Push-pull vs. Line tube preamp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

johnheath

Well-known member
Joined
Jul 31, 2014
Messages
890
Location
Sweden
Hi all…

I do not know if I am expressing myself correctly here but I have built a number of different tube line preamps and I have found some designs that sound really great with very quiet PSU.

A couple of days ago I found a push-pull output transformer (Sowter 8650) among my stuff and thought that I could try to build a push-pull tube preamp and found it to have very fine frequency response and and it sounded really good.

But I also started thinking about why these push-pull preamps are getting more and more rare these days (what I have found out anyway)?

What are the advantages and disadvantages between the two?

Maybe some of you guys can tell something about it?


Best regards

/John
 
Hi, good push-pull output transformers are more expensive and bigger than single end iron for capacitor coupled output stages without DC on the primaries, as far as I know. Cheers Thorsten
 
  Cap coupling has some problems, LF pole of course, high impedance driving the pri at the corner freq means increased low freq distortion.

  Direct coupling caps are nice, but the DC is a problem, push pull configuration has no equivalent DC as they compensate for each side. Also with PP configurations you have some distortion cancelation (2nd harmonic mainly) and I think this is one of the main advantages, as you can get greater linearity starting with a lower linearity stage. PP are more efficient than single ended, and can be biased from class A to class B and all the way in the middle.

  In the old days (impedance matching days) efficiency was a big deal even in signal level stages, as power amplification was expensive and a need. In modern days (bridging impedance days) efficiency hardly matters in signal level stages as delivered power is so low and, avoiding tubes, amplification stages costs cents. Even in tubes you don't need decent driving capabilities in the last stage, you can live with anything below 10k (of driving capability, not output impedance) so it makes sense a less efficient, single ended, cheaper, simpler transformer, big decent cap still cheap compared to a more complex transformer and an extra triode. Single ended, cap, any transformer laying around the bench and let it rock!

JS
 
joaquins said:
push pull configuration has no equivalent DC as they compensate for each side. Also with PP configurations you have some distortion cancelation (2nd harmonic mainly)
JS


thats the thing puts me off about PP. why would u wanna cancel 2nd harmonics! it ur octave high, and sits in the music or sound design perfectly without any adjustment!
however, if were canceling odd harmonics too, it would make sense to have clean line...
but my understanding is , as u stated, mainly second harmonics!
whats happening to 3rd and even harmonics! there is no info for it!

(ps: there are very good use for third.odd harmonics too... i am not skipping that btw)
 
joaquins said:
  Cap coupling has some problems, LF pole of course, high impedance driving the pri at the corner freq means increased low freq distortion.

  Direct coupling caps are nice, but the DC is a problem, push pull configuration has no equivalent DC as they compensate for each side. Also with PP configurations you have some distortion cancelation (2nd harmonic mainly) and I think this is one of the main advantages, as you can get greater linearity starting with a lower linearity stage. PP are more efficient than single ended, and can be biased from class A to class B and all the way in the middle.

  In the old days (impedance matching days) efficiency was a big deal even in signal level stages, as power amplification was expensive and a need. In modern days (bridging impedance days) efficiency hardly matters in signal level stages as delivered power is so low and, avoiding tubes, amplification stages costs cents. Even in tubes you don't need decent driving capabilities in the last stage, you can live with anything below 10k (of driving capability, not output impedance) so it makes sense a less efficient, single ended, cheaper, simpler transformer, big decent cap still cheap compared to a more complex transformer and an extra triode. Single ended, cap, any transformer laying around the bench and let it rock!

JS

Thank you sir

It also seems to be easier in line preamps to add full attenuation in form of potentiometers and HPF since you only work with a single signal?

I get the impression that in a push-pull amp full attenuation and HPF is bit more complex?


Best regards

/John
 
I have had or have Altec 458A, Altec 459A, Collins 6R, Collins 6N, Langevin 117A, Langevin AM5116-B.  To go into esoteric language, all sound bigger/wider than single ended types which sound more center focused and narrow in comparison.  It's said some places the predominant 2nd versus 3rd harmonic difference has a lot to do with that. 

Likewise consider almost any vari-mu compressor is a push-pull amp with more gain than most stand-alone preamps. 

Maybe most people haven't heard an combo SE/PP type in a preamp/line amp context, though almost everyone has heard a Fender or Marshall.    A mixed flavor. 

Many transformers that are not specifically specified for PP use work fine, and work better with PP connection.  Many of those types have center taps available.  Any of the classic UTC inputs A-10/HA-100/LS-10 etc deliver a better response plot connected PP.  Same with output transformers.  In many cases the types advertised as hum bucking windings look better.  I've seen UTC A-24's used to build Langevin 5116-B knockoffs.  We see people using cheap Edcor transformers for PP outputs in tube gear, even though they aren't specified for DC current.  Fine so long as the current is balanced and not excessive. 

Leads to the case of output level:  You will get higher output level from a SE amp with direct feed connection in the output transformer versus resistive feed and cap connected transformer.  The direct feed SE transformer is the most expensive option.  PP with the $12 Edcor will give greater output level, which may prove useful with modern interfacing.  Many of the classic SE resistive feed plate coupled amps struggle to deliver +4dBm with less than 3% distortion, as a 250V B+ will put 100-125V on the output plate through a resistor.  Anemic headroom in modern context.  You can move to the expensive direct feed transformer to raise B+ up around 220V, or you could go PP with the Edcor for similar high B+ and additional grunt from dual output tubes.  Maybe you don't need the grunt, but if you're trying for it with the basic classic circuits it needs to be considered.    Maybe you want the fuzz, maybe you want the clean. 

Rarely seen but used historically is the single ended amp with a PP output stage, one side being purely a dummy stage to deliver DC current balance in the output transformer, no signal to the grid on one side.  This lets you use the inexpensive output transformer to advantage in what otherwise acts exactly like a SE amp. 
 
emrr said:
I have had or have Altec 458A, Altec 459A, Collins 6R, Collins 6N, Langevin 117A, Langevin AM5116-B.  To go into esoteric language, all sound bigger/wider than single ended types which sound more center focused and narrow in comparison.  It's said some places the predominant 2nd versus 3rd harmonic difference has a lot to do with that. 

Likewise consider almost any vari-mu compressor is a push-pull amp with more gain than most stand-alone preamps. 

Maybe most people haven't heard an combo SE/PP type in a preamp/line amp context, though almost everyone has heard a Fender or Marshall.    A mixed flavor. 

Many transformers that are not specifically specified for PP use work fine, and work better with PP connection.  Many of those types have center taps available.  Any of the classic UTC inputs A-10/HA-100/LS-10 etc deliver a better response plot connected PP.  Same with output transformers.  In many cases the types advertised as hum bucking windings look better.  I've seen UTC A-24's used to build Langevin 5116-B knockoffs.  We see people using cheap Edcor transformers for PP outputs in tube gear, even though they aren't specified for DC current.  Fine so long as the current is balanced and not excessive. 

Leads to the case of output level:  You will get higher output level from a SE amp with direct feed connection in the output transformer versus resistive feed and cap connected transformer.  The direct feed SE transformer is the most expensive option.  PP with the $12 Edcor will give greater output level, which may prove useful with modern interfacing.  Many of the classic SE resistive feed plate coupled amps struggle to deliver +4dBm with less than 3% distortion, as a 250V B+ will put 100-125V on the output plate through a resistor.  Anemic headroom in modern context.  You can move to the expensive direct feed transformer to raise B+ up around 220V, or you could go PP with the Edcor for similar high B+ and additional grunt from dual output tubes.  Maybe you don't need the grunt, but if you're trying for it with the basic classic circuits it needs to be considered.    Maybe you want the fuzz, maybe you want the clean. 

Rarely seen but used historically is the single ended amp with a PP output stage, one side being purely a dummy stage to deliver DC current balance in the output transformer, no signal to the grid on one side.  This lets you use the inexpensive output transformer to advantage in what otherwise acts exactly like a SE amp.

Thank you sir for a very detailed explanation

Yes, I actually found the PP I just built to have an "air" to the sound which sounds just great. I also found that, despite having a more expensive PP output transformer, the need for components is less… reducing the overall cost.

The circuit that I made is a blend of known circuits like the ones you mentioned initially in your comment… I also get the impression that you have more "room" for elaboration with feedback and gain without entering strange cut-off and distortion compared to a SE… I might be saying something way off here but as I was fiddling with the circuit that is what I noticed.

Thanks again

Best regards

/John
 
kambo said:
thats the thing puts me off about PP. why would u wanna cancel 2nd harmonics! it ur octave high, and sits in the music or sound design perfectly without any adjustment!
however, if were canceling odd harmonics too, it would make sense to have clean line...
but my understanding is , as u stated, mainly second harmonics!
whats happening to 3rd and even harmonics! there is no info for it!

(ps: there are very good use for third.odd harmonics too... i am not skipping that btw)

  2nd harmonic doesn't sound nice on top of your music... who told you so? It might sound nice on top of a guitar but let the guitar player have a stomp for that.

  Even harmonics gets reduced by this method,  odd doesn't as they sum together as much as the signal does. This is the way a PP works, if you want a circuit to generate loads of even harmonics and very little odds you have to look elsewhere. There is plenty of info about what happens with PP stages, I just didn't mention everything, feel free to google that.

johnheath said:
Thank you sir

It also seems to be easier in line preamps to add full attenuation in form of potentiometers and HPF since you only work with a single signal?

I get the impression that in a push-pull amp full attenuation and HPF is bit more complex?

Best regards
/John

  The discussion started about output stages, you could do level and tone control before getting there. In any case there is a way to control tone and volume right at the PP grids, shunting them with a resistor will attenuate the signal, shunting them with a series RC would cut some highs. Doing so with an RL would attenuate LF.

  Note that the signal is still only one, and "travels" by two path as every signal. In the usual case the signal "travels" by the signal conductor and a reference shield or ground. Here the signal "travels" in two wires, it just happens to be in opposite polarities in each one of them but it comes with a lot of symmetries which help a lot while designing.

JS
 
joaquins said:
  The discussion started about output stages, you could do level and tone control before getting there. In any case there is a way to control tone and volume right at the PP grids, shunting them with a resistor will attenuate the signal, shunting them with a series RC would cut some highs. Doing so with an RL would attenuate LF.

  Note that the signal is still only one, and "travels" by two path as every signal. In the usual case the signal "travels" by the signal conductor and a reference shield or ground. Here the signal "travels" in two wires, it just happens to be in opposite polarities in each one of them but it comes with a lot of symmetries which help a lot while designing.

JS

Thank you sir

I could have given a more detailed information. I was referring to a circuit where you send the signal in two paths all the way… not grounding the negative of the input transformer. I guess that a HPF somewhere in the path must include them both? And likewise with a attenuator… like using a two gang pot?

Best regards

/John

 
joaquins said:
  2nd harmonic doesn't sound nice on top of your music... who told you so?

no one.... that what i use on channel inputs... some clipping circuits (no tube) and some  VST plugs works like PP too, no second harmonics. i sometime use that on master fader, depends on what i am doing!

 
johnheath said:
But I also started thinking about why these push-pull preamps are getting more and more rare these days (what I have found out anyway)?
Push-pull configuration is an elegant way of getting rid of 2nd-harmonic AND, with it, a lot of distortion. But people want to hear what they've been told is "tube sound", or in other words, asymetric distortion.
What's the point of paying a premium for tubes and transformers and not having the dirt?  :eek:
 
abbey road d enfer said:
Push-pull configuration is an elegant way of getting rid of 2nd-harmonic AND, with it, a lot of distortion. But people want to hear what they've been told is "tube sound", or in other words, asymetric distortion.
What's the point of paying a premium for tubes and transformers and not having the dirt?  :eek:

Yes, without knowing all the facts my estimation is that there are thousands of "tube sounds" out there?

But, I must confess that I was impressed with this "new" (for me) design with its fine sound and straight forward amplification without the usual clipping and distortion. And of course it is not far from thinking …why aren't there more modern designs using PP? If I have understood it correctly they often used the old preamps in desks with separate PSU and power amps and gain controls and all.

Today it seems like everything should be placed inside a single box with phantom power, Hi and Lo filters, pads and more but it should still be possible to build pro gear with tubes using PP and all the modern features in a box. I know that a decent PP-output transformer is more expensive than a decent single ended… well, I guess it is up to everybody to choose anyway :)


Best regards

/John
 
johnheath said:
Yes, without knowing all the facts my estimation is that there are thousands of "tube sounds" out there?

But, I must confess that I was impressed with this "new" (for me) design with its fine sound and straight forward amplification without the usual clipping and distortion. And of course it is not far from thinking …why aren't there more modern designs using PP? If I have understood it correctly they often used the old preamps in desks with separate PSU and power amps and gain controls and all. 
I really think the reason is they lack the tube "flavour" that's so easy to get with a SE stage.


I know that a decent PP-output transformer is more expensive than a decent single ended… 
Not true. For the same quantity of iron and copper, a PP xfmr can handle about 4 times the power of a SE. That's for a capacitor-coupled (no DC) stage; it's even worse when there's DC. A good xfmr must have sandwich construction, so there is no penalty for the number of windings in a PP xfmr.
 
abbey road d enfer said:
I really think the reason is they lack the tube "flavour" that's so easy to get with a SE stage.

Not true. For the same quantity of iron and copper, a PP xfmr can handle about 4 times the power of a SE. That's for a capacitor-coupled (no DC) stage; it's even worse when there's DC. A good xfmr must have sandwich construction, so there is no penalty for the number of windings in a PP xfmr.

Ok, nice to know :)

I was just comparing some of the output transformers that I have been looking at and used. But I certainly don't know all that much .

Best regards

/John
 
johnheath said:
Thank you sir

I could have given a more detailed information. I was referring to a circuit where you send the signal in two paths all the way… not grounding the negative of the input transformer. I guess that a HPF somewhere in the path must include them both? And likewise with a attenuator… like using a two gang pot?

Best regards

/John

  No, think more like a U pad. Single gang, 2 fixed resistors (one on each signal path) and a variable one in the middle, reducing the value of the variable generates attenuation, if it has a cap in series HF cut, if it has an inductor in series LF cut. It could have both, the two fixed resistors could have a cap each in series. Many other combinations.

  As I said, it's a single signal, even if you are working with a fully balanced path. The advantage of such thing I'm not sure,  6 extra dB  (in gain and headroom) for twice the tube count doesn't add up by it self. Good CMRR should be obtained in the transformer, trying to improve that with tube matching path doesn't make a lot of sense... I don't know the actual decision why to go that route, maybe just to reduce distortion in a perfectly symmetric path as abbey said.

JS
 
abbey road d enfer said:
Not true. For the same quantity of iron and copper, a PP xfmr can handle about 4 times the power of a SE. That's for a capacitor-coupled (no DC) stage; it's even worse when there's DC. A good xfmr must have sandwich construction, so there is no penalty for the number of windings in a PP xfmr.

It's such a pity that I can't find those super cheap (full audio bandwith) PP xfmr! Reality sucks...😀
 
rock soderstrom said:
It's such a pity that I can't find those super cheap (full audio bandwith) PP xfmr! Reality sucks...😀
Yes; it's a chicken-and-egg situation. Is it because there's not much interest in line-level PP outputs that the manufacturers offer is scarce, or the contrary?
That's a stumbling point for DIY, but for a manufacturer that orders custom xfmrs, a PP xfmr is cheaper than SE.
I have experimented with a bare bones design for a low-gain high-headroom preamp, using a single dual triode in fully balanced topology. The only xfmr I could find that was adequate was the Sowter 9084 (at about £80!), originally designed for the Ampex 351.
 
Back
Top