JFET gain control "resistor" in Weinbridge oscillator

Tim, Yup an oscillator is on the knife edge. You need a gain around the loop of exactly three. At that point the circuit oscillates. This is the same factor of three that forms the (K) gain term in filters. A high Q filter has a K close to three. Now that you've got the gain above three oscillations grow...the problem now is to limit the gain. For the light bulb circuit you need a feed back resistor that is three times the light bulb resistance (at the operaing voltage), The problem is finding a light bulb with a big enough resistance. Just look at I/V specs for light bulbs and pick the ones with the highest resistance. Then you have to find an opamp with enough current to drive the whole thing.

e e

ith

be

"A small pcb layout change could easily

Well maybe not RF but it could very well cpacitive coupling to the front panel... and there have been several pcb redo's where the circuit had turned over to others. I've check the panels and haven't been able to identify any significant change in the ground situation. or other errors. I also don't like the high current draw of the light bulb stabilized oscillator so I'll get back to you if I have similar troubles with the FET version. Concerning the circuit. Check out AN47 at Linear. legg says page

1. I'm using a OPA227 and three of the highest resistance light bulbs I can buy from Mouser. The feed back resistor is about 200 ohms(400 ohms with 200 +200 ohm pot in parallel). The RC net work has a 10k dual ganged pot and switchable caps from 3uF to 10nF... or close to that anyway. I pick off the output from the non-inverting input, (Which gives another RC of filtering) and run opamps at about 15Vp-p. (I think I use all of the 40mA of drive current).

he

ge

with

t

to be

s o g

by

m f

Hi Joerg, I'm sorry I don't know LTspice. I have a copy of electronics workbench that is 10 years old at least. I sometimes try circuits on there if I find that the circuits on my bench are not behaving "properly". This can sometimes help guide me in understanding what going on.

"begins to oscillate, or motorboat." The motorboating is different from the oscillation. The circuit oscillates at 3 kHz, and when some circuits "motorboat" the amplitude is being modulated by about 30% with a frequency that is 1/30th - 1/100th the oscillation frequency. If you listen to it on a speaker it sounds like a motor boat in rough water.

f the

ridge

ou with

t.

use a

from J.

JFET

$4.30

and

is the

.

ly

he

y,

e f

Joerg, I'll check out the specs... I'm not to worried about RF oscillations. The opamp has no gain up there and that the only gain element.

It appears I've "Really" screwed up. The aplication note is number

1. (AN43) All about bridge circuits. I believe the sub- title is "marrying gain and balance" I'm down-loadeding it now, but I live in the sticks with dial-up and this could take a while. OK I've got it AN43 page28-31. The FET is a 2N4338 as I said. Which is good since I understood the circuit with a "2k ohm" FET but not the

60 ohm FET legg was talking about. Sorry, George

be

legg, I screwed up it's application note AN43 about bidge circuits not AN47. See pages 28 to 31... and it is a 2N4338 FET with Ron =3D 2k ohm or so, which is good because otherwise I didn't understand the circuit. Sorry for the confusion, George

Funny "touch me" story:

Back in The Day, I worked on the NATO Seasparrow missile system on a US Navy destroyer (USS John Young, DD93). This was at a time when there was still a lot of analog fire control equipment out there (even electromechanical computing!) and Seasparrow was transitional between that and the new digital world.

There was a box called the Radar Target Data Processor which was basically a 19" rack 4ft deep and 7 feet tall, full of opamps and oscillators. A nightmare, and nobody (including me) really knew how it worked. There were a bunch of lamps on the front that would light up to report error numbers; interestingly the lamps were assembled in a pattern that looked like "FU" if you used a bit of imagination.

So anyway, from time to time something would drop out of lock and the whole front panel would light up like a Christmas tree. There was a time-consuming realignment process you had to go through to get everything corralled. I noticed that when I slid the upper drawer out to start this procedure, if I put my left hand in just the right position on the card cage and held it for a few seconds, about half the time everything would re-lock without having to go through all the rigmarole.

Fast forward to the day I'm being discharged. My seabag is literally sitting on the quarterdeck and I'm in civvies just waiting for 1000 so I can step off for the last time. And what do you suppose? Uh-huh -- the RTDP freaks out. One of the guys comes up and wants to know if I'll look at it?

So I roll out that top drawer, put one hand on each side, tilt my head upward, and say "Be HEALED!" in my best televangelist voice. And all the lights go back to normal. So I closed the drawer, noticed it was a couple of minutes past 1000, and walked out to the pier. Couldn't have scripted a better exit :-)

A few years ago they used DD973 as a target in a torpedo exercise, and I noticed in the photographs of the sinking that they'd stripped every piece of hardware from her except the Seasparrow system.

--
Gordon S. Hlavenka
Join the Revolution at http://www.ronpaul.com

Was wondering what all the fuss was about - with a MHz oscillator, there's little problem with the extremely long time constants used in the AGC loop.

I'm familiar with this problem using thermistors (below 100Hz), for the conventional twin-T oscillator - a 2sec AGC settling time for manually operated equipment is no big deal.

The circuit in AN43 is working at audio frequencies. Vpmax for this part is

"George Herold"

So I'm ready for the next stage solution. The first time I read Jim Williams article on the Weinbridge oscillators (this is also in one of his books.) I didn't quite "get" the integrator FET solution to the gain control problem and the light bulb looked like such a simple idea. (I also loved that it was the begining of HP) But I've advanced since then and am now ready to try this other circuit.

-----------------------------------

How about a new and much better oscillator circuit ???

A simple one that does not use bulbs or strange JFETs to stabilise it.

One that can be made with a single quad op-amp ( eg TL074 ) and no special parts.

One that exhibits no amplitude bounce when adjusting the frequency or range changing.

One that has constant THD of 0.15% from 2 Hz to 10kHz.

One that has flat amplitude ( +/- 1% ) over that same frequency range.

------------------------------

See ABSE ( alt.binaries.schematics.electronic ) for the schem of a " Low Distortion Oscillator " design I developed 20 years back and was published in Electronics Australia magazine in February & March 1989.

The circuit use two all pass filters as a "phase shifting" oscillator with harmonic distortion cancellation ( IC2b). As shown, the output level is 2.5 volts rms - which can easily be increased. Just ask me.

Ignore thermistor TH1 which for very low THD only - the 4 diodes and resistor network is all you need.

Tuning is done with a standard double gang pot with both halves grounded at one end. A simple Schmitt trigger square wave converter is there too, if you want that.

BTW: This the device you are working on ?

...... Phil

But the FET represents a gain element all by itself. It can happily sing the RF-blue somewhere around UHF channel 20 with none of the other parts involved. Seen it. Just like that sports car can get into a crash all by itself on an icy road with no other cars around ;-)

Have you looked into Hughes-Net? That's what the SW guy in town uses because DSL ends a mile from here and he lives another mile down from there. It's expensive, but at leat you get some speed.

Try the Panasonic or some similar small FET. I don't think you'd be happy with a 60ohm FET, too much control loop gain. Also, the J201 looks good and Digikey has then for 10-15c depending on how many. Only TO92 though.

(800k size file)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Just think about it: If a 2SK3372 or J201 works you'll save about $4 per unit. Meaning you've earned a six-pack of the really good stuff like Porter or Octoberfest Ale for every two units :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

You, too?

I left the army two days late because the next guys had to be taught how to run the multiplex truck. Because of a kablouie and some hospital time I returned to my company on the day that was supposed to be my last day. They kind of pleaded with me to hang on but they couldn't pay extra. I said ok. It was weird, 40-50 people standing at attention. All in battle dress and one in jeans (me).

The kablouie was one of those exercises. Weeeeeooouuu ... bang, weeeeeeeoooouuu ... bang. Then Weeeeeeeeeeeee (Lieutenant shouted "Oh s..t! Get down!") ... *KABOOOOM*.

Quite sad.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

k

Legg, Dang, you're blowing me away.... I think you're saying that the J201's are my babies! they looked good to me too. Is the low turn-on voltage important? (is this Vpamx?) I was thinking I needed to do a bit of analysis and get a better handle on the gain. I am far from a FET expert. Can I find Ron vs V gate data on the spec sheet?

cial

ge

ge.

" Low

shed

h

.5

d

at

if

Phil,

Ah you found one of my babies! yup that's it.

Well I don't want to change the fornt panel so I need tuning with a coarse capacitance range and fine dual ganged pot range. I'm slightly worried about the harmonic distortion which "screws up" the phase in the phase shifter. The phase shifter is a phase sequence filter also stolen from H&H. The other potential problem I would have with some "switched" oscillator is the radiated noise. (I'm not sure what your oscillator is, but you mentioned a comparator.) I would like to keep the "leakage" of the oscillator into the front end down to the 10 nV level.... that's about the noise voltage level of the preamp. (AD620). I think the JFET circuit will be OK. I'll have contorl of both the time constant of the intergratng loop and the gain, by changing the feedback vs Jfet resistance ratio. that's the hope any way. George

of the

nbridge

you with

e

ent.

d use a

s from J.

nt JFET

r $4.30

s and

r is the

ge.

ally

the

ney,

the

=A0If

a h

ch

"Only TO92 though. " I love TO92! they plug right into my white protoboard! But Don't worry once it's working on the white protoboard I build a dead bug prototype. George

to be

s o g

by

m f n

Hmm I've spent all my development time on this unit years ago. This is an effort to "stop the bleeding".

"George Herold" Phil Allison"

Ah you found one of my babies! yup that's it.

Well I don't want to change the fornt panel so I need tuning with a coarse capacitance range and fine dual ganged pot range.

** Which is EXACTLY what the design has.

I'm slightly worried about the harmonic distortion which "screws up" the phase in the phase shifter.

** Huh ???

The other potential problem I would have with some "switched" oscillator is the radiated noise.

** What the heck are you on about ?

There is nothing "switched" about the circuit - just a TL074 operating linearly.

Operation is very similar to a Wein bridge, without all the drawbacks.

(I'm not sure what your oscillator is, but you mentioned a comparator.)

** I did NOT mention any comparator !!!!

Have you seen the schem on ABSE or not ??

I would like to keep the "leakage" of the oscillator into the front end down to the 10 nV level.... that's about the noise voltage level of the preamp.

** At least give the schem a look - pal.

Cos you sound like you are 100% off with the fairies.

..... Phil

The minimum resistance specified for the parts is for Vgs=0. The pinch-off voltage is actually negative, with respect to the source. If it were otherwise, the integration capacitor couldn't be a polarized part.

A larger pinch-off voltage just increases the time it takes for the AGC to reach initial control gate levels. The same effect would occur using the 'low distortion' version of the app note.Transconductance seems to shadow Vp in proportion, so the distortion products might increase.

I'll post the Siliconix curves for this series (NPA).

RL

ial

w

hed

ith

d f

ing

Phil, Opps, Sorry, I really set you off. Pleae accept my humble appologies. I did not look at your schematic! My only excuse can be resonding to email late at night after after tipping a few with friends during the Buffalo Bills game. (The Bills won!!!)

I am at work now and will happily look over the circuit. You mentioned a double All Pass phase shifter, which I see as the first two op-amps, but how is the loop amplitude controlled? What stops the oscillations from growing with out bound? OK I think I see when the four diodes start to conduct the gain around IC2a is reduced. If I leave out the thermistor then a signle RV2 pot will adjust the amplitude. I'll try bread-boarding the circuit and see what I think.

When I was first building this whole piece of equipment I first tried using one of those single chip oscillators (ICL8038) (I think these have now gone obselete.) These try and flatten out the triangle wave into a sine wave, But it was hard to get a clean sine wave out and the higher frequency components gave uncontrolled phase shifts when they passed through the "Phase Sequence Filter" The phase sequence filter accepts a sine wave and it's inverse and outputs four sinewaves in quadrature. which is just what I wanted. (No phase locked loop) But it only works for frequencies within it's design band. Frequencies above above the band went through the phase sequence filter with some gain... any distortin was amplified by the phase shifter. (I've forgetten the exact details and would have ot go look at my notes.) .... So I started looking at building my own oscillator and when I found the Weinbirdge I "fell in love". (Does this make me a farie??) So the romance is now winding down and I'll again certainly look at your oscillator. And agian sorry for getting you upset.

George

on

Thanks Legg. I see the curve I need. But Now stating at the circuit I seem to be making a sign mistake in my head. As the amplitude of the oscillations grow I want the resistance of the FET to increase so that the gain of the Weingbridge op-amp decreases. Not to worry it will all become clear once I build it and look at some voltages. George