Valved Wien Bridge Oscillator Thermistors? - UK Vintage Radio Repair and Restoration Discussion Forum

Chris55000 · 3rd Oct 2018, 12:11 am

Hi!

A theory question for everyone!

The circuit shown came from an article in Electronic Engineering magazine, and is simplified to use a double-triode valve of the ECC83 type. The oscillator frequency should be very roughly 1kHz.

A number of designs for valved A.F. oscillators were published in P.W. and W.W., but the stumbling-block with all of them I've seen is they are thermistor-stabilised like the example I've drawn here!

The £1000-questions are:-

1) How is the thermistor value (X1 in my circuit) decided? Guesswork? A data sheet? Estimated from the wien Bridge Osc. Gain being approx. 3?

(Most of the designs used a long-obsolete S.T.C. thermistor I've never seen anywhere!)

2) Is there a modern replacement? Or would lamps be preferable, and what raing/number?

Chris Williams

Refugee · 3rd Oct 2018, 1:36 am

I have seen a circuit with a FET used as a substitute.
I could have a go at looking it up if you get into trouble. Have a look at circuits from the late 1970s when FETs began to get cheaper than those thermiators.

joebog1 · 3rd Oct 2018, 5:37 am

Major stumbling block that one !!! I have umpteen dozen circuits with very nice oscillators, with superb specs, but alas all use that thermistor. My desk draw is filling up with small light bulbs to use as substitutes. SOME of the valved Wein bridge oscillators had very excellent specs and used light bulbs, after all we wouldnt have Hewlett Packard anything if his origional oscillator didnt work.

I have some circuits that use a single bulb, some with two bulbs in series, and some with three bulbs. Ratings for the bulbs are "usually 3 watts @ around 100/105/110 volts, but I also have some with 200 volt bulbs, and rated at 15 watts. Go figure. I "thinks" maybe the best thing to do is what I will do after I clear up a few unfinished projects on the bench already, and that is to make a bench birdsnest circuit and play with various bulbs and various combination of bulbs.
When I start I will post my findings here, but alas thats a ways down the track!!.

Best wishes in your search, I cant find them either!! For that matter I cant find the varistors used in some American guitar amplifiers to supply true vibrato, rather than tremelo, either.

Joe

Diabolical Artificer · 3rd Oct 2018, 6:47 am

Use a bulb Chris. I have some 24v bulb's that might do the job, drop me a PM and I'll send you one. It's mainly trial and error but I think you need to calculate for a closed loop gain of 3. The boffin's here will correct me.

I made a similar Wein bridge oscillator and didn't have the right bulb, so used the heaters of a ECC83 instead, this worked till I got a suitable bulb.

Andy.

Radio Wrangler · 3rd Oct 2018, 7:12 am

Here's one:

http://hpmemoryproject.org/wa_pages/wall_a_page_01.htm

uses a bulb. Also uses a double gang capacitor rather than a pot.

David

MrBungle · 3rd Oct 2018, 8:56 am

I’ve built a few of these to experiment with albeit with lower voltage opamps and FETs. The feedback network is about as unfussy as it gets really. I would stuff a suitably rated bulb in and adjust the feedback resistor network so it doesn’t saturate and that’s it. Job done. I’m not sure you can reliably calculate it up front as each bulb is surprisngly different. I reckon thermistors were used to combat this spread problem. Also thermistors aren’t as microphonic as bulbs are.

On the lower voltage side of things I use the 24 volt panel indicators

lesmw0sec · 3rd Oct 2018, 10:13 am

Horses for courses these days. If you are after the lowest possible distortion, the thermistor approach is hard to beat, but if a stable amplitude and exact frequency is the criterion, with distortion not an issue, I'd go for a direct digital synthesis chip.

kalee20 · 3rd Oct 2018, 11:08 am

It's a question all the text-books skirt around. Frequency and frequency stability is well-covered - amplitude and amplitude stability isn't! How on earth DO you calculate what amplitude an oscillator settles at??

As to the thermistor / light-bulb / whatever - ever since Hewlett and Packard had their original light-bulb moment - the thing must have enough thermal inertia not to change its resistance appreciably during a cycle, else distortion will obviously result - but it must be quick enough to stabilise within a few seconds or so, else the amplitude will 'bounce' excessively. And as thermal inertia is not specified for any light-bulb I've seen, it has to be trial-and-error.

The V-I characteristic, again - during a cycle the thing behaves as a resistance, just V/I, but as this is a function of temperature hence voltage, you'd have to plot graphs showing R against Vrms - and find somewhere where dR/dVrms is high, so that a small change in applied V gives a large change in resistance. Knowing the resistance, you can choose the remaining feedback components to give you your gain of 3; knowing the value of Vrms you'd know the amplitude - and then no doubt you would find this to be too small or too large to be workable, so you'd start again, with a different bulb or thermistor, but knowing which way to head.

At least, that's how I would start about it!

ms660 · 3rd Oct 2018, 11:15 am

Sometimes heterodyne was used in the old days for constant amplitude across the range.

Lawrence.

trh01uk · 3rd Oct 2018, 6:41 pm

Well, I am not sure why anyone is struggling with this one! I just looked up "wien bridge oscillator" on Wikipedia, and it seems to me it tells you everything you could want to know - including how to work out the value of the thermistor for stable oscillation: https://en.wikipedia.org/wiki/Wien_bridge_oscillator

If you scroll down to the bit about the Hewlett oscillator, it becomes clear that the OP's schematic is virtually the same as that one. The main difference is in the position of the thermistor. The article says the oscillation criterion is Rb = Rf/2. Translating that to the circuit here we get R4 = (X1 + R5)/2. Given that R5 (=100k) is >> 2 x R4, that does leave me wondering slightly how the theory is going to fit this exact circuit. Chances are the equations here are rough approximations, and assume the perfect op-amp as in the circuit at the start of the Wikipedia article. Changing R4 to 56k might well make a bit more sense, and leave some room for X1 to have a positive value.

Richard

Radio Wrangler · 3rd Oct 2018, 9:21 pm

"Communications Circuits: Analysis and Design" by Clark and Hess is one of the few books to go into oscillator amplitude analysis.

David

joebog1 · 4th Oct 2018, 5:18 am

Quote:

Originally Posted by lesmw0sec

Horses for courses these days. If you are after the lowest possible distortion, the thermistor approach is hard to beat, but if a stable amplitude and exact frequency is the criterion, with distortion not an issue, I'd go for a direct digital synthesis chip.

IF you are only looking for a "screamer". Its well know that synthesis is FULL of harmonics.

"Well, I am not sure why anyone is struggling with this one! I just looked up "wien bridge oscillator" on Wikipedia, and it seems to me it tells you everything you could want to know - including how to work out the value of the thermistor for stable oscillation: https://en.wikipedia.org/wiki/Wien_bridge_oscillator

If you scroll down to the bit about the Hewlett oscillator, it becomes clear that the OP's schematic is virtually the same as that one. The main difference is in the position of the thermistor. The article says the oscillation criterion is Rb = Rf/2. Translating that to the circuit here we get R4 = (X1 + R5)/2. Given that R5 (=100k) is >> 2 x R4, that does leave me wondering slightly how the theory is going to fit this exact circuit. Chances are the equations here are rough approximations, and assume the perfect op-amp as in the circuit at the start of the Wikipedia article. Changing R4 to 56k might well make a bit more sense, and leave some room for X1 to have a positive value. "

We all know that Richard!!! please point me in the direction of the manufacturer of "said" thermistors!!.

My own theory of why Hewlett Packard units worked so well, was that they had carbon filament bulbs rather than tungsten. We all know about carbon as a resistor, and constantly wail about its temperature co efficient and long term drift.

Yes its essential that one requires a gain of X3!!. otherwise oscillation

1. Ceases
2. Overloads untill the resistance gets hot enough to reduce the gain

Read overshoot and "bouncing" output voltage untill the resistance "stabilises.

Just my thoughts.

Joe

3rd Oct 2018, 1:36 am	#2
Refugee Dekatron Join Date: Apr 2012 Location: Worksop, Nottinghamshire, UK. Posts: 5,553	Re: Valved Wien Bridge Oscillator Thermistors? I have seen a circuit with a FET used as a substitute. I could have a go at looking it up if you get into trouble. Have a look at circuits from the late 1970s when FETs began to get cheaper than those thermiators.

3rd Oct 2018, 5:37 am	#3
joebog1 Banned Join Date: Jun 2015 Location: Mareeba, North Queensland, Australia Posts: 2,704	Re: Valved Wien Bridge Oscillator Thermistors? Major stumbling block that one !!! I have umpteen dozen circuits with very nice oscillators, with superb specs, but alas all use that thermistor. My desk draw is filling up with small light bulbs to use as substitutes. SOME of the valved Wein bridge oscillators had very excellent specs and used light bulbs, after all we wouldnt have Hewlett Packard anything if his origional oscillator didnt work. I have some circuits that use a single bulb, some with two bulbs in series, and some with three bulbs. Ratings for the bulbs are "usually 3 watts @ around 100/105/110 volts, but I also have some with 200 volt bulbs, and rated at 15 watts. Go figure. I "thinks" maybe the best thing to do is what I will do after I clear up a few unfinished projects on the bench already, and that is to make a bench birdsnest circuit and play with various bulbs and various combination of bulbs. When I start I will post my findings here, but alas thats a ways down the track!!. Best wishes in your search, I cant find them either!! For that matter I cant find the varistors used in some American guitar amplifiers to supply true vibrato, rather than tremelo, either. Joe

3rd Oct 2018, 6:47 am	#4
Diabolical Artificer Dekatron Join Date: Jul 2009 Location: Sleaford, Lincs. UK. Posts: 7,662	Re: Valved Wien Bridge Oscillator Thermistors? Use a bulb Chris. I have some 24v bulb's that might do the job, drop me a PM and I'll send you one. It's mainly trial and error but I think you need to calculate for a closed loop gain of 3. The boffin's here will correct me. I made a similar Wein bridge oscillator and didn't have the right bulb, so used the heaters of a ECC83 instead, this worked till I got a suitable bulb. Andy. __________________ Curiosity hasn't killed this cat...so far.

3rd Oct 2018, 7:12 am	#5
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,876	Re: Valved Wien Bridge Oscillator Thermistors? Here's one: http://hpmemoryproject.org/wa_pages/wall_a_page_01.htm uses a bulb. Also uses a double gang capacitor rather than a pot. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

3rd Oct 2018, 8:56 am	#6
MrBungle Dekatron Join Date: Jun 2016 Location: London, UK. Posts: 3,687	Re: Valved Wien Bridge Oscillator Thermistors? I’ve built a few of these to experiment with albeit with lower voltage opamps and FETs. The feedback network is about as unfussy as it gets really. I would stuff a suitably rated bulb in and adjust the feedback resistor network so it doesn’t saturate and that’s it. Job done. I’m not sure you can reliably calculate it up front as each bulb is surprisngly different. I reckon thermistors were used to combat this spread problem. Also thermistors aren’t as microphonic as bulbs are. On the lower voltage side of things I use the 24 volt panel indicators

3rd Oct 2018, 10:13 am	#7
lesmw0sec Octode Join Date: Jul 2009 Location: Carmel, Llannerchymedd, Anglesey, UK. Posts: 1,507	Re: Valved Wien Bridge Oscillator Thermistors? Horses for courses these days. If you are after the lowest possible distortion, the thermistor approach is hard to beat, but if a stable amplitude and exact frequency is the criterion, with distortion not an issue, I'd go for a direct digital synthesis chip.

3rd Oct 2018, 11:08 am	#8
kalee20 Dekatron Join Date: Feb 2007 Location: Lynton, N. Devon, UK. Posts: 7,082	Re: Valved Wien Bridge Oscillator Thermistors? It's a question all the text-books skirt around. Frequency and frequency stability is well-covered - amplitude and amplitude stability isn't! How on earth DO you calculate what amplitude an oscillator settles at?? As to the thermistor / light-bulb / whatever - ever since Hewlett and Packard had their original light-bulb moment - the thing must have enough thermal inertia not to change its resistance appreciably during a cycle, else distortion will obviously result - but it must be quick enough to stabilise within a few seconds or so, else the amplitude will 'bounce' excessively. And as thermal inertia is not specified for any light-bulb I've seen, it has to be trial-and-error. The V-I characteristic, again - during a cycle the thing behaves as a resistance, just V/I, but as this is a function of temperature hence voltage, you'd have to plot graphs showing R against Vrms - and find somewhere where dR/dVrms is high, so that a small change in applied V gives a large change in resistance. Knowing the resistance, you can choose the remaining feedback components to give you your gain of 3; knowing the value of Vrms you'd know the amplitude - and then no doubt you would find this to be too small or too large to be workable, so you'd start again, with a different bulb or thermistor, but knowing which way to head. At least, that's how I would start about it!

3rd Oct 2018, 11:15 am	#9
ms660 Dekatron Join Date: Apr 2011 Location: Cornwall, UK. Posts: 13,454	Re: Valved Wien Bridge Oscillator Thermistors? Sometimes heterodyne was used in the old days for constant amplitude across the range. Lawrence.

3rd Oct 2018, 6:41 pm	#10
trh01uk Octode Join Date: Nov 2010 Location: Ross-on-Wye, Herefordshire, UK. Posts: 1,654	Re: Valved Wien Bridge Oscillator Thermistors? Well, I am not sure why anyone is struggling with this one! I just looked up "wien bridge oscillator" on Wikipedia, and it seems to me it tells you everything you could want to know - including how to work out the value of the thermistor for stable oscillation: https://en.wikipedia.org/wiki/Wien_bridge_oscillator If you scroll down to the bit about the Hewlett oscillator, it becomes clear that the OP's schematic is virtually the same as that one. The main difference is in the position of the thermistor. The article says the oscillation criterion is Rb = Rf/2. Translating that to the circuit here we get R4 = (X1 + R5)/2. Given that R5 (=100k) is >> 2 x R4, that does leave me wondering slightly how the theory is going to fit this exact circuit. Chances are the equations here are rough approximations, and assume the perfect op-amp as in the circuit at the start of the Wikipedia article. Changing R4 to 56k might well make a bit more sense, and leave some room for X1 to have a positive value. Richard

3rd Oct 2018, 9:21 pm	#11
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,876	Re: Valved Wien Bridge Oscillator Thermistors? "Communications Circuits: Analysis and Design" by Clark and Hess is one of the few books to go into oscillator amplitude analysis. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done