OpAmp power amp, would this work?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
After reading about the Nelson Pass, Beast with a 1000 JFETs, and also seeing this multi paralleled Op-Amp power amp:

SoundStageXtreme.com | UltraAudio.com | Blue Circle Audio NSL Stereo Amplifier

it got me thinking.

The power supply is +/- 18V. Using the OPA2134PA can give +/- 35mA at the output, so we could safely use them at 30mA, so a 560ohm resistor is used on each op-amp output, giving you 18 / 560 = 32mA. If you used 50 dual op amps per channel, you could get a max of 3.5A (100 op-amps in parallel), that would be 3.5² x 4 (P = I² R) = 49W into 4ohm load.

Would this circuit work:

OPAMP_Amp.png


Here is a schematic I drew that has 4 op-amps in parallel. Each has a gain of 17.6 and an output resistor of 560ohms to limit the current. R1 and R2 are used at the front end to drop the voltage as the OPA2134 has in input voltage limit of +/- 0.7v. With a pk-pk input of 4v and the volume pot turned fully up, there is only +/- 0.66v at the input to the buffer.

Will this work? And if so, could I simply parallel a load more of the devices?
 
Last edited:
Administrator
Joined 2007
Paid Member
Check out Elektor (October 2010) for Doug Selfs NE5532 power amp using multiple devices. It used around 40 (in total) NE5532's.

Those 560 ohms can come right down in value the more opamps you use (by a factor of several hundred :))
 
Bridging aside, you don't get anywhere near 49 watts of non clipped sine wave with +/- 18v supplies and 4 ohm load. To simplify, lets say the amp can swing completely to the rails, therefore:
(18*(1/sqrt 2))^2/4 = 40.5 watts

With any practical amplifier, you get around 28 watts, maybe as much 35 with a design that optimizes the output swing.
 
OK. My maths was wrong, but 28watts is still a worthy amount of power :)

Just downloaded the Doug Selfs Op-Amplifier and must say it is an interesting read. The Op-Amps used are NE5532 which have an input voltage of +/- the supply voltage, instead of the +/- 0.7 volts that the OPA2134 has. From reading the article, the gain is done at the front end of the amp and all the parallel OpAmps are unity gain, obviously this dramatically cuts down on parts.

I notice from his schematics that only 1ohm resistors are used on the outputs of each OpAmps. I can see how this would work, but you are relying on the current limiting feature of the OPAmps, will this not damage them if you drive them to the limit? If you built this actual amp and had the audio playing and then removed one of the sections of OpAmps, you could suddenly run into overcurrent in the remaining OpAmps and they would start dropping out, what would happen? Would the audio slowly dwindle away, or would you get spikes and horrible noise from the speaker as each one dropped out?

Also, what is IC6 doing in the schematic?

Is the NE5532P as good as the NE5532?
 
On the datasheet for the NE5532 it says that the absolute max ratings are +/-22 V for the supply and 'Input voltage, either input(2) (3)' maximum is VCC±

Oh, my mistake, I just re-read the OPA2134 datasheet and that has a max input of +/-13.

I'm still thinking of having a go at building something like this and the NE5532s are way cheaper than the OPA2134, but are they as good?
 
Administrator
Joined 2007
Paid Member
IC6 looks like some "form" of DC servo. It's used to set the operating point of IC1. Doug must have had his reasons for specifying the OP177 (just ultra precise DC operating conditions I guess... for that its one of the best around) but it seems overkill.
 
Administrator
Joined 2007
Paid Member
:) Yes, the NE5532 has a wider supply range than the OPA2134 if you wanted to run the IC's to the limit.

The 5532 is fine as a "gain block" as here. A wild card could be the OPA2604 (-/+24 volt supply) and the TLE2072 which is slightly less at -/+19 volts but has the benefit of a much higher output current (60 or 80ma from memory)
 
The OPA2604 is more expensive and the TLE2072 is about the same price. Is there any audio quality difference? The extra voltage and output current is obviously a bonus?


I've been looking at prototype boards like this:

http://www.ebay.co.uk/itm/Copper-Pr...?pt=UK_Sound_Vision_Other&hash=item27c81aee5d

I thought I could build 4 or 5 parallel Op-Amps onto them and then stack several boards using header pins to connect them. That way you could add and remove boards at will.

The big question I am having now is the quality of sound that could be expected from an amp built entirely from OpAmps. I mean the build would be fun and the result would look pretty cool, but would it sound any good compared to my LM4780 chip-amp? My chip-amp sounds very nice, but I love experimenting and building different things.
 
Last edited:
I'm still thinking of having a go at building something like this and the
NE5532s are way cheaper than the OPA2134, but are they as good?

Hi,

A handful of chips are better than the 5532 for audio but way more expensive.

The way Self uses them in the 5532 power amplifier probably means
you'll get very little benefit using way more expensive chips in a
different topology, or trying using them in the same topology.

If you have enough op-amps and overdrive the amplifier it will clip
near the rail voltages. If you don't have enough op-amps for the
current it will clip at lower voltage swings somewhat differently.

I'm not sure why such low value current sharing resistors are used.
Pretty sure anything up to 10R or so would work just fine.

rgds, sreten.
 
Cheers Sreten. Is something like this worth building in your opinion. It sure would look cool with stacks of boards populated with OpAmps, but would it sound any good? It would be nice if it sounded as good as my headphone amp I built that uses a single OpAmp. I spose I need to design and build it....... I'm still working on my DAC/Pre Amp, will be finished eventually if I don't keep getting distracted with other projects :)
 
Hi,

I'm one of those people that thinks that basically only badly designed
amplifiers don't sound good, and the merits of decent designs hard
to judge. For sure the 5332 amplifier will have low distortion, but
as to sounding good I don't know, I've never heard one.

Seems to me more a case of it can be done, so here it is.
As usual with Self everything seems to be done properly

Certainly more practical than the "Beast with a thousand J-Fets".

rgds, sreten.
 
Last edited:
I presume the series resistors at the ouput of each opamp is to keep the opamps from dragging each other around, ie. isolation.

They do the same sort of thing with the LM3886 parallel designs, I presume others as well...

Btw, Gilbert also made some amps into womens shoes some years back for CES... :D

_-_-
 
I was looking at this circuit a couple of months ago. I have been trying out a few headphone amp circuits on a breadboard (gonna build one or three when I make my mind up) and I tried a 5532 in a master/slave buffer circuit with equal currents (10 ohm resistors in the prototype; could be tweaked). I was surprised at how well it worked into generic 32 ohm phones. It sounded very clear and dynamic, with good lows and highs. I will try 3x2 (6 opamps) master/slaves in parallel.
 
The reason I originally was looking at using high resistors, 560R, was that it would limit the current the opamp could put out.

Anyway jcx, nice looking heat sink on that amp! How does it sound? I didn't really want to go down the heat sink route if I were to build this, that's the reason for using lots of OpAmps and running them 'cooler'. Do OpAmps generally have built in thermal trip?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.