Topic: V3 vs 722/44 A/D (comments from SD)

[Todd R]

jeez, just admit it Todd, Grace sucks, SD rules! 

Oh yes, I forgot, thanks for setting me straight.

Really, I've owned both SD and Grace products, and I think highly of both firms, as well as their products.  But I really wouldn't take to much stock in that quote from the SD engineer.  And really, bottom line is they had the opportunity to get a new and somewhat improved part since they came to market 2 1/2 years after the V3.  

[scb]

so how does the mytek's ak5394a stack up?

[Todd R]

so how does the mytek's ak5394a stack up?

The ak5394 looks on paper to be a quieter part from a performance standpoint, but is far more power hungry.  Current draw of 665mW, so almost 3x of even the TI 1804.  It's published noise specs use different measurements and do not seem to be comparable.  But its A-weighted dynamic range is 117db minimum, 123db typical, which is quite a bit better than the TI and Cirrus parts.  Anti-aliasing performance at 44/48/96 is the same as the Cirrus and TI parts, but not nearly as good as even the TI 1804 at 192k.  Likewise with the group delay, quite a bit worse than even the TI part.

[scb]

interesting.  the Sound Devices guy claimed that was the best chip out...he said they didn't use it because of the power requirements and the price

[Todd R]

Not surprising, I'd imagine price and power dissipation are the greatest concerns for a design, followed pretty closely by noise/distortion and dynamic range specs.  It might be nice that a part has somewhat better anti-aliasing and group delay, but I wouldn't imagine these performance characteristics are that high up on the importance list.

[Todd R]

Moke,

I believe A-weighting is a means of weighting dynamic range testing/specifications so that they better reflect how the human ear hears perceives sound.  A-weighted numbers are generally better than un-weighted dynamic range specs.

[Brian]

Moke,

I believe A-weighting is a means of weighting dynamic range testing/specifications so that they better reflect how the human ear hears perceives sound.  A-weighted numbers are generally better than un-weighted dynamic range specs.

as usual, Todd is on(or close to) the money

this is a direct quote from the Sound REinforcement Handbook by yamaha:

"The fact that the ear is not linear guided the makers of sound level meters to use as a corrective filter - the inverse of the typical equal loudness contour - when measuring SPL.  The filter has the so-called "A weighting" characteristic
........snip..........
Given the sensitivity characteristic of the ear, the "a-weighted" curve is thge most suitable for low level sound measurement.  Remember that 40dB SPL(@1kHz) is equivalent to the sound of a very quiet auditorium or of the average quiet residence.  In the presence of loud sounds, such as rock concerts, the ear has a "flatter" sensitivity characteristic
..........snip.........
In order for the measured sound level to more closely correspond to the perceived sound level, one would want a flatter respnse from the SPL meter.  This is the function of B and C weighting.  In apparent conflict with this common sense approach, OSHA and most governmental agencies that get involved with sound level monitoring continue to use the A scale for measuring loud noises.  Sonce this causes them to obtain lower readings than they otherwise would, the inappropiate use of the A scale works in favor of those who don't want to be restricted"

what I believe that is saying is that the A scale is an old spec that was used for low sound level devices with dynamic ranges.....like 40-50dB.  it seems gear manufacturers still use the scale to get more desirable results when testing their gear.

[F.O.Bean]

here are some comments from the SD folks on the A/D chip in the V3 vs the 722/744T, taken from the SD support forum. Of course, the A/D chip itself is only part of the story, implementation has a lot to do with how things eventually sound, but I thought it was interesting:

>The V3 uses a PCM1804 from TI, and the 744 uses a CS5361. As a side note - we started this design with the 1804, but it turned out to be such a poor chip in almost every regard, we sold them all back to TI and switched to the much better, cleaner 5361. It is quieter, lower distortion, lower group delay, draws less current, has MUCH better anti-aliasing at higher sample rates, and has no idle tones, and most importantly, sounds much better.<

FWIW.

Steve

marc laughs ass OFF.

Todd laughs ass off upon hearing Marc finding justification when the Sound Devices design engineer claims his component selection is be better than his competitors.   Did you think he'd say, oh I picked crappy parts for my design?

Ok, so I looked these up a bit to compare:

Claim:

- Cirrus CS5361 (722) is quieter - DOUBTFUL, but maybe true.  The CS5361 and the TI PCM1804 both have minimum A-weighted Dynamic range of 106db.  The CS5361 claims "typical" operation at 114db, while the PCM1804 on claims 112db "typical" operation.  Even if these typical values are true, that isn't much in the way of a whopping difference.  [BTW, semiconductor mfgs only need to meet min and max performance levels.  Typical values are only marketing claims and should be taken with a huge grain of salt.]

- CS5361 is lower distortion - again, maybe.  The CS5361 and the PCM1804 both have maximum THD+Noise of -95db.  The CS5361 claims with "typical" operation THD+Noise is no more than -105db; the PCM1804 only claims typical THD+Noise to be -102db.  Once again, even if true, not much of a difference.

- CS5361 has much better anti-aliasing at higher sampling frequencies.  True, but very misleading.  The CS5261 and the PCM1804 have identical anti-aliasing specs for 44k, 48k, and 96k operation.  The CS5361 only exceeds the PCM1804 at 192kHz operation -- only the highest sampling frequency, not "higher" frequencies.  So unless you're recording 24/192, you won't see a difference.

- CS5361 has lower group delay.  True.  Frankly, I don't know what practical effect group delay has.  Guess it's time for more research.

- CS5361 draws less current.  YES.  The cs5361 draws only 135mW, compared to 225mW for the PCM1804.  From our standpoint, this means the 722 will get better battery life.  My bet is that this is exactly why the design engineer chose this part of the TI part.  All the rest is marketing fluff.  Power management, esp for portable products, is a HUGE concern to design engineers.  These are high current draw devices to begin with, much moreso than most of the rest of the components used.  Saving 30% here is a big savings.

- CS5361 sounds much better.  One man's opinion, and a very, very biased opinion at that.


So much for all his claims.  I think the real statement should have been more like this:

The V3 uses a PCM1804 from TI, and the 744 uses a CS5361. As a side note - we started this design with the 1804 way back when we started the 722 project, but since we were way, way, way behind schedule, we were able to use the CS5361 that only came out in late 2004.  This was a great opportunity, since it draws less current and totally saved our asses, since we had already gone way beyond on our power consumption in the rest of the design. We had all those TI PCM1804's sitting around in the stock room which we couldn't put to use since we totally missed our design targets.  Since we had no money coming in from this project and since the TI parts were in high demand and limited supply, we sold them all back to TI which really helped with our quarterly earnings, which were really sagging since we were way, way, way behind with getting the 722 to market.

 

+T

[ethan]

so who is going to mod the v3 with this newer chip and report the results back?

heh....like it's that easy.

[creekfreak]

so who is going to mod the v3 with this newer chip and report the results back?

heh....like it's that easy.

I am holding out for the Hard Drive MOD

[Todd R]

so who is going to mod the v3 with this newer chip and report the results back?

heh....like it's that easy.

Yep, the CS5361 and PCM1804 are completely different chips, so there isn't really a way to run the CS5361 in the V3.

The PCM1804 chip used by the V3 can output DSD data instead of PCM data, though.  Tascam already has a rack mount DSD recorder, and is reportedly working on a portable DSD recorder.  So will Grace be able to mod the V3 to output a DSD data stream?

[MattD]

Yep, the CS5361 and PCM1804 are completely different chips, so there isn't really a way to run the CS5361 in the V3.

The PCM1804 chip used by the V3 can output DSD data instead of PCM data, though. Tascam already has a rack mount DSD recorder, and is reportedly working on a portable DSD recorder. So will Grace be able to mod the V3 to output a DSD data stream?

I'm going to say it's not possible because the pinouts are different in DSD mode than in PCM mode according to the PDF. DSD uses the L/R clock as the Data clock and the Data clock as the Left output. What was the Data output becomes only the Right output in DSD mode. I'm guessing the output circutry of the V3 would have to be completely realigned.

Also, how would it output the DSD data? As far as I know, right now there is no way to digitally transmit DSD from one device to another. Not that I don't think that'd be solved, but until that is, it makes creating a "DSD V3" much harder.

[JackoRoses]

so who is going to mod the v3 with this newer chip and report the results back?

heh....like it's that easy.

Yep, the CS5361 and PCM1804 are completely different chips, so there isn't really a way to run the CS5361 in the V3.

The PCM1804 chip used by the V3 can output DSD data instead of PCM data, though.  Tascam already has a rack mount DSD recorder, and is reportedly working on a portable DSD recorder.  So will Grace be able to mod the V3 to output a DSD data stream?

I guess the smirk was not enough to show the sarcasm

[Genghis Cougar Mellen Khan]

here are some comments from the SD folks on the A/D chip in the V3 vs the 722/744T, taken from the SD support forum. Of course, the A/D chip itself is only part of the story, implementation has a lot to do with how things eventually sound, but I thought it was interesting:

>The V3 uses a PCM1804 from TI, and the 744 uses a CS5361. As a side note - we started this design with the 1804, but it turned out to be such a poor chip in almost every regard, we sold them all back to TI and switched to the much better, cleaner 5361. It is quieter, lower distortion, lower group delay, draws less current, has MUCH better anti-aliasing at higher sample rates, and has no idle tones, and most importantly, sounds much better.<

FWIW.

Steve

So basically, if the chipset sounds better and someone has the opinion that the 722 still doesn't sound as nice, you'd have to assume the pre is that much worse than the v-2/v-3 no?   

I'd still run one... 

[greenone]