quite interesting. Paralleling audio IC amps seems to lower noise...
...Interesting experiment isn'it guys ?this was inspired by the massive paralleleld opamp made of ne5532
It certainly does. Here are some of my thoughts on the issue:
"In a sense, the power output stage here is the ultimate low voltage-noise amplifier. When multiplying opamps to reduce noise, it is rarely economic to go beyond four sections. (Two packages of 5532) However, here we have 64 in parallel to get adequate current delivery. 64 parallel amplifiers will reduce the voltage noise by √64 = 8 times, (-18 dB) and since the output noise of a 5532 voltage-follower is about – 119dBu, the noise from the power stage alone will be a truly subterranean –137 dBu. This is below any means of direct measurement. The current noise is however similarly increased, so a suitably low source impedance would be required to feel the benefit."
This is an extract from the third edition of Self On Audio, which will be published this autumn. It adds all the latest articles to the compilation and has much extended prefaces to every chapter. Lots of new stuff!
I have considered a discrete option. I have bucket loads of BD139/140 I could parallel....
Heed the gypsy's warning...
I tried this with MJE340/350, and as the number of parallel devices increased, the difficulty of taming parasitic oscillations in the output stage grew worse and worse. Eventually I gave up and turned to more promising lines of enquiry.
I certainly don't say it can't be done, and maybe you will succeed where I was an early quitter. If so I'd be glad to hear about it.
In case you have bucket loads of TO92's.
Attachments
somewhere, 10 year ago I bulit the DoZ amplifier (Death of Zen - A new Class-A power amp) at the begining I used 2 paraleled tip3055, to improve heat transmission, I added 2 more pairs .
was not sure but I felt that the sound quality changed in a good sense.
I ended up testing it with 20 pairs of paraleled transistors, and it was the warmest , the most delightful amp I've ever heard.
I do not say that the distortion was particularly low - or high- because I did not measured it, however it sounded great to my ears, very different to the one pair version.
Such a big difference is surprising. I encourage everyone to do the test.
no nasty noises no oscilations, power source was a simple capacitance multiplier, with a main transformer of 27volts as I remember.
I think that successful atempt was due to the stable design of "master" Rod Eliott.
Thanks for the insightful answer Mr Self.
I am surprised that an opamp sound that good to the hears.
After all, there is a lot of amplifiers that have tremendous gain and massive feedback and achieve good figures -on the paper (THD)- but lacks "musicality to the ears.
Paraleled opamps might have sound just like a bad discrete or IC amp. But it is definitively not the case. I have tried paralelling 16 tl074 opamp and they sound good in my headphones, smooth warm sound, nothing to do with usual IC audio amplifiers (tda2003 etc).
I will be looking foward to your opinion
Stereo amp from the early '90s, 2x150W, Belgium design (msrp of ~$10k here back then).
The sides of the amp boards have hundreds of TO-92 transistors in parallel as the output stage, NPN at one flank, PNP on the opposite.
No heatsinks, the back panel has four ventilation slot areas over the entire width in the bottom, middle, and top.
Single toroidal transformer in the bottom section of the amp 'case'.
(Nelson Pass mentioned he did a similar thing with small signal bipolars)
'The beast with a 1000 JFETs'. Actually I think it was more like 1124 JFETs per channel.
Jan
)
Haha thanks, I'm flattered but I'm the type of engineer that doesn't memorise anything if I have a good textbook to refer to.
I have considered I may just be easier to use an opamp to drive BD139, and in turn drive a larger single transistor such as TIP3055.
Although I would obviously lose the noise benefit of multiple devices doing that. Hence my interest in lme49600/lt1210 and their ilk driven by an opamp. Though I'm sure I'll have stability issues at some point ��
In these situations, is it advisable to use devices with higher Ft in the current amplification stages, when extending the opamp feedback loop to include the output devices in a global loop?
Last edited:
Hi,
The stunning noise performance of the output stage is a red herring,
what precedes it will never get near such numbers, its irrelevant.
Noise in a power amplifier is generally determined by its input stage,
and there is no such thing as low noise gain or output stage design.
High Ft output devices are fine if you can control the parasitics.
rgds, sreten.
The stunning noise performance of the output stage is a red herring,
what precedes it will never get near such numbers, its irrelevant.
Noise in a power amplifier is generally determined by its input stage,
and there is no such thing as low noise gain or output stage design.
High Ft output devices are fine if you can control the parasitics.
rgds, sreten.
I don't dispute it. But it's an interesting situation, is it not?
Of course.
Hyphen added to make it clearer. No, there isn't, but maybe for the VAS stage there should be. Some opamps have a significant noise contribution from the stage after the input, if it's another diff gain stage.
But as you say, almost anything you put in front of a standard Blameless amp will be noisier, so it's hardly a priority to investigate this sort of thing. From my recent rant at the Warsaw AES convention:
"A Blameless amplifier with differential pair input has an EIN of around -123 dBu so with a gain of 23 times (+27 dB) the output noise is a low -96 dBu. If you have an input system that consists only of a 5534 configured as a voltage follower, the amplifier output noise goes up 3 dB."
Tell me about it. One of the more difficult problems.
See Small Signal Audio Design, but it will have to be the second edition.
the idea was to divide the board in two parts, the preamp and the massive paralleled section.
Then the paralleled section is split in 2 or 3 boards of 10cmx10cm.
10 of these boards can be made for 20 euros by Smart-Prototyping, shipping included.
Much less expensive than the 60 euros for a single board from elektor!
I have started to design a 10X10cm board , but it is for the TL074, it can currently hold 15 ICs.
If someone wants to improve it, I can attach the fritzing or the gerber files.
Have a nice day!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.