Maybe you should. I am very opening minded (for an engineer) when it comes to the arts which I believe our hobby falls into, ultimately this is about perception, and enjoy the "design" phase but lothe the building phase of the hobby, love the listening phase.
Any way I am still saying that if you have headphones that are 98+ db efficient there isn't much need for a DHT. A high gm triode or pentode connected shunt regulated Spud SET is as close as one can get to "wire with gain." Not only subjective listening but measurement as well, at normal listening levels a WE437 (or WE417A/6C45Pi) has no measureable distortion above the noise floor, also with led bias there are literally no caps in the signal current loop.
Several of us have built different incarnations of the Spud headphone amp. Some follow Ciuffoli series feed approach, others take the Lyn Olsen parafeed approach. So this divide is very similar to the banter we are having regarding transformers.
But unless one needs more than 100 mW, I humbly don't see the need to go into DHT's, we are adding more distortion (see Ale's 6e5P measurements), more noise, more caps in the signal path, more stages. I highly recommend that someone interested in a tube headphone amp take the spud approach, I think Pano's report of excessive hum with his 4P1l-4P1l bear this out.
Building a DHT headphone amp is needed for phones that need big voltage and are inefficient (planar magnetics, old AKG.)
The DHT design is more difficult than a preamp design or an amp design for speakers, because we need a black background and good linearity. This is why if I know I need about a 1/3 a watt I want to design for 3 watts.
Getting a headphone amp to have a dark black silent background is essential, almost all DHT SET amps have hum when you put your ear to the speaker. This is the worst thing to deal with from a headphone amp.
Now powering these 307A's or 300B's, I am going to be pushing the SSHV2 to its limits. I am considering the madia regulator for the output, I've had good luck with these in the past, they are very low noise and require less heat sinking than the shunts.
But as far as DHTness, if one just wants DHTness for efficient phones I would stick with the schematic on the first page (112/26 to 71A/4P1L), it just doesn't afford any extra power than a spud 6E5p.)
I know we spent a lot of time on transformers but I think we learned a valuable insight. A quadruple 8 ohm secondary properly interleaved is the way to go, we get all the output impedances we need: 8 ohms for our speakers, 32 ohms for our grados/denons/orthos/modern akgs, and we get 128 ohm for the senns , beyers, and older AKGs. We get this without asking for a transformer with ridiculous secondary DCR/inductance.
The conflict was a bit painful but we can all spec the custom transformer we need from our favorite supplier and now we can move on to designing the power supply which may also need a custom transformer
Last edited:
Hi Regal,
There are all sorts of pure design reasons for using spud amps just as you say. It may be that with certain trade-offs around extra stages and components it could even sound "similar but different".
What you will come up against in the debate about DHT or indirectly heated is that some people just flat out PREFER the sound of DHTs and believe that to their ears they sound different in quite distinct ways - more dynamic, more transparent, some kind of added magic. These people, and I'm one, hear a subtle veiling of the sound with indirectly heated tubes that just doesn't go away however you use them.
So yes, it all sounds very logical. In practise, however, some of us are simply not going to use anything other than DHTs. I haven't used an indirectly heated tube for 2 years, for instance. I'd have to hear something equivalent to a 26 or 4P1L or 2a3 to want to use it, and so far I haven't, and I've been through a mountain of tubes in my time.
Andy
There are all sorts of pure design reasons for using spud amps just as you say. It may be that with certain trade-offs around extra stages and components it could even sound "similar but different".
What you will come up against in the debate about DHT or indirectly heated is that some people just flat out PREFER the sound of DHTs and believe that to their ears they sound different in quite distinct ways - more dynamic, more transparent, some kind of added magic. These people, and I'm one, hear a subtle veiling of the sound with indirectly heated tubes that just doesn't go away however you use them.
So yes, it all sounds very logical. In practise, however, some of us are simply not going to use anything other than DHTs. I haven't used an indirectly heated tube for 2 years, for instance. I'd have to hear something equivalent to a 26 or 4P1L or 2a3 to want to use it, and so far I haven't, and I've been through a mountain of tubes in my time.
Andy
Thankyou, I am open to the idea of a dht driver, I wish I could find an 841 but they are not available so I think DHT for the driver becomes a long term goal. Short term I am starting with a high gm IDHT triode. I figure if I was going to build a 3 stage amp all DHT amp its going to be an 845/GM-70.
However, the 4p1l has enough gain with a stepup input transformer and a 1:2 step-up IT transformer, it has a low enough Rp to drive a 1:2 IT. Are there issues with magnifying the microphonics of the 4P1l by using a step-up input transformer and then a step=up IT on its output ?
Of course if one had a nice OTL 26 preamp this would not be an issue.
The other option that makes a lot of sense is using the 4P1L driver in pentode mode, possibly a switch to move from triode to pentode depending one whether one is wearing power hungry phones? But again in pentodemode the microphonics are multiplied by 5 in the input/driver position.
If I do use the 4p1L I plan on hiding it and suspending these eastern block socket/shield inside the chassis, I figure if these worked for the Russians they will work for me. My direction is a bit different than yours due to the fact I can't lift much over 20 lbs! Plus I don't have the craftsman ship to do the beautiful top plates like yourself, but man your work is inspiring!
Attachments
I think Ale found that pentode mode sounded inferior to triode mode, so no free lunch there.
I don't know the full story here, since I haven't built a headphone amp. Don't know how many stages are needed.
I wouldn't call what I do craftsmanship!!! Nice of you to say so, though. I've enjoyed metalwork since I learned it in school, and I'm happy as a sandboy launching into chunks of aluminium with my various drills, files and saws. I have little sense of beauty in design, though my son's a designer and I love modern art. I just lay out stuff in a functional way so I can hear all my various projects, and a modular system with several top plates works well for me. No need to breadboard. I just join up the plates. I do get a lot of joy out of my system. I really like it as it is now - 26>4P1L>PSE 4P1L, all in filament bias.
Andy
Well the pentode was used I believe with a humming amp for high efficiency phones (nooffense to plano but not a good baseline.)
Adding the #26 up front makes some since probably adds negligible distortion. This is all IT coupled? I have a slew of #26's that I've never used. Your work with phono=pre's tells me you have a lot of experience with low noise, your approach (lots of iron) I'm sure could make a great headphone amp, but just too much weight for me.
The other thing I am trying to explain is for high power planar headphones that need a lot of voltage swing (40-50 ohms) PSE 4P1L's don't really gain much over a single, if you are interested in planar headphones you can trace a few load-lines and see what I mean. The high B+ and + high grid voltage of a 300B makes it the best choice. But reports are that the 307A can run 20W Pd have me going that direction first (just cause I am a cheap-stake when it comes to tubes.) I recently read a report that Sovtek 300B'sgo into short circuit over 400V due to an exposed grid to plate flaw. Supposedly the JJ 300B's are better at higher B+, even the JJ 2A3-40 may be an good choice.
All DHT parafeed is also a nice option for high efficiency headphones, like Doug's L'esspressivo++. Sennheiser specifies the HD800 as "102 db" I assume that is 102 db/mw. You'd only need about 0.8v to be playing at 105 db, doable with his 71A DHT design. Off the shelf autoformers on the output with 2 db steps are available from Intact Audio.
They would offer a THD at 1W with planar headphones comparable to what we are accustomed to with the spud amps (THd < -70 dB), the issue is supplying a noise /ripple free B+ at such a high voltage. But this is what I understand Mr Cooter built for his planar headphones. I say if you have the ambition go for it would certainly be the lowest distortion SET headphone amp possible.
As far as sticking to all DHT, looks like my DAC should be able to drive a balanced to SE 1+1:1+1 input transformer, these aren't expensive hammond, cinemag both offer nice input transformers. That gains a 4V rms input from a DAC output. It would max out a 307A or PX25 with a single 4p1L. Also a 4p1l has a low enough rp to drive a 1:2 interstage transformer to allow recordings that are 6db low if needed or just 1:1 if not. This would give plenty of power for a nice two stage all DHT amp cap free headphone amp. Best output tube choices are 300B, PX25,307A, PSE 4P1l (if you run them real hard), or 2A3. I think I can squeeze 2 watts out of the 307A so that's where I am starting with a dedicated SSHV2 per output tube and fixed bias via IT, The SSHV2 should have no issue powering a single 307A per channel. The 4p1l driver would not be SSHV supplied since it would have a CCS or Gyrator.
Things are really falling into place with the design, lot of options for different builds.
Honestly still have thoughts of parafeed, but I appreciate all the input so far on what is really the first new look at an all DHT headphone amp in almost a decade. Cant wait to finish this DAC and get back to tubes.
I worked up a spreadsheet for the widest spectrum of headphones I could think of to help builders pick their output tube. Note that I used actual sensitivity measurements rather than believing the manufacturer's marketing" specs ( this is the mistake RS made with the O2.)
I think a 5k primary with 4 separate 8 ohm secondary's and Rod is right, a 300B covers things pretty well, as long as the primary inductance is high enough and the tube Rp is low enough as to make the load insignificant in the bass roll-off calculation. with some headphones like planars and Grados the damping factor is of little importance since they have a flat impedance curve and current delivery is another "myth", headphones are all about voltage. There are some phones with tougher impedance curves but these can be handled with all secondaries in parallel. The interesting thing is how inflated some of the brands sensitivitys spec's are, general trend is that the max power spec is what you want to shoot for.
Hope this is helpful, it works out nicely. I think the data shows that a powerful DHT can be an advantage even for sensitive phones since it's plate is barely swinging and making negligible distortion. But it also can show that a 4p1l output is a good choice for those who have no interest in planars or old AKG's.
I think a 5k primary with 4 separate 8 ohm secondary's and Rod is right, a 300B covers things pretty well, as long as the primary inductance is high enough and the tube Rp is low enough as to make the load insignificant in the bass roll-off calculation. with some headphones like planars and Grados the damping factor is of little importance since they have a flat impedance curve and current delivery is another "myth", headphones are all about voltage. There are some phones with tougher impedance curves but these can be handled with all secondaries in parallel. The interesting thing is how inflated some of the brands sensitivitys spec's are, general trend is that the max power spec is what you want to shoot for.
Hope this is helpful, it works out nicely. I think the data shows that a powerful DHT can be an advantage even for sensitive phones since it's plate is barely swinging and making negligible distortion. But it also can show that a 4p1l output is a good choice for those who have no interest in planars or old AKG's.
This is for dynamics, not an average listening level, its called headroom. I think speaker guys have about the same target's. Many target 115, but this I don't think is needed for an SET.
Better explanation here:NwAvGuy: More Power?
Better explanation here:NwAvGuy: More Power?
That is a clearly written and empirically supported piece of writing meant to advance a subjective viewpoint of how much loudness an amplifier/headphone system should be able to attain. It is based upon using a corner condition in the sub-genre of audiophile recordings of classical music. I can't say that I would use this as a design goal for my purposes but I now have a better idea of what have been the unspoken assumptions that you've been working under and I respect that as we all have different goals.
Yea I have a large collection of raw (un mastered/uncompressed music) with huge dynamics compared to commercial stuff. With the spreadsheet one can change the headroom goal to meet your indicidual goals (that's the nice thing about aspreadsheet), just trying to help people understand the relationhip between what is asked of the output tube/transformer and actual headphone sensitivity/impedances, in no way is the spreadsheet "static", its a tool.
All,
Firstly many thanks to Vegard for sending the 307a sample for tracing. I had very little time today for this and hopefully can do further testing next week. Initially in my view 307a is better to be used with Schade feedback and is not that great in triode mode. Also created a SPICE model which haven't tested so would be great if someone can play with it and report any issues.
307a DHT in triode and Schade feedback | Bartola Valves
Hope this helps!
Cheers,
Ale
Firstly many thanks to Vegard for sending the 307a sample for tracing. I had very little time today for this and hopefully can do further testing next week. Initially in my view 307a is better to be used with Schade feedback and is not that great in triode mode. Also created a SPICE model which haven't tested so would be great if someone can play with it and report any issues.
307a DHT in triode and Schade feedback | Bartola Valves
Hope this helps!
Cheers,
Ale
Thankyou, this rules this one out, I guess this is much worse linearity than even a modern eastern 300B.
Just an update, since Ale ruled out my 307a's I found a pair of RCA 2A3's for $80. Now the DAC is done (ecept designing a case) and I can focus more of the DHT head-amp. I am looking forward to an amp with more power than my little WE417 can provide.
I settled on a transformer its 5k/8+8+8+8. This is a great combination, very versatile and efficient, it has a newer type C core but I am going with this mainly for the expertise in winding (interleave, low DCR, etc.)
On the front will be a cinemag input transformer to give a 4Vrms from the balanced DAC outputs.
Simple single input/driver : a Phillips E182CC. which is basically a super select 5687. Just a single CCS common cathode for each channel from each section. I like this tube because it has a high enough Cg to use the full 4Vrms input. Then coupling capacitor, fixed bias for the 2A3. I just like oil caps so the pragmatisst can shoot me for ordering AmpOhm copper in oil's, they have worked nicely for me in the past.
Power supply is going to be a SSHV2 per channel feeding only the output tubes. I am over sizing the transformer (Edcore) so if I want to upgrade to the 40W 2A3's (basically 2.5V 300b's) it will be easier. Of course coleman regs for the output tubes. The nice thing about the SSHV I you can have a high prefilter B+ and stimy down the voltage with a big resistor without affecting the sound (the first stage of the SSHV is a CCS.)
It will be a 2 chassis build with umbilical cord. I am looking for a chassis that is big enough that I can switch to 841 DHT's should a pair ever find my way. I strongly considered a pair of the 4p1l's with a 1:2 IT but 4 SSHV's and the heat sinking required would be very difficult.
The nice thing about the E182CC is it is a linear tube that really simplifies the build, gain is 24 (much higher than a 5687 or 6n6p), but has a distortion profile that matches the DHT output well enough for good harmonic cancellation which I find to be the real reason for needing a DHT driver when I was number crutching. The new 841 model Ale posted simulates amazing, but these tubes are impossible to find right now.
That's it in a nutshell, pretty simple except the coleman and SSHV regulators and the output transformers, but I hope others are building and if you do share your experience. I would be happy to post schematics but its basic enough I think anyone can figure it out. Key is low noise and flexibility.
I settled on a transformer its 5k/8+8+8+8. This is a great combination, very versatile and efficient, it has a newer type C core but I am going with this mainly for the expertise in winding (interleave, low DCR, etc.)
On the front will be a cinemag input transformer to give a 4Vrms from the balanced DAC outputs.
Simple single input/driver : a Phillips E182CC. which is basically a super select 5687. Just a single CCS common cathode for each channel from each section. I like this tube because it has a high enough Cg to use the full 4Vrms input. Then coupling capacitor, fixed bias for the 2A3. I just like oil caps so the pragmatisst can shoot me for ordering AmpOhm copper in oil's, they have worked nicely for me in the past.
Power supply is going to be a SSHV2 per channel feeding only the output tubes. I am over sizing the transformer (Edcore) so if I want to upgrade to the 40W 2A3's (basically 2.5V 300b's) it will be easier. Of course coleman regs for the output tubes. The nice thing about the SSHV I you can have a high prefilter B+ and stimy down the voltage with a big resistor without affecting the sound (the first stage of the SSHV is a CCS.)
It will be a 2 chassis build with umbilical cord. I am looking for a chassis that is big enough that I can switch to 841 DHT's should a pair ever find my way. I strongly considered a pair of the 4p1l's with a 1:2 IT but 4 SSHV's and the heat sinking required would be very difficult.
The nice thing about the E182CC is it is a linear tube that really simplifies the build, gain is 24 (much higher than a 5687 or 6n6p), but has a distortion profile that matches the DHT output well enough for good harmonic cancellation which I find to be the real reason for needing a DHT driver when I was number crutching. The new 841 model Ale posted simulates amazing, but these tubes are impossible to find right now.
That's it in a nutshell, pretty simple except the coleman and SSHV regulators and the output transformers, but I hope others are building and if you do share your experience. I would be happy to post schematics but its basic enough I think anyone can figure it out. Key is low noise and flexibility.
Regal,
I played recently with Rod's regulators and they are wonderful. Eas to build and to adjust. Just make sure you have a decent size heatsink. I used a flatback standard finned Al heatsink size 100x75x40mm for a regulator that feeds 1.3A and it stays warm to touch. When I used a heatsink 1/2 of the size above it got very hot. Also it is important to make sure you design the raw power supply to fit your needs. I used a 75w 2x9V at 4A to feed two of Rod's regulators on 4.2V/1.3A filaments.
Hope this helps.
I played recently with Rod's regulators and they are wonderful. Eas to build and to adjust. Just make sure you have a decent size heatsink. I used a flatback standard finned Al heatsink size 100x75x40mm for a regulator that feeds 1.3A and it stays warm to touch. When I used a heatsink 1/2 of the size above it got very hot. Also it is important to make sure you design the raw power supply to fit your needs. I used a 75w 2x9V at 4A to feed two of Rod's regulators on 4.2V/1.3A filaments.
Hope this helps.
Thanks for the tip, I'm going to try a common mode choke for the raw supply along with the capacitors , unfortunately it will have to be in the power chassis and share umbilical cord. Did you use a separate raw supply for each tube ? Hopefully can find a domestic dual bobbin transfo supplier like he uses in England.