Well, to begin with I don't know what your default PNP is, but if it is like my simulator's default one, it confers about unlimited gain and bandwidth when loaded with a current source --- that is, the collector admittance is zero, or at least damn close. And of course the ideal output buffer doesn't load that node either. So what we are really looking at is the behavior of the 2SC3503 with perfect everything else --- perfect voltages on the rails, perfect current generators (one of which works with 700mV across it), perfect output buffer, and I guess a perfect voltage source as the generator. No temperature compensation, so the output voltage will vary around 38mV/degree K. With R3 adjusting the output offset voltage, a relatively high sensitivity to fluctuations of a less-than-perfect V2.
But with a real PNP the performance is still good. There is some peaking in the frequency response that needs to be compensated for (47pF Miller cap, or... results in a -3dBr at about 5MHz). With a 2SA1381 for harmonic distortion to 15th I see 9ppm at 10kHz 2V p-p in, and mostly second as one would expect from the asymmetrical design. So IM will be mostly simple sum and difference products. If the generator has 1k in series the distortion at the input is about 2.1ppm and overall about 11ppm.
With 35V peak out R4 will dissipate about 5W, of course this will be rare for music.
Making good current generators is not too hard but will entail some number of additional parts. The nearly-perfect output buffer is tough.
EDIT: With the 47pF 10kHz distortion rises to 24ppm.
Last edited:
Yes, this was the brochure that I mentioned. But thanks for the time to include the link.
25% is consistent with my rather limited information. More authoritative is Gray and Meyer. Just to extract the key point for "fair use".
"The value of rb varies significantly with collector current because of current crowding.15
...
Consequently, the value of rb is reduced, and in a typical npn transistor, rb may decrease 50 percent as lc increases from 0.1 mA to 10 mA"
p. 32 in the 4th edition. The current values are for an IC transistor, of course.
I don't place too much faith in datasheet labels either but I am inclined to accept them in the absence of any evidence of error.
Best wishes
David
Last edited:
I didn't really think I would pull you in Brad! 😀
This was aimed at Jan.Didden and the others who feel looking at IPS & VAS with a perfect OPS is important.
Waddya mean 'near-perfect'? Its perfect perfect 🙂
_______________________
We should really organise a group buy for these LTspice default PNP & NPN transistors. Their performance is superb and you can use them as output devices at 1000W 4R too.
Anyone know the right people at Linear Tech?
Meaning there that realizing an excellent buffer is tough. I'm working on it right now until I get some paying work 🙂
It's a difficult measurement to make, but I suppose they may have gotten a few points and then drew the curve through them. At the highest currents, a pulsed measurement of noise in a well-defined narrow bandwidth at high frequencies could be done before dissipation had heated the device very much.
Just to catch up on a few old posts while I wait for a reply😉
I don't have an ECM8000 but do plan to buy one or some similar measurement unit.
What do you recommend as pre-amp and phantom power source?
I would like to avoid "wall warts" and the like, and keep the whole set-up compact and simple.
The USB stuff is simple but seems to lack flexibility. I would prefer to use the sound-card inputs.
Best wishes
David
Some of the older data sheets (BC109 et al) used to show plots of NFC vs collector current.
I designed my first RIAA equalizer in circa 1980 (yes, I am an old fart) and I remember trying to understand the plots and trying to raise the collector current to minimize noise.
Waly, might be its a different noise mechanism, but I am not aware of any significant others besides 1/f
I am using the ECM8000 with a Terratec DMX 6Fire USB which provides a XLR Mic input and is capable to deliver 48V phantom voltage too. The output quality of this USB sound device is also good enough for most measurements.
Software: ARTA
Br, Toni
There is an excellent discussion in Sze and Ng, Physics of Semiconductor Devices 3rd ed., not restricted to ICs per se, and including many equations and ample drawings of the devices. Unfortunately it would strain the fair use doctrine to scan it and post, and Google preview omits the crucial pages.
In regards to the output stage -
EDN.com/design/consumer/4397264/3/Designing-a-low-distortion-output-stage....
Thx-RNMarsh
EDN.com/design/consumer/4397264/3/Designing-a-low-distortion-output-stage....
Thx-RNMarsh
Designing a low-distortion audio output stage - Part 1: Introduction, the problem with push-pull outputs | EDN
It would have been helpful for KCP to include some bandwidth and distortion results. Perhaps he does this later in the series.
Included in the link to EDN are all 4 parts. It was published a year ago. So perhaps the rest has been published already (?). havent searched for it. but can not imagine an article about low distortion without the distortion data.
Non-the-less, you can adapt his fig 2 topology concept into your CFB amp topology - do it in SPICE, first, and see about it's results.
Thx-RNMarsh
Du..uh! Found part 4
The-Class-i-low-distortion-audio-output-stage--Part-4-
Bit suspicious of the way he 'measures' distortion though. Why not use conventional 50W 8R THD etc. ?
Gotta say that while KC-P is a true pedant in the best sense of the word, my personal dealings with him in Jurassic times reveals he sometimes uses his "it can be shown" & "left as an exercise to the reader" to hide dodgy stuff which he hasn't quite thought through properly or investigated. 😱
The-Class-i-low-distortion-audio-output-stage--Part-4-
Bit suspicious of the way he 'measures' distortion though. Why not use conventional 50W 8R THD etc. ?
Gotta say that while KC-P is a true pedant in the best sense of the word, my personal dealings with him in Jurassic times reveals he sometimes uses his "it can be shown" & "left as an exercise to the reader" to hide dodgy stuff which he hasn't quite thought through properly or investigated. 😱
Last edited:
Could be. but it is a great concept to be explored as it solves a lot of issues [if it can be made to work]. can we try it- try to make it work for us here?
Thx-RNMarsh
Thx-RNMarsh
Some time ago there was a long discussion about Class i (as Kendall Castot-Perry called his ops)and siblings here http://www.diyaudio.com/forums/solid-state/202684-class-i-siblings.html
BR Damir
BR Damir
Hi Guys
Krglee, you might also be suspicious of someone who refers to a simulated circuit as "a practical amplifier". Kendall does that with respect to the final circuit in part-4 of the article.
All the threads about class-i on this forum pretty much denounce its implementations. Some Japanese companies used ICs to control the bias as very tight matching is required of the various semi junctions. Even with custom ICs, they all abandoned the method in favour of easier to implement and lower THD approaches.
Yet, some publishers will print anything that is sufficiently burdened with math as to flummox most readers. Those who wade through realise what they've swam through...
Have fun
Kevin O'Connor
Krglee, you might also be suspicious of someone who refers to a simulated circuit as "a practical amplifier". Kendall does that with respect to the final circuit in part-4 of the article.
All the threads about class-i on this forum pretty much denounce its implementations. Some Japanese companies used ICs to control the bias as very tight matching is required of the various semi junctions. Even with custom ICs, they all abandoned the method in favour of easier to implement and lower THD approaches.
Yet, some publishers will print anything that is sufficiently burdened with math as to flummox most readers. Those who wade through realise what they've swam through...
Have fun
Kevin O'Connor
Swim Swam Swum
Many are the ways. But although I prize high performance in terms of linearity, stability, bandwidth, and low noise, it's apparent that many, especially professional audio reviewers, don't seem to hear the same way as I (see recent Stereophile review of the Croft integrated amp for instance).
Brad
Yes I remembered getting about knee-deep in this as soon as I saw the "class-i". That designation, by the way, at least as an upper-case I, was already taken by Gerald Stanley for his patented switchmode approach, which greatly facilitates interleaving. It was perhaps unfortunate that "I" comes after "H", as this prompted one commentator to disparage it from his bully pulpit before knowing its motivation. Maybe K C-P wanted to have an allusional tie-in to an Apple app.
Many are the ways. But although I prize high performance in terms of linearity, stability, bandwidth, and low noise, it's apparent that many, especially professional audio reviewers, don't seem to hear the same way as I (see recent Stereophile review of the Croft integrated amp for instance).
Brad