Indeed, the Aleph-mini is a nice sounding little amp. More importantly, it is summer-compatible with my current location.
Another quick simulation in SPICE showed that increasing AC current gain seems to lower overall THD. With 50% gain, I get 2nd harmonic at -90dB and 3rd at -119dB. Increasing to 66% then gives 2nd at -97dB and 3rd at -125dB. Higher levels of AC gain improve these numbers even more, but show the 2nd and 3rd harmonics getting closer together in magnitude. Perhaps this is the difference in sound - the relative difference between 2nd and 3rd harmonics.
Of course, there are two issues with my simulations... First is the overly optimistic nature of SPICE (I don't expect to see .002% THD in my actual build). Second, it tells me nothing about the change in sound quality from varying the AC gain. Hopefully time will tell once I get the amp built...
Another quick simulation in SPICE showed that increasing AC current gain seems to lower overall THD. With 50% gain, I get 2nd harmonic at -90dB and 3rd at -119dB. Increasing to 66% then gives 2nd at -97dB and 3rd at -125dB. Higher levels of AC gain improve these numbers even more, but show the 2nd and 3rd harmonics getting closer together in magnitude. Perhaps this is the difference in sound - the relative difference between 2nd and 3rd harmonics.
Of course, there are two issues with my simulations... First is the overly optimistic nature of SPICE (I don't expect to see .002% THD in my actual build). Second, it tells me nothing about the change in sound quality from varying the AC gain. Hopefully time will tell once I get the amp built...
No surprises. Varying the gain of the current source has a substantial effect
on the sound. As you increase the gain, the open loop gain of the amplifier
increases, so distortion goes down and becomes more 3rd harmonic in
character. As you decrease the gain of the current source toward having
constant DC current, the open loop goes down, there is less feedback,
the distortion goes up and is mostly negative phase 2nd harmonic.
Playing with the current source gain (either Aleph or mu-follower circuits)
also impacts the efficiency. If you are simply looking to maximize efficiency,
then the 50% gain figure is close to what you want - the current source is
seen to contribute 50% of the AC output current. With this the actual
efficiency of the amplifier will be 40% at best for real circuits.
When you deviate from 50% you will find that you need to increase the
bias current to have symmetric clipping, but you pick up some interesting
performance aspects:
If you double the current and the gain of the current source, what used to
be the gain transistor has very little work to do, and actually becomes more
like a driver transistor for the current source. The distortion gets much
lower, but higher order harmonic content.
If you double the current and set the current source as constant (0 gain)
then the distortion becomes more purely 2nd and continues to get lower.
This setting tends to sound quite good.
Of course in both cases the efficiency drops by half, so you're getting down
to 15% to 20% instead of 30% to 40%
on the sound. As you increase the gain, the open loop gain of the amplifier
increases, so distortion goes down and becomes more 3rd harmonic in
character. As you decrease the gain of the current source toward having
constant DC current, the open loop goes down, there is less feedback,
the distortion goes up and is mostly negative phase 2nd harmonic.
Playing with the current source gain (either Aleph or mu-follower circuits)
also impacts the efficiency. If you are simply looking to maximize efficiency,
then the 50% gain figure is close to what you want - the current source is
seen to contribute 50% of the AC output current. With this the actual
efficiency of the amplifier will be 40% at best for real circuits.
When you deviate from 50% you will find that you need to increase the
bias current to have symmetric clipping, but you pick up some interesting
performance aspects:
If you double the current and the gain of the current source, what used to
be the gain transistor has very little work to do, and actually becomes more
like a driver transistor for the current source. The distortion gets much
lower, but higher order harmonic content.
If you double the current and set the current source as constant (0 gain)
then the distortion becomes more purely 2nd and continues to get lower.
This setting tends to sound quite good.
Of course in both cases the efficiency drops by half, so you're getting down
to 15% to 20% instead of 30% to 40%
I updated my Aleph 3 with similar parts as recommended. The difference being I used Fairchild FQP3P20 instead of the Toshiba 2sj313. After a short listen over the weekend, I too feel the sound is on the bright side. However, it is noticeably better in terms of noise and the bass is heavier (something I tend to like). I'm going to run it in for a while and see if the sharpness subsides slightly.
Yes the Alaph3 sound with the 9610 is compared to the SJ313 in some aspects worse.
There is only one thing astonishing....
When I hear the 9610 version again alone I am happy, when I hear the SJ version not, I am more bored ....just another amp.
The 9610 makes some things so good that the defective aspects diminish.
There is only one thing astonishing....
When I hear the 9610 version again alone I am happy, when I hear the SJ version not, I am more bored ....just another amp.
The 9610 makes some things so good that the defective aspects diminish.
Dear Gerd, Our brain is an adapting machine, Its change all the time, our mind always say there are some better, For me is a constant, listen another and another speaker, musical version, another amp, some weeks likes one with NFB and another its sound crap. This is the life, sadly and gladly too.
With all the ongoing Aleph talk, this might be the right place to ask... Which output FETs would likely perform better in an Aleph headphone amp build for driving a power hungry 50 Ohm impedance: IRF610, IRF240, FQP3N30 or 2SK2013? Or maybe even some other FETs that I haven't considered yet?
Dear Friends,OK, here we go.
Last month I collected some "junk" Aleph 30's and 60's from the back shelves
of the storage space (the same one you remember from Raiders of the Lost Ark)
with the intention of salvaging the heat sinks drilled for TO-3 devices for use in
another project (yet another Sony SIT amplifier).
Opening them up, I discovered that they were in good enough shape that they
were worth keeping with a little cleaning and repair.
You will perhaps remember that I previously posted an update to the Aleph 3
which incorporated Toshiba 2SJ109 Jfets on the input stage, and while this
is still a worthy update there are two problems with it; first, it's difficult to
get these parts, and second, they only work on the lower voltage amplifiers
which leaves out most of the other Alephs.
However on these I decided to try my favorite small power Mosfets, the
Toshiba 2SJ313's. Apart from a different package and pin-out in some
amplifier models, they will drop right in as replacements, and they need to
be matched for Vgs at 10 mA.
The Toshiba's are preferred, but you can make do with other comparable
modern Mosfets if you can't get these parts.
Noting that some units had zener noise with R105 at 4.7K, I dropped that
value to 3.3K, and you will note that I optionally deleted Q104, the over-
current protection. Also, I moved the chassis ground connection directly
to the AC Earth ground.
The addition of the Toshiba Mosfets made for an immediate improvement
in power supply noise rejection, dropping it typically by a factor of 4 to 6, to
as low as 50 uV unweighted output. This alone justified the effort.
Also, as a tweak, you can replace the 220 uF caps with the silk caps from Elna.
Listening tests showed a significant improvement over the originals, and I
gave copies of these to my business partners for Xmas.
I'd be so obliged if You could give answer to the question . Is it possible to replace the Q105 (MPSA18) with the ZTX 450 without any circuit changing?
Regards,Chelg
Yes any low noise to-92 npn transistor is suitable.
Circuit is relying on Vbe of transistor which is basically the same.