Q5 and Q7 is not really strange, but a fairly common push-pull stage, a voltage follower, a collector circuit in push-pull.
Since the entire ensemble with Q6 and Q8 also has to work as a level shifter, i.e. there is a DC and AC coupling, the dimensioning of the components (their exact operating values) is not quite trivial, but in the end it is.
I have deliberately dispensed with frequency response compensation of any kind in /at this (working) state.
Without a test circuit, a laboratory setup - I can't go any further. For me, the concept has been theoretically chewed over from beginning to end, now the next step is the test setup - afterwards I'll know exactly what to do, not before!
Since the entire ensemble with Q6 and Q8 also has to work as a level shifter, i.e. there is a DC and AC coupling, the dimensioning of the components (their exact operating values) is not quite trivial, but in the end it is.
I have deliberately dispensed with frequency response compensation of any kind in /at this (working) state.
Without a test circuit, a laboratory setup - I can't go any further. For me, the concept has been theoretically chewed over from beginning to end, now the next step is the test setup - afterwards I'll know exactly what to do, not before!
Oh dear, does that basically mean that the diamond configuration promises us no advantages?
I'm just thinking about Steve Dunlap's KRILL again.
In other words: This configuration causes no damage, or rather the NFB corrects the damage and straightens everything out again. According to the motto: crooked times /multiply crooked is approximately straight.
Can we trust MultiCap? The results always look so incredibly good.
If you add a constant or multiply by a constant a non linear function the result will still be the same non linear
function, wether you drive the power transistors by diamond s buffer or by a classical same polarity EF it s the
output transistor that will set the score, namely its output current is a function of the exponential of its Vbe.
function, wether you drive the power transistors by diamond s buffer or by a classical same polarity EF it s the
output transistor that will set the score, namely its output current is a function of the exponential of its Vbe.
I know that 'wahab', I am also very familiar with this situation and the basic connections. I have said and thus expressed something different to you.
But that is not a reason for dispute or argue 😉.
Let's just be happy about the very practical (far away) resistance values of Bernard's design.
HBt.
But that is not a reason for dispute or argue 😉.
Let's just be happy about the very practical (far away) resistance values of Bernard's design.
HBt.
Dear wahab,
are you resting on your laurels or can I persuade you to make concrete suggestions and values?
As we are in the hobby sector, that would only be fair.
So you said the “Blameless amp from Self” has proper limitations and for example a BC140-16 or BC160-16 would lift the design, as well as the use of BC556 & BC546 (B-types probably) ...
Would you like to explain this?
kindly,
HBt.
are you resting on your laurels or can I persuade you to make concrete suggestions and values?
As we are in the hobby sector, that would only be fair.
So you said the “Blameless amp from Self” has proper limitations and for example a BC140-16 or BC160-16 would lift the design, as well as the use of BC556 & BC546 (B-types probably) ...
Would you like to explain this?
kindly,
HBt.
The result was bad, frontend was bad, so output stage did not matter.
This one looks better in sim:
Class A, Bias 1A
I did not try standard EF here.
Have not cared for hf compensation.
Last edited:
I would think that Bernhard s schematic above is close to a good implementation, only thing is to add an active CCS
for the differential, and see if a VAS driven by an EF is not better than a cascode.
Also for a non cascoded bipolar input differentals BC546A/BC556A are better than the B type because the early effect
is less pronounced for the formers.
BC141/161 are to be prefered to BC639/640 if the voltage allow as they are T039 and more rugged thermaly but
they both are good enough as VAS and even as drivers for the TO39, just imagine that circa 1974 Dual made
the mistake to use 2N2904/2218A as drivers in a low power amp with BC141-10/161-10 in the regulated PSU
while they would have been much better as drivers than the 2N.
Wahab, not to forget that I made this circuit in one day
There are many things that could be done, but will it ever be built ?
I put an EF in front of the VAS, 3rd harmonic down another 20dB.
From the pure numbers, high power wins a bit, low power loses a bit, but I also reduced cascode zener voltages...
Output power very clean 20 Wrms, "ok" a good 25 Wrms.
Last edited:
I am at a loss for words, almost speechless, I request that the fourth Michelin star be awarded to Bernard (for the first - and only - time). Munich is always worth a trip, however rocky it may be.
Dear Bernard,Wahab, not to forget that I made this circuit in one day
please allow yourself a few more days to reflect on your morphed proposal - I think that is necessary. Because you had better not confront your client with this proposal. Perhaps the first thing you should do is to cut out the red pencil and pay attention to the availability of components. You might also switch back to B operation (we Germans traditionally call it AB). And so on.
Well, to be honest, your current proposal will definitely not even see a test setup - it will fail in the first team meeting.
Well, your dimensioning (and topology) needs it under all circumstances.
Likewise, of course, that's how it's done, that's the standard.
And your suggestion leaves the A mode at the latest at 16Wrms at static 8Ohm loads & 8Wrms at static 4Ohm loads.
What trick does your quick suggestion use to keep the quiescent current of 1Amps constant?
#
Please don't take offense, but your fourth star will take a lot of convincing.
kind regards,
HBt.
Excuse me wahab,
but the “early” objection no longer works for me - for me it's nothing more than a rhetorical evasion.
Just say that you prefer h22e, i.e. an rce of 55k555 ohms, to 16k666 ohms ...
I have an original, old Siemens data book in front of me.
#
In plain language (no attack),
Your explanations are not enough for me personally. But now the thread is easily drifting into a turbulent and controversial star chef number - and I don't want that.
I reach into the shelf and open the Leitz folder and in front of me is an original, old service manual -> with an amazingly similar approach, almost 1 to 1.
Coincidences do exist.
Coincidences do exist.
Looks like designed with the simulator at a fixed supply voltage.
Increase supply voltage several volts and that thingy will explode.
A clueless design.
Spice simulations help improve basic understanding.
They can NOT replace it.
Last edited:
The L.Stellema concept shows strengths in the idea, but weaknesses in the approach and implementation. I try to put myself in the mindset of the author - and to respect the resulting direction, the route. So it can only be a matter of mitigating recognized weaknesses and emphasizing the strengths of the concept or philosophy.
In the next step, only a laboratory set-up will be able to show how far I will get with my initial thoughts and a few simulations.
There is simply no way around it.
In the next step, only a laboratory set-up will be able to show how far I will get with my initial thoughts and a few simulations.
There is simply no way around it.
A final note:
I am aware that I do not reveal all my thoughts in a draft and that I take many points for granted and self-explanatory, this also includes existing and obvious imponderables of operational safety ...
HBt.
I am aware that I do not reveal all my thoughts in a draft and that I take many points for granted and self-explanatory, this also includes existing and obvious imponderables of operational safety ...
HBt.
Are you interested in the L.Stellema design, a real-life 2024 new version?
(Then I suggest retaining the three assemblies IPS, VAS, OPS as far as possible.)
#195 refers to #191