50W amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
analog_sa said:
I am just curious: what is the motivation behind discussing this topology? Even a chip amp beats it into the ground both objectively and subjectively. And takes half an hour to build. Is it just a learning exercise?

Building chip amps is for beginner, who don't want to understand audio technology, only to get a sounding amplifier. Let's visit chipamp forum: many novices who made a chipamp without any knowledge.
I don't say that chips sounds bad. Not. Maybe they beats these old Lin-topology amps.
But hey, when you are going to design your own amplifier, you will need many knowledge. And you can't get this konowledge by soldering a few passive parts near a power-IC...
So with your words: it's a learning exercise.

On the margin: I hear many time past that chip amps will sweep away discrete amplifiers. And they aren't that successful.
See Cordell, Pass, Dean, etc... And the most of the audio manufacturers, the big firms. They aren't prefer chips.
Chipamp-rage can be misleading!

(Do you know AKSA-55 amplifier by Hugh Dean? That's derived from old Bailey-Lin schematic. Bootstrapped, old-looking circut with modern components. That beats chipamps into ground...)
 
edl said:



On the margin: I hear many time past that chip amps will sweep away discrete amplifiers. And they aren't that successful.
See Cordell, Pass, Dean, etc... And the most of the audio manufacturers, the big firms. They aren't prefer chips.
Chipamp-rage can be misleading!

(Do you know AKSA-55 amplifier by Hugh Dean? That's derived from old Bailey-Lin schematic. Bootstrapped, old-looking circut with modern components. That beats chipamps into ground...)



I know and i agree. Chip amps are mediocre and my comments only referred to the schematic as posted . As a learning exercise it's great.
 
Duo said:
If you use that method of current limiting with the two transistors, which is known as 'foldback current limiting', make absolutely sure to use the diodes across the output transistors from power supply to output as seen in the original schematic. This is because with a reactive load, such as a speaker, if the current limiting kicks in during music playback, the speaker can pull the output transistors into reverse bias and destroy them. The diodes simply drain this current into the power supply and prevent the transistors being cooked should such a thing happen.

It sounds really nasty when the current limitin kicks in, but it will save a power amp from shorted output conditions if designed and implemented properly.

Read this if you would like some in-depth information on the topic.

Testing the Limits
Thank you very much, Duo, that explains the presence of these two diodes (Q734 and Q736), I was also wondering what use it was to have them there.
I also understand the function of the small inductor with the 2.2 Ohm resistor in parallel at the ouput is to drive capacitive loads without oscillations, right?
And the small RC at the output is to prevent ringing?
 
Gigapod: You are correct. The RC network from output to ground is otherwise known as a Zobel network. If you analyze it mathematically for frequency response, you'll find that at some ultrasonic frequency, its impedance drops quickly and works to shunt high frequency current to ground. This can be very useful to prevent wideband amplifiers from oscillating.

Of course, the inductor and resistor in parallel with each other at the output have a rising impedance at ultrasonic frequency in order to decouple the amplifier from the load at such frequencies. The impedance of the inductor rises while the resistor becomes the coupling to the output; don't worry about this affecting audio performance, it should not have a significant effect until well above the audio spectrum.
 
My mods

Hi everyine thanks for all the tips. Well my mods are R11 reduced to 100 hm and a 100nF cap added across Q5.

Another question, would this topology be rugged enough to be upgraded to a 100W amp with

1) C546 replaced with MPSA42 and BC556 replaced with MPSA92
2) Paralled output stage using 2 TIP2955 and 2 TIP3055 with additionnal emitter resistors
3) the PSU upgraded to +- 42V

Your comments are welcome

thanks
 
- there is absolutely no need to replace BC546s to MPSA42. BC546 even performs here better.
- the safe operating area of that MJE2955 and MJE3055 is not enough for a 100W amp, even if two pairs paralelled. Choose something better device. Use a single pair of MJ15003/MJ15004.
- +/-42V idle rails are absolutely fine for 100W-RMS on 4 ohm.
 
Youd need about 50V rails for 100W into 8 ohm.

Q1 (VAS) would need to be MJE350 or similar.
Q6 should be MJE340.
Q7/Q10 (Drivers) would need to be MJE15032/33 or similar.

I would recommend using LED's as the current sink voltage references.

You would best using decent rugged output transistors such as MJ15003/MJ15004, MJ15024/25, or MJ21193/4. If you want to stick with flat pack devices, I would recommend MJL21193/4 devices.

Your amp is quite similar to Doug Self's Blameless design. I would recommend you look at that
 
new amp schematic

Hi thanks again for all the replies. I have posted a new schematic any comments are welcome.

Thanks
 

Attachments

  • newamp.gif
    newamp.gif
    20.4 KB · Views: 783
My suggestions:
- R1 and R2 change 33 to 100 ohm
- Q1 = BD139. 200mW dissipation is too much for MPSA92! BD139 is even faster. Q6 = BD140. BC456 wouldn't enjoy 0,2W...
- C6 = 2.2uF
- C1 will cause turnon-thump, remove that
- change R7 to trimmer-potentiometer + a resistor (in series)
- 47pF capacitor across feedback resistor can cause oscillation
- R4 is too low, DC offset will be high. Change it to 4.3k ohm

Most of my suggestions are already suggessted by Edl, haven't you read it?!
 
you don't need to run your bode plot from "DC to Daylight"....... all you need is to go to (and maybe a little beyond) your unity gain frequency. you're looking for peaking in the frequency response curve that reveals possible instability, as well as the frequency response curve itself.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.