I have ortho headphones that I really like but they are very inefficient. I have run number after number, simulation after simulation. And it just seems not feasible to really deliver the power they need.
The phones are 38 ohms and recommended 2W of power.
So they need ~200 mA of current.
That's over 8V rms across the drivers, to put this in perspective for an 8 ohm driver that's almost 9 watts of power.
I want to design an SET for these phones but it just seems like I am missing something.
It would help a little having a larger secondary, but the resistive loss/output impedance can be a killer.
Where would you start for such a requirement ? I almost think I need a very high voltage tube like a Gm-70 that allows a broad voltage swing?
It is really a challenging design criteria.
The phones are 38 ohms and recommended 2W of power.
So they need ~200 mA of current.
That's over 8V rms across the drivers, to put this in perspective for an 8 ohm driver that's almost 9 watts of power.
I want to design an SET for these phones but it just seems like I am missing something.
It would help a little having a larger secondary, but the resistive loss/output impedance can be a killer.
Where would you start for such a requirement ? I almost think I need a very high voltage tube like a Gm-70 that allows a broad voltage swing?
It is really a challenging design criteria.
Last edited:
As DF96 mentioned, have a transformer built - Edcor can probably do this - with a 38-ohm secondary. Use a 2A3/6B4G in SE mode, or some other tube that delivers the power that you need.
Another suggestion - use a hybrid approach with a tube front-end and a solid-state buffer that can supply the current.
Another suggestion - use a hybrid approach with a tube front-end and a solid-state buffer that can supply the current.
Hi,
The standard for headphones is a 120R source impedance, and you really need
to ascertain if they are designed for this or will work well with much lower
source impedance - I simply don't know the answer to this question.
I used to have some Yamaha Ortho's. Connected them directly to the
output of my amplifier with a 100R wirewound pot in the earth return.
rgds, sreten.
The standard for headphones is a 120R source impedance, and you really need
to ascertain if they are designed for this or will work well with much lower
source impedance - I simply don't know the answer to this question.
I used to have some Yamaha Ortho's. Connected them directly to the
output of my amplifier with a 100R wirewound pot in the earth return.
rgds, sreten.
Edcor will do a great custom job but there will be a one-time design charge of an extra $50 for a non-standard design (no extra charge for subsequent orders of same item).
To add more power without adding any more stages and remain single-ended class "A" you could go to parallel output tubes. Unfortunately, that gives you current and not voltage, so to enjoy that extra power you really need a lower input impedance at the transformer primary. Just going to parallel outputs would make your original output transformer mismatch twice as bad.
However, if you're willing to have Edcor make the transformer or find a suitable one, the fidelity should be really good. Many problems with output transformer design are related to the radical turns ratio. Your high-impedance headphones requiement would make the turns ratio requirement less extreme, and the parallel output tubes would also make the turns ratio requirement less extreme. The result should sound really nice.
To add more power without adding any more stages and remain single-ended class "A" you could go to parallel output tubes. Unfortunately, that gives you current and not voltage, so to enjoy that extra power you really need a lower input impedance at the transformer primary. Just going to parallel outputs would make your original output transformer mismatch twice as bad.
However, if you're willing to have Edcor make the transformer or find a suitable one, the fidelity should be really good. Many problems with output transformer design are related to the radical turns ratio. Your high-impedance headphones requiement would make the turns ratio requirement less extreme, and the parallel output tubes would also make the turns ratio requirement less extreme. The result should sound really nice.
Actually, the lower turns ratio is quite beneficial for your application! Transformers with lower impedance ratios behave closer to ideal.
You may also be able to use a transformer that is off the shelf for that. A 2000R to 8 R is the same turns ratio as a 8000R to 32 R (which would be perfect for an EL84 or 6V6)
Sounds like HE-5's maybe?
If so... this is really correct. HE-5's are specified as 38 ohms and 87.5dBSPL/1mW. If my cyphering is correct 2W will get you 120dBSPL, the "typical" threshold of pain. So if you want loud and/or good headroom, 2W is not unreasonable.
HOWEVER... you also need to consider the source impedance. Your headphones may very well not sound very good if you drive them with a high source impedance. Then again, they might sound great - I do not know these phones well enough to render an opinion. But the driving source impedance will have a significant (maybe even dramatic) effect on the way they sound.
BTW, the old standard of 120 ohm source impedance has been abandoned... most modern high-end headphones (especially the planars) sound terrible with that high of a Zout. Just as with modern multi-way speakers, it's easiest to design the headphones if you assume an ideal voltage source.
Keeping the source impedance low will mean either using a very low Rp tube or a higher ratio transformer. Say you use a 2A3, with an Rp of ~800 ohms and a normal OPT with 2.5k primary impedance. If you use a 40 ohm secondary your source impedance will be about 12 ohms. I think this will be just right for these headphones, but I am not 100% sure.
You can experiment using a solid-state amp with near zero impedance, and put different resistors in series with the outputs to see how this changes the sound.
BTW, I posted some info and a spreadsheet that helps you calculate what power is really needed to drive headphones: Headphone & amp specs
Pete
If so... this is really correct. HE-5's are specified as 38 ohms and 87.5dBSPL/1mW. If my cyphering is correct 2W will get you 120dBSPL, the "typical" threshold of pain. So if you want loud and/or good headroom, 2W is not unreasonable.
HOWEVER... you also need to consider the source impedance. Your headphones may very well not sound very good if you drive them with a high source impedance. Then again, they might sound great - I do not know these phones well enough to render an opinion. But the driving source impedance will have a significant (maybe even dramatic) effect on the way they sound.
BTW, the old standard of 120 ohm source impedance has been abandoned... most modern high-end headphones (especially the planars) sound terrible with that high of a Zout. Just as with modern multi-way speakers, it's easiest to design the headphones if you assume an ideal voltage source.
Keeping the source impedance low will mean either using a very low Rp tube or a higher ratio transformer. Say you use a 2A3, with an Rp of ~800 ohms and a normal OPT with 2.5k primary impedance. If you use a 40 ohm secondary your source impedance will be about 12 ohms. I think this will be just right for these headphones, but I am not 100% sure.
You can experiment using a solid-state amp with near zero impedance, and put different resistors in series with the outputs to see how this changes the sound.
BTW, I posted some info and a spreadsheet that helps you calculate what power is really needed to drive headphones: Headphone & amp specs
Pete
I am working currently on a project of the amp for such headphones. The real problem is, either to find source of such very high quality custom made transformers, or to use topology that allows to use much cheaper mass - produced transformers, otherwise it will be too expensive. Requirements to the amp that drives headphones in terms of sound quality are much higher than to amps that drive speakers. It is not enough to take some tube amp and drive headphones. High sensitivity 2W speakers mask by own distortions flaws of amps that will be well heard through 2W headphones.
Chris has a point though. A higher-quality over-sized off-the-shelf transformer might handle the mismatch and sound as good as a smaller custom transformer with perfectly optimized match.
What tube (and what corresponding transformer source impedance) were you thinking of? Keep in mind I'm no expert to be giving advice without confirmation. But that won't stop me. I'm just in love with parallel SE.
In the Edcor line, I would try the GXSE10-16-1.7K which is the 10-watt version with 1.7K input / 16 output and it's only $16. For that cheap, what have you got to lose?
Then there's their 15-watt GXSE15-16-1.7K with 1.7K input / 16 output that will still get pretty low bass.
Or for a lot more money there's the wide-bandwidth CX line with the 25-watt Edcor CXSE25-16-1.25K (which is a little lower) transformer for single-ended with 1.25K ohms in / 16 ohms out. Drive that with the right paralleled single-ended tubes and you'll have plenty of power even at 40 ohms. Lots more transformer than you need, probably plenty of bass despite any mismatch, that could serve as a nice headphone amp or a nice speaker amp; you might even have enough power to optimize it for speakers and burn off a bit with a resistor across the phones for a better match and damping factor.
But I don't have the experience, know the tubes you prefer, nor did I do the math, so I defer to other opinions.
What tube (and what corresponding transformer source impedance) were you thinking of? Keep in mind I'm no expert to be giving advice without confirmation. But that won't stop me. I'm just in love with parallel SE.
In the Edcor line, I would try the GXSE10-16-1.7K which is the 10-watt version with 1.7K input / 16 output and it's only $16. For that cheap, what have you got to lose?
Then there's their 15-watt GXSE15-16-1.7K with 1.7K input / 16 output that will still get pretty low bass.
Or for a lot more money there's the wide-bandwidth CX line with the 25-watt Edcor CXSE25-16-1.25K (which is a little lower) transformer for single-ended with 1.25K ohms in / 16 ohms out. Drive that with the right paralleled single-ended tubes and you'll have plenty of power even at 40 ohms. Lots more transformer than you need, probably plenty of bass despite any mismatch, that could serve as a nice headphone amp or a nice speaker amp; you might even have enough power to optimize it for speakers and burn off a bit with a resistor across the phones for a better match and damping factor.
But I don't have the experience, know the tubes you prefer, nor did I do the math, so I defer to other opinions.
And Chris makes the best point: go someplace and try those phones on some SET with 16 ohm outputs. Maybe you don't need them that loud. Maybe they have good bass. All the math and all the guessing isn't as good as a listening session. If it's loud enough but lacking bass control, try them with a 120 ohm resistor across each channel of the headphone, making a 30 ohm load on each side, and see how loud they get. The match would be better for the 16 ohm tap, and the loss of 1/4 of the power isn't much on the decibel scale.
Last edited:
Mr Millett is correct on the make , they're HE-5LE's (orthos) and yes they need an amp to be able to swing a full 8VRMS across the drivers to handle transients. (I have a tube mofset hybrid doing the job now and have taken measurement.)
Wattage isn't the issue.
You have to think of this in terms of Voltage.
Someone touched on the ideal solution. I have simulated this and run calculations to death.
To take advantage of the lower turns ratio means using a low Ra tube, at least a 2A3 or equivalent so that the primary can be low.
Sure you can get 2 watts from a 5k:32 45 SET on paper, but the DCR of the secondary will be so high as you never really reach 8Vrms across the voice-coil.
It is a puzzle that seems simple till you really dig into it.
Also think about how much voltage is being swung across the primary to give 8Vrms across the secondary, it means a tube with a high operating voltage.
I would love to use parallel 4P1L's but their operating point and Ra just doesn't allow the voltage headroom on the loadlines.
I believe the 6C4C is the best candidate for a budget tube based on about a million simulations but its even at its limits, a GU-50 may be better.
With headphones you want linearity and a quiet power supply cause you hear everything.
The easy answer is a 300B, but then building a quiet power supply becomes a trick. I've got colemen filament regs so not worried about that just B+.
For those interested in tubes it is an interesting puzzle.
But I think the key is a good 2k:32 OPT, going much higher than that on the primary and 2watts into 38 ohms really isn't as easy as it appears first glance.
I can post my simulations over holiday shutdown if folks are interested. These modern ortho headphones really sound amazing, its a worthwhile group project if there are other headphone freaks out there who want to drive modern orthos with a classic SET.
Wattage isn't the issue.
You have to think of this in terms of Voltage.
Someone touched on the ideal solution. I have simulated this and run calculations to death.
To take advantage of the lower turns ratio means using a low Ra tube, at least a 2A3 or equivalent so that the primary can be low.
Sure you can get 2 watts from a 5k:32 45 SET on paper, but the DCR of the secondary will be so high as you never really reach 8Vrms across the voice-coil.
It is a puzzle that seems simple till you really dig into it.
Also think about how much voltage is being swung across the primary to give 8Vrms across the secondary, it means a tube with a high operating voltage.
I would love to use parallel 4P1L's but their operating point and Ra just doesn't allow the voltage headroom on the loadlines.
I believe the 6C4C is the best candidate for a budget tube based on about a million simulations but its even at its limits, a GU-50 may be better.
With headphones you want linearity and a quiet power supply cause you hear everything.
The easy answer is a 300B, but then building a quiet power supply becomes a trick. I've got colemen filament regs so not worried about that just B+.
For those interested in tubes it is an interesting puzzle.
But I think the key is a good 2k:32 OPT, going much higher than that on the primary and 2watts into 38 ohms really isn't as easy as it appears first glance.
I can post my simulations over holiday shutdown if folks are interested. These modern ortho headphones really sound amazing, its a worthwhile group project if there are other headphone freaks out there who want to drive modern orthos with a classic SET.
Last edited:
A few years ago Electra-Print wound me a set of SE output transformers with a 5K primary and pair of 8 ohm secondaries which could be wired in parallel for 8 ohm loads and in series for 32 ohm loads.. All of my headphones are 32 ohm and work great with this 700mW amplifier.
A 45 running (250V/35 - 40mA) into 5K will get you 2W..
A 45 running (250V/35 - 40mA) into 5K will get you 2W..
You can definitely get more than 8V from a 45. The ratings on the datasheet are quite conservative considering its 12W plate dissipation. But if you don't think it's enough, you can use a 300B. Electraprint can make you a custom 3k:32 transformer. Probably a speaker transformer into 8 ohms would be enough, but I'm not familiar with these headphones. At full power, it would certainly melt any headphone I know, including AKG K1000.
Getting a 300B quiet is not hard. I use it for my headphones (which are far more efficient than yours). You just need regulated DC supply for the filament. If your headphones are really less efficient at ear level than speakers are from 1 meter away, then noise won't be an issue anyway.
Getting a 300B quiet is not hard. I use it for my headphones (which are far more efficient than yours). You just need regulated DC supply for the filament. If your headphones are really less efficient at ear level than speakers are from 1 meter away, then noise won't be an issue anyway.
Last edited:
The efficiency is in mW not watts, so they're not that inefficient
Tell me more about your regulated DC B+, which design did you go with ?
thanks
The efficiency is in mW not watts, so they're not that inefficient
Tell me more about your regulated DC B+, which design did you go with ?
thanks
Mine is just George Anderson's TubelabSE with the power transformer in a separate enclosure. You can find the schematics and PCB on his website. It can be wired p2p as well.
The filaments are regulated DC but the B+ is unregulated CLCLC with two 2H chokes. The WE417A driver tubes have CCS supply so power to the gain stage is very clean. B+ is 400V biased at 80mA.
I have Denon and Sennheiser headphones. I just use the speaker's 3.5k:8 transformers and it still gets too loud with the volume pot at 9 o'clock. Sounds amazing though.
Thanks for the info, if I had Senns or Denons this would be easy
I've got a little spud WE417 SSHV B+ Electra-Print OPT SET that I love for my efficient Grados. These ortho headphones are such oddballs but they sound amazing if you haven't tried the ortho Hifimans or LCD2 you may want to your amp design would probably be a good match.I just did a test, with my 6DJ8+Mofset hybrid. Turned the amp up to normal comfortable level and listened to a song, then I played a 0dbs 500hz test tone. I measured 6.5 VRMS across the transducers with my Fluke.
So a little math. 6.5Vrms = 18.7V p-p. With a 3.5k:16 OPT that equates to a 276V swing across the primary. That is a big long load line for any SET tube, nearly impossible.
Now with a custom 2.5k:40 OPT I would need 150V swing across the primary. Still a healthy task.
So specific question for anyone, what DHT(s) would you choose for normal listening levels that required a 150V wide load line across a 2.5K primary ?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.