colpits AGC stability problem

Hi, I've made a 1ghz+ wide range colpits VCO wich although oscillates well I'm trying to control the amplitude to reduce harmonics.

The problem is I cant get the AGC stable. When it starts oscillating it instantly builds up to an uncontrolled amplitude, the bias circuit either cuts off completly yet it still oscillates or it stops suddenly.

The problem is only apparent with a transistor with low Ccb, a higher Ccb transistor is stable but this severly restricts the tuning range.

Ive tried various things of wich anti series pin diodes in the feedback was best although this does add a lot of inductance to the feedback path.

Is there an easy way to stablise the amplitude without clipping ?

Colin =^.^=

Reply to
colin
Loading thread data ...

Have you looked at the stability of your feedback loop? Searching on proportional-integral-differential (PID) control should give you some kind of clue. or working out the Bode plot of your feedback loop.

It sounds as if the gain of your feedback loop (proportional term) is much too high, and you will probably have to use an integrator to inch up the gain around your oscillator until it is just oscillating. Try and work out some way of making the loop gain around your osicllator some more of less linear function of your gain control signal - makes stabilising the loop a lot easier.

--
Bill Sloman, Nijmegen
Reply to
bill.sloman

How are you frequency compensating the gain feedback control loop?

Reply to
John Popelish

was

It's only a WAG but from lower frequencies I've seen oscillators with two very close but distinct modes of oscillation. In your case, the Ccb seems to be keeping these apart but the level of ALC is causing a switch from one mode to the other. Maybe try ALCing using the supply rail. john

Reply to
John Jardine.

(1) Confucius say, "Picture/schematic worth 1000+ words" ;-)

(2) See my MC1648 for an example...

formatting link

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

well

Ccb

to

Yes ive experienced this too, its a bit annoying, but it occured without ALC when tuning through the range. When tuning up in frequency it would drop down part way through then carry on going up, it was quite dependant on supply voltage.

Colin =^.^=

Reply to
colin

well

was

Well I tried quite a few circuits so wasnt sure wich one to post, that chip is quite nice, would be ideal if it would go over 2ghz.

The bias control I have used is found here in figure 4b in this rather interesting document ...

formatting link
although my oscillator is grounded emitter.

Fortunatly Ive managed to get it much more stable now, by adding a 200R resistor in the oscillator feedback so it doesnt burst into oscillation so abruptly, doesnt seem to cuase any other noticble problems so far.

thanks Colin =^.^=

Reply to
colin

Have you considered one of the "stub" ceramic resonators? I've seen those used quite a lot in GPS and WiDeFi stuff.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

Not in this instance no, as I'm trying to get a wide as tuning range in one go as possible wich is about 1 octave, or a bit more at lower frequencies, using varactors.

A Yig equivalent would be nice but without the bulky coil.

Colin =^.^=

Reply to
colin

Since you refered to Ccb, I will assume this is a bipolar transistor. The gain of a bipolar transistor increases when the collector current increases. Bipolar transistor oscillators often have a tendancy for the collector current to rise when oscillation starts. As far as the AGC servo is concerned this is putting a positive feedback inside the servo loop. This can make the servo very hard to make stable.

To make your life easier, you will want to make the time constants in the bias circuit short. You don't want them to have a cut off frequency much below the F/Q of the tuned element.

Typically, in the small signal case, the bias control does not control the amplitude of oscillation. It controls the rate of change in the amplitude. This means there is a 90 degree phase lag between the control and the amplitude. The circuit that measures the amplitude also has some phase lag. If you add another 90 degree phase lag in the servo circuit, you will have a servo loop that oscillates. Normally you would include a zero in the servo circuit to prevent this oscillation.

--
--
kensmith@rahul.net   forging knowledge
Reply to
Ken Smith

[snip]

Then decreases once IKF is reached.

A common AGC method, used in the far past, for RF AMPLIFIERS was to INCREASE collector current to REDUCE gain, yet INCREASE signal handling.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

Hi, thanks for your detailed reply.

Yes, and yes this positive feedback is the crux of the problem, as the bias is increased slowly and it starts to oscillate, the increased AC voltage on the base gets rectified and draws current from the DC blocking capacitor as it makes the base voltage average more negative wich also then increases current through the bias resistor suddenly bursting into life and at full amplitude. When reducing the bias gradually at some point the amplitude falls therfore increasing the average base voltage and hence reducing the voltage accros the base resistor and it stops oscillating.

It behaves just like a schmit trigger relaxation oscillator! I think I would be better off if I could find a suitable FET with extremly low capacitance.

With a voltage, resistor and bias blocking capacitor this certainly seems to be the case, plus some hysteresis.

I have found a constant current base bias does greatly improve the control of collector current, hence gain and to some extent limit of amplitude. however what I am still a little puzzled by is that even though the bias current source had turned fully off the transistor was still oscillating at full steam, somehow it was getting base bias internaly, maybe the negative base voltage was so great it cuased reverse current in the emitter base junction ?

Yep, only one pole can be allowed to dominate.

The problem was the speed at wich it was bursting into oscillation, the risetime of the detector output was very fast. This would have to be very fast indeed to overcome the positive feedback especialy through the amplitude detector as well. Alternative is an inner faster loop with feedback from collector current.

The base capacitor was 220pf bias resistor 10k but changing it to 10pf/1k made no difference, although this is a UHF oscillator I didnt want to go very low as it is designed to cover a very wide range and it was a struggle to get it to oscillate at the lowest frequency.

A 2n3906 transistor now controls the base bias more closly by monitoring the collector current (instead of an op amp). However this was still not fast enough without reducing the speed at wich oscillation built up, reducing the RF loop gain with a 200R slowed this down enough to make it stable, surprisingly the oscillator still seems to work ok at .5-1ghz I just hope it still works at 1-2ghz. interestingly the base bias voltage now falls as the bias is increased.

A less demanding outer opamp loop can now control the level quite easily.

Ive now got a few more loops to deal with, Laser output power control, modulation depth control (wich is part of a larger loop), final Gaas Fet output stage current and drive level, unfortunatly the hard part is they all intereact. At least they shouldnt be as dificult as the oscillator loop for wich simulation was unhelpfull.

Colin =^.^=

Reply to
colin

In message , dated Sat, 26 Aug 2006, Jim Thompson writes

Philips made specific transistors (BF198, 'forward AGC type') that did this, but for other types, the effect occurs at too high a collector current to be desirable. This is a case where all transistors are not the same, and a careful selection of device is necessary.

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
 Click to see the full signature
Reply to
John Woodgate

That would actually make good sense for receiver stages, although less so for oscillator.

Intersting, I didnt realise bf198 was made for that.

The collector current settles to about 2.5ma wich is quite low, but is follwed by a buffer to give .5Vpp The BFR340 transistor I used still worked very well at well above its rated collector current, it would probably go pop long before it lost much gain, although now ive switched to a START544 SiGe device.

Colin =^.^=

Reply to
colin

I prefer differential stages for oscillators, then AGC is actually _energy_ control into a finite-Q tank.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

Yep. I don't think I've even seen "forward AGC" mentioned for at least 20 years.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

Is this with ecl or do you have an example ? sounds tricky with discretes at >1ghz.

Colin =^.^=

Reply to
colin

In message , dated Sat, 26 Aug 2006, Jim Thompson writes

Most of us here are dinosaurs, after all.(;-)

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
 Click to see the full signature
Reply to
John Woodgate

Nope. I'm a Pterodactyl ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

It's custom PECL, with inductors on-chip, or the ceramic shorted line off-chip.

But pretty much as illustrated in the MC1648, but with more precise level comparison.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.