Thanks for the replies to my previous message, I think I understand the LTP based circuit now.
I like the simplicity of the new circuit, it will be interesting to hear what it sounds like. A few things occured to me:
The bias point seems to be very sensitive to temperature (my sims show a change of 3V between 20-100 degrees) so ac output coupling sounds like a good idea, I guess we could also use an op-amp servo if we wanted dc output coupling with a more complicated circuit.
R2 allows power supply noise into the signal path and we have lost the benefits of the virtual earth of the other circuit so will it hiss? (My main motivation for wanting to build a headphone amp is the unbearable hiss on the headphone out of my main amp)
This circuit will generate much more heat that the other one, we the case itself be a big enough heatsink?
The overall gain will be dependant on the individual device characteristics, will they need careful matching between channels?
Seeing the differences betweens the simulation results Nico and Hugh got I thought I'd have a go at modelling it in gnucap (not very user friendly) to see what it said. The voltage gain is 10.8dB with a phase shift of -3.6 degrees @20kHz, it is 1dB down around 170kHz and 3dB down around 320kHz. At 20Hz the phase shift is +4.2 degrees due to C1, dropping to below 1 degree @80Hz. The harmonics were virtually identical to those Hugh quoted. I used a BC547B and a BD140 for the transistors. I wonder if the difference in the results between the simulations is down to differences in the transistor model data - I tried two different models for the BD140, one from ON semiconductor had an unrealistically high hfe and gave very different results to those above which used the fairchild model. In the end it is what it sounds like that matters.
It has not been above 10 degrees here in glasgow for a while and we have had a couple of frosts recently so some class A heat would be welcome! Although I don't think I could cope with 39 degrees.
Best wishes
Phil
I like the simplicity of the new circuit, it will be interesting to hear what it sounds like. A few things occured to me:
The bias point seems to be very sensitive to temperature (my sims show a change of 3V between 20-100 degrees) so ac output coupling sounds like a good idea, I guess we could also use an op-amp servo if we wanted dc output coupling with a more complicated circuit.
R2 allows power supply noise into the signal path and we have lost the benefits of the virtual earth of the other circuit so will it hiss? (My main motivation for wanting to build a headphone amp is the unbearable hiss on the headphone out of my main amp)
This circuit will generate much more heat that the other one, we the case itself be a big enough heatsink?
The overall gain will be dependant on the individual device characteristics, will they need careful matching between channels?
Seeing the differences betweens the simulation results Nico and Hugh got I thought I'd have a go at modelling it in gnucap (not very user friendly) to see what it said. The voltage gain is 10.8dB with a phase shift of -3.6 degrees @20kHz, it is 1dB down around 170kHz and 3dB down around 320kHz. At 20Hz the phase shift is +4.2 degrees due to C1, dropping to below 1 degree @80Hz. The harmonics were virtually identical to those Hugh quoted. I used a BC547B and a BD140 for the transistors. I wonder if the difference in the results between the simulations is down to differences in the transistor model data - I tried two different models for the BD140, one from ON semiconductor had an unrealistically high hfe and gave very different results to those above which used the fairchild model. In the end it is what it sounds like that matters.
It has not been above 10 degrees here in glasgow for a while and we have had a couple of frosts recently so some class A heat would be welcome! Although I don't think I could cope with 39 degrees.
Best wishes
Phil
Changing the subject.... from weather to business.
I have assembled a single channel of the SE design (it took me less than half an hour). I find it more objective to test something in a monaural configuration as you are not distracted by effects (of course playing on both cans and the stereo signal mixed).
Although five hours does not constitute in-depth listening tests but I think I have formed some opinion of each topology.
I used the various disks for evaluation. Tracey Chapman, I find this a very revealing album and although it is a small ensemble of backing instruments, it has some difficult passages and can overload an amp easily.
I also listened to tracks from the following artists:
Acoustic Alchemy, Brian Bomberg, Diana Krall, Dire Straits, 3rd Force, Marcus Miller, Pat Coil, Bob James, Deep Purple, Joe Cocker, Moody Blues, The Who and Rare Earth. Sorry I don't like classical.
I think that it is very difficult for any one person to speculate how others would perceive an amps voice for lack of a better description. Personally I prefer amps that do not add any colouration (distortion of any kind) to the program material, but it was suggested that preference is given to a valve sound for whatever reason.
However I am an extremist and very critical listener when I listen to equipment rather than music.
The extremist view - listening to the amps
In my honest opinion neither amp has a valve sound, in fact it is rather unlikely that they would have, but the single ended design tends toward this kind of warm laid back sound, while the original amp sounds what one may describe as faster or urgent.
Listening to the original amp, I feel (with some of the recordings) the same anxiousness as I would if stuck in a traffic jam on a hot day without air-conditioning, in other words I wished the song would finish.
The single ended design is much more relaxing and easier to listen to for the same tracks however, the current limitation make for extreme transients produced horrible clipping. Current need be set to minimum 180 mA to cope with transients even using my HD800s whgich are relatively high impedance.
Although I enjoy Marcus Miller bass slam in some of the solos, I actually could not listen to a whole track on the original design, it became irritating after a short while almost as if the amp did not want to follow the program.
I was more forgiving listening on the SE amp because it was more tuneful, still not great but okay.
The resolution of the SE amp was better in that minute detail was audible while the original amp masked detail quite easily in that some instruments just became a blend of others and the detail was lost. A triangle being struck in the background could be heard on the SE while there was no hint of the triangle in the original amp.
On voices the original amp outperformed the SE in that the SE warmness did not go well with either male or female voice, the SE did not make it sound chesty but it was not real and the original design did a better job of it.
Now listening to music
The final listening was to use all the previous music and listened to it at normal level while writing my trip report, in other words I was actually listening to music not trying to be critical about anything.
Both amps sounded pleasing most of the time but the SE was easier to listen to. Some tracks as explained above on the original design actually made me stop working and waited for the song to finish because I could not concentrate on what I was doing, it was objectionable (hard word). Keeping in mind that the tracks I chose I believe would be quite a stringent test.
The final test and probably what would be very common:
I have several thousand recordings on my PC (MP3 format) and both amps perform well using this lossy format there was nothing objectionable about either and I am still listening to the SE without feeling tired - I do listen to music continuously while working. The SE amp is probably my favourite at this time as it has an easy relaxed way of conveying the program material and only gets stressed if the dynamic range gets out of hand, but this can be overcome by increasing the bias current.
In my opinion this does not say that the one amp is better than the other they are quite different in the way they perform for whatever reason I do not know, but I prefer the sound of the SE over that of the original.
What I will do next if anyone is interested is running the original amp in SE mode, but that will have to wait until tomorrow, I am going to watch the rugby now.
Kind regards
Nico
I have assembled a single channel of the SE design (it took me less than half an hour). I find it more objective to test something in a monaural configuration as you are not distracted by effects (of course playing on both cans and the stereo signal mixed).
Although five hours does not constitute in-depth listening tests but I think I have formed some opinion of each topology.
I used the various disks for evaluation. Tracey Chapman, I find this a very revealing album and although it is a small ensemble of backing instruments, it has some difficult passages and can overload an amp easily.
I also listened to tracks from the following artists:
Acoustic Alchemy, Brian Bomberg, Diana Krall, Dire Straits, 3rd Force, Marcus Miller, Pat Coil, Bob James, Deep Purple, Joe Cocker, Moody Blues, The Who and Rare Earth. Sorry I don't like classical.
I think that it is very difficult for any one person to speculate how others would perceive an amps voice for lack of a better description. Personally I prefer amps that do not add any colouration (distortion of any kind) to the program material, but it was suggested that preference is given to a valve sound for whatever reason.
However I am an extremist and very critical listener when I listen to equipment rather than music.
The extremist view - listening to the amps
In my honest opinion neither amp has a valve sound, in fact it is rather unlikely that they would have, but the single ended design tends toward this kind of warm laid back sound, while the original amp sounds what one may describe as faster or urgent.
Listening to the original amp, I feel (with some of the recordings) the same anxiousness as I would if stuck in a traffic jam on a hot day without air-conditioning, in other words I wished the song would finish.
The single ended design is much more relaxing and easier to listen to for the same tracks however, the current limitation make for extreme transients produced horrible clipping. Current need be set to minimum 180 mA to cope with transients even using my HD800s whgich are relatively high impedance.
Although I enjoy Marcus Miller bass slam in some of the solos, I actually could not listen to a whole track on the original design, it became irritating after a short while almost as if the amp did not want to follow the program.
I was more forgiving listening on the SE amp because it was more tuneful, still not great but okay.
The resolution of the SE amp was better in that minute detail was audible while the original amp masked detail quite easily in that some instruments just became a blend of others and the detail was lost. A triangle being struck in the background could be heard on the SE while there was no hint of the triangle in the original amp.
On voices the original amp outperformed the SE in that the SE warmness did not go well with either male or female voice, the SE did not make it sound chesty but it was not real and the original design did a better job of it.
Now listening to music
The final listening was to use all the previous music and listened to it at normal level while writing my trip report, in other words I was actually listening to music not trying to be critical about anything.
Both amps sounded pleasing most of the time but the SE was easier to listen to. Some tracks as explained above on the original design actually made me stop working and waited for the song to finish because I could not concentrate on what I was doing, it was objectionable (hard word). Keeping in mind that the tracks I chose I believe would be quite a stringent test.
The final test and probably what would be very common:
I have several thousand recordings on my PC (MP3 format) and both amps perform well using this lossy format there was nothing objectionable about either and I am still listening to the SE without feeling tired - I do listen to music continuously while working. The SE amp is probably my favourite at this time as it has an easy relaxed way of conveying the program material and only gets stressed if the dynamic range gets out of hand, but this can be overcome by increasing the bias current.
In my opinion this does not say that the one amp is better than the other they are quite different in the way they perform for whatever reason I do not know, but I prefer the sound of the SE over that of the original.
What I will do next if anyone is interested is running the original amp in SE mode, but that will have to wait until tomorrow, I am going to watch the rugby now.
Kind regards
Nico
Just as a matter of interest, I used an LM317 as current generator for the SE amp. Tomorrow I will place the same current generator in the original design.
Maybe I was a little hard on my choice of words in the posting above, but I am trying to describe what I hear and feel. Others may have a totally different view and I would suggest aothers who has the components at hand also try it out and give their opinions.
Maybe I was a little hard on my choice of words in the posting above, but I am trying to describe what I hear and feel. Others may have a totally different view and I would suggest aothers who has the components at hand also try it out and give their opinions.
Nico,
Many thanks. This is kinda what I suspected would happen; I've built the push pull circuit, know the sound, and agree. The SE I've also built, but not taken off the test bench yet.
My feeling is that when we are using headphones, we are really listening actively and privately to music. Therefore the presentation should be non-fatigueing and laid back, wearing headphones is not always comfortable anyway and we need every advantage we can get.
So I would go for the SE design.
Now, heat. A high Z headphone will need higher voltage, certainly, I'm surprised it also needs high current. At 180mA, and using a 24V 1A smps supply we really need to dissipate around 4.5 watts. I suspect a case, even of 220mm outside dimension, would get pretty warm, so we might need to put in a 6C/watt heatsink. This will complicate the design, as it cannot use the entire rear, since room must be left for power and source input.
Power supply would definitely benefit from a choke of some sort after the regulator, just to remove any HF artefacts more effectively than a cap alone.
To improve the sound of the PP amp we could simply introduce nested feedback coming straight off the VAS collector. I will devise this later today. This will remove the urgency from the sound, render it more organic, and enhance imaging. It also reduces global feedback to less than 20dB.
John,
Good idea, can you do it, Nico? It might take a bit of time, and I don't want to abuse your good graces.......
Cheers,
Hugh
Many thanks. This is kinda what I suspected would happen; I've built the push pull circuit, know the sound, and agree. The SE I've also built, but not taken off the test bench yet.
My feeling is that when we are using headphones, we are really listening actively and privately to music. Therefore the presentation should be non-fatigueing and laid back, wearing headphones is not always comfortable anyway and we need every advantage we can get.
So I would go for the SE design.
Now, heat. A high Z headphone will need higher voltage, certainly, I'm surprised it also needs high current. At 180mA, and using a 24V 1A smps supply we really need to dissipate around 4.5 watts. I suspect a case, even of 220mm outside dimension, would get pretty warm, so we might need to put in a 6C/watt heatsink. This will complicate the design, as it cannot use the entire rear, since room must be left for power and source input.
Power supply would definitely benefit from a choke of some sort after the regulator, just to remove any HF artefacts more effectively than a cap alone.
To improve the sound of the PP amp we could simply introduce nested feedback coming straight off the VAS collector. I will devise this later today. This will remove the urgency from the sound, render it more organic, and enhance imaging. It also reduces global feedback to less than 20dB.
John,
Good idea, can you do it, Nico? It might take a bit of time, and I don't want to abuse your good graces.......
Cheers,
Hugh
Last edited:
Hi Hugh,
the rugby is over and South Africa lost to the French. It is not a problem to make the original design back to SE which is what I have had before changing it. I was happy with it then, but I would like to compare it myself to the simple SE design which I must say does have a comfortable sound and yes I would agree with you that headphones sitting on the ears are uncomfortable after a while and it would make one more critical in the long run.
I should have it up and running again in its original form probably in an hour or so. I do agree that it ran pretty hot.
I must also say that one does not suffer from earache with the HD800 or any headphone that covers the ears rather than sit on the ears. I must also point out that the HD800 is very revealing, and rated as the best headphones available.
I have another set, HD 600 and also HD595 which I would give a bash since they have a completely different sound and probably more in line what most enthusiasts have.
I would also like to point out that near clipping is much easier detected on headphones than on speakers and the SE design when the CCS runs out of current does not clip, the halve wave simply falls apart completely and all kinds of cr@p is generated. This is unfortunately the downfall of any SE design, it works great to a point and then completely crashes unlike other designs that gradually clips.
Also keep in mind that we all agreed that the patch cord is capacitive and could draw unexpected peak currents. (I did not place the 120 ohm series resistor) which is probably there to soften the Q of the transducer system.
Nico
the rugby is over and South Africa lost to the French. It is not a problem to make the original design back to SE which is what I have had before changing it. I was happy with it then, but I would like to compare it myself to the simple SE design which I must say does have a comfortable sound and yes I would agree with you that headphones sitting on the ears are uncomfortable after a while and it would make one more critical in the long run.
I should have it up and running again in its original form probably in an hour or so. I do agree that it ran pretty hot.
I must also say that one does not suffer from earache with the HD800 or any headphone that covers the ears rather than sit on the ears. I must also point out that the HD800 is very revealing, and rated as the best headphones available.
I have another set, HD 600 and also HD595 which I would give a bash since they have a completely different sound and probably more in line what most enthusiasts have.
I would also like to point out that near clipping is much easier detected on headphones than on speakers and the SE design when the CCS runs out of current does not clip, the halve wave simply falls apart completely and all kinds of cr@p is generated. This is unfortunately the downfall of any SE design, it works great to a point and then completely crashes unlike other designs that gradually clips.
Also keep in mind that we all agreed that the patch cord is capacitive and could draw unexpected peak currents. (I did not place the 120 ohm series resistor) which is probably there to soften the Q of the transducer system.
Nico
:
Cancel that, I placed 1 minute of sample recording of each amplifier for download on the web site below:
Index of /temp
Anyone can download and listen to it. I will not say which file belongs to which amplifier so this is really a blind test, and up to you to decide.
I am really pushed that you hear the difference through loudspeakers with a a PC sound card because I cannot but here it is. The upload should be complete in the next half hour from this posting.
Kind regards
Nico
Cancel that, I placed 1 minute of sample recording of each amplifier for download on the web site below:
Index of /temp
Anyone can download and listen to it. I will not say which file belongs to which amplifier so this is really a blind test, and up to you to decide.
I am really pushed that you hear the difference through loudspeakers with a a PC sound card because I cannot but here it is. The upload should be complete in the next half hour from this posting.
Kind regards
Nico
http://digisec.co.za/temp/Recording Nov 14 2009 10 57 22 AM.wav
the wav in the address bar confirms it is an audio file.
But when I right click "save as" it says it's a movie file.
I looked at the quick time settings and it's far too detailed for me to start altering the defaults, without knowledge.
the wav in the address bar confirms it is an audio file.
But when I right click "save as" it says it's a movie file.
I looked at the quick time settings and it's far too detailed for me to start altering the defaults, without knowledge.
Last edited:
Andrew, I am not certain how to help you. Let me get this in order, you double click on the file on the web-site and the PC should ask you to open, save or cancle. Select SAVE. Another box should open for you to select where to save it. Choose a folder. You may want to change the name like 1.wav to force the file name and extension.
Once saved as 1.wave or whatever you chose, quicktime should open and play it if wav is one of the default settings. This is a basic raw windows type file and I do not see why you should experience problems.
If I saved it as MP3, this is a lossy medium and probably sound junk.
Nico
Once saved as 1.wave or whatever you chose, quicktime should open and play it if wav is one of the default settings. This is a basic raw windows type file and I do not see why you should experience problems.
If I saved it as MP3, this is a lossy medium and probably sound junk.
Nico