Back from dreamland Nico! Send me your email, I would like to write to you!
I confirm all your statements about low distortion and poor sound. This causes sharp criticism and has come to regard as religion.
The CFP makes a fine headphone amplifier but to improve THD (while maintaining the harmonic profile) you should reduce the collector resistor on the master from 2k2 to 330R.
Hugh Dean
I confirm all your statements about low distortion and poor sound. This causes sharp criticism and has come to regard as religion.
The CFP makes a fine headphone amplifier but to improve THD (while maintaining the harmonic profile) you should reduce the collector resistor on the master from 2k2 to 330R.
Hugh Dean
It does! 😀
Partial Harmonic Distortion: 0.022820%
Total Harmonic Distortion: 0.074398%
@Michelag you can still use 16 ohm headphones using a supply voltage of 19 V without changing anything else.
Now both magicians of this circuit are in this thread I would like to thank Hugh and Nico for this great design!
Last edited:
How about naming it the 'AAS (Aspen Audio Synthesis) headphone amp' (see https://www.diyaudio.com/community/threads/diy-class-a-headphone-amp-suggestion.415623/post-7753352)?
I now have this:
Perhaps the resistor (R13) of the Zobel filter should be adjusted to the impedance of the headphones used.
I now have this:
Perhaps the resistor (R13) of the Zobel filter should be adjusted to the impedance of the headphones used.
Last edited:
I don't think the zobel is needed and R10 could be replaced with 10k. Headphones are easy to drive and this amp uses local feedback
To cope with a slight tendency to instability with fast transistors I would encourage using a 22pF from base to collector of the slave (Q2) although the TIP is a slow device.
As you raise the supply voltage the thd drops further; a 24V supply runs very well.
HD
To cope with a slight tendency to instability with fast transistors I would encourage using a 22pF from base to collector of the slave (Q2) although the TIP is a slow device.
As you raise the supply voltage the thd drops further; a 24V supply runs very well.
HD
Like this?
I noticed using 16 ohm headphones with a supply voltage of 19V the 3rd harmonic was bigger than the 2nd one. Increasing the supply voltage to 24V corrected that:
I do not know. It must be very simple but trying to calculate that I always end up with an unbelievable value...
I see, but here the review also states:
100 dB is getting dangerous very quickly (after 15 minutes of continuous exposure there is a health risk), so I really don't understand where the crazy 1 - 4 W comes from.
Assuming you really want to risk your auditory sense and listen for 15 minutes with 100 dB and furthermore assuming 12 dB peaks, you would not need more than 16 mW peak.
(Edit: removed unnecessary part)
100 dB is getting dangerous very quickly (after 15 minutes of continuous exposure there is a health risk), so I really don't understand where the crazy 1 - 4 W comes from.
Assuming you really want to risk your auditory sense and listen for 15 minutes with 100 dB and furthermore assuming 12 dB peaks, you would not need more than 16 mW peak.
(Edit: removed unnecessary part)
If you want more power you could either use a basic CCS (this would change the circuit considerably) or reduce your power resistors values, this will generate more heat so good cooling would be required, I would consider the resistor below because it is easily heatsinked (and would look really cool 😎 ) This is the tradeoff with simple Class A.
This is the calculation I was missing.
So I guess this actual amp with 24V will be just fine.
For 16R cans, use this formula which is based on HIGHEST Vpp before clip at the load:
[Vpp]exp2 / 128 (note the divider factor with 8R speakers is 64, so that with 80Vpp output you have 100W). This assumes a sine wave; not strictly true of course, but useful for design.
So, for your 24V supply, lets take it to about 15Vpp. This would be 1.75Wrms into 16R.
But you will find that at this current your load will be peaking at 15/2 over 16 = 470mA. This is no issue for the positive half cycle, but not the negative half cycle, which needs the same current but is restricted by the very low voltage at the collector of the slave at lowest peak. You normally need about 4V minimal at the collector at clip, so there is the real risk that the current drawn from load to ground might turn on the slave transistor.
If this happens, the music stops.........
So, to keep the slave always on, you must pass a little more current through it at idle so that at negative half cycle where the voltage at the collector of the slave is very low. So, for 470mA max, you would need to set the current slave to 500mA.
It is after all Class A, there are some pitfalls........ many designers have two resistor strings from the CFP to ground, switchable, one fo >100R cans and another for <100R.
HD
[Vpp]exp2 / 128 (note the divider factor with 8R speakers is 64, so that with 80Vpp output you have 100W). This assumes a sine wave; not strictly true of course, but useful for design.
So, for your 24V supply, lets take it to about 15Vpp. This would be 1.75Wrms into 16R.
But you will find that at this current your load will be peaking at 15/2 over 16 = 470mA. This is no issue for the positive half cycle, but not the negative half cycle, which needs the same current but is restricted by the very low voltage at the collector of the slave at lowest peak. You normally need about 4V minimal at the collector at clip, so there is the real risk that the current drawn from load to ground might turn on the slave transistor.
If this happens, the music stops.........
So, to keep the slave always on, you must pass a little more current through it at idle so that at negative half cycle where the voltage at the collector of the slave is very low. So, for 470mA max, you would need to set the current slave to 500mA.
It is after all Class A, there are some pitfalls........ many designers have two resistor strings from the CFP to ground, switchable, one fo >100R cans and another for <100R.
HD
You may be interested those are mentioned here: https://diyaudioheaven.wordpress.com/headphones/measurements/audeze/lcd-xc/
No surprise there is no correction filter needed for those... 🙂
Actually, other source linked by @widea states Efficiency: 93dB @ 1mW
Yes, but my fiio M11 I'm not able to make them sing, neither with my cavalli Jones tube headphone amp (but this is for Z>200ohm, so ok), I'm quite baffled with them now. Bought used.
One may wonder why they were sold... I am afraid there is only one way to find out... What do other reviews say about these headphones?
Did you contact the manufacturer?
That they be the best of the best, creme de la creme, and so on. There is nothing wrong with them, they're sold as studio cans, I thought of a super mega wow sound, yet I prefer the 100€ Sony mdr7506...
And now I'd like to try a senn 6xx...
Look, there's plenty of these cans for sale, looks like audiophiles get tires very soon and need to have the next desire fulfilled.
And now I'd like to try a senn 6xx...
Look, there's plenty of these cans for sale, looks like audiophiles get tires very soon and need to have the next desire fulfilled.