Russian specialist kickass the doodle
Maxim@iXBT vs. Mark@3DSoundSurge

(special version, not for publishing)

Hello, Mark!

My name is Maxim Liadov (maxim@ixbt.com). At present time, I am a technical writer and "Multimedia" part editor on the best Russian hardware web source, iXBT Hardware (www.hardware.ru). Recently, I have translated personally one of my very interesting materials into the English and have posted it on www.digit-life.com as "Updated Santa Cruz Review" (the first version was very bad, because It wasn't translated in co-authorship with me). Some info about myself: I have High Technical education (Moscow State Technical University) in Military Digital Automatic Control Systems (I'm also Soviet Army Officer of Anti-Aircraft Defence). In other hand I have two music educations: clarinet/sax and piano/synthesizer. With PCs from 1987 and with music from 5-th years old.
I was very interested when reading your comments on http://www.3dsoundsurge.com. Let me give you more precise definition to the some fine moments:

Digi-Life has updated their sound quality focused review of the Turtle Beach Santa Cruz but most of my comments posted on the first version still stand.
Unfortunately, I have not seen your previous comments. In next time a good idea send it to me personally.
As above, we only have so many hours in a day to do all our site work part time and once upon a time used to contact webmaster but I seen in your comments that you correctly criticize the journalists calling the Santa Cruz Live killers when its got lots of driver issues and is competitive in some areas but not a "killer". Basically we gave up other than a few select sites who have shown they actually care about being accurate and won't react in anger to comments. I am pleased to add you to that "list".
Thanks. :) The secret is that I'm not presumptuous teenager-writer. I am normal man who just is trying to help other people in their choices. For me the truth is more important then personally pride.

Basically I don't follow the logic of some of their sound quality conclusions and a few other comments.
It is possible because English is not my first language, as you can see. :) Contact with me and I will always find a time to give you more large comments.
I sympathize with the difficulties of translation, especially when you do technical writing such as your review as the necessary detail may not always come across. I really should have added that in my comments that odds are it was because of the translation and your English is certainly better than my Russian :) Please find some more specific comments down in your email.

When it comes to the tests I can add that we are starting to use the Spectra soft test suite (the review at least in part used some Spectra product) and my analog loopback results (which is what all their testing is based on) are different than their.
In the original (Russian) version there was a part "Measuring with the help of SpectraLAB", where were provided lots of information about test setting's configurations and why exactly. May be I will translate into the English this my material too (if it is needed for readers).
Well I am not sure if most readers will want it but I think it's good to have that material to offer as a reference for those who do.
I think that too. For those who are really interested in this things there are many another good technical sources and SpectraLAB's Help.

I am not confident enough in my own results to publish them yet. However, I will say that my preliminary results give the edge to the Santa Cruz over the Live in A-B loopback testing of S/N ratio and frequency response with the gap wider for 44.1 kHz (vs 48 kHz) tests but apparent in both cases.
Live! is not one card "always and forever" in its hardware. It may be provided by Sigmatel STAC9721 quad channel codec or Creative CT1297-TAT + Philips 1330A stereo codecs. This things not depended on Live!'s model numbers (4620, 4670, 4760, 4830, etc). "You have to see it for yourself" (с) by Matrix. :) Which is had you used for your tests?
CT4620, CT4760 and the new 5.1 version (I forget the number). Like I said, I am not satified with my tests yet but in identical A-B testing the Santa Cruz has better s/n and much better frequency response than the 3 Live's I did loop-back testing on.
The card model number is not important in general. The very important is the codec's number. The best Live! card must have Creative CT1297-TAT + Philips 1330A two channel stereo codecs (this is chip's marking solder in the card). Please find it and making sure "what you have". I have this ones on my CT4830 (2000 year manufacturing).

Again, we need to do quite a bit more familiarization with our test suite before I am ready to offer any hard numbers or insist that my numbers are correct. The other bit worth noting is when it comes to game audio quality, MP3 playback and DVD playback via the cards analog outs the important numbers are the Digital to Analog numbers but those can't be tested using loopback testing. We will also adding a reference audio card to our test suite in order to get at those results along ADC results.
This thing in my plans too. By now I have found digital PCI oscillograph for more fine measurement then given by any "loopback method". My true idea is not "to kick" Santa Cruz. Only to get REAL comprehensive examination of all its features. My English version article is my own, non-commercial working to start up our new English project.
I also wondered why you did not publish any frequency response tests since those are of particular interest to readers.
Because I always had full Live! FR analysing in my Russian Live!Platinum Review. You can read :) or just can see on the pictures here (there are English legends on it): www.hardware.ru/multimedia/sbliveplatinum.html. To the point, CT4760 with STAC9721 has more worse characteristics then my CT4830 with two codecs. :)

Getting back to the review, you may want to compare their Live results to results posted at PCAVTech, where they also have tests of many other cards to compare against.
The first is, there aren't tests 48->48 (as measured in my material) on PCAVTech for SB Live!
Good point!
It is not fear. There is a saying: "The man have not mistakes that who do nothing." :)

The second is that I said in my own article:
"I remind once again. Given figures are not absolute value, they are given only for comparison of one card with another one and strongly depend on measurement method."
No disagreement there. It's all any of us can hope to do since (1) there is no standard sound card measurement technique and (2) our test beds vary. That does not change that I am getting better s/n numbers on the Santa Cruz than the Live though.
The bad thing is that I have not translated test conditions (such as SpectraLAB settings) because have not enough time. Although always criticize that things.
On the numbers just for our discussion I am getting between 72 and 75 dB 48 kHz loopback and was around 81 dB for the Santa Cruz (can't find my results right now).
Was Live! microphone pre-amps being as un-plugged and as muted? What is about recording level in Live mixer? It must be ~90..100%. I always use CTlauncher, Volume tab, and for the real time tuninng of both signal and recording levels. You can send to me your spectrograms for my "critical look". :)

Absolute value is not available without calibration from external ideal source.
And since the PC audio all measures in different ways there is no standard for us to go from anyway so just standard techniques and consistent methodology is all we can manage.
My remote Sent-Peterburg colleagues from www.softjoys.ru are having possibilities to take more accurate measurement of Live! (the "old is gold" model within two codecs). I will translate for you something from their material http://www.ixbt.ru/multimedia/sblivehomest.html:

Announcement SBLive! analog audio characteristics
			RATED LINE OUTPUT		FULL SCALES OUTPUT
Line Output Dynamic			1.0 Vrms		1.40 Vrms
Frequency Response at -1dB			10Hz to 44kHz		10Hz to 44kHz
Signal-to-Noise Ratio (A-weighted)			96 dB		100 dB
THD + Noise (A-weighted)			0.002%		0.002%
The cardmakers often choose measurement method that to give them the more good results for the their cards. Therefore it is necessary to comparison the card with another same cards. Audio characteristics in detail There is professional testing system: Test conditions: there is only one un-muted source in mixer all effects off tone control set to 0 dB (50 %) setting up 4 speakers out configuration Table of results: Test: PC - Digital - Analog Line output Input source Ideal Wave file of OdBFS @ SR=48KHz Output level @ Line-out 1Vrms SNR (@Idle channel noise) 90dBr Dynamic Range 94dBr THD + Noise @ -3bB 0.007% Frequency Response @-20dBr 20Hz - 20KHz, +0.05 dB /- 1 dB   Test: A - D - S/PDIF Output @ SR=48KHz Input source @ Line-in 1Vrms @ 997Hz SNR (@Idle channel noise) 92dBr Dynamic Range 92dBr THD + Noise @ -3dB 0.004% Frequency response @ -20dBr 10Hz - 21KHz, +0.2 dB / -1dB   Test: A - D - PC - D - A @ SR=48KHz Input source @ Line-in 1Vrms @ 997Hz Output level @ Line-out 1Vrms SNR 87dBr Dynamic Range 91dBr THD + Noise @ -3dB 0.008% Frequency response @ -20dBr 10Hz - 20KHz, +0.05dB / -1dB
As we can see the characteristics a little worse then announcement, but still very good. The very bad result is PC - Analog output - Analog input - PC, where SNR is 87 дБ, TND+N is 0.008 %, but it is professional studio level! Comparison with other cards There are measurements consist on SpectraLab program, through himself test source. Sound card distortions (THD), -3 dB 1 KHz signal S/N ratio -3 dB 1 KHz signal SB Live! с Midiman Flying Calf A/D на входе S/PDIF 0.00197 % 85.36 дБ Event GINA 0.00219 % 79.96 дБ SB Live! 0.00383 % 81.65 дБ TB Montego 0.00412 % 74.87 дБ

As you may see (and really to my surprise too – I find this things just in this time then I wrote this e-mail to you), his "0 dB signal" test has 91 dB on serious measurement system and my test has 90.1 dB for codec's loopback. His 87 dB and my 87.5 dB for "-3 dB signal" are also near. And to the end his 81.65 dB in SpectraLAB's in-out test is near to my 80,8 дБ SBLive! Value front and 82,5 дБ SBLive! Value rear.Therefore I hope that you become familiar with SpectraLAB and will test all rights at short time ago.

More about measurement methods you may read in this outstanding book: "Electronic measurement and instrumentation", K.B.KLAASSEN, IBM Almaden Research Center, Cambridge University Press. Or one can be used something another good metrology folio.
Thanks! I will look for it. I have been using the Audio Measurement Handbook by Bob Metzler as a reference. He is one of the founders of Audio Precision Inc.

A quick question for you.  In the prior email you said to be sure the record level is around 90 percent. Why is that important? 

As you know, codec on Live! have 18 bit for many things. One of that is a gaining quiet music signal in digitaly. What is "good thing" your may hear in WinAmp's Equalizer - Preamp slider. It is not very good (in " " is conversely mean in Russian, not sure that you consider also) . Let me quote a Sigmatel 9721 spec (it's the codec of your SBLive!):

Analog Mixer Input Gain Registers (Index 0Ch - 18h)
These registers control the gain/attenuation for each of the analog inputs. Each step corresponds to approximately 1.5 dB. The MSB of the register is the mute bit. When this bit is set to 1 the level for that channel is set at -dB. Register 0Eh (Mic Volume Register) has an extra bit that is for a 20dB boost. When bit D6 is set to 1, the +20 dB boost is on. The default value is 8008, which corresponds to 0 dB gain, bost off, and mute on. The default value for the mono registers is 8008h, which is to 0dB gain with mute on. The default value for stereo registers is 8808h, which is 0 dB gain with mute on.

Therefore it is advisable to set 0 dB gain level. In this mode there is no "digital gain" and must be the best test characteristics.

I see from tests of Fortemedia's FM801 card that for 48 kHz output it can make a big difference in s/n (60 dB vs 78 dB) where as with the Live there is basically no difference. Why is that?  Should it not be as simple as testing at the same level below full scale (-3dB for example). May be it is that I have talken above. Another codecs have other 0 dB gain level in mixer. The default levels may be set than RESET command set to codec's mixer (I use for this case "AC'97 Mixer" programm by Alex Mina). Usually default level is setting right after driver installation. Because of it, a good idea is to write one time those values to a paper.

And if you have another questions - you're welcome! :)

Maxim Liadov

Senior Technical Editor,
Digital Audio section of
iXBT Hardware