The use of "A" weighting in A/D & D/A measurements is a bit of snake-oil marketeering. It hides the HF artifacts present in delta-sigma conversion. There is a big difference in specs for a given DS device measured unweighted and "A" weighted. The philosophical argument is that "A" weighting better approximated the MTF of the ear and therefore carries significance because it will better represent the listener's perceived performance. We have found that it also hides noise at the very high end of the audio spectrum (and beyond - check that out too). This HF garbage can play havoc in other parts of a processing chain. When one looks at AF converters, one really needs to look at SNR, floors, etc. out past the nyquist frequency (40KHz+) of the card in question.
It is quite difficult to maintain passband performance with analog filtering downstream from the card to fix "close-in" noise due to slope factor, especially on DACs. While one can easily Pspice a model, they are difficult to physically realize or manufacture with any repeatability. As a result, it is best to find cards, converters or whatever that don't exhibit these issues and avoid comparison of devices based upon the snake-oil specs. Lee Pedlow Systems Engineering Sony Electronics, Inc. San Diego, CA CONFIDENTIAL This email is covered by the Electronic Communications Privacy Act, 18 U.S.C. 2510-2521 and is legally privileged. This e-mail is intended only for the personal and confidential use of the individual(s) to which it is addressed and may contain confidential information. If you are not the intended recipient, you are hereby notified that you have received this document in error, and that any review, distribution, copying or disclosure is not authorized. If you have received this communication in error, please notify the sender immediately by telephone and destroy the message. -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jim Lux Sent: Saturday, September 24, 2005 2:48 PM To: Tayloe Dan-P26412; Ahti Aintila; FlexRadio@flex-radio.biz Subject: Re: [Flexradio] Audiocards, USB, etc. At 02:24 PM 9/24/2005, Tayloe Dan-P26412 wrote: >This 29v pk-pk sound card range is the essence of my concern about >sound cards. The gain must not be constant. > > >The TI part, as specified on page 7, has an input range of +/-2.5v, >centered around 2.5v or 5v peak to peak. This is the maximum input > >Lets look at the Wolfson part. On page 6, the input rage of this part >is specified to be 2v RMS which can also be expressed as +/-2.8v or >5.6v pk-pk. This seems a bit odd because the part has a supply voltage >of only 5v, and I would not expect the input to be larger than the >supply range. If there were a onchip voltage divider? More likely, that spec has more to do with where some diode forward biases. The A weighted vs. non-weighted difference is probably significant since >we probably use these in an unweighted manner. Thus the TI part is >probably really a 109 db dynamic range part. Almost certainly. I am curious as to why they use A weighting, which is really for evaluating SPLs. >Ideally, you should be able to get 6 db of range for every bit in the A/D >converter. Thus, it would seem that 24 bit A/D converters would be capable >of 144 db of dynamic range (there might be a n-1 factor in here, 138 db, I do >not remember). This is obviously not the case with real converters. I have >seen real converters approach this only when they are running very slow, such >as a 10 Hz sampling rate. Since we want 48+ KHz sampling rates, we get less >conversion accuracy and a smaller dynamic range. high performance converters usually have a ENOB number, which rolls all the errors into one number. ENOB is always < converter bits. >The good news is that to the extent A/D converters improve (more dynamic >range), >simply buying a better sound card should be all that is needed to upgrade the >performance of the receiver. Assuming that the A/D is actually the limiting factor. I suspect that things like nonlinearity in buffer amplifiers or in the mux/QSD, as well as reciprocal mixing of the LO phase noise will actually limit the performance. There's also the "total power" problem.. A wide open receiver gets the noise power for the entire band (even if much of it is filtered out after the downconversion), so a strong out of band signal far away can raise havoc still (or, at least require very high IP3 numbers for the system). If you were to just ballpark it and assume something like a 10dB NF for a 50MHz band, you're looking at -174 + 10 + 77 = -87 dBm input power, just for the noise. Add in some strong 60 and 120 Hz sources, etc, and a bit of selectivity in the front end starts to look like a good idea. It's remarkably easy to get microvolts or even millivolts of 60 Hz into a system. Digital clocks will also be a problem. 0.1 V of switching hash on the power line isn't a real big deal for digital logic, but you'd better have pretty darn good power supply noise rejection in the analog part of the system. >- Dan, N7VE > James Lux, P.E. Spacecraft Radio Frequency Subsystems Group Flight Communications Systems Section Jet Propulsion Laboratory, Mail Stop 161-213 4800 Oak Grove Drive Pasadena CA 91109 tel: (818)354-2075 fax: (818)393-6875 _______________________________________________ FlexRadio mailing list FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz
