Hi guys!

It's been a while since I tested the C90 against the Krell. For those that have read the previous comparison between these two, you already know that the Krell took the C90 for a ride.

https://www.clublexus.com/forums/sho...threadid=31166

Now the tables have been turned.

The players (no pun intended) - a highly modded 4 grand cd player (with 500+ dollars of parts) versus a (then) stock Sony C90. The C90 is no slouch when it comes to car audio and it enjoys a good reputation as a fine sounding head unit roughly based on the flagship XESZ50.

The C90, after mods, is now the reference. Yes, it's WIPING KRELL ON THE GROUND.

Before I get into that, here's some of the facts. BOTH use the Burr Brown PCM1702 20 bit converter. The converter itself, at it's ANALOG end has a settling (recovery time) rate of 200 nanoseconds. Now the op amp input (depending on the op amp) will range anywhere from 550 to 2000 nanoseconds. The D/A converter, at it's analog output, is LITERALLY going faster than the op amp can recover. Result? Smeared or glossed over sound.

Even the acclaimed OPA627A (burr brown) op amp, at 13 to 15 dollars each has a settling time of 550 nanoseconds. Clearly there is something missing here with the numbers.

The stock C90 (and xesz50) is using a Analog Devices OP275 op amp. For the car audio world, this is considered extremely good. For the REAL world, this BARELY "cuts the mustard". The Alpine F#1 Status pieces are using 4558 op amps. THESE STINK in REAL world terms...and it's a 7 grand setup!!! I also have a feeling they're surface mounted chips so they're not going to be easy to replace.

So what is an audio guru supposed to do? Modify! The stock op amps are taken out. Mill max sockets and HIGH SPEED op amps are put in. (I'm not saying which ones!) How fast? Slew rate (how fast it will REACT to a quick signal) is well over 125 volts per microsecond. Stock is 22 volts per microsecond. You can't have an op amp that's "too fast" for audio. Settling time for the stock pieces are 200 nanoseconds. Modded pieces are better than 100 nanoseconds. This assures that the op amp will be ready when the D/A fires off the next analog signal. Bandwidth is exceptional for these also.

Now here's the thing. You can't just drop them in and expect results right away. Components need "burning in". When I first dropped in the components, the player sounded HARSH. After about 30 hours of play, the sound started to mellow out. The player when listened to now is EXTREMELY polished, as if a good 3 angle valve job was done as well as some good Extrude honing.

How good is it now? You can hear ALL the mistakes, clear as day. Even most commercially produced CDs will exhibit mistakes in their processors or equipment. You'd be surprised on how many "for the general public" recordings are out there. There are probably only a couple of CDs that I can recommend now as reference material. But one of them sticks out as a TOP performer with regards to audio engineering in the studio. But, I'll get to that later.

C90 and Krell go up head to head. The Krell enjoys the wider soundstage due to it's discrete (transistors...lots of them) output. The C90 is using just the op amps with no further "processing". Depth in sound...C90. Transients and response to treble, especially guitar stings...DEFINITELY the modded C90. No question at all. Resolution of picking out any errors? C90. By a good margin. Dynamics? C90. Pretty much a clean sweep. The experiment went so well that I'm modding the Mc processors, the amps and the Velodyne controller!

Even the stock high end Mark Levinson, Sonic Frontiers, Wadia and Krell won't use an op amp of this caliber. They're happy with a cheap OPA2134 or OPA604 Burr Brown. Definitely not a high speed amp.

But, for the record, the Krell is still equipped with the OPA627 Burr Brown op amps. These are the ones that take 550 nanoseconds to completely settle. The C90 enjoys the high speed op amps. Rest assured, there will be a part 3 to this matchup! The Krell will get the high speed op amps and then it's time for some serious listening.

As for the recording that I'm using - it's produced by Doug Sax and Robert Hadley of The Mastering Lab in Hollywood, California. These are the same guys that re-engineered Pink Floyd and other reissues. THESE GUYS ARE D$MN GOOD, and it takes alot for me to say that! There is absolutely no tonal "questionables" at all. Guitar strings are exact. Bass notes do not have any bloat or overhang, and it's a VERY new recording.

Some may not like the artist (I know I didn't at first) but the CD is so well recorded that it made it as THE TOP reference CD for me. (Yes, beating out Hell Freezes Over, assorted HDCD recordings and pretty much all of the test discs out there.

It's the album "This Way" by Jewel Kilcher, which has the track "Standing Stil". If it can withstand my extreme audio nitpicking then you know it's extremely well recorded. Give it a listen.

Percy
Doing the crazy thing of modding.

 

Sounds like someone has too much money on their hands. I've still got a freaking stock cassette player in my ES300.

Re: "Before I get into that, here's some of the facts. BOTH use the Burr Brown PCM1702 20 bit converter. The converter itself, at it's ANALOG end has a settling (recovery time) rate of 200 nanoseconds. Now the op amp input (depending on the op amp) will range anywhere from 550 to 2000 nanoseconds. The D/A converter, at it's analog output, is LITERALLY going faster than the op amp can recover. Result? Smeared or glossed over sound."

I can't keep up with you on some of the electrical aspects since I'm not an EE major yet (bastards @LSU won't even let me take basic circuit classes until I finish 4 more math classes), but is this a case a gimmick backfiring on a company considering that CD's are recorded @16bits anyway? Honestly, I can't tell the difference between a 1 bit d/a and a 20 bit (after having a/b'ed the Nakamichi 40z and 45z), but I was just curious as to whether or not you think the "smeared" sound you hear could have been avoided using a cheaper part.

 

Bababooey,

You really don't have to keep up with the electrical aspects. Let's put it this way...if there was a garden hose spilling out 1 gallon of water per second and you only had the capacity for 1/2 to 1/4 gallon then you'd lose lots of water. Same aspect with the d/a and op amp.

CDs are recorded at 16 bits, at least most of them are. The 18 and 20 bit converters are used to overcome thermal noise. If you had a 16 bit d/a trying to play back a 16 bit recording then you'd only get out 14 bits, maybe 15. This is where the 18, 20 and 24 (overkill) converters come in handy.

If the system is properly designed you would more than likely NOT be able to tell a 1/20 bit versus a ladder 20 bit converter. All depends on how well engineered the analog stage and power supply is. Garbage in, garbage out. It's that easy.

Percy

 

Thermal noise?

 

Noise generated by heat. Or in this case, it's the noise generated by all the electrons impacting each other. If this didn't happen, electronics wouldn't emit any heat. But they do!

If you look at CCD cameras you can see a visual result from thermal noise. They usually appear as pixels that lighter or miscolored from the rest of the field. Not to be mistaken as dark pixels or "dead" pixels. That's why high end astronomy cameras (long exposure times) have a built in Thermo Electric Cooler. Brings things down to 30 degrees C below ambient temperature.

Now for D/A and any digital application, the heat reduces the overall effectiveness of the converter. The bare minimum I would recommend in a ladder type network is 18 bit. 2 bits can be "lost" to thermal noise. Now if its the Delta sigma type (for example, the Crystal CS4329 or the Crystal CS4390) then it would be a 1 bit (bitstreaming) and 20 bits out, or 24 in the case of the 4390.

You're going to find out in digital that everything looks good IN THEORY. Nyquist looks good on paper (2x sampling points of the frequency will give you an exact representative of the frequency) but in practical applications this isn't always true due to circuit design or components. The math is good, but once when you get to the actual design of the circuit it usually has problems.

Early bitstreaming d/a's had a problem with high frequency noise. You can take an old 1984 cd player (wow...that's almost as old as most college students...now I feel OLD!) and it will have a high frequency harshness. This was due to the converters and the power supply. They didn't give a care about the power supply or the analog output stage. The converter did it's job but Nyquist didn't predict anything about a bad power supply!

Then came the ladder networks. (Don't ask...you'll learn about these as your education in digital circuits goes on.) They had their problems too. And hybrids like the CS4329. (Look it up...you'll learn a bit.) http://www.crystal.com

Bottom line...everything in theory looks good. What really happens when the circuit is built and in production is an entirely different story. The engineers LAUGHED at the guy who mentioned "what about the power supply"? Look who's laughing now.

Take those math courses. They're good for a basic knowledge of how things work, but they're NOT the last word. The math minds can bark out figures all day (It's called THEORY) and they still won't mean a darn when the circuit is actually built. IN THEORY everything should work out nice and happy. In practical application it usually is a different story.

Percy

 

Percy, there are a lot of process engineers who will bless your name saying that underperforming pixels on CCDs are primarily due to thermal noise.

I forget who it is who said it but the quote I remember is "In theory, there is no difference between theory and practice. In practice....." And if you really want to talk about differences between theory and practice, let's get into photonic modelling versus device performance.

 

Ron,

Photonic modeling vs device performance already sounds NASTY! I'd be willing to be that there are alot of guys shaking their heads at the actual results!

Percy

 

Percy - Unfortunately I am one of them.

 

Percy,
From your posts, you seem to stress the importance of measurements (channel separation, THD,... etc....) What are your thoughts on tube amps and LPs...Tube amps don't have the low THD as solid state but they certainly sound good to me.

I have a friend here (same as me, BS in EE) who just does NOT believe in tube gear and LPs and high end stereo. He basically said that ANY cd player with a D/A with CD's encoded in the same bit rate should sound exactly the same.....He dismisses high end audio and if you even mention the difference between lamp cord and transparent cable XL reference, he'd just laugh. "It's just wire....." ...the typical "a watt is a watt" engineer who just refuses to even try and listen to a nice system....of course he thought my mcintosh amps and dyn speakers were better than his civic stock stereo but he thinks it's only because of the speakers. We all know this is BS but it's just frustrating...
I brought up the point where a stereo reviewer (robert harley) said that the sampling at 96kHz (nyquist states 40kHz should be plenty for human) produced a more "open" sound even though we can't hear past 20kHz. Oh man...did i get some lecturing...(i'm not sure if it produces a more open sound but I have heard a 24/96 system and it was VERY nice)

Like you said, everything in theory and on paper looks nice but we (those few of us who care and could hear) know it's never that easy. I wish I can be more knowledgeable just so I can have some kind of an argument when these nonbelievers start spewing physics and DSP...they have a point, i just don't know enough details for a rebuttal....my ears tell me differently..and I know Jadis/B&W silver signature sounds better than Jenson/cerwin vega....
gotta vent..good to have someone like you on my side

 

amlin423,

What we have here are two different examples of engineers. You are open minded and actually LISTEN INTO music and gear. The other guys is just spewing numbers and facts WITHOUT listening or looking INTO the whole picture. There's alot more going on than *just* the D/A converter, or *just* the wire.

A person can quote digital theory all he wants but it's not going to get anywhere far in the real world. I can understand his "wire is a wire", "d/a is a d/a" and "watt is a watt" attitude since that's what he's been taught. I'm guessing that his "wire is a wire" attitude comes from his training and teaching. But, each wire (depending on how it's wound, size of conductor, etc) has different capacitance and resistance as well as inductance. Will this make a difference in overall sound? Definitely. You can take his lampcord versus the Transparent Audio Ref XL remark with a grain of salt. There are obvious differences and if you have the right equipment, they are measurable.

His mentality of the stock Civic system versus yours is highly dubious to say the least. Shows a LACK of open mindedness and flexibility, and that's being NICE about it. This opens up a whole new can of worms! No way in he** will the stock Civic have the same channel seperation, signal to noise and power output that your amp does. Stocker versus aftermarket. Gee...wonder who will win this one. Whap him upside the head on that one for me. :P If it doesn't sink in I have a "for stubborn engineers" hammer ready in my lab. Let me know if you need to borrow it.

24/96. I'd really like to see this experiment done. Use the same timing appropriate for the converter, the same analog output stage, and the same digital input receiver (use one that's capable of 24/96 and backwards compatible with 16/44.1 sources). Have a rapid A/B comparison between the PCM1702K (a 20 bit d/a) and the PCM1704K (a 24 bit d/a capable of 24/96), both made by Burr Brown. My guess is you won't be able to hear the difference. Nyquist will still hold true. I'm still not recommending a 16 bit converter due to thermal noise.

BUT HERE'S THE KILLER....If you take most of the recordings from 15 years ago versus the ones from 5 years ago versus the ones that were made during the last year, you should notice the difference in recording. Most operating consoles still have OP AMPS in them (and some are REALLY CRAPPY), and different power supplies that may not have better filtering (higher levels ac ripple may still get through, affecting the d/a converters). The result is that these artifacts are transferred onto the recording. Best guess is if you were to take 2 channel stereo, one at 20 bit and the other at 24/96 using equally built power supplies, fast response op amps, a good analog stage and all of the goodies, you really won't be able to pick out the difference. It's not the d/a's, it's the support components.

Tube amps...I've only listened to one of these and it was a Conrad Johnson. Very SMOOOOTH. It's one of those amps you can listen to on a very tiring day and it won't hurt your ears. Yes, I know there's quite a bit of distortion going on and the bass wasn't very well controlled at all, but for those instrumentals (guitar strings, orchestra) it worked wonders. Vocals were quite good too! But, in all honesty, I'd rather have some well engineered solid state (krell FPB) so I can listen into the details. There's a bit missig in the tube amps, but they do sound QUITE good!

LP...mention that to some of the younger members and they'd probably ask "What in the heck is that?!?!" I haven't played the LP collection that I have (small collection though) in well over 15 years (mostly "classic rock"..Eagles, Boston, Billy Joel, etc). But I do remember that the vocals sounded more "real" than with the digital sources at the time. More depth and body to them. The signal sources didn't have to go through a bunch of crappy op amps or a noisy power supply with d/a conversions being affected. Those were the days.... (popping back into reality) But those pops, scratches and noise got to you after a while if the record was in bad shape!

The recent HDCD recordings (the well engineered ones, not the "hacked together" ones, are probably the closest you can get to an LP in the overall "body and depth" of sound. From what I've been told, it's a very simple filter. But, the guy who told me that wasn't revealing what HDCD is actually filtering or boosting as that remains an industry secret. (darn!) The only NON HDCD recording that I can wholeheartedly recommend (due to it's engineering) is Jewel's "This Way" cd. If you don't like the voice, just ignore it and listen to the guitar strings. Outstanding.

Percy

 

Dunno, I don't see how you can hear a difference between an electron that had a little trouble getting to your amp/speakers vs one that got their fairly easily. I'm going to have to go with the EE on this one. I sure as hell can't hear a difference among RCA's or power cables, and it doesn't appear that anyone else could when given the challenge.

http://momentum.soundillusions.net/a...tsaug2001.html

Richard Clark has also said that he'd be willing to do his amp challenge, except using the same amp with different cables. Go get your $10,000 if you can truly hear the difference. The only difference I have found in RCA's is their ability to reject noise, but I doubt that this particular test is done in the car as RC himself has written articles regarding RCA placement so as to avoid picking up engine noise.

 

Bababooey,

RC's challenge is so bogus (with so many "safety" rules) that it's not even worth the effort. There was another link on this before on the site. No modded amps allowed, no test tones, no home amp versus car amp. Absolute BS. I can pretty much GUARANTEE that if I had you in front of a Mark Levinson high end amp or a Krell Full Power Balanced amp versus a Jensen or Phoenix Gold, you would be able to pick them out WITH EASE. His "watt is a watt" attitude doesn't hold water. If it did, why all these restrictions?

Wires. I mentioned this one before. Go out and compare a MIT350 interconnect with a M1500 Monster M Series. The difference was enough for an entire shop (Tweeter Corp, formerly known as United Audio in Schaumburg) to change over to Monster after one of my testing sessions with all the salespeople (and some customers) present. They still sell MIT but the Monster has better results. Note that this is INTERCONNECT, not speaker wire. There are also differences in speaker wire but it all depends on how well made they are.

The little electron still has to follow capacitance, resistance and inductance. To put it simply, if you take a wire (any old lamp cord will do,or even a high end wire) and coil it up, you'll still have a wire. But that wire will act more like an inductor rather than a straight wire. With interconnects and speaker wire, the way that they're wound internally will also make a difference.

Go find a shop and listen to both. If you have a high end amp shop nearby, go listen to the Krell FPB or high end ML. You don't need a set of "golden ears" to pick them out, that's why RC had all of his "rules" to cover his a**.

Percy

 

I haven't been able to tell the difference between amps so long as they're putting out the same amount of power and have the same xover slope. Of course you can say if the challenge is BS if you take his opinion as being that "all amps are equal" rather than "a watt is a watt." You can't compare a 500 watt Krell amp and a "500 watt" Jensen amp as the Jensen is very obviously overrated. You would have to significantly clip the Jensen to reach the rated power (or as close to it as it could get), which would make it easy to distinguish to anyone who isn't deaf or close to it. What he's trying to say is that it's hard to justify buying a $1200 Brax amplifier that puts out it's rated 100x2 vs. buying a quality product like Zapco or US AMPS for about 1/5 the price and puts out the same power unclipped- or, even better, spending $300 bucks for a quality product that puts out twice the power and has the headroom and increased output. There really is none. THD, slew rate, and s/n ratio are all marketing jokes as any respectable amplifier is going to meet the requirements to the point that any "better" rating would be pointless (i.e. I can't even name an a/b amp with a THD over 1).

 

***Butler Tube amp has THD over 1 percent - it's a tube amp and not all amps (or watts) are the same. Same with Milbert tube amps - thjey're also over 1 percent I believe. They're (Butler tube amp) is using the tubes in the preamp section only and the real work of amplification is still with discrete semiconductors. Alpine MRV F series has over 1 percent when bridged and most solid state amps will do this. But I'll have to agree to a certain point that THD, Slew Rate and S/N are "jokes" depending on how they measured them. Most amp companies will measure at 1khz for the s/n, where it is the greatest and also at 1khz for THD since it's the lowest number. McIntosh will measure at the HIGHEST point from 20 to 20khz for THD (that 0.005 percent is the HIGHEST number for the entire audio bandwidth!) and at the LOWEST point for the signal to noise ratio, from 20 to 20khz. And GUESS WHAT...the measurements can be CONFIRMED and REPEATED on your test bench! Car Stereo Review has done this and also Auto Sound and Security. They all confirmed the low THD numbers and the high S/N numbers.


Percy

 

I know you can match output with an O-scope. I've done it before with my Arc 2100 and old POS Sony XM430, and I couldn't hear the difference at 20 watts. Of course there isn't the price difference of the Brax-Zapco/USAmps, but almost everyone would consider them to be in different leagues. I assumed that you were saying that you could hear a difference between a "200 watt" Jensen and 200 Brax amp each putting out around max power.

What's the big deal if it doesn't use the same quality components so long as it performs as well audibly? When was the last time you listened to a setup and thought to yourself, "Sounds like some nice transisters on that amp" or "Man, those MOSFETS are doing a great job"? And JL makes very well regulated amps for about 1/3 the price of Brax if that's what you're looking for.