John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Jump to Latest
#95,486
RNMarsh said:
Not asking for OPS cmos devices. IPS, drivers etc. application.


-RNM
Click to expand...

An interesting issue. 1 watt into 8 ohms is perhaps as high as 110 dB for a really high efficiency horn loaded compression driver at 1M.

Just to allow for maximum hearing sensitivity allow the noise to be 120 dB bellow that. So an output noise level to achieve that would be 2.83 uV.

Many pro amplifiers have a gain of 20 but to simplify things call it 28.3. Thus 100 nV should do for the bandwidth of 20 to 20,000 hertz. Or a noise figure of 700 pV per sqrt hertz. (But not really as human hearing is not flat at any sound level but particularly not very flat at low levels. )

Now as that would be an absolute worst case, many high accuracy loudspeaker systems are a least 20 dB less sensitive so 7 nV would be very good and allowing for room noise that could reasonably go up to 21 nV.

Of course that is only available acoustic signal to noise levels. There are also going to be limits from the signal source and preamps.
 
Last edited:
#95,487
For drivers the preceding gain is, roughly, (1000 / 0.97) = 1031. So to achieve an input referred noise of 3 nV/rt.Hz you'd want a driver transistor whose noise is comfortably less than 3.09 microV/rt.Hz. I'll bet more than half of all currently manufactured, widely available "complementary CMOS transistors" meet that spec. More likely 80% or even higher.
 
#95,488
Well that is good news for potential CMOS use in some audio places. Most of the development of CMA circuits today are using cmos.

Who makes a good cmos transistor complimentary pair?
Any CMOS IC that are reasonable for trying in audio app?

??


Many interesting possible circuitry with some benefits to audio.

Maybe like this or infinitely more to review for potential improvements for audio.
Patent US4958133 - CMOS complementary self-biased differential amplifier with rail-to-rail ... - Google Patents
Patent US6278323 - High gain, very wide common mode range, self-biased operational amplifier - Google Patents
http://www.ti.com/lit/ds/symlink/opa725.pdf

THx-RNMarsh
 
Last edited:
#95,491
To minimise the effects of high voltage variation causing microscopic changes to the resistor characteristics (as the output swings rail to rail across the resistor/s).

It is a contributory source of a quarter of a half of one percent of nothing to the total distortion of the amp 😉

That said, resistors do have maximum voltage ratings, even when the dissipation is way below the devices maximum rating.
 
#95,493
Bazes's self biasing arrangement in 6,278,323 et al, is helpful if you're trying to design a semi-precision amplifier within a CMOS digital IC, whose fabrication process lacks well-characterized, well-controlled resistors and BJT's (for reference voltage generation). Amplifier designers working with discrete devices at the board level, on the other hand, can use 1% or 0.1% resistors , zener diodes, million-pF capacitors, and high quality discrete transistors to make truly excellent bias generators for cascodes. This results in higher output impedance and less headroom loss (greater output swing) for the amplifier.

I'd expect to see this kind of amplifier used inside the Phase Locked Loop of a digital chip's clock generator circuitry, or in the bias network of a 4-PAM serial I/O module, which sends 2 bits at a time over a single wire, by encoding in quaternary (4 discrete voltage levels).
 
#95,495
Mooly said:
To minimise the effects of high voltage variation causing microscopic changes to the resistor characteristics (as the output swings rail to rail across the resistor/s).

It is a contributory source of a quarter of a half of one percent of nothing to the total distortion of the amp 😉

That said, resistors do have maximum voltage ratings, even when the dissipation is way below the devices maximum rating.
Click to expand...

I was just looking at some Vishay-Dales RND (IIRC) series which are leaded devices. They are quoting 5uV/V voltage coefficient, so in the context of an ultra low distortion amp, not to be ignored. So I believe they would series a whole lot of devices like this to minimize this effect and of course the power dissipation to minimize the resistance changes due to thermal coefficients.
 
#95,496
"the right way" is to use all equal value, indentical brand, rating resistors for the feedback divider, n+1 resistors for a noninverting gain of n

to the degree that all parameters and thermal environments match then everything ratios out - including the nonlinear terms
 
Status
Not open for further replies.
Top Bottom