♫♪ My little cheap Circlophone© ♫♪

[Elvee]

Originally Posted by padamiecki

Dear all,
please remind me how to use asc file to get from it working schematic in ltspice, my ltspice reads it as txt file

You right-click on it, choose "save as", and edit out the .txt (one of many methods possible).

Single ended with the offset correction of an LTP:

Q2 is in a CFP, and operates at constant current.
Q1 sees low impedance tail strongly following Q2.
Q1 gets full amplification rather than usual half...
Q1 and Q2 balance at same current and voltage

.

Clever scheme.

I find that C4 can be removed completely.

If you want to go even further in that direction (linearizing the LTP), another method is to use a crossquad.
The beast is difficult to tame, but it's worth the trouble:
THD @1W 1KHz can easily be pushed deep below the ppm threshold

Other than thermal backbond and fuses, can't figure a benign
way to soft limit current into a shorted load.

That wouldn't be very difficult by inserting current-sensing resistors in the supplies of the output stage, and it would interfere very little with the normal functionning of the amp.

[padamiecki]

Originally Posted by Elvee

You right-click on it, choose "save as", and edit out the .txt (one of many methods possible).

...cannot make this in Win7,
txt extension cannot be deleted...

[Elvee]

Originally Posted by padamiecki

...cannot make this in Win7,
txt extension cannot be deleted...

Then save it complete with the .txt extension, and the come back to it, and edit the filename to remove the .txt.

[padamiecki]

Originally Posted by Elvee

Then save it complete with the .txt extension, and the come back to it, and edit the filename to remove the .txt.

as above,
cannot make this...
there is no txt extension in the name, but the file has txt in the properties...
in XP I was able to make it, but under Vista or W7 impossible... or I do not know how...
I will try to make this in win commander or similar,
thanks
PS Total commander has option to change the extension, I solved the problem, sorry for offtopic

[powerflux]

i've read the whole topic but i feel like i still missing something.

what about the thermal effects and biasing? normally, the biasing transistor and/or diodes or whatever should be on the same heatsink as the output BJT's. in the PCB layout posted before they weren't even close.

Originally Posted by Elvee

requires no adjustment, no matching, no thermal compensation and yet has a rock-solid thermal stability

maybe i don't understand the circuit correctly (i'm fine with the other claims, there are more simple amplifiers like the ESP P3a that don't need matching or adjustment to work forever), but how does this stay thermally stable?

[Elvee]

Originally Posted by powerflux

i've read the whole topic but i feel like i still missing something.

what about the thermal effects and biasing? normally, the biasing transistor and/or diodes or whatever should be on the same heatsink as the output BJT's. in the PCB layout posted before they weren't even close.

maybe i don't understand the circuit correctly, but how does this stay thermally stable?

You are indeed missing something: the whole point of the Circlophone topology is the complete riddance of thermal problems, compensation, etc:

The servo stabilizes the quiescent current, but it goes further: it actively controls the output stage so that both sides always remain active in a "warm" class AB, having most of the attributes of the full class A, including a complete absence of crossover artifacts.

It is extremely easy to build, offers excellent performance even with 2N3055's, requires no adjustment, no matching, no thermal compensation and yet has a rock-solid thermal stability.

The circuit is completely under servo control, thanks to the current-sensing resistors, R8 and R11 and their associated circuitry.
The servoing action controls the quiescent current, without needing such messy things as Vbe multpliers or adjustable resistors, but it also goes further, and forces the output stage to remain in class A under large signal conditions.

in the PCB layout posted before they weren't even close.

Which demonstrates the concept does indeed work as advertised: ask odysseybmx

[powerflux]

Originally Posted by Elvee

The circuit is completely under servo control, thanks to the current-sensing resistors, R8 and R11 and their associated circuitry.
The servoing action controls the quiescent current, without needing such messy things as Vbe multpliers or adjustable resistors, but it also goes further, and forces the output stage to remain in class A under large signal conditions.

i understand now, i missed what to look for. the 'servo' is quite big (relative to the rest)

thank you. i will build this amp in the following weeks. i've got a +/-32V SMPS from a Linn integrated system that has been lying around for months waiting for something interesting; and that has just been found

[kenpeter]

Originally Posted by powerflux

i've read the whole topic but i feel like i still missing something.

what about the thermal effects and biasing? normally, the biasing transistor and/or diodes or whatever should be on the same heatsink as the output BJT's. in the PCB layout posted before they weren't even close.



maybe i don't understand the circuit correctly (i'm fine with the other claims, there are more simple amplifiers like the ESP P3a that don't need matching or adjustment to work forever), but how does this stay thermally stable?

In my case: the SRPP biasing transistor should be on same sink as close
to Q8 as possible. Q8 gets a little hotter than its mate because it gets
no help from base to emitter current leak. Schottky diode stack should
mount to a separate sink. When SRPP transistor gets heated, but diode
voltage drops are cooler ambient, bias is shunted away from JLH splitter.

Total bias available for sharing by both output transistors is modulated.
This is where it behaves like the circlophone, yet completely different.

I would bolt collector of hottest transistor Q8 directly to live sink, with
no electrical insulating pad, only grease. Q6 and Q7 would require pads.
If you prefer to drill a TO-18 hole and stuff Q6 into it, then I guess Q8
would need a pad too. Only one transistor can be in direct contact....

[padamiecki]

Originally Posted by Elvee

If you want to go even further in that direction (linearizing the LTP), another method is to use a crossquad.
The beast is difficult to tame, but it's worth the trouble:
THD @1W 1KHz can easily be pushed deep below the ppm threshold

unbeliavle how important are inductors and 1.5 ohm resistors

[kenpeter]

TO-18? What was I thinking? They make 2n2222 in TO-92 plastic too.
Much better that Q8 be the one to have metal to metal contact...