Please comment "my" design

[TroelsM]

Hi all.

I've been working on this design for a while and I would like any comment's you've got.

I guess that the schematic is know material to most of you and the only odd thning is the VI-limiter around Q1, Q12, Q14 and Q25 and the components around them.

The supply for the output (Vcc36/Vss36) is a 620W torroid. The 42V and 12V supply´s are regulated.

I'll build 4 of these (lot of components, -I know..) and bridge them to make at stereo amp of approx. 2*175W

The prototype (without the VI-limiter) is working fine and is very stable. I removed the input filter and feed the amp with a squarewave and there where no signs of overshoot or peaks onthe output.

I've got a complete PSpice model of the amp if any of you want it?

Most of the design is taken from: "Introduction to electroacoustic & Audio amplifier design" by Leach, and from schematics found on the www

The things I´m wondering about:
-What difference would base-resistors in the diff.-amp do?
-How do I calculate the resistor values around the output-stage?

More to come. I hope to make PCB's in a week or so.

TroelsM

[hienrich]

hi TroelsM,

theres no problem with your work
the output stage is just a bit messy. why not making it the driver stage and with the output devices paralleled.


but this thing will work, if you want your output stage to be that way
then no problem
one more thing
you need not have two supply rails, for the VAS and the output stage respectively
these type of supply rails are recommended for FET output stages mostly.



nice job,

hienrich

[TroelsM]

Hi.

The many driver transistors are there to distribute the load of the TIP's. I dont know if it's necessary. Maybe I'll remove some of them and use 1 driver for all 3 TIP's.
I can't afford to much space for cooling of the drivers and I guess that 3 drivers will be able to get rid of the heat better than 1 would?
How will the current-sharing between the TIP's be affected by 3 or 1 driver? I guess that 1 driver is the best solution, but what do you think?
I was hoping that I did'nt have to measure the Hfe og all the transistors, but will it work?

The higher supply for the pre-stage is not necessary, but i'ts my understanding that i'ts the best solution for a number of reasons.

TroelsM

[deleveld]

Your design look fairly reasonable but there are a couple of unusual points I noticed:

Your output base resistors (R11 etc) are quite low and your drivers are working fairly hard to get the outputs into conduction. Using low values here help output device turn-off but if you are using TIP devices (they not fast) then I dont know why you need 33 ohms there. You have slow output devices and then you use a technique to make them as fast as possible, but you take the disadvantages of the technique. If you are worried about output stage bandwidth then dont use the TIP devices. I dont think output stage bandwidth should be a real consideration considering the inductor at the output. It is a lowpass filter anyway. Personally, I'd use a 100 ohms there (placed around the emitter resistors) and have 1 driver (maybe a different type) for all 3 output devices. Less pieces to worry about...

I think R4 and R50 dont serve any purpose at all. I know it looks 'better' but it makes no difference. If i remember correctly the Douglas Self site mentions this somewhere.

Personally I dont understand the fascination many people have with these 'double LTPl' input stages. I think a normal LTP with a current mirror is really the best. But I admit the double LTP stages look nice on schematic.

Using multiple supplies for 1 channel amp always seems to me to be bad design. Better to redesign the amp to use 1 supply. Its a very expensive way to get an amp that is just a little bit easier to design.

These are just some things I noticed. The design looks good and will probably work just fine.

Doug Eleveld

[deleveld]

Oh yeah and your comment about driver emitter resistors: You dont need to match output devices if you use only1 driver. If you place the emitter resistors around the output emitter resistors. The voltage drop over the output emitter resistors helps balance output device current for different output device gains.

Doug Eleveld

[destroyer X]

But tell me, that's the question appeared into my mind.

How it sounds.... in comparison with other known sound you have or used to compare.

When i see those double simetrical...i always imagine some 6 dB reduction of bass.

That's the reason of the question.

related the circuit...the darlingtons you constructed was great...also the heat distribution in many output transistors were clever.

But tell me...the sound...please.

nanabrother@yahoo.com

Carlos

[AndrewT]

Hi,
the current protection on the VAS also limits dissipation on the output protection. You can change Q12&14 to fast, low power types. Q12 & 14 also require stopper diodes in series from the drive lines (collectors of q11 &15)
I don't know what D5&6 are doing?
I have seen C4 & C5 replaced with a single cap NOT connected to the output. You should leave lots of room and extra holes in the pcb for different sizes of C4&5 to change the time constant of the VI protection. You may want to go as high as 100uF to allow peaks to pass through without limiting.
I agree the idea to ditch the integrated darlingtons.
regards Andrew T.

[darkfenriz]

hi TroelsM
i like it very much
including r4&r50- they equalize R*Cmiller

including separete rails-they always improve sound and you can get rid of "dual mono" and run normal stereo

including separete drivers-they work at 3 times lower currents compared to one driver, it is good with more output bjts to be driven by lower power (read:faster and higher beta) drivers

the only thing I am wondering about is frequency compensation..
enough??

regards

[darkfenriz]

one more thing:
do you use regulated supply fo VAS?
I think it is worth doing and easy even with LM317/337

[uli]

Hi TroelsM,

IMHO the time constant of the DC-servo is too low. Its just
100ms which is 1.6 Hz. Better use 470n caps.
The TL071 is not the best choice for servocircuits, its a poor,
outdated device. I would use a fetopamp with lower offset
voltage.

Its a pleasing design though.

Uli