Playing with valve oscillators

[Miguel Lopez]

Hello everybody

I've been out of touch on valves due to time restrictions, but I want to share with you a small project that i just recently started.

I have always had plans to build a second valve oscilloscope and a powerful audio amplifier (with 807s). Mr Time had been implacable, so I've been a little distanced from valves for a while. But now I want to work on a project which is a device to be used with those both bigger (and future) projects.

It is my intention to build a signal generator with two fixed outputs: one at 1kHz and the other at 25kHz. The first one will be to test valve amplifiers while the second it's intended to help in the design of the EHT generator for the scope.

This way I can work with valves (which is something I love), and at the same time I can build a useful tool for future projects. It's a win-win situation. On the other hand as Mr Time is always flying away from me, I want to use parts that I have built before for other abandoned (or modified) projects. I will use the chassis that I built several years before for the EHT generator for my scope. You can see it here: https://www.vintage-radio.net/forum/...5&d=1347901046
on this thread:https://www.vintage-radio.net/forum/...t=81184&page=8

I also want to experiment with new circuits and so, I want to build the 1 kHz signal with a phase shift oscillator, while the 25kHz will be an L-C oscillator. This way a can "play" with valves too. I have tested those circuits in my valve tester board (which I call "High Tension Bread Board" or HTBB)

The circuits that I want to build are shown on the attachments and what I have done till now is there too.

[Diabolical Artificer]

Good stuff mate, did they oscillate as expected and how accurate were they?

Andy.

[Radio Wrangler]

You don't need a lot of gain to make an oscillator work. Those circuits ought to be quite happy with a 12AU7 or 12AT7 if you find those less expensive.

David

[ukcol]

Hi Miguel,


First published in the southern hemisphere under the title "Radiotron Designer's Handbook" and here by Wireless World under the title "Radio Designer's Handbook" this book is considered by many (including me) to be the "bible" on the subject of valve radio electronics.

It can be download free of charge here :

https://www.google.co.uk/url?sa=t&rc...-rdd_6Oc5L7RAI

and from a number of other sources using you favourite search engine.

There are also plenty of hard copies available for sale from various sources if, like me, you prefer to work with a real book.

Of particular interest to you will be Chapter 24 (page 947) "OSCILLATORS".

[Restoration73]

If you need a higher output, 12BH7 has the same pin arrangement.

[Miguel Lopez]

Originally Posted by Diabolical Artificer

did they oscillate as expected and how accurate were they?

They oscillate very well on the HTBB. Yesterday I mounted the LC oscillator on the chassis and it oscillated well too. I'm using 6H2P double triodes.

Originally Posted by Radio Wrangler

You don't need a lot of gain to make an oscillator work. Those circuits ought to be quite happy with a 12AU7 or 12AT7 if you find those less expensive.

Have to disagree here. I have tried the same circuit with different valves. I have tried ECC81, ECC82, 6N1P, ECC85, 6N23P. None of them oscillate. I have only make it oscillate with ECC83 and 6N2P

Originally Posted by ukcol

First published in the southern hemisphere under the title "Radiotron Designer's Handbook"

I already have it. Very good book indeed.

Originally Posted by Restoration73

If you need a higher output, 12BH7 has the same pin arrangement.

I can give it a try in the HTBB. The chassis is wired for 6N2P which has a different pinout.

[kalee20]

Originally Posted by Miguel Lopez

Have to disagree here. I have tried the same circuit with different valves. I have tried ECC81, ECC82, 6N1P, ECC85, 6N23P. None of them oscillate. I have only make it oscillate with ECC83 and 6N2P

If you are referring to the 'classic' phase-shift oscillator, 3 identical RC stages (or CR stages...) then you need a voltage gain of 29 minimum to make it oscillate. It's a bit tedious doing the maths - but at zero-phase shift this is indeed the attenuation in the network (it turns out to be 6x5 - 1).

So a 12AU7 definitely won't work! Unless you cheat and get yourself a bit of extra voltage gain with a transformer too!

[Miguel Lopez]

Originally Posted by kalee20

the 'classic' phase-shift oscillator, 3 identical RC stages (or CR stages...) then you need a voltage gain of 29 minimum to make it oscillate

That's right. You can add a fourth CR stage, and then the required gain falls to about 18, and with five stages it is even lower. But I want to keep as simple as possible.

Here a couple of pictures of what I did yesterday:

[G8HQP Dave]

ECC81/12AT7 ought to work OK for a phase shift oscillator, provided that the phase shift network does not have too low an impedance. Your network has low impedance; it looks like it was taken straight from a BJT design.

[Miguel Lopez]

Originally Posted by G8HQP Dave

Your network has low impedance

That's why I connected it via a cathode follower. The network is not loading the oscillating valve. Maybe with some heavy Math, I could make it oscillate using a 12AT7, but fun is against Math so I use "trial and error" method

[Diabolical Artificer]

"fun is against Math so I use "trial and error" method " Yep, quicker and easier to pop another valve in, the only snag being when it doesn't work, you have no idea why till you do the math's in some cases.

Andy.

[kalee20]

You can even make an RC oscillator using a cathode follower. You need an RC network with a gain >1 at a particular frequency, but it turns out (surprisingly) that this is indeed possible using not very many components!

[Miguel Lopez]

Originally Posted by Diabolical Artificer

"fun is against Math so I use "trial and error" method " Yep, quicker and easier to pop another valve in, the only snag being when it doesn't work, you have no idea why till you do the math's in some cases.

Have to agree with you at this point Andy. I made this Excel sheet to calculate the values of the components, but it gives me extremely low capacitor value for a frequency of 1kHz so I didn't use it anymore. It is based on a circuit which do not use a cathode follower, so the feedback is taken directly from the anode of the oscillating valve, loading it. This loading effect is taken into account in the calculations.

[Radio Wrangler]

I think the Wien bridge RC oscillator needs a gain of about 3 to make it go.

The problem with the three section phase shift oscillator is that to get the oscillator to work, you need 180 degrees of phase shift. Theoretically a 2-stage network can give 180 degrees, but only at the expense of infinite attenuation. A three stage network gets 180 degrees and although it needs a fair amount of gain, at least it's now finite. A 4 stage network is less extreme and so on with more stages, with the gain requirement reducing, but giving diminishing returns.

David

[G8HQP Dave]

More stages could mean higher effective Q, although Q for most such networks is rather low. I seem to recall reading that the Wien bridge has a Q of 0.25? Or maybe that is the twin-T CR notch filter.

[Miguel Lopez]

Originally Posted by kalee20

You need an RC network with a gain >1 at a particular frequency

With passive components only? Doesn't that opposes the law of conservation of energy, or even more isn't that a "perpetum mobile"

[G8HQP Dave]

That is what I thought when I first came across this idea. However, it is possible to get a small voltage gain from a passive CR circuit. Power loss of course.

I seem to recall you have to use a CR circuit sitting 'on top' of another pair of circuits.

[Argus25]

One issue with phase shift oscillators, apart from the required gain, is the amplitude of the output waveform is very dependent on the overall gain of the active device/s. With fewer gain stages,or only one, less degenerative feedback can be used, so the gain becomes very dependent on the properties of the active device and less so on the external circuit components. So with a valve or transistor, temperature & power supply variation can have significant effect.

Also as the amplitude increases so does the distortion and as the amplitude drops, the waveform is better quality, but it might not oscillate. This is where OP amps come in handy with a lot of feedback to lower the gain and make the circuit less dependent on the properties of the active devices.

So to make a really stable phase shift oscillator I think it should require a number of heavily degenerated gain stages, and still enough overall gain to oscillate, rather than less gain stages that have a higher individual but less stable gain with temperature variations etc.

A single transistor though can have some degeneration (unbypassed emitter resistance) so that it will make a fairly stable phase shift oscillator with repeatable performance for a number of similar transistor specimens. But I don't know if this is true for most triodes.

Admittedly all feedback oscillators require stabilzation of some sort but self (grid current) biased types ameliorate a lot of the problems. One thing I like about frequency synthesizer designs is the nice uniform amplitude output over a large range of frequencies, quite awkward with L-C oscillators without a feedback system around them.

A good lesson in the way to level stabilize a VFO over an enormous range of frequencies with a feedback loop is had by looking at the design of Tektronix's SG503 levelled sine wave generator. The output level is detected and used to control the emitter current of the oscillator transistor itself.

[Miguel Lopez]

Originally Posted by kalee20

You can even make an RC oscillator using a cathode follower. You need an RC network with a gain >1 at a particular frequency, but it turns out (surprisingly) that this is indeed possible using not very many components!

One must learn something new everyday. This is my learning for today. Very interesting
https://www.circuitlab.com/editor/#?id=aug8xf

[Radio Wrangler]

An oscillator has to have at least a little bit of excess gain for it to start and for the signal to grow. Beyond a point, the gain has to reduce to establish a stable operating level, else it will give square waves.

Lots has been written on controlling and stabilising the frequency of an oscillator. Very little on its amplitude.

David