Headphone question?
One would think that the headphone driving problem begs for a class A solution due to the low power requirements. It is true that a lot of successful low power class A designs have been done but there is some magic with class D that can’t be denied. There are some chips available that possibly could work but most are made for 5 volts or less and this gives no headroom for the higher impedance phones. Also most are open loop types, not good! Does anyone have any ideas? How about an all discreet UcD type design? Anyone feel like doing a PCB?
Roger
One would think that the headphone driving problem begs for a class A solution due to the low power requirements. It is true that a lot of successful low power class A designs have been done but there is some magic with class D that can’t be denied. There are some chips available that possibly could work but most are made for 5 volts or less and this gives no headroom for the higher impedance phones. Also most are open loop types, not good! Does anyone have any ideas? How about an all discreet UcD type design? Anyone feel like doing a PCB?
Roger
Transformers?
This could be done but at what cost? Quality transformers are expensive and would be bulky. This would indeed solve several problems like the fact that headphones have a common ground and are single ended in the way they connect up. If you could find something surplus that would offer 1:1 windings with a 1 watt or so power handling it would be worth a try. The transformers would also offer additional filtering of the switching frequency. Good luck on your search.
Roger
This could be done but at what cost? Quality transformers are expensive and would be bulky. This would indeed solve several problems like the fact that headphones have a common ground and are single ended in the way they connect up. If you could find something surplus that would offer 1:1 windings with a 1 watt or so power handling it would be worth a try. The transformers would also offer additional filtering of the switching frequency. Good luck on your search.
Roger
stlblue said:
Yeah, I know. Things can be hard to find on this forum, not sure why.
Sure wish we could get a sticky post at the top to direct people to the Wiki/FAQ.Class-D is a bad idea for headphones for several reasons.
1) More power than you really need. Headphones require very little power compared to normal speakers, so many of the problems of power amps are not much of a concern for headphones. As stated above, Class-A amps would seem a good choice here.
2) Tons of noise. Class-D amps are noisy, by headphone standards. If you plug headphones directly into a class-D amp, even a small one, you'll hear what I mean! This noise is not a problem with normal speakers as they are so much less sensitive than phones.
3) The output filter. Switching amps use an low-pass output filter the is usually designed to work with a 4~8 ohm load. Headphones are often 32 ohms or more, so the filter response will be wrong.
4) H-Bridge design. The small Tripath chips used in the Sonic, Charlize, Fenice 20, AMP6, AMP3, etc. are bridged. Thus the left and right grounds can not be tied together. You would have to rewire your headphones to dual mono.
For the same money, size and trouble, a good Class-A amp, solid state or tube, even opamp, should give you better sound.
Hope that helps.
Re: Headphone question?
If someone can come up with a schematic I'd be willing to do the PCB layout.
1) We could design the amp to be low power.
2) We could use an LCLC output filter or some other measures to combat noise.
3) The output filter could be designed specifically for a higher impedance load.
4) H-Bridge isn't a problem if we could implement a single ended UCD design.
Michael has some good points, but if a low powered discrete UCD type design is built specifically for headphones it could work very well! Any takers? At this level of power all of the components could easily be surface mount and switching frequency could probably be increased above 1MHz for even easier filtering.
These are all just ideas of course
I haven't heard any good class a amps to compare to but class d definitely has magic compared to class ab!sx881663 said:
If someone can come up with a schematic I'd be willing to do the PCB layout.
panomaniac said:
1) We could design the amp to be low power.
2) We could use an LCLC output filter or some other measures to combat noise.
3) The output filter could be designed specifically for a higher impedance load.
4) H-Bridge isn't a problem if we could implement a single ended UCD design.
Michael has some good points, but if a low powered discrete UCD type design is built specifically for headphones it could work very well! Any takers? At this level of power all of the components could easily be surface mount and switching frequency could probably be increased above 1MHz for even easier filtering.
These are all just ideas of course
The headphone question
All valid points;
1) The lower power issue is one to think about. I personally wouldn’t want the amp to have voltage headroom issues with either the input or output side. High impedance phones can require quite a bit of drive voltage so I would say +/-12v should be considered a minimum.
2) The output filter is potentially a real problem, as stated they are designed for a specific load. I guess a fixed load could be added to narrow down the range of loads it could see but I would think at the expense of added noise. With a double filter configuration I don’t see how a UcD type feedback arrangement could be worked out. Once again a transformer could be the answer as an additional filter and to break the ground paths.
2a) This lower power amp could be run at 1mHz or even more and this should simplify the filter and noise issues. A UcD type design with all low voltage surface mount parts would be real nice. Looking at the Ztec offerings for output devices, their N and P type devices in the same package could make the drive single ended and a lot simpler. No level shifting or bootstrapping required. A chip type comparator could be used. Remember we only need about 1 watt max. More could be nice to drive small speakers. Maybe 3 watts max?
3) What impedance do you design for? Phones vary from around 32 ohm to over 600 ohm. With a 32ohm fixed load added the range would be from 16 to 30 something and this is doable as we are dealing with 4-8 ohm range right now.
4) Yes, this is why the UcD was brought up in the first place.
This is quite a project with the end result a real question. The results could be worth it or it could end up a waste of time. Sounds like a worthy DIY project to me!
If there is enough interest a new thread should be started (UcD type headphone amp?) with stated specific design goals.
Roger
All valid points;
1) The lower power issue is one to think about. I personally wouldn’t want the amp to have voltage headroom issues with either the input or output side. High impedance phones can require quite a bit of drive voltage so I would say +/-12v should be considered a minimum.
2) The output filter is potentially a real problem, as stated they are designed for a specific load. I guess a fixed load could be added to narrow down the range of loads it could see but I would think at the expense of added noise. With a double filter configuration I don’t see how a UcD type feedback arrangement could be worked out. Once again a transformer could be the answer as an additional filter and to break the ground paths.
2a) This lower power amp could be run at 1mHz or even more and this should simplify the filter and noise issues. A UcD type design with all low voltage surface mount parts would be real nice. Looking at the Ztec offerings for output devices, their N and P type devices in the same package could make the drive single ended and a lot simpler. No level shifting or bootstrapping required. A chip type comparator could be used. Remember we only need about 1 watt max. More could be nice to drive small speakers. Maybe 3 watts max?
3) What impedance do you design for? Phones vary from around 32 ohm to over 600 ohm. With a 32ohm fixed load added the range would be from 16 to 30 something and this is doable as we are dealing with 4-8 ohm range right now.
4) Yes, this is why the UcD was brought up in the first place.
This is quite a project with the end result a real question. The results could be worth it or it could end up a waste of time. Sounds like a worthy DIY project to me!
If there is enough interest a new thread should be started (UcD type headphone amp?) with stated specific design goals.
Roger
Re: The headphone question
1) +/-12v would be good voltages to design around.
2) I should have explained my LCLC filter idea. Just use the standard UcD configuration and add another LC filter (with a cutoff frequency maybe 2-3x the first one) to filter out higher frequency junk.
2a) N and P type FETs and a chip comparator would be perfect for this sort of application.
3) Doesn't the UcD design have load independent frequency response (after the first LC filter)? The second LC filter after the feedback node could be 'tuned' for a 600ohm load, or would it be possible to take feedback after the second LC filter and bring it back to the non-inverting input?
I'd sure be interested. I don't know enough to help design the amp, but I'd be glad to help with the PCB layout.
1) +/-12v would be good voltages to design around.
2) I should have explained my LCLC filter idea. Just use the standard UcD configuration and add another LC filter (with a cutoff frequency maybe 2-3x the first one) to filter out higher frequency junk.
2a) N and P type FETs and a chip comparator would be perfect for this sort of application.
3) Doesn't the UcD design have load independent frequency response (after the first LC filter)? The second LC filter after the feedback node could be 'tuned' for a 600ohm load, or would it be possible to take feedback after the second LC filter and bring it back to the non-inverting input?
sx881663 said:
I'd sure be interested. I don't know enough to help design the amp, but I'd be glad to help with the PCB layout.
panomaniac said:1) More power than you really need. Headphones require very little power compared to normal speakers, so many of the problems of power amps are not much of a concern for headphones. As stated above, Class-A amps would seem a good choice here.
Certain headphones could use that kind of power ie: the AKG K1000/340
2) Tons of noise. Class-D amps are noisy, by headphone standards. If you plug headphones directly into a class-D amp, even a small one, you'll hear what I mean! This noise is not a problem with normal speakers as they are so much less sensitive than phones.
I've never heard anyone from head-fi.org complain about the T-amp being noisy with their K701/1000/340s yet, actually.
3) The output filter. Switching amps use an low-pass output filter the is usually designed to work with a 4~8 ohm load. Headphones are often 32 ohms or more, so the filter response will be wrong.
Wouldn't a 4/6/8 ohm resistor in parallel solve that problem?
4) H-Bridge design. The small Tripath chips used in the Sonic, Charlize, Fenice 20, AMP6, AMP3, etc. are bridged. Thus the left and right grounds can not be tied together. You would have to rewire your headphones to dual mono.
I know I'm wiling to do that
For the same money, size and trouble, a good Class-A amp, solid state or tube, even opamp, should give you better sound.
Not necessarily, especially when the cheapest headphone amps recommended for the K1000 and K340 cost $300 or more.
Balanced out you have with the T-amps already. Balanced in is another matter...
Could you elaborate on that? Thanks (I have a source capable of balanced out)
I'm also quite interested in using a T-Amp for headphones. I've got a pair of Grados running off a Millet Hybrid and that's all well and good, but the Millet doesn't have the clarity I'd like and the other solid state options are rather pricey. An adapted T-Amp would be an affordable alternative.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.