Forums

Old school Push pull audio amplifier transformer flux managment

Started by mike daniels December 11, 2021
I have a switched mode power supply background and an looking at something like this for an old school audio amp design.

https://www.circuitstoday.com/push-pull-amplifier


For SMPS pushpull we always used current mode control to assure the flux in the core does not walk all the way into saturation.

Looking at these designs, none monitor or measure the current to assure the flux does not walk.    Many of them are biased class AB so there is always a small DC current through the windings.   if they are not perfectly balanced, that would cause the core to bias to one side.  if the first sign wave peek pushes it further into that side, it would saturate the core for that half cycle while the opposite half cycle resets it lower.     This of course would cause distortion on the output signal.   I'm looking at adding a DC balancing circut with opamps and a reference to assure the split in bias current is as close to perfect as possible.  The old SMPS trick of adding resistance is undesirable since that would increase the output impedance.    

in short, for you guys that have made a few of these back in the day, what did you do to assure the flux in the output transformer did not wander to the saturation zones due to transistor mismatch or DC bias mismatch?  

I do realize you can tune the dc bias manually but will that hold up over years of use?   

On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels
<dmcousin2000@gmail.com> wrote:

>I have a switched mode power supply background and an looking at something like this for an old school audio amp design. > >https://www.circuitstoday.com/push-pull-amplifier
Nice example of thermal runaway. And random transformer phasing. And distortion. -- Father Brown's figure remained quite dark and still; but in that instant he had lost his head. His head was always most valuable when he had lost it.
On 12/11/21 10:34 AM, jlarkin@highlandsniptechnology.com wrote:
> On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels > <dmcousin2000@gmail.com> wrote: > >> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. >> >> https://www.circuitstoday.com/push-pull-amplifier > > Nice example of thermal runaway. And random transformer phasing. And > distortion. >
That worked OK in those "all Japanese six" radios I've seen in the history books; I don't think they built them with emitter resistors either pushing 1/4 watt they weren't gonna run away. A cap across the speaker coil made a Zobel that was good enough for rock n roll I guess when the battery was providing a few ohms of impedance Sine oscillator plus the same circuit with parallel PNPs with Rs in the emitters and base made an ultrasonic power oscillator in this old chestnut: <https://www.ebay.com/itm/Industrial-Transistor-Circuits-by-Allan-Lytel-1st-Ed-third-Printing-1969-/254560767963>
On 12/11/21 9:08 AM, mike daniels wrote:
> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. > > https://www.circuitstoday.com/push-pull-amplifier > > > For SMPS pushpull we always used current mode control to assure the flux in the core does not walk all the way into saturation. > > Looking at these designs, none monitor or measure the current to assure the flux does not walk. Many of them are biased class AB so there is always a small DC current through the windings. if they are not perfectly balanced, that would cause the core to bias to one side. if the first sign wave peek pushes it further into that side, it would saturate the core for that half cycle while the opposite half cycle resets it lower. This of course would cause distortion on the output signal. I'm looking at adding a DC balancing circut with opamps and a reference to assure the split in bias current is as close to perfect as possible. The old SMPS trick of adding resistance is undesirable since that would increase the output impedance.
Don't think you need to add any resistance the transistors will have plenty already, they're not being hard-switched.
> in short, for you guys that have made a few of these back in the day, what did you do to assure the flux in the output transformer did not wander to the saturation zones due to transistor mismatch or DC bias mismatch? > > I do realize you can tune the dc bias manually but will that hold up over years of use? >
I don't think I've ever seen any special precautions against flux-walking taken in schematics of those old circuits from the time, the transistors aren't ever switched hard on, and the internal impedance of germanium transistors run class B or AB I think provided enough negative feedback across the winding to keep it in check, it's not like a switching converter where milliohm resistances are connected in series on each half-cycle. In class B there's an automatic "dead time" too There's a schematic of a 200 (?) watt ultrasonic power amplifier in this book that uses a variant of that circuit if you can find a scanned copy online (I'll upload later if not), they don't do anything special IIRC. <https://www.ebay.com/itm/Industrial-Transistor-Circuits-by-Allan-Lytel-1st-Ed-third-Printing-1969-/254560767963>
On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels
<dmcousin2000@gmail.com> wrote:

>I have a switched mode power supply background and an looking at something like this for an old school audio amp design. > >https://www.circuitstoday.com/push-pull-amplifier > > >For SMPS pushpull we always used current mode control to assure the flux in the core does not walk all the way into saturation. > >Looking at these designs, none monitor or measure the current to assure the flux does not walk. Many of them are biased class AB so there is always a small DC current through the windings. if they are not perfectly balanced, that would cause the core to bias to one side. if the first sign wave peek pushes it further into that side, it would saturate the core for that half cycle while the opposite half cycle resets it lower. This of course would cause distortion on the output signal. I'm looking at adding a DC balancing circut with opamps and a reference to assure the split in bias current is as close to perfect as possible. The old SMPS trick of adding resistance is undesirable since that would increase the output impedance. > >in short, for you guys that have made a few of these back in the day, what did you do to assure the flux in the output transformer did not wander to the saturation zones due to transistor mismatch or DC bias mismatch? > >I do realize you can tune the dc bias manually but will that hold up over years of use?
For audio, you're working with a pretty low Q in the magnetic circuit, a typical peak to average power setting that's ridiculous, defined load impedances and a host of other factors that don't allow you to get anywhere near the point where anything but stray capacitance and primary inductance are likely to dominate the design. If you add to that the common requirement for matching in push-pull circuits and gapping in single-ended (yech) circuits, it's not an issue that produces dramatic failures as frequently as it will in higher frequency power-only delivery circuits, where you can destabilize and/or blow up your circuit in a few switching cycles. RL
On 12/11/21 12:39 PM, legg wrote:
> On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels > <dmcousin2000@gmail.com> wrote: > >> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. >> >> https://www.circuitstoday.com/push-pull-amplifier >> >> >> For SMPS pushpull we always used current mode control to assure the flux in the core does not walk all the way into saturation. >> >> Looking at these designs, none monitor or measure the current to assure the flux does not walk. Many of them are biased class AB so there is always a small DC current through the windings. if they are not perfectly balanced, that would cause the core to bias to one side. if the first sign wave peek pushes it further into that side, it would saturate the core for that half cycle while the opposite half cycle resets it lower. This of course would cause distortion on the output signal. I'm looking at adding a DC balancing circut with opamps and a reference to assure the split in bias current is as close to perfect as possible. The old SMPS trick of adding resistance is undesirable since that would increase the output impedance. >> >> in short, for you guys that have made a few of these back in the day, what did you do to assure the flux in the output transformer did not wander to the saturation zones due to transistor mismatch or DC bias mismatch? >> >> I do realize you can tune the dc bias manually but will that hold up over years of use? > > For audio, you're working with a pretty low Q in the > magnetic circuit, a typical peak to average power setting > that's ridiculous, defined load impedances and a host of > other factors that don't allow you to get anywhere near > the point where anything but stray capacitance and > primary inductance are likely to dominate the design. > > If you add to that the common requirement for matching > in push-pull circuits and gapping in single-ended (yech) > circuits, it's not an issue that produces dramatic > failures as frequently as it will in higher frequency > power-only delivery circuits, where you can destabilize > and/or blow up your circuit in a few switching cycles. > > RL >
Did they try to match the transistors on that topology, like tubes? I think most physical examples of this topology outside battery radios were scrap before I was born.
On Saturday, December 11, 2021 at 7:37:36 PM UTC, bitrex wrote:
> On 12/11/21 12:39 PM, legg wrote: > > On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels > > <dmcous...@gmail.com> wrote: > > > >> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. > >> > >> https://www.circuitstoday.com/push-pull-amplifier
That is an old circuit, ca 1960's,... https://worldradiohistory.com/BOOKSHELF-ARH/RCA-Books/RCA-Transistor-Manual-1964.pdf pages 344 - 360. Such an "old school audio amp design" will have limited fun, and indeterminably educational - you'll have to know how to interpret its imperfect behavior. Finding those transformers might be difficult or expensive... try www.hammfg.com/electronics/transformers/audio Anyway, its not hi-fi by today's standards, and won't be stable. Skip these "learn about circuits" examples: they are very basic, only meant to teach, not so practical. Professional-design power amps using discrete bipolar transistors do exist, though I'd skip past the germanium and early silicon BJT eras... and skip transformer coupling. Simply no need for those since good NPN-PNP complementary power BJTs appeared. Today, a hobbyist can grab one of the audio class-AB power ICs and get pretty good results. Quickly. Or better yet, a class-D module... cheers, RS
On Sat, 11 Dec 2021 11:30:31 -0500, bitrex <user@example.net> wrote:

>On 12/11/21 10:34 AM, jlarkin@highlandsniptechnology.com wrote: >> On Sat, 11 Dec 2021 06:08:01 -0800 (PST), mike daniels >> <dmcousin2000@gmail.com> wrote: >> >>> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. >>> >>> https://www.circuitstoday.com/push-pull-amplifier >> >> Nice example of thermal runaway. And random transformer phasing. And >> distortion. >> > >That worked OK in those "all Japanese six" radios I've seen in the >history books; I don't think they built them with emitter resistors >either pushing 1/4 watt they weren't gonna run away. A cap across the >speaker coil made a Zobel that was good enough for rock n roll I guess >when the battery was providing a few ohms of impedance
Nice damping factor too. -- Father Brown's figure remained quite dark and still; but in that instant he had lost his head. His head was always most valuable when he had lost it.
 dmcous...@gmail.com wrote:
========================
> I have a switched mode power supply background and an looking at something like this for an old school audio amp design. > > https://www.circuitstoday.com/push-pull-amplifier > > For SMPS pushpull we always used current mode control to assure the flux in the core does not walk all the way into saturation. >
** P-P audio amplifiers are nothing like SMPS.
> Looking at these designs, none monitor or measure the current to assure the flux does not walk.
** Simply not an issue.
> Many of them are biased class AB so there is always a small DC current through the windings. > if they are not perfectly balanced, that would > cause the core to bias to one side.
** P-P audio output transformers with many E and I shaped laminations create a distributed air gap that stops this. Core balance can become an issue if a toroidal type is used - which is quite rare in audio amps.
> I'm looking at adding a DC balancing circut with opamps and a reference to assure the split in bias current is as close to perfect as possible.
** Not necessary.
> in short, for you guys that have made a few of these back in the day, what did you do to assure the flux in > the output transformer did not wander to the saturation zones due to transistor mismatch or DC bias mismatch?
** Match the Hfes of the output pair. The point you are missing is that transformer core saturation is only an issue at the lowest operating frequencies - generally under 50Hz for a high quality design. Higher frequencies are barely affected by a bit of magnetic imbalance. Cos they swing the core so little. ...... Phil
 bitrex wrote:
==========
> > I think most physical examples of this topology outside battery radios > were scrap before I was born. >
** Not true at all. Still used for 100V line PA systems & emergency evacuation amplifiers. Such amps must run from a 12 or 24V back up battery and output 70 or 100V rms of audio. Of course using silicon transistors, good bias stabilisation and sharp limiting of low frequencies to preclude audio transformers saturating. ...... Phil