The Aleph Design Reloaded

nothing is stopping you , when Revelations are in question

:)

though , that's not Miller cap , more antioscillation thingie
Could you please clarify , I didn't understood ..
I have seen many amps /preamps where input LTP stage is connected to VAS transistor's base/gate and there is a capacitor connected to base and collector of VAS transistor .. in explaination it was given to be for handling Miller effect at high frequencies ..
But in any of the aleph circuits or in ba-1 or ba-2 circuits, I haven't seen any such capacitor on the VAS MOSFET ..
Could you please give some inputs regarding this ..
 
Last edited:
Official Court Jester
Joined 2003
Paid Member
well , being aware that I'm not even a deck-boy on this ship , I'll try .........

it is more matter of looking at Goat - is it better from front , or from back side :clown: ?

some amp designers/tweakers are using Miller capacitance term speaking of sort of problems , while other designers/tweakers are using "oscillation" term , addressing their concerns/intentions ...... while ,evidently, family of exact same cures is used in practice

however , Miller capacitance , at least to my understanding, is brought when frequency linearity is objected , while oscillations is used when "amplification phase" is of primary concern thus overall gain stability

stability per se , when circuit is translated from paper to physical world , is complemented with actual physical layout (pcb , wires etc.) so , in s some cases , Papa was able to squeeze through without compensations , with amp having more than enough clean phase behavior , while - in some other cases - phase compensation was/is necessary

explanation can be , probably , found in overall simplicity of his amps (read - minimal number of stages), and in fact that he's either talented , or versed , or both - in solving PCB puzzles
 
to play with the stock aleph 3 bias is it R113 to replace with a variable resistor ? what value :) ?
459110d1421179834-aleph-design-reloaded-aleph-3-jfet-gif
 
Member
Joined 2003
Paid Member
R105...is the change to 3.3K applicable to 2SJ313 inputs only, or can you use it with IRF9610 as well? Does the change to 3.3K from the usual 10K in Aleph schematics a function of the rails, or can you apply it to all Alephs? I've simulated various R105 (3.3K, 10K, 2K, 1K), and found there isn't much of a change in mA of the current source. Just trying to understand this partiular change.

Thanks
John
 

Attachments

  • aleph.png
    aleph.png
    29.9 KB · Views: 460
I've had a set of matched SJ313's sitting around for a couple of years. I've just been waiting for an opportunity to install them into the Alephs. I finally got that opportunity this week. In doing so I discovered an issue with the Alephs that I was unaware of. When I built them I used boards that I had procured through a group purchase many, many years ago. The boards were not the best design and it appears that the fellow that built them mixed some new & old features. I didn't know this and used the original Aleph 5 schematic to build my mono amps with. When I built the amps I didn't have a way to test them aside from a scope and DVM. I set up the AC current gain the recommended way and expected that they were correct.


A few years ago I bought a QA400 and I used it to test my Aleph mono's and I noticed some anomalies, but I dismissed it as my novice testing techniques. More recently I bought an AP System One and prior to beginning this upgrade I planned to perform before and after testing. Well the before testing showed some weirdness going on. Primarily in the THD vs Freq plots. I had a huge dip around 1 kHz dropping to close to -80dB and abnormally high noise on either side of that freq. I tried to check the ACIG and it was way off at 1kHz. I remember that I had set it up initially at 100Hz so I checked it at 100Hz and it was about where I set it up at around 55%. I checked various freq's and the ACIG was different at every frequency. I checked the second amp and it was doing the exact same thing, so now I know its a circuit or parts problem.

I finally pulled the predriver boards out and inspected and the boards were designed with the later feature of Q5's base tied to the collector with C7. I assembled these with the belief that they were as the original schematic with Q5 collector tied to emitter with a C7 value of .047uF when it should've been .001uF due to the change in position. So I was slewing the hell out of Q5. :)

I made the change to .001uF caps and that solved the problem. ACIG settings are now constant regardless of frequency.

Anyway, in all of the rush to fix my problems I forgot to run pretests while the 9610 diff pairs were still in, so I only have post tests with the SJ313's. :(

I've only had one afternoon to listen to the new Alephs, but based on the issues that were present I cannot give pre versus post Aleph Reloaded comments as my Alephs were reloaded massively. They do sound spectacular on the Vandersteen 2Ci's though. :D

My tests of the amps with the SJ313's are attached. Unweighted S/N came out at -90.5dB left & -92dB right.


Aleph_R&L_Ch_THDvsPwr_SJ313_diffpair_final.jpg

Aleph_L_Ch_THDvsFreq_2SJ313_repaired_1W.jpg

Aleph_Freq-Resp_SJ313.jpg
 
OK, here we go.

Hi Papa
Would you consider this part (see pdf) better than Vishay IRFP9610?
https://www.diodes.com/assets/Datasheets/ZVP3306A.pdf

This one (see link below) looks like it has a zero temp coefficient at the currents we are interested in.
https://ww1.microchip.com/downloads/en/DeviceDoc/VP2106 B082313.pdf

Toshiba/Harris/Fairchild are getting harder to come by, and Jfets aren't always the most convenient to use on a PCB that requires a higher voltage part (eg Aleph 2).

A list of alternative good input stage P-Channel mosfets would be worthwhile putting together - Preferably through hole.
 
I had compared a number of P-channel fets for this. After narrowing down half a dozen or so, I compared them to the SJ313, and a few other storied P-ch fets incl Sj76. They were put into a classic diff pair with (iirc) about 10-20 mA drain current through each of the two fets. The distortion was then measured open loop at 1vrms out. High and Low impedance sources to compare.

The best one was a cheap Infineon device BSP-92 which measured equal or lower distortion than any of the other fets at 1khz and 10khz. It’s not thru hole (sot-223) but you’d want to mount them on a hacked copper lug heatsink anyway so making a thru hole equivalent wasn’t much work.

I found that most of the smaller To-92 T.H. fets couldn’t compare as they had lower transconductance/ p-diss less than a watt @40 deg so it seemed pointless. hth
 
I found that most of the smaller To-92 T.H. fets couldn’t compare as they had lower transconductance/ p-diss less than a watt @40 deg so it seemed pointless. hth

I was thinking of paralleling a pair of TO-92 devices similar to J2 (so 4 for the diff pair), the parts I was looking at have a nice tempco so matching isn't even that critical, like it might be with say 2SJ74.

Paralleling will kill two birds with one stone with the additional benefit of lower noise, so 3 birds with one stone.

Not bad. I will have a nice roast bird dinner.

Hahahaha