Transistor oscillator.

This article...

doesn't give a relationship between transistor parameters and switching frequency, but then it's not a common-base circuit they use.

joe

What frequency limit do you mean? For most uses the rule of thumb is probably Ft/100, but there are some radio applications that push that to Ft/10, and at one point in the 70's, I am told, (and maybe still, for all I know) there were microwave applications that considered a transistor "useful" at any frequency where it had power gain was considered useful, and that frequency could actually be above Ft.

My first thought would be to build a circuit that automagically finds the point where the phase shift in the current gain is 90 degrees. Since you have modern electronics to draw on for your test circuit, make some high- impedance source to pump current into the transistor base, put a really small collector impedance on it and amplify the snot out of it, then find the point of 90 degrees phase shift.

--
www.wescottdesign.com

Sure, it's common to test a new fabrication recipe for transistors with a ring oscillator. Three or more (odd number) of inverters, negative feedback connected in cascade (series) (which is why it has to be an odd number).

Oh, you'd have to modify the scheme for non-identical units. And, it's intended for biased inverting amplifiers (so depends somewhat on resistor values).

Hi, An interesting question, but very vague. An oscillator is defined by its oscillation frequency, and if a transistor is operated at its frequency limit, then one would have to find a transistor whose frequency limit is the same as your frequency of interest. Also, your suggestion of a capacitor connected to the base sounds more like a "Joule thief" design, a very inefficient one.

A ring oscillator is not a very useful circuit anyway -- I have never seen a practical circuit triggered by a ring oscillator.

For good reasons: an inverter in a ring oscillator doesn't necessarily execute a full logic swing, it doesn't have a load similar to a working gate, nor any external wiring (if the ring oscillator is an on-chip construction in an integrated circuit). It does, however, run at a high frequency set by internal delays and gain-versus-frequency of the individual stages. Barkhausen's criterion gives you a good handle on the stage delays at unity power gain, being

t_stage_delay = period/(2 * N)

[snip] [snip]

I have made VCO's from ring oscillators, just vary VDD,VSS ;-) ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

Maybe build a single gain stage with a socket for the transistor, and look at where the gain rolls off...?

George H.

Maybe build a single gain stage with a socket

AFAICR: the blocking oscillator period is determined by the time it takes the core to saturate - not how fast the transistor can go.

Basically something along the lines of a "figure of merit".

I'd like an oscillator in which the capability of the transistor is what sets the frequency.

That could well be the least bad option - but could arise that I don't have

3 identical transistors to make the ring.

Would that work with 2 modern transistors and see what frequency the test transistor holds it back at?

With mostly PNP germanium transistors, I could make up the numbers with a couple of BF541 or similar.

The question wasn't as vague as you seem to think.

The capacitor on the base is decoupling - because its a common base stage.

The capacitor strapped from collector to emitter isn't some weird mystery - that configuration is pretty much standard on the RF stage of small bug MIC circuits.

Usually the LC tank determines the frequency - I was wondering if I could get away with just an RFC in the collector circuit, or would it resonate with the feedback capacitor.

That was pretty much the plan.

But I'd like it to be self oscillating, and do so at the maximum frequency the transistor allows.

Second to last post at this site(click on image to straighten it out):

Two routes might work: use a square wave or pulse generator, and compare input and output rise/fall on an oscilloscope - it's dirty, and the oscillation frequency is not the data you collect, but delay and rise times.

Or, consider a delay-line oscillator: just couple the output of a one-transistor inverting amp into a delay line to the input (base). Oscillator period difference from twice the delay time is the transistor contribution.

Oh, and for this scheme, you'd want to Baker-clamp the transistor; there's no use getting tied up in saturation effects if you want a 'maximum frequency' indication.

So what sets the frequency in that circuit?

They are often used in random number generators.

That's what I was trying to describe, however poorly (the picture inside my head is clear! Really!).

Unless you have a hundred transistors or more, it may be easier to make a manual current gain test setup and manually measure Ft.

If I remember correctly, the current gain of a transistor goes as a low- pass filter, with a pole at Ft/beta. By the time that you get down to Ft the phase shift will be 90 degrees, or perhaps even more (if there are more gain-robbing processes that just the canonical Ft one).

You could try to make some automatic gizmo that varies frequency until the Ft/beta point is hit, but I don't know of any existing thing that you could buy.

Maybe do a search in case some hobbyist magazine published some such tester?

--

Tim Wescott 
Wescott Design Services 
http://www.wescottdesign.com