OTL amps, my pick for the tube used.

@bobheinatz, here is what I can say at the moment. The design is past the prototype stage and a few pilot amps will be built for listening tests.These will be monoblocks with moderate power and so far have been tested on a pair of Quad 63 with great results. I am going to have the prototypes in my system shortly on my Quad 57s. Ideally these will work quite well with 16 and 8 ohm box speakers and can include an external autoformer if you have speakers with lower or more difficult loads.

I can say while Roger studied the H3aa circuit as part of his research, this OTL will have more in common with the earlier Futterman designs which were simpler and in the process Roger was able to further reduce the parts count in half. A Pentode/Triode switch will also be available.

IIRC Roger mentioned the Futterman amplifiers have 7 adjustments with two pages of instructions to set up the amp. Roger simplified this by installing a meter with a 6 position switch so you can view the current in each tube and easily see if one is becoming weak or running too hot. There is also just one bias adjustment as the tubes are supplied computer matched. The only other adjustment is the bias for the driver stage and that is done on the same meter.

The 633C initially became popular for its unusual shape and very low cost. Even in small quantities they could be had for $13 including socket. Very unusual to sell a tube with a socket. They actually recommenced replacing the socket with each tube renewal!

Now I see M Mathews is offering them for $142 so this is either new production or NOS. I was told that there were warehouses everywhere with this tube stacked to the ceiling. Don't worry, never will run out, I was told.

This tube is very close to our 6336 made only by Tungsol and equals 3 x 6AS7 in parallel. It has other features and I recommend take the time to read the 3 paragraph description. The number is strangely similar.

https://frank.pocnet.net/sheets/127/6/6336A.pdf

I recently worked on a VK-75 amp which uses 4 of the 6C33. When turning the amp over to work in it the tubes keept falling out of their sockets. Replacing the sockets in most 6C33 amps I have seen is not always easy... but it should be. 

On to the tube I use, the 26DQ5 has all the attributes one can desire in a OTL output tube. Those are

• High Current per tube
• High overload current. RCA advertised 400% for a few seconds.
• Low screen voltage, ability to run in Triode
• Octal socket holds the tube firmly. Much better than wirepins.

The thing that has always concerned me is the cathode flaking I see with Pass Tubes.That tells me they are not up for high peak currents like Sweep tubes are. I have to be very careful while curve tracing these Pass Tubes so as not to flake the cathode. It lowers the peak current and can cause quite a bias shift.

When you get right down to it Sweep Tubes are the only choice for making OTL amplifiers. 

They actually recommenced replacing the socket with each tube renewal!

The original Russian spec sheet called for 750 hours! In practice I've seen the tube outlast the socket.

Now I see M Mathews is offering them for $142 so this is either new production or NOS. I was told that there were warehouses everywhere with this tube stacked to the ceiling. Don't worry, never will run out, I was told.

I got this same story- but was also told by Mr. Mathews that the only supply there would ever be was NOS. So I'm not buying the story that New Sensor controls the tooling, but if that actually turns out to be the case (we've not seen any 6C33s made past the 1990s) it would be a simple matter to set the production up to make the tube for the socket used by the 7241- then the sockets would last decades instead of 100s of hours.

The thing that has always concerned me is the cathode flaking I see with Pass Tubes.

The 6C33 is a pass regulator tube but does not seem to have this problem. IME, the key to making **any** power tube last is 'pre-conditioning'. This is the practice of lighting up the filament circuit for an extended period (often several days) prior to applying B+. Each power tube type has a certain minimum amount of pre-conditioning time that is optimal for the type. Once properly pre-conditioned, the life of the tube can be doubled, and premature failure due to arcing (caused by cathode coating failure) is dramatically reduced.
In the construction of an OTL, there are several factors that should be considered in the choice of power tube, and they are fairly practical. The cost of the tube vs power output is one factor- closely related is availability (which should be considered in any design FWIW). Reliability is important as well as things like plate resistance (which will relate to how many tubes are needed to do the job) as well as the filament current the final tube bank is going to need to make it run.

Availability is a major concern because power tubes fail. While we can get 10,000 hour service lives out of our tubes, knowing that the tube will still be there down the road has been important, since our amps are pretty reliable. So since the inception of Atma-Sphere we've had an internal design rule called the 20-Year Rule, which simply states that since our gear is built to run about 20 years without service (other than tubes)- after which time the filter capacitors may need replacement, its a Good Idea to design the amp for tubes that are available and in production. Since we've now been in business 43 years, this rule has really worked well for us- we've seen competition that didn't take this idea into account and consequently had to close down. In the twilight of the tube era (which ironically has lasted longer than its heyday) it reasonable to expect that the diversity of tubes will dwindle so we've always been conservative about tube types. Its helpful when you have more than one source for tubes!

The ultimate OTL tube in many regards is the 7241 but it's expensive and rare- not a good choice. The 6C33 is a lot cheaper and with slightly over 1/2 the power output capacity of the 7241, still a good choice, but its prodigious filament requirements (and resulting heat production) causing it to eat sockets makes it less reliable.

Sweep tubes could be an excellent choice, hampered only by availability as they are decades long out of production- except for the PL509 and its variants. They have good filament requirements though and are otherwise pretty reliable, although linearity isn't great- but that can be improved by wiring them in triode. 

The 6AS7G is the tube we chose as most on this thread know. But rather than use the American types which indeed proved to have fragile cathode coatings, we used the Russian variant, the 6N13C (marked '6H13C' or sometimes '6AS7G'). This tube has reasonable filament requirements, is easy on the sockets, is inexpensive and over the 43 years we've been in business, proven pretty reliable- such that we have always had a 1-year warranty on the power tubes. Linearity is also good- it is a triode after all- and it has a fairly low plate resistance. Since we were uninterested in building an amplifier with a lot of feedback, linearity in the output section was/is pretty important to our design, as keeping distortion low without feedback is a bit of a trick.

I prefer my OTLs of the non-cyclotron, low waste heat, non-autoformer variety.

So do I. A cyclotron OTL would be a thing to behold though if it didn't make you blind :)