SET v. Pushpull


I did a search and didn't find any threads on this topic.  I am a newbie to tubes - I have a Tavish Adagio phono stage, and I am looking to add an integrated tube amp into the analog chain in my system - I have never owned one, so my knowledge level is low.  Right now I am running KEF LS50 speakers, but they are a little small for my listening area so possibly looking at larger speakers - I was very impressed by the Joseph Audio Pulsars I recently heard, as one example.

Anyway, most of the integrated tube amps in my price range are seem to be push pull, although the Line Magnetic is SET.  At a high level, I understand the push pull design gives higher output power than SET, but also higher potential for distortion.  

Any practical experience with the pros and cons of each?  And how important is this in the overall scheme, in real world applications?  Say, compared to the choice of output tube?

Thanks,


Tom 
tgr
Many excellent replies! 
You really want to do some listening to a variety of tube amps. Many tubes sound quite different and of course designs of a given tube do as well. There are a number of push pull amps that allow you to roll EL34s, 6550s and a few other variants.

SETs are a broad group and many are very low power(e.g. <10w). You really need a super friendly speaker(e.g. horn) to get the most of those. It all comes down to matching speaker+amp. 

A couple of personal experiences might be of help. I loved the sound of a number of SETs I heard, but I didn't like many of the horns and other speakers that were the best fit. That led me to 845 SET monoblocks from Dehavilland which run at 35w and allow you to open up your speaker options to many conventional speakers in the 93db neighborhood. Silverline La Folias were my best match with them. Merlin VSM Series also a good match, though lighter in the bottom end. 

When I had the opportunity I ran a pair of Atma-Sphere MA1 OTLs, which are another entirely unique class of tube amps that do many things better than most others again IF AND ONLY IF paired w/the right speakers (i.e. moderately efficient but more importantly with a friendly impedance curve. Some folks say they conquer that issue using items like Speltz Autoformers, to open up the speaker options to theoretically almost any impedance. Personally, the autoformers were not my cup of tea. OTLs had speed and clarity superior to most of the other tube amps I've heard of all other types. 

For me you've got to find the speaker you want first and then ask what amps will be a good fit with it, and consider a wide variety of topologies based on listening. If you can't audition, buy used so you can change gears without losing your shirt. Cheers,
Spencer
Post removed 
Tom,

You have been given very good input here. Almarg's points about impedance was definitely spot on (as usual in his posts): both as a key speaker attribute when considering to match with SET, and in showing your current speakers aren't the best fit with SET. The main points were around impedance variability over the frequency range. I also want to point that impedance magnitude is very important for SET matching as SET amps are constant current sources.

Most other amps are constant voltage sources, so as you halve the speaker impedance they deliver twice as much power, or half as much power as you double impedance. You've probably amps specs saying 200W into 8ohm and 400W into 4ohm. Here's the simplified math behind that:
Power (P) equals voltage (V) times current (I). P=V*I.
Also V=Z*I, where Z is impedance.
You can replace in the first one: P=V^2/Z. If an amp is constant voltage, power is inversely proportional to speaker impedance. Double Z, halved P.

On the other and SET are constant current, so in P=V*I you can replace into P=Z*I^2. As you double speaker impedance the amp delivers twice the power.

Look into speaker designs by SET amp manufacturers. For example Coincident Technology. Their speakers are 95dB-ish sensitive and 10 ohm, low variation impedance. Just looking at the 95dB sensitivity might lead you to believe they aren't good match for a 300B 8W, but the designer had that amp in mind when designing. BTW, note these aren't horns. While horns are indeed a good match for SETs, not only horns are suitable for SETs.

Also, another consideration often brought up by Ralph from Atmasphere: SETs produce extremely low distortion provided you run them at say 25% of their rated power, and low distortion up to 50% of their rated power. Distortion grows fast as you approach rated power. So while it's true SETs produce more distortion, that is true at their rated power, and a well designed system that incorporates SET should take into consideration running them as easily as possible. Hence it's also true SETs produce very low distortion, with the added benefit of producing more pleasing harmonics when they do.

Bottom line: it's all about execution. No blanket statement can be made. SETs could be better or worse. Your existing speakers aren't a good match for SET. But you also mentioned they are too small for your room. Maybe you want to go SET knowing you will replace your speakers with SET friendly later. Or replace your speakers first and then find an amp to match.

Sorry for the length!
@Lewinskih01, thank you kindly for the nice words. And for the generally excellent and informative post. But a little of it isn’t quite correct.

SETs are not constant current sources, which of course would mean that they provide an amount of current that remains essentially constant as load impedance varies, as long as they are operated within the limits of their maximum voltage, current, power, and thermal capabilities. They, and also push-pull tube amps, are not constant power sources, either. What is fair to say is that to a very loose approximation most of them behave as constant power sources, and how loose that approximation is depends on their output impedance. While a few others that have unusually high output impedances loosely approximate behaving as constant current sources, to a degree that depends on their output impedance.

As we have said, nearly all solid state amps maintain voltage that remains essentially constant into varying load impedances (for a given input voltage to the amp), as long as they are operated within the limits of their maximum voltage, current, power, and thermal capabilities. (Certain First Watt amps that have been intentionally designed to have high output impedances are notable exceptions). While SET and push-pull tube amps, when operated within their capabilities, will deliver more voltage and less current as load impedance increases, while delivering less voltage and more current as load impedance decreases. But the product of voltage and current in those cases will not remain constant for most such increases or decreases.

To see that, consider a tube amp to be an ideal voltage source (having an output impedance of zero, and outputting a voltage V) in series with a resistor, denoted as R1, that is equal to the amp’s output impedance. And consider the speaker impedance to be R2.

Per the voltage divider effect the voltage appearing across the speaker will be (V x R2)/(R1 + R2). The current provided to the speaker will be V/(R1 + R2). Plugging various values for R1 and R2 into those formulas, and calculating the product of the voltage and current provided to the speaker, will illustrate my point.

For example, consider an amp having an output impedance of 2 ohms (which is actually a good deal higher than the output impedance of the 4 ohm tap of the Line Magnetic SET amp that was referred to earlier), and speaker impedance that increases from 4 ohms at one frequency to 8 ohms at another frequency. You will find that both the current and the power that are delivered into 8 ohms are significantly less than what is delivered into 4 ohms, assuming the amp is operated within the limits of its maximum capabilities into both impedances.

Best regards,
-- Al

 One thing I would point out is that some of the distortion from push pull comes from the crossover region which can be more objectionable to your ear. SET distortion is mostly even ordered and usually is perceived as pleasant. But you do want to run SET very lightly in which case it’s one of the lowest distorting topologies around.  In my experience, running very efficient easy load speakers, I preferred SET over push poull and 0TL over SET by no small margin. 
 And in my experience SET running a proper load is not warm and fuzzy. It’s quick light and dynamic. I just found O TL to offer an even better picture into the musical event. Hard to describe.