A Tribute to Julius Futterman

Woe is Futterman

I have found Julius Futterman to be greatly misunderstood by those who write about his circuit designs. I have researched Futterman circuits extensively and the vast majority of the time I encounter an explanation it is absolutely incorrect. In 1954 Julius Futterman published a concise explanation and instructions on how to build his 8 watt all triode amplifier (Journal of the Audio Engineering Society Vol. 2, #4). This 5-page article is quite easy to find, and I recommend it to anyone who would like to understand the finer points of the design. This circuit is painfully simple, and the explanation could not be clearer. I assume those who do not understand his designs have only viewed the later schematics of the H3-AA which is quite complex. There are a few pundits who have the tools to gaze upon a schematic and see the essence of it, and I will tell you my interpretation shortly, but first I need to debunk a couple myths.

It is often said that Futterman amplifiers could only be serviced by Julius himself. Well my very first encounter with a Futterman amplifier was in 1974 when an unequal pair of H3s appeared on my shop bench along with a schematic and two pages of adjustment instructions. I believe the instructions were typed by Julius himself and they were quite detailed. There were 7 adjustments to be made in a definite order. In addition, there was a test socket to facilitate the measurements which only required a multimeter. The procedure was not a problem for me and in my opinion any clear-headed technician should not have had any trouble interpreting the instructions. Suffice it to say I was able to complete the required work on the amplifiers. So much for Myth #1.

Bootstrapping

Now I will now let you in on what 99% of those who write about the Futterman OTL circuit do not know and in the process debunk Myth #2. It is commonly stated that the drive to the upper and lower output tubes is unequal. This is simply not true and Julius in all his writings explains this. When I was 14 and making transistor amplifiers for fun I learned about bootstrapping. Only those who fooled with early transistor amplifiers would know this, but Julius knew as well. Those not in the know incorrectly referred to this as positive feedback. However, the reader would have no idea what positive feedback was and frankly its use is misleading in this case.

Bootstrapping is the practice of coupling, usually through a simple capacitor, the output of the amplifier to a pair of resistors in the driver. In Futterman’s OTL circuit this allows the output voltage, whatever it may be, to be added to the driver voltage and thus make the drive to the upper tube equal to the lower. Those who do not see the bootstrap ascertain that the drive is unequal and go to great lengths to attempt to make it close to equal all the while fussing that Julius got it wrong. Well, Julius got it right and it is referenced quite clearly in his published writings. However, those who wrote incorrectly about the concept most likely did not know of these writings or ignored them. Instead, they look at the famous H3-AA schematic and miss it entirely.

OTLs, RMS, and Music Power

In 1982, during the early years of Music Reference, there was a need for a power amplifier. I am the kind of guy who wants to design and make his own output transformers but as of yet did not know how. Therefore, I designed the ultimate DC coupled OTL based on Futterman’s work but greatly expanded. By the time I was done and had built the prototypes I thought this is no amplifier for me to produce so I sold the design to Counterpoint and it became the SA-4. It used 8 x 6LF6s with a servo and no output capacitor. It won several awards but not many were made. It was expensive for the time at $7500 a pair.

My next encounter was with Julius himself. I was visiting Harvey Rosenberg and company at NYAL on another matter and thought I would give Julius a try. I knocked on his door and this slight gentleman asked me to return in an hour as he was about to have his sack lunch. I spent about 3 hours with Julius, saw the amplifiers, the pressure cooker, the transformer winder, all of it. Many years have passed since that visit and during that time I learned how to make excellent output and power transformers for the RM-9 and RM-10, but now and then would make an OTL just for fun…. and then put it away.  

In 1959 Julius Futterman wrote an article detailing a 38 watt OTL circuit that was published in Electronics World (Electronics World, May 1959: https://www.americanradiohistory.com/Archive-Electronics-World/50s/1959/Electronics-World-1959-05.pdf) and my latest OTL design takes its cue from this design and not the H3-AA that most people know. Harvard Electronics and Tech Instruments, both of New York, built very similar H3-AAs. The NYAL OTL series is also of the same circuit. It appears to me that when sweep tubes (horizontal output tubes for CRT televisions) became more powerful in the 25-inch color sets Julius moved to them. First with the 6FW5 which has no plate cap as seen in the earliest Harvard amplifiers. In that application 6 tubes were used per channel. Later the chassis was re-punched for 4 pieces of the various wire pin tubes with plate caps that were more powerful so 4 could do the work of 6. Many of these amplifiers were modified for 4 tubes leaving 2 sockets empty. The rest of the amplifier is unchanged though the screen regulator tubes do vary. 

Now, here is where Julius and I depart. RMS power, a most meaningless measure, became the rule at this time. Julius got caught up in the hype and as a result his circuit design became more complex. In addition, more RMS power meant the amplifier ran the tubes harder and reliability suffered. Let me be clear on this, we do not listen to RMS power and it is only because of the abuse of power specifications in the 1960s that the FTC stepped in and laid down the law.

Had I been on the FTC committee I would have pushed for Music Power because we listen to music. Music power is simply short-term RMS power. Music does not drag the power supply down like continuous RMS does. If we get anywhere close to continuous RMS power we are clipping the amplifier excessively. However, Julius gave in, boosted the power, and got 90 watts RMS into a 16 ohm load, but only for a few seconds. You do not test the RMS power of an OTL amplifier and leave for lunch. You measure it fast as you can. Better yet you measure it with tone bursts which are an accurate representation of music.

Tubes we use and tubes we do not…

The Futterman OTL circuit uses a particular type of tube called a sweep tube. It is capable of 1 amp of current per tube and we use 3 to pull up and 3 to pull down. This gives us an output current of 3 amps into any load. Since power is current squared times the load this means the output power roughly doubles as the load doubles, and 16 ohms is an optimum load for such an amplifier though it does quite well into 8 ohms. At 4 ohms most of the energy is going into the tubes and not the load so this is not recommended.

[Now allow me to digress a bit before I get back to tubes. Some years back I designed and made a very limited run of the Music Reference OTL-1 that had a built-in matching autoformer for low impedance loads. An autotransformer is quite a different animal than a push-pull output transformer as it has no split primary and no switching transients. The autoformer allows the amplifier to preserve all its wonder when used with low impedance speakers. The power is markedly increased as the amplifier sees its ideal 16 ohm load all the time.]

As the years passed the 6LF6 went out of production but other tubes appeared. Through some luck I found that one of my tube suppliers had gotten stuck with several thousand 26DQ5 octal based sweep tubes, perfect for an OTL. A little here and a little there and eventually I bought up the entire stock with the hope of someday making a simple, reliable, and easy to operate OTL. I am proud to say that with my latest design this is what I have done.

There is another type of OTL circuit I must touch upon here with some trepidation. I have extensive experience with the Circlotron OTL which is fraught with problems. Where the Futterman amplifiers have very good damping (in excess of 40) the Circlotron has an extremely low damping factor. Low damping results in the speaker’s frequency response and bass control being severely modified and the listener does not get to hear the speaker as the designer intended. Instead they get to hear something else entirely, which they may like but is far from correct.

The other problem with most Circlotron OTL amplifiers is that they use the wrong type of output tube. Power tubes are designed quite differently for different applications. The sweep family of tubes was designed for very high peak currents and the cathodes hold up with no sparking even under high current conditions. Circlotron OTL amplifiers use an entirely different tube known as a series pass tube that was designed for regulated power supplies. These have quite low current capabilities and under heavy current one can see sparks jumping off the cathode as cathode material is expended. This loss of cathode material reduces the power output of the amplifier and can result in shorted tubes. For comparison, the 26DQ5 tube is capable of 1 amp of current where the 6AS7 used in a Circlotron circuit is rated for only 1/4 amp. It obviously takes more 6AS7 tubes to do the job and not do it as well. 

The Circuit

As I mentioned earlier Futterman’s 1959 design is the inspiration for my latest OTL amplifier, but there are differences. To start, it has less than half the parts and is switchable from Pentode to Triode mode (again, Julius was in a power race, so this feature likely did not even occur to him). As mentioned previously the Futterman H3 circuit had 7 adjustments, two pages of instructions, and required a sensible person very familiar with a meter. This does not describe today’s audiophile who mistakes “bias” with “idle current”. When we adjust an amplifier, we are always measuring idle current while adjusting bias.

[I digress again, but I posted a thread on Audiogon and 200 people could not make this distinction!]

If you test 100 tubes for bias to achieve a set idle current you will find the result makes a bell curve. The range of this curve is such that tubes from opposite ends will produce idle currents that vary almost 10 to 1. For this circuit the tubes are run at 100 mA for a dissipation of 16 watts on a 24-watt tube. The grid voltage range for that is 35-45 volts, therefore randomly selected tubes could easily run as little as 10 mA and as much as 120 mA. Julius had this problem also and with all due respect I have to wonder if he even thought to address the issue as there is no provision for measuring the idle current of each tube. To solve this problem, a built-in meter is included with a 6-position switch that allows one to check the current in each tube. In this way one can find a weak tube or a tube that is running too hot. There is thus one bias adjustment as the tubes are supplied matched. The other important adjustment is the bias for the driver stage. This is quite critical and therefore is available on the same meter to check from time to time. Now we have an amplifier with a built-in meter and only 2 adjustments where the Futterman OTL circuit had 7 adjustments, no meter, and two pages of somewhat cryptic instructions.

Finally, I am pleased to announce that we have passed the pre-production stages of development with this new design and are now building production models. So, I present this yet to be named OTL as a tribute to Julius Futterman, and perhaps more importantly, my way of giving the audio community a simple, reliable, and easy to operate OTL that in Music Reference tradition they can enjoy for many years to come.

[Source: circa 2018]