Discrete LM3875 ?

[ash_dac]

Hi,

Has anyone built a discrete LM3875 amp to compare to the chip version?

I was considering using 2SC5200 / 2SA1943 output transistors and BC547 input transistors. I'm not exactly sure what transistors I would use for the cascode, and driver section.

[Leolabs]

Think that is a simplified diagram.

[ash_dac]

Originally posted by Leolabs
Think that is a simplified diagram.

I'ts detailed as an equivalent schematic (excluding active protection circuitry)

What else is simplified other than the current sources ?

[Onra]

Patents, may be

Regards,
Onra

[rabbitz]

There's some info on this in the Chip Amps forum and I think the result was not as good if I remember correctly.

[Tony]

i suppose you can discrete wire an equivalent, but then you can not hope to capture the feature for which the chip was built in the first place, that is integration, and compactness...

[destroyer X]

And this will reduces the capacitances you normally have inside chip constructions.

This will produce a better sound, as you can change a lot of compensation capacitor values...those used to avoid oscilations....they are also sound killers.

I would be happy to see some discrete design, as those NPN output transistors are nice because you can use 2 units with the same characteristics, when NPN and PNP are never the same.

If someone has a detailed, complete schematic, please, let me know, as i have strong beliefs that some discrete will work better than the chip, and i will be happy to confirm my beliefs.

regards,

Carlos

[ash_dac]

Hi,

I have put the simplified circuit into LT Spice.

I am a little confused about the current mirror in the cascode. Why is there a npn transistor there ?

[Luke]

Isnt the main benefit of gainclones the small signal path? surely this benefit will be lost in a discrete design. still very interested to hear someones experiences. Then maybe we can wack on a t net and valve buffer stage, maybe even a snuber supply

[destroyer X]

Small distances inside can produce the capture of signals because of electromagnetic energies produced when an electric signal travels into wiring, or chip interconection pins and wires, also the waffer internal construction.... wires together can be understood as two capacitor plates...having small capacitances and tuning the circuit to some megahertz...also coupling can exist without the oscilation.....because too small input to output leads distance...at least 1 inch is needed to keep them safely apart, some sensitive circuits use to start oscilations with the base metal lead.....reason why some scopes show in the screen some spurious, many of them related the internal reference oscilators, others are beeing captured by radio transmitters that we use to have in our homes....electronic baby sitter, wireless telephones, automated garage doors, elevators machine controls, computers and many things produce signals that can be captured.

To avoid those possible oscilations cause inside chips, some capacitors are used, some inside the chip and some bigger outside and some of them represents enormous troats to eat oscilations, to send em to earth.

All those things goes slowing the entire system...reason why those chips are strange in the high end, or, at least, do not have impressive high end reproduction.

Well.....i would be very happy to discover that i am wrong or rigth...do not matter to me if i am a good analist or not...i want to KNOW!... to finish with my own doubts....only doing the LM3875 in a discrete way.... and you man, are near to make this possible to me.

Also i think that there are a lot of circuits inside the chip that they do not publish.... the protections, the automatic self adjustments...those things we cannot find in the published schematics, and they are some of the sonic destroyers i suppose...... so...doing the whole thing without the secrets, we will have a better reproduction...OR NOT?

regards,

Carlos