[Roger A. Modjeski]

Rather than develop one OTL I am developing a series of them to fit different needs. I would appreciate your input on what you would find most desirable. These will be sold direct and enjoy higher value for that.

After considering and experimenting with many topologies it appears the Futterman configurations work the best. That basic configuration was used by me in the Counterpoint SA-4 in 1984. I have simplified that design and made it more reliable.  Futterman,  Peterson and many others were working on similar stacked output tube configurations in the 1950's. These are excellent configurations and do not have the faults claimed by other makers of OTL amplifiers. Here is but one example of how Futterman's work is often mis-understood. 

http://www.stereophile.com/tubepoweramps/1197trans/index.html

OTL amplifiers have a great advantage in that they solve many of the problems of making a good output transformer, something I have spent years doing. Although our output transformers are better than most they still have the problem of re-combining the positive and negative phases of the output tubes equally at high frequencies. This makes load stability difficult, limits feedback, distortion and damping factor. In single ended amplifiers there is only one phase (hence single ended) somewhat reducing the problem and perhaps one of the things that people like about SE amps. However their distortion is large and although it is claimed that second harmonic distortion is benign it means there is large intermodulation distortion which is far from benign. Its IM distortion that muddies up the sound as volume is increased. We do make SE amplifiers and they sound great as long as used in the lower power range. This is well documented in reviews and measurements in Stereophile.

Now on to something complete different. Although not a new idea, there are many advantages in using a low ratio matching transformer with OTL amplifiers. I have mentioned this before in this forum and found most listeners find the idea acceptable. These types of transformers are made in an entirely different way, can be toroidal and have very wide bandwidth. The Zero transformer is one example. I will be providing a custom transformer for these amplifiers which will increase their power by up to 8 times the direct power. They are especially useful in the smaller amplifiers.

One member recently asked in this forum "Why doesn't anyone make small OTL amps?"... good question. It's all about the number and type of output tubes. Because output power is directly a function of current and impedance it makes the most sense to use a tube that provides the most current per tube. Since output power is determined by current squared doubling the number of tubes quadruples the power. Therefor its is tempting to use a lot of output tubes. The SA-4 and the NYAL OTL-3 both used 8 output tubes per channel to achieve 100 watts. However there are other ways to achieve more power with fewer tubes.

The smallest OTL that can be made uses 2 output tubes per channel and one driver. It produces very clean 3.2 watts per channel into 8 ohms. It has a mono switch which allows it to make 12 watts into 8 ohms, 4 times the power. This is because in mono the current doubles and thus the power quadruples. For many of us 3 clean watts are sufficient for 90 db listening levels.  For louder levels, an external 2 to 1 matching transformer can be employed. This is another way to double the current and there is plenty of voltage available. Then the little stereo amplifier would be 12 watts per channel in stereo and 48 watts in mono. These are all RMS continuous values. For music where peaks are usually the limiting factor one can approach 100 watts in mono with a 3 to 1 transformer. One advantage of these OTL amplifiers is there is plenty of voltage for peaks. When the amplifier clips it runs out of current rather than conventional tube or transistor amps which typically run out of voltage.

All too many of the recent OTL amps have been built on a PC board. This is a big mistake. Over time the solder connections between the PC board and socket pins deteriorate and are the largest cause of problems. In addition the heat from the tubes often slowly cooks the PC board making brown circles under the tubes, peeling traces and are almost impossible to repair. The OTL amps we are making are all hand wired. A small PC board for the solid state servo and DC protection circuit is employed but produces no heat.

Prices range from $1,900 for the small stereo/mono to $8,500 per pair for the 100 watt mono blocks with 8 output tubes per channel.

[steve f]

I've owned a couple of OTL amps of various design. I even remember Harvey pushing NYAL products wearing some kind of kimono at a SCES years ago. Generally speaking, I like them, except for the high output impedance, no negative feedback, circlotron types. Please tell us more about your designs.

[airhead]

A correction and a question.  First the correction:  NYAL OTL3's use 6 output tubes per channel, not 8 (although George Kaye offered an upgrade to 8 for a while).  I know this because I have a pair, which I've owned since the eighties.  I think the output is around 80 watts per channel, but I've never measured them of course.  I use them to drive some very inefficient but very high impedance speakers (Stax F83's).  They work well, except in the bass region; the speakers, oddly for an ESL, have low impedance in the bass.  I've fixed that problem with the aid of Roger's wonderful subwoofers and crossover.  The system as it is now sounds, well, fantastic.  The OTL 3's use a lot of feedback but run quite cool.  They are built on a PC board, which still looks okay to me.  Before the Futterman amps I tried the Counterpoint SA4; I have to say that the Futterman's work much better for me with my speakers than the SA-4 did. 

Now the question.  I recently read the article linked to below by Roger.  What is the misunderstanding in that article?  Is it Rozenblit's analysis of the circuit and his redesign?  Or elsewhere in the review?  I have never heard the T8's, but there are aspects of his design philosophy that seem appealing (low stress and cool operating range of the tubes, less fancy packaging…..).  I also wonder if Steve F can clarify more his experience with amps…..did he try the Atmasphere's for example, and not like them?  Which did he try?  The Atmasphere's tend to run very hot, I'm told.  I have a friend who has a pair he sometimes uses with his Soundlabs, but I haven't heard them. 

[steve f]

Airhead, I guess my hatred of typing shortened my answer way too much. I've owned three different OTL amps over the years and have spent time  with a few more. I still have one I use regularly. It's a Transcendent Sound SE-OTL built from a kit. 1.5 WPC of Class A single ended bliss. I wish it were a bit larger though. Hang on to those OTL 3's!

I'm not an amp guy. I've built a few kits over the years but never designed one. I am a speaker guy though. Many different types have been designed and built in my shop. I prefer high efficiency multi-way designs that are easy loads for most amps. I have built an ESL, planar, various boxes, and open baffle designs. Okay, that's me and where I come from.

A long time ago, I had as an extended loaner a used Futterman H3a. I guess it was built in the 60's. I had it for about three months in the late 70's, and it was a golden oldie even back then. I don't recall much about the amp except that it had a clarity that other amps lacked in my experience. Needless to say I was hooked. I have heard several NYAL amps at trade shows. Harvey was a real showman. One year at a SCES, he wasn't there, and people didn't hang around to listen because they came to see Harvey  perform.  I had a small used Atmasphere which was probably kit built by someone. I believe it ran about 20-30 WPC. I got rid of it because it didn't give me the sound the Futterman amp did. I played around with a couple of others, now long out of business.  I had a used Transcendent T-8 stereo amp for awhile. I couldn't keep tubes in it. The problem wasn't the amp; the tubes I had were prone to infant mortality, and failed often.  I almost gave up on OTL amps. Since I have speakers in the 95-100 DB sensitivity range, I tried the smallest TS OTL which uses a 6H30PI for a driver and four 6C19PI tubes for each channel. I had two of the amps run as mono-blocks but sold one as there wasn't any need for two with my speakers. I kept the other. If it had 10 WPC I would consider my amp buying days over.

I am of the opinion that history has been unkind to Julius Futterman. I am not aware of any amps burning up, oscillating, or any of the negatives attributed to his amps, or any other modern OTL amps either. I never had a problem, except for certain output tubes. I never saw the Counterpoint except as a static display.

Here's what I expect out of an OTL. Reliability. great sound, inexpensive or at least long lasting tubes please.  I know that to get Class A performance, only 3-4 WPC are possible. AB amps are fine. I don't want any auto-formers. I look forward to Roger's designs, as I'm sure he will raise the bar.

Steve

[airhead]

Dear Steve,

Thank you very much.  Indeed i intend to hang on to my OTL3's for as long as possible.  However I am afraid I do have add a bit of evidence to support their iffy reputation; one of mine did catch fire some years back.  Thanks to advice from George Kaye, I had it repaired and retuned, and it works fine now.  I think the other amp still has the same set of tubes it first came with.  I agree about the amplifier's clarity; nothing else I tried came close:  (Counterpoint SA4, Audio Research D250, Bruce Moore something or other…..).
The 6LF6's it uses are very hard to find now; for  a while Kaye was offering to change the circuit to use EL509's (used by Transcendent), but he  won't do that now
because he says the currently available EL509's are unsatisfactory.   

I guess I'm not really a hifi guy; I have barely changed anything in my system since the eighties, except for necessary replacements and upgrades. 

[Ericus Rex]

Awesome, Roger!

When can we see pics of these babies?

[Roger A. Modjeski]

Awesome, Roger!

When can we see pics of these babies?

I am exploring some unusual formats for the physical design. Some are entirely made of wood. I still have a few more ideas to explore in that area and will put up some pics when I have a few more of my ideas in physical form.

Sonicaly the smallest amplifier is done and produces about 5 very clean watts into a 8 ohm load and 10 watts into 16, 20 into 32 ohms.  An optional auto-transformer will allow users of any speaker impedance to get the higher powers. An auto-transformer is easier to make and has not the problems of a push-pull transformer. It can also be outside the feedback loop which is a godsend to stability.

As the circuit has evolved and been simplified I will say it is the shortest path amplifier that I have ever built and has the fastest response I have ever achieved. Because of this, the amplifier has no increase in distortion at either the low or high end of the spectrum which spans 1 Hz to 150 KHz. There is no way to do this with conventional tube amplifiers. The amplifier is internally DC coupled and other than the offset-nulling effect of the servo is a DC amplifier.

[steve f]

Wow!

[OzarkTom]

NYAL, lots of great memories and not so great. I sold several sets of OTL 3's back then and had trouble with every one. I sure wish it had been different.  Harvey was a very colorful man. One year at the CES show, Harvey had a huge stack of OTL 1's and a pair of Quad ESL63's. Everyone that heard the system immediately looked for subwoofers. The bass and dynamics were tremendous, one of the best at the show. Harvey was all smiles and having a field day.

The best memory was finding my longtime friend Rex. Rex saw my ad for NYAL's in the paper and came buy and bought a of 3's in 1982.  We have been friends ever since.

OTL's, my favorite amps. And these are to be made by Roger, one of the best in the business.

[Roger A. Modjeski]

I recall seeing pictures of stacked chassis at NYAL exhibits. I believe the amplifier circuits were pretty much the same with the extra boxes being power supply additions. The biggest problem with OTL amps of that day were that they were made on PC boards which are not a good place to support power tubes. All the OTL amps of the day used wire pin tubes whose sockets did not perform well over time and temperature. I have chosen octal socket tubes and furthermore mounted them on the chassis. Reliability has been a big issue with OTL amps and I am paying much attention to the reliability of mine.

[PET-240]

Have both the Transcendant Beasts and Atmasphere M60 clones in build for a comparo, goody, just when I had things settled.....

[Roger A. Modjeski]

Have both the Transcendant Beasts and Atmasphere M60 clones in build for a comparo, goody, just when I had things settled.....

Are they up and running? Any observations so far?

[Duke]

Sonicaly the smallest amplifier is done and produces about 5 very clean watts into a 8 ohm load and 10 watts into 16, 20 into 32 ohms.

Correct me if I'm wrong, but I think those numbers imply a fairly high output impedance.   That looks like current source behavior to me, so I'm guessing very little global negative feedback is used.   Not asking you to divulge, just saying, "hmmmmmm!"

By any chance, have you experimented with speakers that present a 32 ohm load? 

What would you consider the "optimum" load to be (never mind whether anyone is building such a speaker)?  Again if you'd rather not say, no problem. 

[JohnR]

Duke, consider what would happen if an amplifier's maximum output power was limited by current capability, not by voltage swing. (This is independent of output impedance and doesn't make the amp a current source.)

[Duke]

Duke, consider what would happen if an amplifier's maximum output power was limited by current capability, not by voltage swing. (This is independent of output impedance and doesn't make the amp a current source.)

That's the sort of behavior Roger's describing, I think.  Maximum current stays constant (is the limiting factor), and voltage goes up as the load impedance goes up.  This is the opposite of voltage-source (or "constant voltage") amplifier behavior, which we see in solid state amps.

I would have called that "current source" behavior, or "constant-current" behavior, but maybe my wording is wrong.

I'm under the impression that the degree to which an amplifier exhibits current-source or voltage-source behavior is related to the ratio of load impedance to source impedance (speaker impedance to amplifier output impedance), but I may be wrong, having only a little knowledge of the subject, which is of course a most dangerous thing. 

[PET-240]

One article I found enlightening was by Ralph Karsten from AtmaSphere, will see if I can find a link, was regarding the output measurement paradigm, it may be on his website, looking forward to these though!

Chuz,

Drew

[JohnR]

That's the sort of behavior Roger's describing, I think.  Maximum current stays constant (is the limiting factor), and voltage goes up as the load impedance goes up.  This is the opposite of voltage-source (or "constant voltage") amplifier behavior, which we see in solid state amps.

I would have called that "current source" behavior, or "constant-current" behavior, but maybe my wording is wrong.

Hi Duke, I think you are conflating the maximum output with the transfer function...

[PET-240]

Here you go,http://www.atma-sphere.com/Resources/Paradigms_in_Amplifier_Design.php
Not trying to take focus off Roger and his thread, so apologies if it offends, not intentional.

Chuz,

Drew.

[Duke]

Hi Duke, I think you are conflating the maximum output with the transfer function...

Let me reword then:  Current stays constant, and voltage goes up as the load impedance goes up.  This is the opposite of voltage-source (or "constant voltage") amplifier behavior, which we see in solid state amps.

Maybe I'm missing something?

PET-240, I'm familiar with Ralph's paper.  Am I misunderstanding it?  From his description, it sounds to me like Roger's amp is beyond even "power paradigm", and more like "current paradigm", which Ralph doesn't mention, but his S-30 exhibits such behavior to a certain extent. 

[JohnR]

Let me reword then:  Current stays constant, and voltage goes up as the load impedance goes up.  This is the opposite of voltage-source (or "constant voltage") amplifier behavior, which we see in solid state amps.

Maybe I'm missing something?

Let me reword too You are conflating "maximum power output vs load impedance but independent of input voltage" with  "power output vs load impedance given a constant input voltage (and while in a linear operating region)".

To put it another way, "constant current" and "current limited" are not the same thing. You can have a current-limited voltage source.

For argument's sake, suppose you have a current source that delivers 5A out for 1 V input. Into 4 ohms, you get 100 W. Double the load impedance to 8 ohms, current stays the same, you get 200 W. (Provided the voltage limit is not reached). That is what you are thinking of.

Now suppose you have a voltage source that produces 20V from 1 V input. Into a 4 ohm load, you have 100W. Into an 8 ohm load, the current halves, and you have 50 W output. Hopefully we agree so far

Now suppose that this voltage source was limited to producing 5 A output. And let's say it's maximum output voltage (provided current limit is not reached) was 100V. So with 1 V input, with the 4 ohm load, we are at maximum power output - 100W. When we double the load impedance to 8 ohms, however, we have halved the current output. So we are no longer at maximum output. We also have to double the input voltage to 2V to get back to maximum output. So, with 2V input, we have 5A and 40V on the output. 200W.

So, with a current limited voltage source, the maximum power output increases as impedance increases. (Up to a point, of course, until the voltage limit is reached.) That does not make it a current source.

Now, I don't know the specifics of Roger's amp but I have run into this with another OTL. (Which I only half built due to the issue of power output suddenly become much less than expected when the real current capability of the output tubes I bought was discovered.) I was simply trying to point out that equating the increasing maximum power output with high impedance or current-source behavior is not correct. FWIW Roger did say in the OP "When the amplifier clips it runs out of current rather than conventional tube or transistor amps which typically run out of voltage." and "This is because in mono the current doubles and thus the power quadruples."

 

J