Using an existing PSU I built in a different project, parts I have now and a spare chassis, I did a quick test of DHT headphone power amp with AC filament - one tube per channel, parafeed, no hum pot, just using a dedicated center tapped filament transformer for each tube. All through a HD6xx headphone. The goal is just to see how bad the hum is.
45 tube has just a tiny bit of hum when there is no music playing - definitely something I can live happily with.
46 tube (triode connection) has a bit more hum - borderline on what I can no longer accept.
Both 45/46 tubes were with filament transformers of decent quality (and potted).
3D6 tube in triode mode: more hum than 46. This is with Hammond 266 series transformers.
The B+ supply circuit is the same (140-150v range), and very well filtered for all three output tubes.
45 was the very pleasant surprise. 3D6 needs more work apparently.
45 tube has just a tiny bit of hum when there is no music playing - definitely something I can live happily with.
46 tube (triode connection) has a bit more hum - borderline on what I can no longer accept.
Both 45/46 tubes were with filament transformers of decent quality (and potted).
3D6 tube in triode mode: more hum than 46. This is with Hammond 266 series transformers.
The B+ supply circuit is the same (140-150v range), and very well filtered for all three output tubes.
45 was the very pleasant surprise. 3D6 needs more work apparently.
Interesting. Did you use 1.4 V or 2.8 V for the 3D6 filament? 2.8 V with filament CT grounded is the hum-canceling connection. 1.4 V is not.
I had no hum issue with my push-pull 3D6 circuit.
I had no hum issue with my push-pull 3D6 circuit.
This is really interesting circuit. Thanks for posting sser2
A few questions:
A few questions:
- If I use DC on the filaments, can I connect 1 DC supply to both PP tubes in a channel at the same time? Taking it further, can I connect all 4 tubes with 1 DC supply? In other words, most optimally, do I need 1, 2, or 4 isolated filament supplies? I'm still really confused about how the signal is related to the directly heated element, and if there is mixing etc.
- For the center tapped transformer, bias point, can I connect both channels to the same point? (The minus 3.2-4V). I have -24V on hand. I'm also wondering if I can create this point from a voltage divider reference, or if it has to be low impedance.
- All said and done, what is the gain of this circuit? (Taking into account each tubes mu, the transformers, etc)
- How well do the tubes have to be matched? Distortion no doubt related to matching, no?
Thank you for your interest, Archan.
1. My original schematic shows 3D6 filaments tied together and powered from the same DC source, a NiMH battery. The battery can be replaced with a DC supply, but design of such supply is a non-trivial task, especially for a headphone amplifier. I used a 6 ah NiMH cell from a Honda Insight battery. A power tool cell or A-type NiMH cell can also be used, but Panasonic-made Honda cell is the best, it has very low internal resistance of about 30 mOhms. All 4 tubes can be fed from the same NiMH cell. The battery effectively shorts the filaments and grounds both filment ends.
2. For center-tapped AC, all 4 tubes can be fed from the same filament winding, if that is your question. Bias can be from from an external negative source through voltage divider, but resistance from grid to ground should not exceed 50k, as this is fixed bias. Battery bias is the best because it has very low grid to ground resistance.
3. Gain is about 5-6. This is more than enough with a standard 2 vrms source. Overall gain can be tweaked by the ratio of the input transformer.
4. As the circuit works in Class A, precise tube matching is not necessary. Better matching will give the benefit of better cancelation of the 2nd harmonic. DC currents of PP tubes should be well-matched for good low frequency response. With either battery bias or external source bias, means should be provided for DC balancing.
1. My original schematic shows 3D6 filaments tied together and powered from the same DC source, a NiMH battery. The battery can be replaced with a DC supply, but design of such supply is a non-trivial task, especially for a headphone amplifier. I used a 6 ah NiMH cell from a Honda Insight battery. A power tool cell or A-type NiMH cell can also be used, but Panasonic-made Honda cell is the best, it has very low internal resistance of about 30 mOhms. All 4 tubes can be fed from the same NiMH cell. The battery effectively shorts the filaments and grounds both filment ends.
2. For center-tapped AC, all 4 tubes can be fed from the same filament winding, if that is your question. Bias can be from from an external negative source through voltage divider, but resistance from grid to ground should not exceed 50k, as this is fixed bias. Battery bias is the best because it has very low grid to ground resistance.
3. Gain is about 5-6. This is more than enough with a standard 2 vrms source. Overall gain can be tweaked by the ratio of the input transformer.
4. As the circuit works in Class A, precise tube matching is not necessary. Better matching will give the benefit of better cancelation of the 2nd harmonic. DC currents of PP tubes should be well-matched for good low frequency response. With either battery bias or external source bias, means should be provided for DC balancing.
Speaking of filament battery, a simple circuit can be added that trickle-charges the battery when the power is off. NiMH can be trickle-charged indefinitely, it doesn't hurt the battery. I see no disadvantages in filament battery, only good things.
AC-DC supply's #1 problem is SS rectification noise, which is especially important with such low voltage. Then there are issues of supply's output impedance. Tying all filaments together will require low output impedance, but voltage regulator behavior WRT audio signal across the filament is problematic, and big capacitor across filament doesn't sound good. With high impedance supply a la Rod Coleman, a separate supply for each filament is necessary.
AC-DC supply's #1 problem is SS rectification noise, which is especially important with such low voltage. Then there are issues of supply's output impedance. Tying all filaments together will require low output impedance, but voltage regulator behavior WRT audio signal across the filament is problematic, and big capacitor across filament doesn't sound good. With high impedance supply a la Rod Coleman, a separate supply for each filament is necessary.
For the 2P29L preamp I built, I used Vicor DC DC converters and ripple attenuators, and had no issues with noise on the filaments. I'm a big fan of SMPS with tubes, although I know I'm in the minority. That preamp sounds terrific, so I'm tempted to explore that route further, and possibly refine the approach.
I agree though, the battery is an attractive way to go too. I never used on an audio circuit, but I have trickle charged an NiMH using a 10m45s. I had a timed fast charge mode also that could be used, by changing the resistor on the CCS chip that worked quite well. I used to design interactive exhibits for museum exhibits, and had some elements that were battery driven, but haven't yet used a battery in an audio circuit yet.
I agree though, the battery is an attractive way to go too. I never used on an audio circuit, but I have trickle charged an NiMH using a 10m45s. I had a timed fast charge mode also that could be used, by changing the resistor on the CCS chip that worked quite well. I used to design interactive exhibits for museum exhibits, and had some elements that were battery driven, but haven't yet used a battery in an audio circuit yet.
I bought a case of 3D6’s about 15 years ago. A quick look at the specs made me interested in them although I don’t remember why.
I may have to revisit them given all the great stuff above.
Peace,.
Tech
I may have to revisit them given all the great stuff above.
Peace,.
Tech
I just purchased a few 3D6's on ebay, and curious to test them out on some simple circuits.
This might be a silly idea, to use a PP transformer output circuit like this, for a line stage, but I am thinking about creating a gain stage to go into a First Watt M2, which is solid state buffer basically. According to this thread, input impedance of that circuit is 24K if I recall correctly.
https://www.diyaudio.com/community/...in-class-a-b-and-maybe-a-power-whammy.390636/
If I were to pursue such an idea, what kind of transformer would I look for? 15K:15K PP interstage? Does an interstage contain enough inductance that I don't need the additional 60H choke? Cruising for some vintage transformers, the L is not often stated.
This might be a silly idea, to use a PP transformer output circuit like this, for a line stage, but I am thinking about creating a gain stage to go into a First Watt M2, which is solid state buffer basically. According to this thread, input impedance of that circuit is 24K if I recall correctly.
https://www.diyaudio.com/community/...in-class-a-b-and-maybe-a-power-whammy.390636/
If I were to pursue such an idea, what kind of transformer would I look for? 15K:15K PP interstage? Does an interstage contain enough inductance that I don't need the additional 60H choke? Cruising for some vintage transformers, the L is not often stated.
PP topology for a line stage is a common approach. Such line stage (or an amplification stage in an integrated amplifier) is called differential stage. All op-amps use differential topology. Tube implementation of this approach was used in Lynn Olson's Aurora amplifier. Lynn used transformer-coupled differential stages as voltage amplifier and driver.
3D6 is a good candidate for such use, along with other DH tubes such as 2P29L, 1J29B, and 1P24B. All have amplification factors around 8-10.
A line stage output transformer is something that old timers called "two plates to line". Typically it is 30K CT: 600. UTC made such transformers in the A, HA, and LS series, need to look up for exact models.
The purpose of the 60H choke is not to compensate for insufficient inductance of the output transformer. This choke acts similar to CCS: it enforces strict Class A operation and symmetry of the differential stage.
3D6 is a good candidate for such use, along with other DH tubes such as 2P29L, 1J29B, and 1P24B. All have amplification factors around 8-10.
A line stage output transformer is something that old timers called "two plates to line". Typically it is 30K CT: 600. UTC made such transformers in the A, HA, and LS series, need to look up for exact models.
The purpose of the 60H choke is not to compensate for insufficient inductance of the output transformer. This choke acts similar to CCS: it enforces strict Class A operation and symmetry of the differential stage.
Thanks for your response sser2. Still a lot to learn about transformers, hence my interest in this project. I don't like to build anything that doesn't have at least some mystery and terrain to learn/explore 🙂
The choke confused me because I don't typically see that on a normal PP output tube stage.
Here is the link to the Aurora you mentioned. It looks like he used a Lundahl 1635, but not sure if that would work in my application connecting to a 24K input impedance SE SS stage.
At some point a PP line stage ceases to make sense, when the step down ratio defeats the additional gain from the second tube, and since low output impedance for headphones is not of concern.
The choke confused me because I don't typically see that on a normal PP output tube stage.
Here is the link to the Aurora you mentioned. It looks like he used a Lundahl 1635, but not sure if that would work in my application connecting to a 24K input impedance SE SS stage.
At some point a PP line stage ceases to make sense, when the step down ratio defeats the additional gain from the second tube, and since low output impedance for headphones is not of concern.
I agree. The step-down plate-to line transformer is to reduce the output impedance, with gain sacrificed.
Seems like the UTC A-18 or Tamura A-351 would work for line stage?
- The UTC secondary can be put in parallel for a 20K output, correct?
- And can the Tamura reversed so that the 40K is the primary and 10K secondary? Is 40K too high for this tube? I have 190V on hand for B+
The A-18 is an interstage transformer. High output impedance is unacceptable for a stand-alone line stage.
The line stage would be going into an M2 First Watt power stage. So it's not to be a universal type line stage. More as a gain stage within an integrated amp.
M2 stage is 24K input impedance:
https://www.diyaudio.com/community/...in-class-a-b-and-maybe-a-power-whammy.390636/
Would it be acceptable considering this criteria? The question on that one is can the secondary windings be put in parallel, and we'd have 20K out, no?
What about the Tamura? Is 40K primary too high?
M2 stage is 24K input impedance:
https://www.diyaudio.com/community/...in-class-a-b-and-maybe-a-power-whammy.390636/
Would it be acceptable considering this criteria? The question on that one is can the secondary windings be put in parallel, and we'd have 20K out, no?
What about the Tamura? Is 40K primary too high?
If everything is on a single chassis, that's integrated amplifier. Interconnect of any considerable length is the problem with high output impedance.