I'm no expert , but I have 1990's B&W that have a difficult impedance - and gone through this entire routine .
From what I understand , its wild swings in the phase angle that cause problems for amps .
IMO is all about having an amp with a low Zout and getting current to the load ( voice coils ) .
Recall , voice coils need current to build up a magnetic field .
The many recommendations here , and plant 10 's recommendation , to run 2 Aleph J's in parallel ,
effectively doubles the number of output devices , which 1/2 's the output impedance of the amp .
Your suggestion of running dual die output devices will effectively 1/2 the Zout of the amp .
However , the Ciss will also be doubled , so can the jFET differential pair effectively charge and discharge
the doubling in Ciss ?
Recall, the total Zout of an amp is dependent on both the Zout of the power supply and the Zout of the amp itself .
If you are determined to stick with single ended Aleph J , then decreasing the Zout of the power supply will help .
So connecting more power supply caps in the power supply should help and reducing the impedance of the power cabling inside the amp will help .
The power cabling inside the amp , is in the same order of magnitude as the ESR of power supply caps .
Plus switching to short length 10 awg speaker cables will help .
Make sure you speakers are well anchored , using a wide base , and spikes , will help .
Ultimately though , IMO , ZM's recommendation of running an amp with a [ source ] follower output stage is the way to go .
Of all the output stages , a follower output stage has the lowest Zout .
.
what I said, mentioning emiter follower , was strictly in context of using output mosfets (either number of them or type) with heavy interelectrode capacitances - so emiter follower being nested in between input stage (LTP) and modulated mosfet gates, as way of having those gates firmly driven by some Cojones
reverting to follower OS - I didn't mentioned that
original poster is asking for simple, effective but ideal solution ....... not to change an amp
reverting to follower OS - I didn't mentioned that
original poster is asking for simple, effective but ideal solution ....... not to change an amp
The issue with adding an emitter follower drive stage is the temp coef of a bi polar transistor .
The temp coef of the base - emitter junction is - 2mV / C , then this is connected to Vgs of the MOSFET .
So as the bi polar warms up , it changes the Vgs , so Vout will have a considerable DC offset until the amp is completely warmed up .
This opens open a complete can of worms trying to solve the temp coef drift .
Also, typically a CCS is connected to the emitter follower using a partner transistor .
If a 3rd set of IRFP240's are added to each channel of the Aleph J , instead try
However , depending on the impedance curve of your speakers , magnitude and swing in the phase angle ,
you are probably miles better off building a BA-3 and BA-1 with the SE output stage .
.
The temp coef of the base - emitter junction is - 2mV / C , then this is connected to Vgs of the MOSFET .
So as the bi polar warms up , it changes the Vgs , so Vout will have a considerable DC offset until the amp is completely warmed up .
This opens open a complete can of worms trying to solve the temp coef drift .
Also, typically a CCS is connected to the emitter follower using a partner transistor .
If a 3rd set of IRFP240's are added to each channel of the Aleph J , instead try
- selecting 2sj109's with a high Idss . To max linearity and drive current , set Id to 70 % of Idss .
- try reducing the gate resistors from 220R to 100R , which will reduce the RC time constant - and hopefully no oscillation
- or doubling up the 2sj109's
However , depending on the impedance curve of your speakers , magnitude and swing in the phase angle ,
you are probably miles better off building a BA-3 and BA-1 with the SE output stage .
.
Last edited:
A lot of great suggestions here in this forum...
@stretchneck
There's no replacement for displacement.
More MOSFETs won't help, higher quiescent current above a certain point won't help, and higher supply rails (above +/- 25V DC, especially if left unregulated) are not advisable.
With no simple option to move the class A (single-ended) transfer point higher, AND with no option to get more voltage gain out of input JFET... what other option is there, but to increase the displacement?
Nevertheless, Aleph J in its original form, with 2 out of 3 crucial elements maximised/optimised (the voltage rails and the quiescent current, where the 3rd element - the voltage gain... is what it is...), will drive 4-ohm speakers with 0 issues, especially if a little tender loving care (and a few mods) are applied. Give it a go.
Paralleling of modules will help... However, you'll have to stick to using the C1...
The main Aleph J thread is the source of information and knowledge. Read it, study it, build the amp, play with it... and then look at the circuit diagram again.... this will bring a new set of ideas... it takes time, but the results are oh-so-good.
@stretchneck
There's no replacement for displacement.
More MOSFETs won't help, higher quiescent current above a certain point won't help, and higher supply rails (above +/- 25V DC, especially if left unregulated) are not advisable.
With no simple option to move the class A (single-ended) transfer point higher, AND with no option to get more voltage gain out of input JFET... what other option is there, but to increase the displacement?
Nevertheless, Aleph J in its original form, with 2 out of 3 crucial elements maximised/optimised (the voltage rails and the quiescent current, where the 3rd element - the voltage gain... is what it is...), will drive 4-ohm speakers with 0 issues, especially if a little tender loving care (and a few mods) are applied. Give it a go.
Paralleling of modules will help... However, you'll have to stick to using the C1...
The main Aleph J thread is the source of information and knowledge. Read it, study it, build the amp, play with it... and then look at the circuit diagram again.... this will bring a new set of ideas... it takes time, but the results are oh-so-good.
well, as always, best solution is never ideal, it is nothing more than bunch of compromises
another way is using depletion mosfet for intermediate follower , then having issue of decreased OLG due to lower value of drain resistance at JFet
or regular Nchannel small mosfet, again - greater value of JFet drain resistance resulting in higher OLG, but that easily solved with simple RC based level shifter cell; that being decided as issue or not, only after measuring a prototype
another way is using depletion mosfet for intermediate follower , then having issue of decreased OLG due to lower value of drain resistance at JFet
or regular Nchannel small mosfet, again - greater value of JFet drain resistance resulting in higher OLG, but that easily solved with simple RC based level shifter cell; that being decided as issue or not, only after measuring a prototype
Great stuff! Love these posts.
So now I know that I should start by building an ultra stout Aleph J with low resistance/impedance everything , I can see how that goes and then either add a BA1 as a follower (since using an intermediate follower isn’t ideal in my case) or run Aleph J paralleled.
The only question left in my head now was those initial questions about paralleling Aleph J and any special considerations.
So now I know that I should start by building an ultra stout Aleph J with low resistance/impedance everything , I can see how that goes and then either add a BA1 as a follower (since using an intermediate follower isn’t ideal in my case) or run Aleph J paralleled.
The only question left in my head now was those initial questions about paralleling Aleph J and any special considerations.
You can also adjust the AC current gain of the Aleph current source.
The original has ~ 35% if my memory serves me correctly. This nets 25W into 8 ohms and 12 into 4 ohms.
By bumping it to 66% then you will get a lot more, closer to 50W into 4 ohms. It will be more push-pull-ish than a stock Aleph J since the current source contributes a lot more current but the voltage gain is still purely single ended. If I remember correctly the Aleph J X uses higher AC current gain.
66%, however, is quite high, but if you combine it with lowered rail voltages and increased current then bumping to say 50% is not a bad thing to gain more power.
The original has ~ 35% if my memory serves me correctly. This nets 25W into 8 ohms and 12 into 4 ohms.
By bumping it to 66% then you will get a lot more, closer to 50W into 4 ohms. It will be more push-pull-ish than a stock Aleph J since the current source contributes a lot more current but the voltage gain is still purely single ended. If I remember correctly the Aleph J X uses higher AC current gain.
66%, however, is quite high, but if you combine it with lowered rail voltages and increased current then bumping to say 50% is not a bad thing to gain more power.
Low impedance...
More output devices.
also drivers with enough current.
And of course a transformer that has the current.
Difficult loads are difficult loads.
Listening levels / speaker load need same voltage regardless.
nothing changed.
So dont see how weaseling the rail voltage will
help. Does technically limit current potential on paper.
But output voltage is voltage, load is a load
nothing changed.
more devices, more current
More output devices.
also drivers with enough current.
And of course a transformer that has the current.
Difficult loads are difficult loads.
Listening levels / speaker load need same voltage regardless.
nothing changed.
So dont see how weaseling the rail voltage will
help. Does technically limit current potential on paper.
But output voltage is voltage, load is a load
nothing changed.
more devices, more current
The answers do exist on these pages, just well hidden in ~500 pages of posts... wouldn't it be great if there was some sort of forum AI that would summarise and provide key links on the front of each topic 
So, biasing - Aleph J can be biased upto 2.5A, but in reality 400mV is deemed relatively toasty for a 4U modushop. So some fan assistance / CPU coolers / heat pipes are in order.
I just need to double the standard bias, like what Eric did. 200w idle is probably, 50w into 8 ohms and 25w into 4ohms class A which satisfies my power needs. Easy and cheap, if you ignore the electricity bill.
So, biasing - Aleph J can be biased upto 2.5A, but in reality 400mV is deemed relatively toasty for a 4U modushop. So some fan assistance / CPU coolers / heat pipes are in order.
I just need to double the standard bias, like what Eric did. 200w idle is probably, 50w into 8 ohms and 25w into 4ohms class A which satisfies my power needs. Easy and cheap, if you ignore the electricity bill.
Absolute and necessary reaction needed!I have an interesting variation which I developed 20 years ago, not previously disclosed.
I think it will make an interesting DIY project, so I have placed on the to-do list.
I call it UltraMu.
