i think the issue there would be current hogging, unless you can force the parallel tube to share currents equally..
If the amplifier employs a separate power transformer for the B+ of the power tubes, all that is really needed is a fuse in the transformer's primary. Having been building OTLs of the Circlotron persuasion for the last 50 years I can tell you this is quite effective in the case of a tube failure for protecting loudspeakers! The only exception is if the speaker in question is a lower powered unit unable to handle the power of the amp, in which case a series fuse is a good idea on account of accidents with sources or the volume control of the preamp.
Long time I wasn't here, now read all missed comments, WOW Gentlemen, great conversations. I love this community, so great to see people intelligently discussing such valuable topics after a longer Facebook session.. ahh.. a relief. Thanks for being/remaining humans 
Now to the OTL and especially my route:
1. Finished my OTL headphone amp (6N1P - 6N6P). Needed to convert AC heating to DC and fully eliminate residual hum. Dead silent now, uh. That was a hard work. First noise was gone, then hum too. Fantastic sound..
2. All parts for Tim's OTL arrived, finally.
A double matched set of 6N2P-ER, 6J32P and the 6S33S-V-s.
3. KoonW proposed 1 or 2 pages before, the full elimination of input stage. Now that I have the input tubes, I'm considering the same) Why to build it at all when I already asked for a sensitivity-decreasing mod, and now it's obvious for me: why do I need input stage at all ? With DACs of 2Vrms (RCA) / 4Vrms (XLR) output, I could land directly on the EF86-s. Question is, what to do with the feedback and how much the original behavior (and sonic parameters) of the original design can be kept, apart from taking out the full input stage and those triodes' characteristics. I really don't need high sensitivity at today's DAC voltages, actually, the PRE-stages are now all implemented in DAC-s, why to attenuate and amplify again, right ?
So I think, in the original design the input is due to historical reasons there, for that time's Tape and CD player output.
So, where to begin (eliminating the input stage) and what would be the approx. input sensitivity when going directly onto the EF86-s ? Has anyone done this already ?
4. Chassis is a tiny little bit crowded and uh.. super heavy. This isn't a tank, this is a black hole, I really have concerns if the 3mm aluminum chassis can hold the 3 big trafos for a longer time.
I have a dedicated transformer for heating the output tubes, 2x 12V. Looking at the schematics, one is at GND potential, the other is at -150V which is still okay with regards to insulation.. because I thought I might move this heater transformer into a separate chassis and wire the 2 together via thick twisted cable(s). The heaters are microcontroller started anyway and the transformer itself could exist in a separate aluminum chassis with heatsinks.. (45°C at non-summer room temp). Any considerations, suggestions regarding this solution ? Maybe an additional thick wire to connect the two chassis' grounds ?
It's only a consideration yet, I'll probably try all 3 transformers in the existing chassis.
Now to the OTL and especially my route:
1. Finished my OTL headphone amp (6N1P - 6N6P). Needed to convert AC heating to DC and fully eliminate residual hum. Dead silent now, uh. That was a hard work. First noise was gone, then hum too. Fantastic sound..
2. All parts for Tim's OTL arrived, finally.
A double matched set of 6N2P-ER, 6J32P and the 6S33S-V-s.
3. KoonW proposed 1 or 2 pages before, the full elimination of input stage. Now that I have the input tubes, I'm considering the same
) Why to build it at all when I already asked for a sensitivity-decreasing mod, and now it's obvious for me: why do I need input stage at all ? With DACs of 2Vrms (RCA) / 4Vrms (XLR) output, I could land directly on the EF86-s. Question is, what to do with the feedback and how much the original behavior (and sonic parameters) of the original design can be kept, apart from taking out the full input stage and those triodes' characteristics. I really don't need high sensitivity at today's DAC voltages, actually, the PRE-stages are now all implemented in DAC-s, why to attenuate and amplify again, right ?So I think, in the original design the input is due to historical reasons there, for that time's Tape and CD player output.
So, where to begin (eliminating the input stage) and what would be the approx. input sensitivity when going directly onto the EF86-s ? Has anyone done this already ?
4. Chassis is a tiny little bit crowded and uh.. super heavy. This isn't a tank, this is a black hole, I really have concerns if the 3mm aluminum chassis can hold the 3 big trafos for a longer time.
I have a dedicated transformer for heating the output tubes, 2x 12V. Looking at the schematics, one is at GND potential, the other is at -150V which is still okay with regards to insulation.. because I thought I might move this heater transformer into a separate chassis and wire the 2 together via thick twisted cable(s). The heaters are microcontroller started anyway and the transformer itself could exist in a separate aluminum chassis with heatsinks.. (45°C at non-summer room temp). Any considerations, suggestions regarding this solution ? Maybe an additional thick wire to connect the two chassis' grounds ?
It's only a consideration yet, I'll probably try all 3 transformers in the existing chassis.
Build the output section temporarily and give it a test,that should tell you whether you need the IP section or not.
Re tfmr weight,you could always put an L strut under them to give extra support, like an acoustic guitar strut.
Re separate PSU, if you go down that route I have some big military connectors I think that may help there, one potential problem though, separate PSU's can lead to some issues with noise hum. The last amp I built used a separate PSU, I had a problem with noise I never entirely resolved when siting grounds/chassis ground/mains earth. Just something to bear in mind. Also it's a lot more wiring and fabrication.
If you do decide on a separate PSU and need a hefty connector drop me a PM, Andy.
Re tfmr weight,you could always put an L strut under them to give extra support, like an acoustic guitar strut.
Re separate PSU, if you go down that route I have some big military connectors I think that may help there, one potential problem though, separate PSU's can lead to some issues with noise hum. The last amp I built used a separate PSU, I had a problem with noise I never entirely resolved when siting grounds/chassis ground/mains earth. Just something to bear in mind. Also it's a lot more wiring and fabrication.
If you do decide on a separate PSU and need a hefty connector drop me a PM, Andy.
Hi Guys, seriously I wonder if there's anyone who's made a version of the Tim Mellow's OTL without the phase splitter and just rely on a differential XLR input that will provide the required "true" and "complement" signal needed by the push-pull output stages. Won't that be even lower distortion with one less stage given that these days modern high quality DACs can provide 2V to 4V peak to peak XLR outputs?
Also wonder if there are better tubes to use than the EF86 and ECC83/12AX7 that can change the distortion profile to have negligible higher order distortion compared to the 3rd or 2nd order(if it exists) and also allow a reduction in the global feedback used. It's okay if the output impedance rises to 0.8 Ohms or so, just have to find some loudspeakers with impedance ~8Ohms or more across all freqsuencies
Hi "nickng", I am the one trying to understand and implement the possibility of balanced input of Tim's OTL. Till now, there were only discutions on that, but as I have already built more than two monoblocks and use them for 6 years very happy, I am not sure it can be done.
The main problem I can see is repositioning of the feedback reaction as well as the input requirements on the two EF86. There is someone to explain how can be done theoretically as well as in practice?
By the way, I asked already Tim about it, and he was very gentle in answering a lot of questions, but about this matter, he only says it is possible but not how.
The main problem I can see is repositioning of the feedback reaction as well as the input requirements on the two EF86. There is someone to explain how can be done theoretically as well as in practice?
By the way, I asked already Tim about it, and he was very gentle in answering a lot of questions, but about this matter, he only says it is possible but not how.
If the differential design was robust, yes.
If you want to see an amplifier like that look at this thread https://www.diyaudio.com/community/threads/what-tubes-for-a-otl-tube-amp.161112/
If the output section is floating per the amp seen in the link above, feedback can be done with a pair of resistors from the output and mixed with the input signal in much the same way as is done in an opamp.
welcome back ...
if you drop the input stage, you loose the built-in DC servo function which keeps DC near zero at the speaker terminals;
instead you probably rely on a balance pot in the cathodes of the EF86s;
how often do you want to re-calibrate DC offset ?
remember, grids and cathodes of EF86s are at -330V, while speaker terminals are near zero;
which means that a simple DC feedback loop is not possible;
by the way, open loop differential gain from the grids of the EF86s to 8 ohm loaded output is about 24dB;
so if you needed only 8dB closed loop, you'd have some 16dB remaining for AC feedback;
applied before the coupling caps to the EF86 grids of course ...
A while ago i ordered PCB set and transformers for aresaudio :- Kit including 4 PCBs for the realization of the OTL amplifier with the 6C33C version 2 + 2 endings to build the OTL 4X. The filament transformer is rated at 500W and has 4 X 12v 9A outputs for the 6c33's. Would it be possable to use each output to drive two 6c33's like in the OTL 8X, as it is the same PSU, and from what i understand is the same transformer that is used in the OTL V4 or does the V4 have two of them ?
My turn soon. Got the parts, or 99% of them, and PCBs from Aresaudio in Italy.
Considering putting the transformers inside the cabinet - all 3 of them. I imagine toroids don't give off a lot of heat - am I wrong?
Also nervous about the grounding - but must determine layout of cabinet first, it's from Hammond, 2U tall, 19", 305mm deep.
The transformers are from Airlink in the UK, especially the 625VA one was "cheap". And enoromous.
I am also trying to understand the circuit and the DC points. No luck with that so far!
Considering putting the transformers inside the cabinet - all 3 of them. I imagine toroids don't give off a lot of heat - am I wrong?
Also nervous about the grounding - but must determine layout of cabinet first, it's from Hammond, 2U tall, 19", 305mm deep.
The transformers are from Airlink in the UK, especially the 625VA one was "cheap". And enoromous.
I am also trying to understand the circuit and the DC points. No luck with that so far!
What I have so far is a wooden chassis mock up. As I wanted to just see what / if I would work. I currently only have 4 6c33 tubes and sockets as my original intention was to use 2 tubes per chanel. I don't know if the tubes are any good. They are NOS and all same date code. So I have mounted the four sockets on a wooden chassis and they are wired as a 4 tube mono, as I can then just experiment with just 2 tubes and plug the other tubes in for more power. Then buy more tubes and sockets if all is good. Have you built the protection board yet? The pads for the relays have incorrect spacing! Ares said they new about this, and to bend the relay pins to fit, but then they don't sit nice, so I drilled the holes bigger and fit the relays all nice and flush, only to discover a load of o/c tracks due to me drilling the through plating out, so nw has jumper wires on it. Looks crap but works, will sort this out at a later date.
Thanks for replying! My first build will be the startup-board. I am pretty sure I managed to put a relay in - tried one.
I also got, by asking, a schematic which shows how to wire the startup-board. I have to ask Ares for everything, they send me one drawing at the time.. the PSU board they sent is for the 8-tube version, but it will work for 4 tubes, I just have to reverse-engineer at little bit.
Any experience with toroidals, do they run hot - can they be built in?
I also got, by asking, a schematic which shows how to wire the startup-board. I have to ask Ares for everything, they send me one drawing at the time.. the PSU board they sent is for the 8-tube version, but it will work for 4 tubes, I just have to reverse-engineer at little bit.
Any experience with toroidals, do they run hot - can they be built in?
I've not run the toroides for any amount of tie under load. I don't think they run as hot as ordeary txfmrs. I so far have built the protection pcb the power supply and one driver board, but have not had much time to work on it, hope to do some more towards end of this week. Hopefully might get it wired enough to get it going and if so will run it under load at weekend. I will let you know about transformer temperatures, mine came from Italy ares audio recommended them heater transformer has 4 x 12v@9Amp and two 6.3v and a 12_0_12 at a couple of amps. About 500watts total. 8 6c33 heaters is about 320watts but I only have 4 to test at moment. 160watts.
A tip, FWIW. You'll find the toroids to run cooler if you use stainless non-magnetic bolts to secure them (this is assume that they are mounted by a large washer and a single bolt through the middle). If you do this, do not use a stainless nut!! It will gar and become impossible to remove.
We found that regular bolts are a magnetic short inside the toroid (they aren't supposed to work that way, but life does not always agree with what we think its 'supposed' to do...) and so will run hotter than the toroid by quite a lot.