I've recently started spending a lot of time with headphones and was thinking about having a custom amplifier built for them. Most commercial headphone amplifiers are basically prosaic preamps, so I was thinking of acquiring something more interesting. Has anyone here built a headphone amp based on the 45 before? I'm probably looking for something that'll drive headphones with impedances around 32 to 50 ohms or so. I know the 45 provides more power than I would need, but darn, the 45 is nice. Well of course, other tubes might be nice (71A, 01A, or something) as well? Hmm...
I used my 45 SE amplifier as a headphone amplifier until quite recently when I built a dedicated headphone amplifier around 5842 and 71A. I used a little attenuation because I could hear a little 120Hz buzz from the AC heated 45 filaments. (DC fully regulated)
Pick a low noise, high transconductance triode like the 5842 or triode connected D3A if you need more gain. Use regulated and well filtered dc on the 45 filament (constant current if you can)
My design uses unregulated B+ and bias supplies, 5842 with red led + diode bias, choke loading capacitively coupled to a 71A operated with fixed bias at ~20mA into 5K:32/8 ohm custom made output transformers. I drive my 32 ohm headphones directly and the design is silent - no hiss, hum or buzz. To avoid magnetic coupling between the output transformers, driver load chokes and the power transformer I have built it on two chassis.
The noise floor of the headphone amplifier appears to be significantly below the noise floor of my Amber 3501A audio analyzer, and is in territory where I wouldn't trust my computer sound card either.
I am pleased overall with the sound, and plan to invest in some really good Grado headphones in the near future to go with it. It was originally intended to be cheap, and destined for my office, but I overkilled on this design so it stays home. Next version will be very cheap and cheerful.
In short the 45 will make a great headphone amplifier output tube, the increased headroom will only result in even better linearity at normal listening levels. Just be careful to minimize noise contributors, keep currents toward the high end of the range, choose a driver as I discussed above and make sure the B+ feeding it is absolutely clean - a few mV or less of ripple is reasonable and achievable with a simple CRCLC type filter. Choke loading and parafeed are your friend in this regard, providing very good PSRR.
Splurge on the best output iron you can afford, depending on your performance target. (That's how mine ended up so far over budget, but no regrets
)
Jack Elliano of Electra-Print wound the output transformers for my amplifier to my specification, and I am very pleased with the result.
Pick a low noise, high transconductance triode like the 5842 or triode connected D3A if you need more gain. Use regulated and well filtered dc on the 45 filament (constant current if you can)
My design uses unregulated B+ and bias supplies, 5842 with red led + diode bias, choke loading capacitively coupled to a 71A operated with fixed bias at ~20mA into 5K:32/8 ohm custom made output transformers. I drive my 32 ohm headphones directly and the design is silent - no hiss, hum or buzz. To avoid magnetic coupling between the output transformers, driver load chokes and the power transformer I have built it on two chassis.
The noise floor of the headphone amplifier appears to be significantly below the noise floor of my Amber 3501A audio analyzer, and is in territory where I wouldn't trust my computer sound card either.
I am pleased overall with the sound, and plan to invest in some really good Grado headphones in the near future to go with it. It was originally intended to be cheap, and destined for my office, but I overkilled on this design so it stays home. Next version will be very cheap and cheerful.
In short the 45 will make a great headphone amplifier output tube, the increased headroom will only result in even better linearity at normal listening levels. Just be careful to minimize noise contributors, keep currents toward the high end of the range, choose a driver as I discussed above and make sure the B+ feeding it is absolutely clean - a few mV or less of ripple is reasonable and achievable with a simple CRCLC type filter. Choke loading and parafeed are your friend in this regard, providing very good PSRR.
Splurge on the best output iron you can afford, depending on your performance target. (That's how mine ended up so far over budget, but no regrets
)Jack Elliano of Electra-Print wound the output transformers for my amplifier to my specification, and I am very pleased with the result.
Wow, that sounds simply spectacular. Thanks so much for the help.
Hmm, a 5842 to 45 or 71A sure sounds like a great idea. Haha, building the thing will probably prove quite tricky for a novice like me. By the way, do you think it would be possible to build such an amp for under a grand initially? (Use cheap irons now and then get good irons later as my budget increases.)
Hmm, a 5842 to 45 or 71A sure sounds like a great idea. Haha, building the thing will probably prove quite tricky for a novice like me.
By the way, do you think it would be possible to build such an amp for under a grand initially? (Use cheap irons now and then get good irons later as my budget increases.)
Eclair said:Wow, that sounds simply spectacular. Thanks so much for the help.
Hmm, a 5842 to 45 or 71A sure sounds like a great idea. Haha, building the thing will probably prove quite tricky for a novice like me.
My sense, especially if you are not experienced, is that you will spend a lot more than that. My first big project, which was nowhere as complicated as a 45, involved throwing out as many parts as I used ... more, and cost a small fortune.
My inclination would be to start with a cheap tube with the same basic topology, and then convert to the expensive tube and expensive iron at the same time. The iron and topology, after all, probably are more important to the overall sound quality than the tube.
A good tube to start with for experimenting is the 6W6GT. It sounds very good and is similar to the 45 as far as power output, and is very cheap. Indeed, you may like it enough that you won't bother moving on to the 45 for a while.
I agree on the 5842. Also, look into the 6688 triode strapped. It is electrically about the same, cheaper, and I think sounds better. All of these high Gm tubes can oscillate easily, so some real care in layout is in order.
For topology, I'd look into direct coupling. With headphones, the level of detail is much higher with phones which means that you'll need to be spending a ton on coupling caps to even approach direct coupling in performance. The DRD circuit, or a parafeed version, would be my preference.
Finally, I'll agree with Kevin on the custom Electra-Print iron. I use a custom would 5K:32 transformer with a simple 7788/E810F spud amp for phones and it is very good (the pic has a 5842, but I've settled on the more powerful 7788). Sowter also makes headphone iron.
Ah ha, great advice! Perhaps I should give the 6W6GT a go initially. I guess I should probably start from the bottom and gradually go up the ladder. Direct reactance drive and parafeed sound like great choices as well.
As a side question, what about RPower's 71A direct coupled design?
(Schematic attached.)
As a side question, what about RPower's 71A direct coupled design?
(Schematic attached.)
Attachments
Oh, I've got another quick question as well. Is it a bad idea to connect headphones with an impedance of 32 ohms to 8 ohm outputs? dsavitsk, you mentioned that you were using a custom transformer with 32 ohm outputs. Does that transformer allow the amplifier to work with headphones of slightly higher impedances (50 ohms for example)? If so, about how high do you think it can go before the performance starts to degrade?
Eclair said:
Looks like a reasonable design except for the 1000uF cathode bypass cap. This will have to be of exceptional quality not to audibly degrade the sound quality, and bypassed with multiple smaller caps. (I made this mistake in one of my own designs, and ultimately found a moderately expensive BG electrolytic with several smaller BG in parallel to be acceptable.) You'd be better off with an LED chosen to provide the correct operating point. You can tweak the bias voltage to get the right operating point for both stages as they are interdependent being dc coupled.
Eclair said:
I also have 32 ohm taps on my amplifier which I use to drive 32 - 600 ohm phones without any problem so far. The 8 ohm tap can be used to drive 8 - 32 ohm headphones (or higher) without any problems at some loss in available voltage swing, which with most headphones is not going to be an issue.
Hi,
My 45 (driven by choke-loaded 5842) uses Tamura F-475 and I derived a headphone out from the 16 ohm tap. Sounds very good, but the amp is big for a headphone amp!
ps.
If you are interested, and inspired by Kevin's article, I made a 417A/5842 headphone amp, that spells simplicity and great sound.
My 45 (driven by choke-loaded 5842) uses Tamura F-475 and I derived a headphone out from the 16 ohm tap. Sounds very good, but the amp is big for a headphone amp!
ps.
If you are interested, and inspired by Kevin's article, I made a 417A/5842 headphone amp, that spells simplicity and great sound.
I assume that the thread starter is the same person that emailed me asking the same question. If not, the info still applies.
I just got back from a 3000 mile road trip (I picked up a few hundred pounds of tubes), and finally caught up on most everything. The email asked about modifying a TubelabSE for headphone use.
There is a page on my web site that details the extreme modifications that were done to a TubelabSE to build a custom amplifier for a very particular (and well financed) customer. He had previously purchased a TubelabSE amplifier with 45's for use with speakers. I added a headphone jack and a switch that put the headphones in parallel with a 10 ohm resistor across the output of an otherwise unmodified amplifier (5842 driving a 45), while disconnecting the speakers. I feel that this worked very well, but it could not get very loud with his high impedance (600 ohm) phones. The owner wanted another amp to add to his already large collection. He wanted it only for phones, and since he already had several DHT based amplifiers and this one would only need to make 1/4 watt at most, I suggested to try some tubes smaller than the 45 like the 30, 31 and the 01A.
I set up a TubelabSE board on a bench power supply, and tried several different tubes and transformers. The customer made a 500 mile trip (twice) to listen to each combination with his personal headphone collection that spanned from 32 to 600 ohms. He reached his personal audio nirvana with 01A tubes and multi tapped UTC LS57 OPT's that were meant for a 20 watt P-P amp.
Were these modifications really necessary? In this customers mind, they were. These were extreme modifications for an extreme customer, but they are easilly duplicated by an experienced builder. The details are here:
http://www.tubelab.com/Headphone.htm
Many TubelabSE users have used their unmodified amps with headphones. I have recommended a 10 ohm resistor in parallel with the headphones just to keep the 45 operating with its intended load, but many users simply connect their phones to the speaker terminals.
Keep in mind that the amp must be really quiet for use in this manner since the phones are far more efficient than speakers, and your ears are really close to them! I recommend using a choke and a supplemental capacitor in the power supply. All filaments should operate on DC. The layout and lead dress should be optimized to reduce hum.
I think that any of the designs mentioned here would produce satisfying results. If I was building a headphone amp using a triode wired pentode (6W6) I would give some consideration to the 6V6GT. It sounds really sweet in triode mode. At the power levels used with headphones transformer saturation is not a problem, so there is no need for huge OPT's, which may actually cost in the detail department. The $19 Edcor XSE15-8-5K transformers might just be ideal. I have built an amp like this using a SimpleSE board. It sounds real good with speakers, and with my (cheap) Sony headphones. I have been known to listen to the amp with the phones and the speakers hooked up at the same time. The phones provide most of the sound, while the speakers provide bass enhancement, and some vibratory realism.
I just got back from a 3000 mile road trip (I picked up a few hundred pounds of tubes), and finally caught up on most everything. The email asked about modifying a TubelabSE for headphone use.
There is a page on my web site that details the extreme modifications that were done to a TubelabSE to build a custom amplifier for a very particular (and well financed) customer. He had previously purchased a TubelabSE amplifier with 45's for use with speakers. I added a headphone jack and a switch that put the headphones in parallel with a 10 ohm resistor across the output of an otherwise unmodified amplifier (5842 driving a 45), while disconnecting the speakers. I feel that this worked very well, but it could not get very loud with his high impedance (600 ohm) phones. The owner wanted another amp to add to his already large collection. He wanted it only for phones, and since he already had several DHT based amplifiers and this one would only need to make 1/4 watt at most, I suggested to try some tubes smaller than the 45 like the 30, 31 and the 01A.
I set up a TubelabSE board on a bench power supply, and tried several different tubes and transformers. The customer made a 500 mile trip (twice) to listen to each combination with his personal headphone collection that spanned from 32 to 600 ohms. He reached his personal audio nirvana with 01A tubes and multi tapped UTC LS57 OPT's that were meant for a 20 watt P-P amp.
Were these modifications really necessary? In this customers mind, they were. These were extreme modifications for an extreme customer, but they are easilly duplicated by an experienced builder. The details are here:
http://www.tubelab.com/Headphone.htm
Many TubelabSE users have used their unmodified amps with headphones. I have recommended a 10 ohm resistor in parallel with the headphones just to keep the 45 operating with its intended load, but many users simply connect their phones to the speaker terminals.
Keep in mind that the amp must be really quiet for use in this manner since the phones are far more efficient than speakers, and your ears are really close to them! I recommend using a choke and a supplemental capacitor in the power supply. All filaments should operate on DC. The layout and lead dress should be optimized to reduce hum.
I think that any of the designs mentioned here would produce satisfying results. If I was building a headphone amp using a triode wired pentode (6W6) I would give some consideration to the 6V6GT. It sounds really sweet in triode mode. At the power levels used with headphones transformer saturation is not a problem, so there is no need for huge OPT's, which may actually cost in the detail department. The $19 Edcor XSE15-8-5K transformers might just be ideal. I have built an amp like this using a SimpleSE board. It sounds real good with speakers, and with my (cheap) Sony headphones. I have been known to listen to the amp with the phones and the speakers hooked up at the same time. The phones provide most of the sound, while the speakers provide bass enhancement, and some vibratory realism.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.