R21? and Current Source AC Current Gain

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Call this new old ground if you like,

But I recently measured the in/out effect of R21 (Aleph 5) on the AC voltage across the negative source resisters R64-66 in the negative half of the amp and got about 110mv with R21 in and about 387mv with R21 out in one channel and 414mv in the other.

(This assumes 10 volts across an 8 ohm load as per Mr Pass previous thread).

My understanding is that R21 (453R) in circuit should halve the AC voltage across R64-66 so the current source is doing half the work.

I then checked my board and everying and all appears to be in order, the supplies are 35 volts and the DC bias voltage is about 625 mv across each of R64-66.

The only difference to the published design is IRF240's and an NTE substitute for the MPSA18.

Can any one explain why I am getting this wide variation outside specification, and what effect it will have on the performance of the amp?.

We anyway I put in some muliturn trim pots in and tweeked the Voltage of R64-66 for 50% . The value of R21 is now about 600R.

I have played a few tracks and blow me down I think its better!

Has any one had similar experiences?

best regards

I've been tweaking the resistor values in the R&D Aleph-X circuit, and have been playing exactly the same games. I don't remember exact numbers, but seem to recall similar experiences with my Alephs and the Mini-A.
In my case, I'm using IRF644s for the Alephs and in the R&D Aleph-X. I requested new, contiguous tubes of parts when I ordered; what I received was anything but. There's a wide difference in the Vgs in the devices and that has an unfortunate effect on predictability of the circuit's behavior. I believe that there are differences between the IRF644s and the stock devices, which would also account for more variance. To make life even more interesting, my water-cooled heatsinks keep the outputs much, much cooler than the production Alephs, so bias and such were also effected by a different thermal environment. (Contrary to popular opinion, MOSFETs are sensitive to temperature.)
The short version of what happens is that the amp won't develop full current into the load if the current souce is a bit shy. Premature clipping will result when the amp current limits. Once you balance everything out you'll have a better sounding amp.

R21 and Current Source AC Current Gain


Thanks for your experiences, its good to see some else has fiddled with these bits.


I wonder if in my case with originally nearly 1/2 the expected voltage across R64-66 (ie 25% not 50%) the current source was doing more than its fair share of the AC output current or is it the other way around?.

The IRF 240 fets I selected from my batch a group which had a VGS of about 3.73 on average, but the variation of the overall population went up to 5.85 at the extremes.

So as you suggest I can well image the effect this may have overall on overall circuit behaviour.

I expect in the production Aleph, the fets were taken from batch quantities within tolerance and this would therefore never by an issue.

The Aleph 5 sounds silky clean and as smooth as velvet.

It will be interesting to see how my Aleph 2 measures up.



PS You may recall I purchased a pile of 15,000 uf 63 volt capacitors from a trader on Ebay. They arrived in mint condition unused! (total US$40 for qty 70)

I just tried them out in the stereo Aleph 5 with 120,000 uf (4+4 in parrellel), there is noticably less hum. The Aleph 2 will be interesting with 120,000 per mono block (I have thermistors handy)
Hi Macka,

I'm probably a week or so behind you. Just finished my Aleph5.
I didn't change R21 yet but played with R19 to get the bias right (2.8A was a bit too much for my heatsinks). I tried to understand the way r19 influences the bias and came to the conclusion that the hfe of the transistor (Q5 I believe) is quite important. The lower this gain is the higher the bias current will be for a given constellation because the voltage drop over R15 (1k) will be higher. I had to reduce R19 to 47k (from 221k) to get the bias down to 2.1A.
I didn't use a ztx450 but a BC550C wich could be the cause.

Another thing I don't understand yet is that when I calculate the current source gain with the formula given in the patent I only get 25% for both Aleph 5 and 4. Nelson's shure there's a mistake somewhere........

R21 and Current Source AC Current Gain



I just answer your more recent post before reading your reply in this post.

I now understand what is causing the higher bias in your Aleph 5... the BC550C.

I have asked Nelson the same question about the effect of R21 and the measured value (I get only 25% with R21 in circuit also).

He indicating the gain of the circuit is slightly higher than 50% to compensate for the slightly lower than optimal bias and if you crank your DC bias up a tad, this will work fine at 50%.

So I assume with the MPSA18 for Q5 which I now have in the circuit the DC bias was a bit lower than optimal for 60 watts/8 ohms and they raised the AC gain gain of the current source to compensate. (meaning it some how shares some of the work)

I'm still not sure where this leaves us diyers as the IRF244s in the production Aleph may also behave differently to the more readily available IRF240's which I'm using.

I have tried changing R21 to give 50% and the value works out to 660 ohms. If also changed the sound character in the subtle way, there is more pronounce high frequency detail and sounds a bit harder and less forgiving.

When R21 changed back to the nominated value of 453R the sound is more relaxed and in some ways easier on the ear with a warmer valve like charactor but with the precision and power of solid stage.

Without reading too much into this and assuming the diy version is functioning correctly it could be the value of R21 determined initally based on design formulae and was later chosen during development based circuit performance and on listening trials.

Please note this is only supposition as there are mostly likely a number of variable which are unknown.
R21 and Current Source AC Current Gain

I had the day off and did some more experiement with the current source.

Firstly, I put in the MPSA18, now both channels are identical in behaviour for DC and AC bias.

I measured the following AC current gain setting for several drive levels using a sine wave and 9 ohm load.

The values for R21 in /out where R21 =453R were:

Level Volts Voltage in / Voltage out for R21 Ratio for
Voltage on R64-66
8 0.039 / 0.188 0.20
10.2 0.24 / 0.222 0.24
15 0.10 / 0.334 0.29
17.6 0.123 / 0.390 0.31

This suggest the action of R4 comes into play at higher current levels as explained in the Patent.

However the 8 volt level suggests the negative drive side is only sharing 20% and the current source 80%.

Does that sound right?
This is a bit puzzeling as I am not sure what value or level is meant to resemble theoretical current gain.

Does anyone have any ideas?

My calculation of the current gain setting using the schematic values is :

Aleph 5 current gain equals = 0.117 * 1000/ 0.333*453 = 0.78
from the formula R4*R3/R1*R5 = current gain setting.

Does that sound correct?

Where R 4 = 0.47R/4
R3 = 1K
R1 = 1.0R/3
R5= 453R

I not sure if the above a definately correct so please use rubber bullets!

What all this really means God (and the One and Only) alone only knows!

My own feeling is that if we know whats definately meant to be happening then the DC bias and AC current gain setting can be manipulated quite easily.

best regards

Dynamic current duo

This looks like a job for all you spice modelers. Don't forget that the feedback circuit senses current into the load which means it is a function of the load impedance of the speaker. This is not a simple resistance. Also the gain of the dynamic current source is a function of the BJT beta and input impedance. Anybody for a good jfet diff pair instead of the common emmiter BJT here?

The one and only
Joined 2001
Paid Member
As I commented to macka privately, the formula ignores
the transconductance of the Mosfets in the output
stage, using a simplified assumption that the transconductance
is very high, and not dealing with parallel transistors.

This results in an apparently higher gain than found in the
real circuit.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.