John Curl's Blowtorch preamplifier part III

[waltzingbear]

don't they call that confetti?

Cheers
Alan

 

[dadod]

Not much to add here, since there are only 15V rails there are several complementary IC processes that could do this function exactly (you would not do it this way though). The current sources for one are not temperature stable (might not matter much here). The discrete part is not doing much work so the major benefit is the separation of power out to another chip. I think Jan did this with an AD844 which has access to the gain node.

The NPN/PNP datasheets are a cut and paste of each other which I find a little hard to believe in reality. Operating at widely different Vce's would reduce the benefit somewhat anyway.

With this buffer the rails can be increased to 18V and with LME49610 up to 22V.
Why do you think that discrete part is not doing much? It provide FB of 40dB up to 20kHz.
What do you mean with "The NPN/PNP datasheets are a cut and paste of each other" ?
I built similar version with discrete output buffer GainWire mk3 and no problem with NPN/PNP real transistor there.

 

[billshurv]

High end consumers apparently have the wherewithal to buy what they want, and they happen to like their carefully crafted HD in fully discrete form.

And they love a good story. The reviewers love it too as they can fill 2 pages with the design methodology before they have to start actually reviewing. After all its a luxury good, so like a nice watch, posh car, cat sh*t coffee its about making the purchaser feel good. And a discrete design does that, thanks mainly to conditioning from magazines.


Benchmark get away with evil 8-legs as they are studio first consumer second.



I do hope all the above is wrong btw

 

[Bonsai]

I disagree Bonsai. I have tried both. Discrete wins.

Its because you weren't listening carefully enough John.

 

[Bonsai]

-----------------------------------

Hi,

Let me back up a second.... of course you can add an out-board power stage... bipolar or fet and get a lot of drive to low Z loads. but, that isnt just a ic opamp.

Now with the IC output buffer -- I measured it long ago by itself. It only has good dc specs -- ac is terrible. So bad, in fact, it can ONLY be used in a high negative (GNFB) feedback opamp configuration. Dont know how that combination would sound but I think you can do better with a discrete output stage added on to ic opamp.

I did say with small signal and high Z loads the IC opamp is generally fine. At least the newer CFA types are interesting to me. Maybe for HPA, the newer CFA you mention sounds very good with low Z loads and with Planar Magnetic type. ??

On the other hand, I dont know that 4 cheap transistors and 4 more for a really good output buffer isnt even cheaper than a really good ic opamp with 4 transistor out board buffer. AND, have total control over FB levels/margins etc.

THx-RNMarsh

If you run a discrete buffer open loop I expect the AC performance (and DC performance) is not sterling. The LME49600 was designed to operate within a GNF loop - its ultra wide BW etc. You have thermal shutdown and current limiting etc which make it fool proof IMV.

You can have total control over the overall loop gain - I just put the buffer in the loop but there are other ways to do it if high feedback is not your thing.


You're talking $ and cents for a high end product. Why? The semi content in any HP amp you care to name won't be more than $30~$40 and a finished high end system will sell for $100's or more. In a high end line level/HP type product the semi cost is not the where the money - its mostly in the housing

 

[scott wurcer]

What do you mean with "The NPN/PNP datasheets are a cut and paste of each other" ?

Just what I said, the ON Semi datasheets change NPN to PNP and all numbers are identical and the graphs are simply copied over with the signs reversed.

 

[hitsware]

Or else they've achieved perfect symmetry of processes.

 

[chris719]

And they love a good story. The reviewers love it too as they can fill 2 pages with the design methodology before they have to start actually reviewing. After all its a luxury good, so like a nice watch, posh car, cat sh*t coffee its about making the purchaser feel good. And a discrete design does that, thanks mainly to conditioning from magazines.


Benchmark get away with evil 8-legs as they are studio first consumer second.



I do hope all the above is wrong btw

You can get away with op-amps in high-end too, just depends how you market it. Several years ago on another forum, the Ray Samuels Dark Star headphone amplifier was all the rage. I want to say it was a $3500+ amp using OPA541 power op-amps... it was also very well reviewed. Ludicrous overkill power output, large aluminum chassis, separate PSU box.

 

[Mark Johnson]

In my hazy memory the BCxxx series of devices were branded "Pro Electron" and originated in Europe. At least that's what National Semiconductor's Discrete Devices Databook said.

Here are datasheets from two European manufacturers of BC550-560 , namely Philips and Siemens. The NPN curves are indeed different than the PNP curves, as you would expect.

_

 

[simon7000]

I suspect someone who had a few of them could quickly verify the matching of the transistors in hardware. I suspect the software simulation models just might use the perfect curves and get misleading wonderful results.

 

[Bonsai]

I think Jeff Rowland stuff uses IC’s in the power amps and he charges an arm and a leg and gets rave reviews.

Who knows what the real story is here? It certainly nothing to do with discrete is better than IC that’s for sure.

 

[Mark Johnson]

Pass Labs' current production headphone amplifier is called HPA-1. It sells for $3500 and weighs 14 pounds (6.35 kilograms). What's inside is discrete amplifier made of 8 transistors: 6 in TO-92, 2 in TO-220. Some folks here on diyAudio have reverse engineered the circuit schematic and skilled search-users can find those threads in less than a day, I'm sure. Reviewers seem to love the sound it produces. Max output before distortion shoots to the moon (i.e. onset of clipping) is 8.5V RMS, i.e., a sinewave that swings ±12 volts.

 

[simon7000]

don't they call that confetti?

Cheers
Alan

Or a fire starter!

 

[chris719]

Interested people should also look at the posts by jcx. He's posted several schematics over the years for composite amplifiers that should drive distortion down to nothing for the most part.

Shaq888 (Sergey888) here also created a very nice composite headphone amplifier called The Crocodile.

 

[gerhard]

Just what I said, the ON Semi datasheets change NPN to PNP and all numbers are identical and the graphs are simply copied over with the signs reversed.

I once had a Spice model for the MTP2P50E that had the pinch off voltage
to the wrong side.

I did an AF power amplifier for geophysics featuring 1KVpp out. They feed that into the soil
and measure the potential every 20 meters. You better wear rubber boots there and
keep your hands in your pockets.

 

[john curl]

Jeff Rowland makes very pretty boxes, that use IC's mostly. Rich people buy them for their looks, mostly as well.

 

[john curl]

Mark, the designer of that discrete headphone amp is looking for work, any ideas? Apparently he did too good, and his associates got rid of him.

 

[scott wurcer]

Could we get on the same page with respect to "accuracy" vs. controlled amount of HD2/HD3? Sorry you can't have it both ways.

 

[Markw4]

Mark, the designer of that discrete headphone amp is looking for work, any ideas? Apparently he did too good, and his associates got rid of him.

Yep. Lives not far from here.

 

[Markw4]

Could we get on the same page with respect to "accuracy" vs. controlled amount of HD2/HD3? Sorry you can't have it both ways.

I prefer accuracy myself. Some people seem to prefer the controlled HD2/HD3 thing.

Actually, I believe 'you can't have it both ways' at the same time, but you can have it both ways at separate times by writing to a couple of dac registers.