Bias Oscillator Partial Solution - UK Vintage Radio Repair and Restoration Discussion Forum

Clay1905 · 14th Jan 2026, 7:33 am

Hi Folks,
The continuing story of the Pyrox bias oscillator not running:

Looking at the schematic, I thought that the impedance (less than 1 Ohm DC) of the bias coil would appear as a short across the secondary of Tr2. By way of experiments, I first tried another coil of similar impedance, to be sure there was no short to ground in the bias head coil. This killed the oscillator dead.
I tried an R.F. coil with substantially higher impedance and the oscillator runs happily. Same deal with both the R.F. and bias head coils in series. No big surprise there.

So, question time.
What's the better way to increase impedance in the bias head circuit? I could wind a coil and add that to the circuit, but what about adding a capacitor in parallel to the head coil? If my understanding is right, then if the capacitor is selected so that it and the head coil form a circuit that is resonant at bias frequency, then the impedance should be very high. But it this a good idea in this application?

Keen to hear some thoughts,
Clay.

Clay1905 · 14th Jan 2026, 8:08 am

and here's a picture

kalee20 · 16th Jan 2026, 10:30 pm

Nobody else has commented - and I've been pondering a couple of days!

Fundamentally, something seems wrong: it presumably worked in the first place, so trying a modification doesn't seem right (unless something has been knowingly changed for something different, as the 'best available option' and it's a case of making it work).

If connecting the erase head kills the oscillator, it almost seems as though there is a shorted turn in the erase head. It may still measure around the right DC resistance, but it'll be a huge load at HF (and it won't erase very well either).

Clay1905 · 16th Jan 2026, 11:32 pm

Hi Kalee,
Indeed, this thing is weird, and I cannot be sure that the circuit is complete. The recorder came to me with strong evidence of tinkering. Cold solder, dirty poorly made joints, and even a capacitor 10 times too big in the oscillator. So it's well within the realm of possibility that there is something missing too.

Since writing, I have made a few more tests.
An external coil with similar DC impedance to the bias coil kills the oscillator. It's no easy job to measure, but the coil is about 0.6 ohms. Replacing this with a coil of 46 ohms gets the oscillator going strongly.
I briefly toyed with a little trimmer capacitor in parallel with the bias coil. I don't think there's sufficient impedance in the coil for any capacitor to charge significantly, so gave that idea away promptly.
Adding a 100 ohm variable resistor in series with the bias coil, I had the oscillator start at about 45 ohms. However, the bias signal was not present at the resistor / coil junction. This was with the resistor on the 'hot' side of the coil.
Disconnecting the bias coil from the internal head ground, I connected a 47 ohm resistor in series with the bias coil on the 'earthy' side. Once again I got a strongly running oscillator. I believe this indicates that the bias coil isn't shorted to ground internally.
No load, the oscillator produces about 1V p to p. With minimum impedance to run, about 0.5V
Is this the way to go? Should I be looking for maximum impedance to push the voltage up as far as possible?
Like you, I'm not at all keen to modify. As you say, It worked before. My goal is to get the machine as close to the original as I can. No service bulletin or schematic isn't a help there. It seems that every man and his dog had a different idea on how biasing should work, so comparing circuits isn't worth the time.
Clay.

Clay1905 · 18th Jan 2026, 10:36 am

Got it!
Ass-u-me etc.

Something in my mind asked, what if the oscillator circuit should be balanced, with the transformer coils and associated capacitors of equal values?
In a previous episode, it was established that the oscillator would start with C4 10 times smaller, under no load conditions. Replacing C5 in a similar manner not only has the oscillator start with only the erase & bias coils in circuit, the output is at about 4.8V peak to peak. Much healthier looking.
I'll post results of a trial recording soon.
I'm not in KAOS, so yippie-I-Ay-Ki-Oh!

kalee20 · 18th Jan 2026, 1:59 pm

Good show!

Making an oscillator is easy.

Making said oscillator the frequency you want, the voltage and current you want, the waveform purity you want, the stability you want, isn't.

Trying to fix a circuit when you're not sure what the voltage and current should be, nor the frequency, and when the load might be faulty or changed, the coil may have been changed, the capacitors not their original value, possibly even the valve had been swapped... oh and the circuit reconfigured, is starting the design process all over again!

ben · 19th Jan 2026, 12:36 am

I had a Tandberg recorder which had had a burnup in the oscillator. The tech that worked on it replaced a resistor with one ten times higher in value, presumably after misreading the diagram. I discovered that when investigating why recording/erasure was so weak.

So, apart from taking the time to understand how a circuit works, it also pays to check each and every component in the circuit concerned against the diagram, especially if someone has 'had a stab'!

Clay1905 · 19th Jan 2026, 2:10 am

A little more god news. The erase coil is working well now, and able to properly blank a wire.
However, actual recording remains elusive. The V.U. meter dances about in time with the input, but very little seems to happen at the wire. If I hook up the 'scope to the voice coil terminals, I see what looks just like the theory diagrams of signal + bias.
I think something is pulling down the output from the current amplifier, or similar. I have to make some sort of starting assumption to test. The assumption is that needing the volume control set to above 3/4 to get significant needle movement on the V.U. meter isn't right.

Ben,
If I had any sort of documentation at all, the last few months would have been a lot less angst ridden than it has been.
These 'Pyrox' brand machines were manufactured under license according to a patent held by The Armour Research Company. It seems that neither company was forthcoming with anything beyond advertising copy. And I'd be surprised if there were ever more than a hundred ever made.

14th Jan 2026, 7:33 am	#1
Clay1905 Tetrode Join Date: Sep 2020 Location: Blue Mountains City, New South Wales, Australia. Posts: 76	Bias Oscillator Partial Solution Hi Folks, The continuing story of the Pyrox bias oscillator not running: Looking at the schematic, I thought that the impedance (less than 1 Ohm DC) of the bias coil would appear as a short across the secondary of Tr2. By way of experiments, I first tried another coil of similar impedance, to be sure there was no short to ground in the bias head coil. This killed the oscillator dead. I tried an R.F. coil with substantially higher impedance and the oscillator runs happily. Same deal with both the R.F. and bias head coils in series. No big surprise there. So, question time. What's the better way to increase impedance in the bias head circuit? I could wind a coil and add that to the circuit, but what about adding a capacitor in parallel to the head coil? If my understanding is right, then if the capacitor is selected so that it and the head coil form a circuit that is resonant at bias frequency, then the impedance should be very high. But it this a good idea in this application? Keen to hear some thoughts, Clay. __________________ The higher it goes, the fewer.

16th Jan 2026, 10:30 pm	#3
kalee20 Dekatron Join Date: Feb 2007 Location: Lynton, N. Devon, UK. Posts: 7,926	Re: Bias Oscillator Partial Solution Nobody else has commented - and I've been pondering a couple of days! Fundamentally, something seems wrong: it presumably worked in the first place, so trying a modification doesn't seem right (unless something has been knowingly changed for something different, as the 'best available option' and it's a case of making it work). If connecting the erase head kills the oscillator, it almost seems as though there is a shorted turn in the erase head. It may still measure around the right DC resistance, but it'll be a huge load at HF (and it won't erase very well either).

16th Jan 2026, 11:32 pm	#4
Clay1905 Tetrode Join Date: Sep 2020 Location: Blue Mountains City, New South Wales, Australia. Posts: 76	Re: Bias Oscillator Partial Solution Hi Kalee, Indeed, this thing is weird, and I cannot be sure that the circuit is complete. The recorder came to me with strong evidence of tinkering. Cold solder, dirty poorly made joints, and even a capacitor 10 times too big in the oscillator. So it's well within the realm of possibility that there is something missing too. Since writing, I have made a few more tests. An external coil with similar DC impedance to the bias coil kills the oscillator. It's no easy job to measure, but the coil is about 0.6 ohms. Replacing this with a coil of 46 ohms gets the oscillator going strongly. I briefly toyed with a little trimmer capacitor in parallel with the bias coil. I don't think there's sufficient impedance in the coil for any capacitor to charge significantly, so gave that idea away promptly. Adding a 100 ohm variable resistor in series with the bias coil, I had the oscillator start at about 45 ohms. However, the bias signal was not present at the resistor / coil junction. This was with the resistor on the 'hot' side of the coil. Disconnecting the bias coil from the internal head ground, I connected a 47 ohm resistor in series with the bias coil on the 'earthy' side. Once again I got a strongly running oscillator. I believe this indicates that the bias coil isn't shorted to ground internally. No load, the oscillator produces about 1V p to p. With minimum impedance to run, about 0.5V Is this the way to go? Should I be looking for maximum impedance to push the voltage up as far as possible? Like you, I'm not at all keen to modify. As you say, It worked before. My goal is to get the machine as close to the original as I can. No service bulletin or schematic isn't a help there. It seems that every man and his dog had a different idea on how biasing should work, so comparing circuits isn't worth the time. Clay. __________________ The higher it goes, the fewer.

18th Jan 2026, 10:36 am	#5
Clay1905 Tetrode Join Date: Sep 2020 Location: Blue Mountains City, New South Wales, Australia. Posts: 76	Re: Bias Oscillator Partial Solution Got it! Ass-u-me etc. Something in my mind asked, what if the oscillator circuit should be balanced, with the transformer coils and associated capacitors of equal values? In a previous episode, it was established that the oscillator would start with C4 10 times smaller, under no load conditions. Replacing C5 in a similar manner not only has the oscillator start with only the erase & bias coils in circuit, the output is at about 4.8V peak to peak. Much healthier looking. I'll post results of a trial recording soon. I'm not in KAOS, so yippie-I-Ay-Ki-Oh! __________________ The higher it goes, the fewer.

18th Jan 2026, 1:59 pm	#6
kalee20 Dekatron Join Date: Feb 2007 Location: Lynton, N. Devon, UK. Posts: 7,926	Re: Bias Oscillator Partial Solution Good show! Making an oscillator is easy. Making said oscillator the frequency you want, the voltage and current you want, the waveform purity you want, the stability you want, isn't. Trying to fix a circuit when you're not sure what the voltage and current should be, nor the frequency, and when the load might be faulty or changed, the coil may have been changed, the capacitors not their original value, possibly even the valve had been swapped... oh and the circuit reconfigured, is starting the design process all over again!

19th Jan 2026, 12:36 am	#7
ben Dekatron Join Date: Aug 2006 Location: Madrid, Spain / Wirral, UK Posts: 7,681	Re: Bias Oscillator Partial Solution I had a Tandberg recorder which had had a burnup in the oscillator. The tech that worked on it replaced a resistor with one ten times higher in value, presumably after misreading the diagram. I discovered that when investigating why recording/erasure was so weak. So, apart from taking the time to understand how a circuit works, it also pays to check each and every component in the circuit concerned against the diagram, especially if someone has 'had a stab'! __________________ Regards, Ben.

19th Jan 2026, 2:10 am	#8
Clay1905 Tetrode Join Date: Sep 2020 Location: Blue Mountains City, New South Wales, Australia. Posts: 76	Re: Bias Oscillator Partial Solution A little more god news. The erase coil is working well now, and able to properly blank a wire. However, actual recording remains elusive. The V.U. meter dances about in time with the input, but very little seems to happen at the wire. If I hook up the 'scope to the voice coil terminals, I see what looks just like the theory diagrams of signal + bias. I think something is pulling down the output from the current amplifier, or similar. I have to make some sort of starting assumption to test. The assumption is that needing the volume control set to above 3/4 to get significant needle movement on the V.U. meter isn't right. Ben, If I had any sort of documentation at all, the last few months would have been a lot less angst ridden than it has been. These 'Pyrox' brand machines were manufactured under license according to a patent held by The Armour Research Company. It seems that neither company was forthcoming with anything beyond advertising copy. And I'd be surprised if there were ever more than a hundred ever made. __________________ The higher it goes, the fewer.