the transistors at 30 volts too.

Shouldn't be a problem. You may have to dink with the bias resistor to get the transistors right. Also watch for the transistors getting hot. Most TO-92 transistors are 200mW devices. At 30V, 200mW is an average current of only 6.7ma.

I read the thread on the tube vibrato. They're going OK, but the info on the Magnatone is incorrect. I reversed the Magnatone a few years back, and it reduces to a univibe-style phaser. The varistors in the Maggie don't switch as one of the guys said, but instead have a broad, softly curved variable resistance. The LFO moves them around on this resistance curve. I redid the circuit with LED/LDR packs using tubes and it worked fine.

[frequencycentral]

Thank you R.G.

Ok, we have life.

I have replaced both transistors with BC547's. I removed R29 (100R) and took Q7's emitter directly to ground.

I only need one speed pot, so i have replaced R32 and the attached 100K pot with a 100K resistor.

It works well at slow speeds, but when I turn the speed up oscillation stops - it doesn't recover unless i switch the circuit off and on again.

Are my transistors too low gain?

Shouldn't be a problem. You may have to dink with the bias resistor to get the transistors right.

I am encouraged by your comments! I hope to have a submini tube vibrato/phaser soon!

[R.G.]

Ok, we have life.
I have replaced both transistors with BC547's. I removed R29 (100R) and took Q7's emitter directly to ground.

All right! Good work.

I only need one speed pot, so i have replaced R32 and the attached 100K pot with a 100K resistor.
It works well at slow speeds, but when I turn the speed up oscillation stops - it doesn't recover unless i switch the circuit off and on again.
Are my transistors too low gain?

The Phase Shift Oscillator (PSO) is an iconic oscillator in the EE biz. It is the circuit usually used to introduce the concept of feedback oscillation in undergrad courses.

The phase shift per stage is caused by the series capacitor and the resistor to ground. Ideally, there would be a buffer after each cap/resistor section to prevent them loading each other, and all three resistors to ground would change in synchronism, like a three section pot, and the amplifier would have an infinite input impedance, not loading at all.

Such a "perfect" PSO would need a gain of only 8 to oscillate. If you have additional gain available, you can remove the buffers between sections to get the PSO as you have hooked it up, with the three phase shift sections of a cap and resistor each just connected. This needs a gain of about 29 as I remember. Still not huge.

Finally, if the resistances to ground are un-matched, it takes more gain to make it oscillate. The advantage you get there is that you can use only two or even one of the resistors to ground as a speed control. What you lose in doing that is that the necessary gain from the amplifier is bigger yet, and that the range of possible frequencies is limited. A PSO with all three resistances matched and commonly variable has an unspecified-ly wide range. When you fix one of them by letting it simply be the input impedance of the imperfect amplifier, and vary two of them as you might with a dual pot, then the range is smaller, but still quite wide. When you fix two of them and only use one for a speed control, the range is more restricted, and it may take larger gains to get full range out of the single pot.

If it quits at some point, you have changed the combined phase shifts into a place where the gain needed is too high, and it dies out. You can fix that two ways: more gain or make another resistor vary. The dual-section speed pot varying two resistors gives a nice result. You can also just make the trannys have more gain. However, increasing gain also increases the distortion of the sine wave that comes out. So you have to juggle the two effects.

[frequencycentral]

OK, thanks again R.G.

I'm really having fun and learning with this one.

Here's where i am at right now:



The resistor values make the oscillation stable at both ends of the pot. The fast end is certainy suitable for vibrato, though I don't think the slow end is suitable for slow phasing. I hear what you are saying about fixing two of the resistors and only use one for a speed control restricting the range. I will try implementing a dpdt switch to increase two of the values simultaneously for slower speeds.

[R.G.]

The classical thing to do is to switch the capacitor values to switch ranges. You could use a CMOS switch like the CD4066 to switch another 1uF in parallel with each one  that's there now and that would halve all of the frequencies.

[frequencycentral]

The classical thing to do is to switch the capacitor values to switch ranges. You could use a CMOS switch like the CD4066 to switch another 1uF in parallel with each one  that's there now and that would halve all of the frequencies.

R.G. - you ae ACE.

I just 'published' and new circuit "Vibracaster" as a result of your advice with the PSO. I'm sure people will like it.

I have found throught experimenting that the three caps don't 'have' to be the same value - but would that cause asymmetry in the output? I don't have a 'scope to test this.

[R.G.]

The only thing that the circuit really cares about is that the total phase shift from the collector back to the base is 180 degrees. Any combinations that do that will work. "not working" usually results when the resistor gets too big to make up enough phase shift or so small that it pulls the feedback signal too low to sustain oscillation.

Making caps non-identical works, and might offset non-identicalities in the resistors.

[frequencycentral]

Hi R.G., I hope you spot this thread again, and can give me your help once again.

I used the PSO in my Vibracaster - really I was just testing how the PSO interfaced with tubes, but I thought it might be interesting for other people. So I ended up perfecting the Vibracaster and then building one myself. Then I got sidetracked when my 5672 pentodes arrived and ended up building my PentaDriver . Now I'm back to the project for which I originally wanted the PSO - a low voltage tube vibrato. Here's where I'm at so far:



I'm pleased with all the tube stages, though I have to add the phase shifting network between the phase splitter and the mixing triodes. If I apply the PSO to either mixing triode's cathode I get what I need. What I'm having problems with is the 'last piece in the jigsaw' - inverting the PSO to apply to one of the mixing triodes.

Please note that I changed the configuration of the PSO at the output transistor slightly - I replaced a 4K7 resistor with a 47K, and made the 4K7 pot into a resistor. And the depth pot of course.

How do I add the 'last piece of the jigsaw'? I tried adding the phase splitter you suggested originally - but I think whatever I have done to the output of the PSO to make it interface with my tubes means I have to change the splitter too. I just don't have the test equipment or the knowledge to work this out without help!

[R.G.]

Just offhand, the setup on the tubes looks very odd. But then I'm not used to 6111 tubes. This will require some thinking. Which I will do.

[frequencycentral]

Just offhand, the setup on the tubes looks very odd. But then I'm not used to 6111 tubes. This will require some thinking. Which I will do.

The tubes' setup works for me, I used a similar setup in the Vibracaster and it works good. It is the LFO entering the cathodes which bothers you? I'm aware that the other way is to have the LFO entering at the grid, but my arrangement works fine.

Whithout the phase shifting network I guess it would work as a panner if I took seperate outputs from the mixing triodes - once the PSO splitter is up and running.

I'm getting a 30db volume drop when I mix the outputs of the two output triodes.

[frequencycentral]

Cheeky bump!