The Aleph-X

Jump to Latest
#2,962
Hi Ian,

I always use a 50-200 ohm resistor in series with the JFET sources on my symmetrical input differential pairs to linearize the gain which reduces distortion. Low parisitics are important here, so either a surface mount or 1/8 watt film resistor work best. I also use zener diodes plus bypass capacitors to bias my cascode circuits.

I attached an interesting JFET-Bipolar_cascode input circuit from a diyAudio thead showing these ideas.


If this group is open minded, a Super-Symm design using the newer ring emitter bipolar output transistors from Sanken and OnSemi would be a good challenger to original all MOSFET AlephX thinking. OnSemi adds diodes on the biploar output die for rapid thermal tracking.
 

Attachments

  • sourceresistors.jpg
    sourceresistors.jpg
    47.7 KB · Views: 1,188
#2,963
Your idea of including source resistors is one I have seen used elsewhere, e.g. Borbely. This may well be a good idea but is not something I have experimented with in practice. Have you made any listening and/or measurment comparisons of the two approaches?

I also notice that your circuit connects the cascode references to the source resistors rather than earth (as per my schematic). This is intuitively better but I was concerned about then having to allow for the effects of the reference currents through the diff pair 'tails'. Presumably you have not found this to be a problem?

Hopefully the group is open minded and from my perspective there is plenty of room for alternative ideas. Perhaps we can all build something slightly different and compare the results? We will need to keep the basics constant though otherwise it will be difficult to know to what to attribute any audible differences.

Ian.
 
#2,967
The amount of CCS current is entirely up to you. Choose a wattage at which you wish to leave (more-or-less) pure single-ended operation and enter (more-or-less) push-pull operation and set the CCS accordingly.
Want a rough figure? Shoot for 1A from the CCS. Maybe as much as 2A.
Don't think in terms of percentage of overall power--think in terms of listening room volume achieved before kicking in the push-pull overdrive. Everyone can set theirs differently if they want, depending on how loud they listen, efficiency of speakers, what kind of music, etc. Or use a switch/pot to vary the proportion according to your mood...or whatever proportion of 2nd/3rd distortion suits you.

Grey
 
#2,970
Tazzz said:


Anyway I just had this statement by NP in mind...
Click to expand...


In which he says it's 'pretty arbitrary.' I think you'll find my suggestion of an amp or two completely sufficient to meet Nelson's criterion of 'the first watt or so'...at least for any reasonable load impedance. For that matter, if all you want to do is hit 1W, you can actually decrease my suggested current.
No, there's no magic ratio.
You're trying too hard. Build it and be happy.

Grey

P.S.: If you're absolutely determined to obsess about this current source thing, remember that the current can be changed with a single resistor value. It does not require a different PC board, different topology, or the slaughter of a virgin goat under a full moon. Just change the resistor and--Presto!--you've got a higher (or lower) current from your current source.
Note: No goats were harmed during the building of any amp in this thread.
At least not at my house.
I take no responsibility for what others may do under a full moon.
 
#2,971
GRollins said:



In which he says it's 'pretty arbitrary.' I think you'll find my suggestion of an amp or two completely sufficient to meet Nelson's criterion of 'the first watt or so'...at least for any reasonable load impedance. For that matter, if all you want to do is hit 1W, you can actually decrease my suggested current.
No, there's no magic ratio.
You're trying too hard. Build it and be happy.

Grey
Click to expand...


Arbitrary with a catch:

"However unless you're planning on full-power single-ended Class A, you also need to have a substantial push-pull bias to give a clean transition, otherwise the distortion waveform suffers a glitch, and this is reflected in the sound."

The A1.4 uses only 200mA for the P-channels, and I´m not sure that counts as substantial..


P.S.: If you're absolutely determined to obsess about this current source thing, remember that the current can be changed with a single resistor value. It does not require a different PC board, different topology, or the slaughter of a virgin goat under a full moon. Just change the resistor and--Presto!--you've got a higher (or lower) current from your current source.
Note: No goats were harmed during the building of any amp in this thread.
At least not at my house.
I take no responsibility for what others may do under a full moon.
Click to expand...


Anyway I will shoot for 2A PP and 1A SE as a maximum that will suit my heatsinks (derated) and the number of devices I will need to spread the heat evenly across them.
 
You must log in or register to reply here.
Top Bottom