Aleph-X builder's thread.

The decision needs to be made whether to base your design on 1) what Nelson states clearly and unambiguously, 2) what Nelson states ambiguously, and 3) what others come up with as supposition and through the use of tenuous logic. I think there is too much of 2) and 3) going on. Just MHO.

I really don't want to rain on the parade but there should be more of a focus on essentials rather than minutiae. My comments should be taken as constructive advice.

Graeme
 
Relative DC offset change...

Relative DC offset changed to 100mV from no transformer at input 50mV.
Why this happen?
I want more gain...so I changed input resister 10K to 100 ohms.
The result was increased relative DC offset to 1V.
I dont knnow why.
 

Attachments

  • input.jpg
    input.jpg
    70.3 KB · Views: 1,117
Nelson Pass said:

It's really pretty simple. By driving the center point of the bias
voltage, the front end can effectively swing about 4 more volts
on each peak. The resistors from each supply function to provide
current to the bias voltage generator and load the front
end with a known resistance.

On a circuit like this, that 4 extra volts can be the difference
between 100 watts and 144 watts.

Also, keep in mind that the output devices want some extra
volts left over between Drain and Source on peaks, something
on the order of 5 volts or so.

It occurs to me that this is only necessary if a single supply is used for both FE and output stage. Is there a sonic advantage to using a single supply (as opposed to other considerations such as lower cost, weight and complexity)?

Ian.
 
Nelson Pass said:

Also, keep in mind that the output devices want some extra
volts left over between Drain and Source on peaks, something
on the order of 5 volts or so.

I forget to ask about this in my last post...

If we allow an additional 5 volts for the output MOSFETs, doesn't this mean that if the amplifier is driven into clipping (ignoring for a moment the desirability of doing this), it is the FE that will run out of voltage first? Is this a better scenario that having the output clip? Or is it simply a case of clipping is bad however it occurs?

Ian.
 
Fair point - clipping is clearly something to avoid anyway.

I'm still curious though - in your experience is there a sonic or performance advantage to using a single supply (as opposed to other considerations such as lower cost, weight and complexity)? I can see there is no direct need for a seperate supply if 5v is held in reserve at the output, but migh not a higher FE supply provide greater linearity? Or maybe this effect is swamped by non-linearity in the output.

Ian.
 
The one and only
Joined 2001
Paid Member
It depends on the design. Some circuits, like the Aleph 1-5 and
30 and J have to retain the same rails.

But as a rule, it doesn't hurt to have some extra volts on the
front end. The question is only whether they can really take
advantage of it. If you have to figure on throwing away 5 volts
for the output devices, then you only need to deliver Gate voltages
within a volt of supply. In our case, we drive the center point
of the bias circuit, so our front end only needs to swing within
5 volts of supply. It's still pretty linear at that point, and in truth
we make larger allowances than that.
 
Hallo,

I am doing the UGS boards at the moment.
I would like to ask if there is an (sonic) advantage to select the 2SK389BL and 2SJ109BL to each other for getting the same gate source voltage on n-side and p-side?


Ian, it could be a good idea of you to manage the stand by bias voltage by shunting the bias control resistor.
......I don't know if the front end is shut down - I interpret it from post # 1451 of Mr Pass. Also some pictures of X and X.5 amps look a bit like this approach.


Regards
Dirk
 
Thanks!!!

Then I will try to do that first - have disconnected my soldering iron...
The problem with this dual jfets is the stupid pin out. Have no socket or connector for it. TO220 fets are a lot more easier to match. I use a nice connector with the right pitch, but the jfets don't fit... :-(

Mr Pass, what is the reason / advantage that you choose 475R (think it is - can't see the exact colors) instead of the usual 221R at the gates of the TO247s in the new models?

Dirk
 
noisefree said:

Ian, it could be a good idea of you to manage the stand by bias voltage by shunting the bias control resistor.
......I don't know if the front end is shut down - I interpret it from post # 1451 of Mr Pass. Also some pictures of X and X.5 amps look a bit like this approach.


By shunting the bias resistor, do you mean shorting the upper of the bias control resistors? If so, then I would agree that this will shut down the bias generator which is part of the equation for implementing standby. With the bias generator disabled (i.e. little or no voltage drop across it) the output will float at around -4v absolute due to Vgs of the upper output devices, i.e. those that are connected to the positive supply. These will be biased on by the SE bias resistors as the gates will be at roughly zero due to the potential divider formed by the 10k resistors connected to the supplies. Something like a resistor in parallel with the 10K to the positive supply will be required to lift the gates up to +4v in order to arrange for zero at the output.

With the output stage configured in this way, it will behave as a low power SE source follower and hence will still amplify any input signal which is not particularly desirable in standby. Switching off the FE supply will help significantly although some portion of any input signal will still find its way to the output via the feedback resistor. Probably not an issue in practice since there will be very limited voltage or power but it may be easier to short the input instead.

I’ve no idea how Nelson does it but I am planning on going with something like this approach.

Ian.
 
Hi steenoe,

thanks a lot, really great tip!!!

Had forgotten that I had used this sockets in my first matching tests for two years. The jfets fit very well in them, but for the wide TO220 legs I prefer my ???-connector, which also allows me to put them on a 50°C hot heatsink for faster and better matching.
OK, all jfets are matched and two nice pairs soldered in...
UGS ready!
I'm curious about to power them up...


Hi Ian,

have wrongly interpreted your idea. Thought you mean switching a resistor in parallel to the cathode-reference-pin resistor for getting different bias voltages.
There has to be an easy trick to handle the stand by. An 30.5 has four bias generators but only one relay with dual contacts. So, switching four resistors in the VBgenerator or four inputs is not possible.
Also I hope to get a tip about the output gate resistors.
Why 475(?)R ? Why 150R in a F4?


Dirk
 
noisefree said:
Hi Ian,

have wrongly interpreted your idea. Thought you mean switching a resistor in parallel to the cathode-reference-pin resistor for getting different bias voltages.
There has to be an easy trick to handle the stand by. An 30.5 has four bias generators but only one relay with dual contacts. So, switching four resistors in the VBgenerator or four inputs is not possible.
Also I hope to get a tip about the output gate resistors.
Why 475(?)R ? Why 150R in a F4?


Proabbly my fault, words are a poor way of conveying intent with respect to circuits. I'll post a schematic of my current thinking when I have a spare moment.

One relay with dual contacts for four bias generators certainly constrains what can be done. I have some ideas but will need to think about this a bit more.

I agree a tip about the gate stoppers would be nice but I suspect you will get the standard answer about it being a compromise between HF response and stability. If 475R is used, then I would guess that this is to improve stability or perhaps remove a spike at around 1MHz, rather than resorting to frequency compensation by paralleling the feedback resistor with a capacitor. I think Nelson said explicitly that he doesn't do the latter in the AX.5. 150R is used in the F4 because the JFETs cannot provide much current to charge the output MOSFET capacitances and hence a lower value is required to get the desired HF response.

Ian.
 
Hi guys,

I have a short question about the power supply in a BZLS.
It has nothing to do with this thread but don't wanne open a new one for this stupid question:

Is it allowed to use a single transformer with two separate equal secondary windings and two bridge rectifiers soldered in?
It is connected in the same way as in the original only with the difference that the secondaries are coming from one transformer instead of two.

The normal way to create a symmetric voltage with one transformer would be to connect the two secondaries in series and use the middle as ground.

(Had used this BZLS board before with a self build board which packed discret diodes and gyrators (for dropping down the dissipation of the regulators a bit), but wanne throw it out and use it in the original constellation.)

Regards
Dirk
 
Official Court Jester
Joined 2003
Paid Member
noisefree said:
Hi guys,

I have a short question about the power supply in a BZLS.
It has nothing to do with this thread but don't wanne open a new one for this stupid question:

Is it allowed to use a single transformer with two separate equal secondary windings and two bridge rectifiers soldered in?
It is connected in the same way as in the original only with the difference that the secondaries are coming from one transformer instead of two.

The normal way to create a symmetric voltage with one transformer would be to connect the two secondaries in series and use the middle as ground.

(Had used this BZLS board before with a self build board which packed discret diodes and gyrators (for dropping down the dissipation of the regulators a bit), but wanne throw it out and use it in the original constellation.)

Regards
Dirk

two secs , each one with own graetz , and ground made between minus of one graetz and plus of second graetz ?

if that's the case - sounds good to me :clown: