| Product Range |
 INTEGRATED |
|
Model IS Model IS Ref Model IT |
|
PREAMP |
| Model PRE 6/2 |
|
AMPLIFIERS |
|
Model A80 Model A80 Ref Model AP150 Model MAP |
|
CABLING |
|
Model CLR Model CNR Model CMR150 Model CHR & CX |
|
INFORMATION |
| Company |
| Technologies |
| News |
| Press |
| Tech Specs |
| FAQ |
|
CONTACT |
| Distribution |
| Contact us |
Back to the list
Some thoughts on the AES and the value of its preprints
march 1998
Hi-Fi Link House Magazines ltd, Dingwall Av, Croydon CR9 2TA UK
tel : (44) 181 686 2599 fax : (44) 181 781 6046
abonnement France, Carte Visa acceptée : tel (44) 1454 620070, fax (44) 1454 620080
E-mail : HI-FINEWS@lhm.co.uk
|
... The third of some of the notable preprints I have spotted in the past two years is from France. The Author (Hephaistos, from Lavardin Technologies) begins with the well known position, that conventional alias ''classical'' tests of audio electronics do not tally with widespread listening experience, and cites simple test circuits he has devised wich can be arranged to prove the point. The defective thinking he identifies is that classical testing revolves around an implicit hypothesis ; that distortion characteristics are immuable and simple. But in reality, they are usually unstable and complex, often varying according to a music signal wich is itself complex and even quasi-chaotic. The changes, futhermore, often have time constants that give rise to 'memory'' phenomena. The fact that different sources of memory arise in components, ICs, and even in particular circuit topologies (eg, as spike saturation or V-I protection recovery characteristics), will come as no surprise to technically experienced readers. The trouble with testing for memory effects is rather like that of jitter in 20+-bit systems : equipment wich has less of the effect than the smallest amount you are trying to measure is probably not in existence. You will have to make it yourself, and without any test equipment to test your efforts against. After achieving this, the author is reporting errors signals that are only 60dB down with one well known transistor-based topology. Simple tube circuits can readily have lower memory, but the Author then relates that transistor circuits with high feedback can also be made with even lower memory, yielding sonics that even tube fanatics will prefer. |
to AES/lavardin preprint Back to the list