ultra low noise audio design

Marantz! About 50 years ago (the time of high-fidelity craze, mono) they came out with a preamplifier that was better than the sensitivity of testing instruments in various laboratories. So reports were coming out with stories but no numbers!

Have fun

Stanislaw.

Elektor has done some designs which use multiple transistors in parallel for low noise.

"ilClod"

Since the topic simply does not exist - neither do any publications.

......... Phil

Homer J Simpson wrote: : Elektor has done some designs which use multiple transistors in parallel for : low noise.

Parallel coupling does not affect the noise temperature, but I guess the audio guys are groping for a low noise matching resistance i.e. voltage noise. Similar characteristics are desirable in SQUID readout amplifiers ( 1-ohm generator with an effective 10K noise temperature ). The best low-frequency numbers AFAIK come from discrete bipolar transistors such as the SSM2210. One example is the bipolar/JFET composite amplifier presented by Drung et al in Superconductor Science and Technology vol 19 no 5 p S235. Their amplifier has dc-to-50MHz bandwidth, 0.33nV/rtHz voltage noise above 1kHz and 0.8nV/rtHz at 0.1 Hz. I feel this is at the edge of what is achievable with commonly available components.

An easier choice is the venerable LT1028, a number of which can be paralleled as shown in the Linear Tech AN-21 application note. In the SQUID amplifier folklore the LT1028's in a ceramic DIP have lower 1/f corner than the plastic ones. The brand new AD8599 appears to perform equivalently to the LT1028, with better low-frequency characteristics than the good old AD797.

Regards, Mikko

Mikko S Kiviranta wrote: : no 5 p S235. Their amplifier has dc-to-50MHz bandwidth, 0.33nV/rtHz voltage : noise above 1kHz and 0.8nV/rtHz at 0.1 Hz. I feel this is at the edge of : what is achievable with commonly available components.

I should have said "...*close* to the edge of what is achievable.." before someone points out that the 1/f corner can perhaps be pushed with chopper stabilization, or you can couple even more SSM2210's in parallel, or suggest using ac modulation with a matching transformer (like Koch et al in Rev Sci Inst vol 67 p 2968 - perhaps even that approach can be tweaked into audio use by some quirck)...

But now that I think of it I suppose Drung could make more money by selling his amplifiers to audiophiles as turntable moving-coil preamps than selling them to the SQUID community. Surely everyone initiated can hear a clear difference between a 20MHz and 50MHz roll-off.

Regards, Mikko

With regard to using FET's in low noise audio designs you probably can't go past Erno Borbely's designs.