What an OTL is, and why you should care

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#1
In audio, the acronym OTL has been around since the 1950s. It refers to a tube amplifier usually, one without an output transformer.

In the world of audio, tubes are often docked for poor bandwidth and high distortion by the advocates of transistors. But it turns out that a lot of that distortion can be traced to the presence of an output transformer, which also limits bandwidth. IOW if you can get rid of the transformer, you have the possibility of reducing distortion and increasing bandwidth.

This is indeed a fact, not just a possibility, but like anything else in life comes at a price. In designing such an amplifier, it is rapidly seen that only a few types of power tubes are suitable for OTL service. Such tubes usually have high plate current capacity, reduced plate resistance and are fairly capable of operating with good linearity at lower plate voltages (some examples are the 6AS7G and variants, the 6C33, EL509, EL519/PL519, along with some out of production types like the 6LF6, 6336 and 7241). Of course, the output transformer serves to match the high impedance of the typical power tube to the low impedance of a loudspeaker, and removing it, even with the high current tube types mentioned, usually means that the resulting amplifier is going to have a fairly high output impedance when compared to a transformer-coupled amplifier. For this reason many OTLs often employ high amounts of global negative feedback, in order to allow the amplifier to operate more as a voltage source which is typically required in the operation of many loudspeakers (although this is not by any means a universal truth- more on that later).

There are two main types of OTLs in service today. The Futterman (named for Julius Futterman) circuit and its variants represents one approach; the primary identifier being that the output section uses a 'totem pole' approach, wherein the output circuit is push-pull, but one tube is driving the speaker with its plate connection, while the other uses its cathode. This approach means that the output impedance varies depending on whether the amp is pushing or pulling so to speak, and this usually requires a fair amount of feedback to linearize the circuit. The second type of circuit employs an output circuit known as a Circlotron, wherein there are two banks of power tubes balanced against two power supply banks of equal polarity in a bridge circuit. This approach has a lower output impedance than a totem pole and lower distortion as it is a true symmetrical push-pull circuit (and is more expensive to build due to the need for two separate power supplies).

Both circuits have special driver circuit requirements as usually the grids of such power tubes (and bear in mind, usually there are a number of power tubes in parallel to increase current capacity of the output circuit) have a considerable amount of capacitance and in the case of the Futterman, also have unequal drive requirements (to deal with the fact that the output section is asymmetrical).

If a similar amount of feedback is used in these circuits, the distortion will be similar or lower than is seen in most traditional transistor circuits. Bandwidth will also be considerably wider; its no problem at all to have full power bandwidth from as little as 1 Hz to well over 500KHz. Usually bandwidth has to be limited to prevent problems with RFI and oscillation, not unlike a solid state amplifier. So in this regard, OTLs are quite successful!

More in succeeding posts...
 

steve williams

Site Founder, Site Owner, Administrator
#2
Ralph

thanks for starting what I consider a most interesting topic esp having heard your OTL amps and the sound was wonderful

I am going to "stick" this thread as a source of reference
 

LL21

Active Member
Dec 26, 2010
10,549
2
38
#3

c1ferrari

Member Sponsor & WBF Founding Member
May 15, 2010
2,025
1
36
#5
It has been my great fortune to obtain an all Atma-Sphere system to drive the MegaLine III loudspeakers.
The Atma-Sphere will enchant 6SN7 tube connoisseurs :p
Looking forward to reading future installments.
Thanks, Ralph :cool:
 

LL21

Active Member
Dec 26, 2010
10,549
2
38
#6
I think one of the amateur questions (ie, from me) is about the power handling of an OTL. I have read that into tougher loads, people voice concerns about OTLs which as you say don't have the 'translation' of the tubes higher impedance vs the speakers lower impedance. Thanks for helping explain: how OTL can be designed around this...or how to select speakers that will not trigger this issue (if I am correct in understanding it can be an issue).
 

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#7
^^ You have put your finger on the primary issue with OTLs! OTLs have a reputation of being load sensitive and this is true depending on the amp itself. Generally speaking the OTLs that use pentode power tubes will have greater difficulty driving lower impedance loudspeakers. On top of that, for many people the low-powered OTL that also drives 4-ohm speakers is sort of the Holy Grail in OTL-land. The simple fact is though that the smaller you make the OTL, the less efficient it becomes unless you also increase the impedance of the speaker. With smaller OTLs (in the case of our 30-watt amplifier for example), although it can drive an 8 ohm speaker, it works much better and makes more power driving 16 ohms. OTOH, we have an OTL that makes its maximum power (over 500 watts) into 3 ohms and thus is comfortable driving almost any speaker made.

Other concerns are the quantity of power tubes and the heat produced. Generally speaking the heat is related more to the class of operation than it is the number of tubes- Class A runs hotter and makes more heat. FWIW, most OTLs are class AB and really don't produce that much heat. An exception are OTLs (or any amplifier actually) that are based on the Russian 6C33. That power tube has a very hot filament and simply makes a lot of heat on that account. By contrast most of the other tubes that are used with OTLs can be in Standby all day (no B+ applied) and they will be only slightly warm- such that you can touch them without any concern of burning yourself.

The thing that OTLs bring to the table is they are the most transparent amplifiers made. That's it in a nutshell. Quite literally, there are no tube amps and there are no transistor amps that express greater transparency. So for those that want to get closer to the music, sorting out how to get an OTL into their system is a good way to do it. This is where you hear all the stories about equipment matching. OTLs generally work better driving higher impedances. I don't think 8 ohms is that much of a problem, but you would be surprised how many people want to take a small OTL and drive Magnaplanars or other low impedance speakers! Quite often in these situations, not only is the impedance of the speaker problematic, but so it the power level. IME many lower impedance speakers are also less efficient, and efficiency always plays a large role with tube amplifiers whether OTL or not- for the simple reason that tube amplifier power is more expensive than transistor amplifier power. If you wonder why Class D has been a big push in the last decade, that latter point is why- amplifier companies make a lot more money on Class D amps than they do on traditional transistor designs.

So tube amps in general need speakers that are easier to drive if you want to really get the best results from the dollars invested. If you keep this simple fact in mind, you do better with all tube amps and OTLs in particular. Here's another universal fact: Its one thing if an amplifier can drive a speaker, but if the amplifier does not have to work hard doing it, its going to have lower distortion and will sound better. Due to this universality, I don't really see the point of 4 ohms or less in speaker design in high end audio, if sound quality is your goal.
 

ack

VIP/Donor & WBF Founding Member
May 6, 2010
5,149
6
38
Boston, MA
#8
Thanks Ralph. A couple of additional comments we should perhaps add: 1) the other problem with transformer-coupled tube amps is transformer hysteresis; 2) Regarding OTLs being the most transparent, we should probably discuss the effect of capacitors in the signal path (including possible output-coupling capacitors). Regarding capacitors, I think the wider SS amplifier scene will soon be changing in that respect; the following is from the Spectral 400RS web page (their amps have always been DC-coupled, as far as I know; and now: ) The new generation amplifier systemboard of the DMA-400 features advanced RF layout techniques for ultra stable high frequency operation without ferrite stabilization or capacitors in the signal path). I am not aware of any other SS manufacturer making such a claim, and even theirs leaves a little bit of haze in interpretation: does the "systemboard" also include the driver stage which sits in its own board?!? I don't know.
 

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#9
Actually there is more to it than that regarding transformers. When you have an OPT (OutPut Transformer), the voltage is stepped down from the high voltage on the power tubes to the low voltage/high current of the loudspeaker. People often don't think of this, but if you don't have to make so much voltage in the amplifier, you also don't need as many gain stages. So in an OTL theoretically you could have less gain stages and in practice we are able to build OTLs with only one.

Not all OTLs have to have coupling caps at their outputs. Our amps for example are direct-coupled, and the driver circuit is direct-coupled to the output tubes. IOW if care is taken in the design, coupling caps can be minimized and I know of at least one design that has no coupling caps in the circuit at all. IME I think there is a good argument for a coupling cap somewhere in the circuit, as by doing so there is a low frequency pole is introduced into the amplifier circuit, limiting its low frequency bandwidth to a frequency higher than the frequency pole(s) in the power supply. In this way IM can be reduced, since if the amp can go lower than the power supply (IOW to DC) it will also have the ability to modulate the power supply. This is seen at times in older amps where the filter capacitors are failing, resulting in a thumping sound known as 'motorboating'.

Stability has definately been a factor in past OTL designs. In fact years ago it was such a problem that people used to think 'its an OTL so its going to blow up'. Its taken some decades, but that is no longer the common wisdom. OTLs can be as stable as any other amplifier. In our case we don't use negative feedback, so our amps tend to be unconditionally stable- IOW will not oscillate regardless of input condition or load on the output.
 

Peter Breuninger

[Industry Expert] Member Sponsor
Jul 20, 2010
1,231
0
0
#10
Great work, Ralph.
 

ack

VIP/Donor & WBF Founding Member
May 6, 2010
5,149
6
38
Boston, MA
#11
In our case we don't use negative feedback, so our amps tend to be unconditionally stable- IOW will not oscillate regardless of input condition or load on the output.
And you can still get low enough output impedance? How? By virtue of using multiple tubes in parallel? Something else?
 

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#12
^^ The question probably should be 'how low does the output impedance of the amplifier have to be in order to get excellent performance from the speaker?'

Obviously 4 ohm speakers are challenging for an OTL with no feedback and not a lot of power to begin with. In practice, we find that the use of feedback has no effect on the power output into a particular load impedance whatsoever. In fact it should not; if it did it would violate a fundamental rule of electricity known as Kirchoff's Law.

Our larger OTLs (MA-1, MA-2 and MA-3) are more comfortable with 4 ohm loads. But there is more to it than that- the real question is whether the amp is behaving like a voltage source. Now it happens that not all loudspeakers are designed with a voltage source amplifier in mind, nor are all amplifiers voltage sources. A good example of an amplifier that gets a lot of rave comments, but is nowhere near a voltage source is just about any SET amplifier. They measure poorly but get nice comments, and this has been going since the SET reappeared in the marketplace in the early 1990s! One need not look much further than that to know that there is more to this than just voltage response. It turns out that most tube amplifiers without loop negative feedback will tend to operate more like a power source than a voltage source, and there are a good number of speakers designed with that sort of behavior in mind. You can read more about that at this link: http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php

However to answer the question generally speaking any OTL will employ paralleled power tubes to reduce its output impedance. It happens that we use triodes, which tend to be lower output impedance on their own, and the Circlotron output circuit helps too, as the output impedance tends to be about 1/2 that of a totem pole circuit employing the same type and number of power tubes.

A certain amount of pragmatism must be employed when solving any engineering problem. IMO being pragmatic is important in audio- the less made up stories and assumptions one carries as a designer, the more likely a simple solution may present itself. As an example I like to point to the issue of lower impedance loudspeakers in general. Whether OTLs even exist is unimportant when we examine the low impedance issue- that is to say certain things come to light very quickly. First of all, all amplifiers, tube or solid state and class D, have higher distortion when driving 4 ohms as opposed to 8 ohms. Its easy to see the in the specs. Its also easy to hear- because the types of distortion that typically appear are higher ordered harmonics and intermodulations, both of which are audible to the human ear in small amounts.

Since this is very much the case one might ask why there are so many 4 ohm speakers, and if you look at the history of audio the reasons why go back decades, to a horsepower race that looked good on paper several decades ago ('look- if you cut the impedance in half, the power of the amp is doubled!!') but in practice was only to the benefit of those that placed a higher value on the specs than they do on the sound actually emerging from the speaker itself. These days we have become so used to this that quite literally we don't think about it.

And this is what I mean by being pragmatic. Not only do amplifier distortion specs suffer, but also the speaker cable itself becomes much more critical, as it becomes a significant portion of the source impedance driving the loudspeaker. By comparison, 16 ohm speakers are relatively non-critical of speaker cables. I'm old enough to remember the old days when a length of zip cord from the hardware store did the job, but back in those days I had 16 ohm speakers. That was part of the reason I got away with it. With a 4 ohm speaker, the cable should not really exceed 4-5 feet for best results and attention should also be paid to the overall gauge, purity of materials and also geometry, all of which affect the characteristic impedance of the cable (which affects to some degree the efficiency of the cable at delivering the signal to the speaker).

Amplifiers of course make less distortion into the higher impedances; quite literally if you are a speaker designer and want your speaker to sound smoother and more detailed at the same time, raising the impedance while maintaining other parameters equal will do the trick, regardless of the amplifier used. Now if we are talking 8 or 16 ohms, all of a sudden OTLs are quite practical as long as the speaker does not also have excessive inefficiency issues. Something I often tell people because it is a simple truth is this:

"If you are interested in sound quality above all else, your amplifier investment dollar will be best served by a loudspeaker that is at least 8 ohms or more. If sound pressure is your goal, then you have a weak (3 db) but valid argument for a 4 ohm speaker, if you have a solid state amplifier."

Now a side benefit of higher impedances is that generally speaking, tube amps (OTL or not) will put out more power into a higher impedance. Not only that, but quite often they run cooler and draw less power from the wall because the output section is operating more efficiently- less of the power made by the amp is being converted to heat (if only in the OPT). Its something to think about. This is why I don't think there is a good argument for lower impedance speakers in high end audio regardless of the amplifier used. The harder you make your amp work, the more distorted its output will be and that translates directly to how close to the music the end result will be.
 

BruceD

VIP/Donor
Dec 13, 2013
897
0
16
#13
Well put and explained Ralph---Yes indeed I was first made aware of your excellent Amps driving my good friend David Magnan's Stacked Quads some years ago

My interest in OTL's furthered by another old pal Jon Syder--who was Harvey Rosenbergs right hand man .

Keep up the great work!

BruceD
 
#14
I was looking for Atma-sphere thread and found this old post. However, I still need to get some tips on how to preserve the 6as7G tubes.

Is it better to leave them on (filament heating) or I should just turn them off? When not listening, I will not engage the B+ voltage.

I read somewhere, in order to prolong the life span of 6as7G, it's better to leave them ON and when we replaced these tubes, we need to have a 72 hours of preconditioning session which I have diligently followed.

Thanks for the advice in advance.

In audio, the acronym OTL has been around since the 1950s. It refers to a tube amplifier usually, one without an output transformer.

In the world of audio, tubes are often docked for poor bandwidth and high distortion by the advocates of transistors. But it turns out that a lot of that distortion can be traced to the presence of an output transformer, which also limits bandwidth. IOW if you can get rid of the transformer, you have the possibility of reducing distortion and increasing bandwidth.

This is indeed a fact, not just a possibility, but like anything else in life comes at a price. In designing such an amplifier, it is rapidly seen that only a few types of power tubes are suitable for OTL service. Such tubes usually have high plate current capacity, reduced plate resistance and are fairly capable of operating with good linearity at lower plate voltages (some examples are the 6AS7G and variants, the 6C33, EL509, EL519/PL519, along with some out of production types like the 6LF6, 6336 and 7241). Of course, the output transformer serves to match the high impedance of the typical power tube to the low impedance of a loudspeaker, and removing it, even with the high current tube types mentioned, usually means that the resulting amplifier is going to have a fairly high output impedance when compared to a transformer-coupled amplifier. For this reason many OTLs often employ high amounts of global negative feedback, in order to allow the amplifier to operate more as a voltage source which is typically required in the operation of many loudspeakers (although this is not by any means a universal truth- more on that later).

There are two main types of OTLs in service today. The Futterman (named for Julius Futterman) circuit and its variants represents one approach; the primary identifier being that the output section uses a 'totem pole' approach, wherein the output circuit is push-pull, but one tube is driving the speaker with its plate connection, while the other uses its cathode. This approach means that the output impedance varies depending on whether the amp is pushing or pulling so to speak, and this usually requires a fair amount of feedback to linearize the circuit. The second type of circuit employs an output circuit known as a Circlotron, wherein there are two banks of power tubes balanced against two power supply banks of equal polarity in a bridge circuit. This approach has a lower output impedance than a totem pole and lower distortion as it is a true symmetrical push-pull circuit (and is more expensive to build due to the need for two separate power supplies).

Both circuits have special driver circuit requirements as usually the grids of such power tubes (and bear in mind, usually there are a number of power tubes in parallel to increase current capacity of the output circuit) have a considerable amount of capacitance and in the case of the Futterman, also have unequal drive requirements (to deal with the fact that the output section is asymmetrical).


If a similar amount of feedback is used in these circuits, the distortion will be similar or lower than is seen in most traditional transistor circuits. Bandwidth will also be considerably wider; its no problem at all to have full power bandwidth from as little as 1 Hz to well over 500KHz. Usually bandwidth has to be limited to prevent problems with RFI and oscillation, not unlike a solid state amplifier. So in this regard, OTLs are quite successful!

More in succeeding posts...
 

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#15
IMO/IME, there really isn't much benefit to running the filaments when the amp is not in use, unless you are doing so to retain some of the warmed up character of the the (since the driver circuit is fully active if the filaments are lit). If you are looking for longest tube life, just warm the amp up at least a minute before applying B+.
 

bonzo75

Member Sponsor
Feb 26, 2014
8,849
44
48
London
#16
IMO/IME, there really isn't much benefit to running the filaments when the amp is not in use, unless you are doing so to retain some of the warmed up character of the the (since the driver circuit is fully active if the filaments are lit). If you are looking for longest tube life, just warm the amp up at least a minute before applying B+.
Do you have any customers running your amp on Martin logan CLX?
 

Atmasphere

[Industry Expert]
May 4, 2010
806
0
16
St. Paul, MN
www.atma-sphere.com
#17
Yes. They are using an outboard device called the ZERO (see www.zeroimpedance.com).
ML speakers are built with a low impedance in an attempt to make them more compatible with solid state amps. The problem is that ESLs in general are incompatible with solid state unless a bit of current feedback is employed in the amp. This is because the impedance curve of an ESL is not also a map of its efficiency, unlike a box speaker.
 

sbo6

Member
May 19, 2014
693
0
16
Round Rock, TX
#18
^^ The thing that OTLs bring to the table is they are the most transparent amplifiers made. That's it in a nutshell. Quite literally, there are no tube amps and there are no transistor amps that express greater transparency. if sound quality is your goal.
Thanks for the additional information. Can you explain how you verified your statement above is fact?
 

Ron Resnick

Site Co-Owner, Administrator
Jan 25, 2015
5,051
29
48
Beverly Hills, CA
#19
Sorry to interject but Ralph is a titan in the high-end audio industry, and we are grateful for his participation on WBF.

Please consider it verified. I promise you that every OTL amplifier I have ever heard manifests a crystalline transparency which I have not heard from any other amplifier topology.
 

JackD201

[WBF Founding Member]
Apr 21, 2010
10,989
8
38
Manila, Philippines
#20
Yes. They are using an outboard device called the ZERO (see www.zeroimpedance.com).
ML speakers are built with a low impedance in an attempt to make them more compatible with solid state amps. The problem is that ESLs in general are incompatible with solid state unless a bit of current feedback is employed in the amp. This is because the impedance curve of an ESL is not also a map of its efficiency, unlike a box speaker.
Hi Ralph

What is the difference between an OPT and having an Autoformer in the path? Were the Circlotron Electro Voice amps "OTLs" with Autoformers built in? I am currently test driving a Circlotron amp that has OPTs. The first thing it reminded me of was your amps driving Tonians at a friend's place which had the somewhat unique ability of delivering sound events "fully formed" at very, very low sound levels. In other words that "comes to life" thing comes in much earlier than usual. You could listen at much lower SPLs without sacrificing intelligibility. I suppose it is because the distortion is lower but also because power delivery is very, very linear. Very little haze.

Thank you in advance.

Jack
 

About us

  • Founded in 2010 What's Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals we enjoy learning about new things and meeting new people and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing