Newbie Project - Tweaking the SoundBlaster Live! 24-bit

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Greetings to one and all - this is my first post on the forums, though I've been lurking for some time. :) Apologies for any idiocies that follow...

I recently aquired, for a princely sum, the aforementioned sound card, and it certainly doesn't sound bad. Before any creative-bashing goes on (don't worry - I'll do that soon - see below) you must note that as far as I can tell the Chaintech AV-710, everyone's first choice on the cheap, is rather hard to aquire through retail channels in the UK.

Obviously, I'm interested in fiddling with the damned thing. :)

Does anyone have any particular experience upgrading sound cards? :)

Here's what I've come up with so far...

Changing the Op-Amps

This morning I performed a survey of the parts. Of all the op-amps mounted on the PCB...none of them have a slew rate greater than 7V/us! And the headphone output op-amp (JRC4556A) has a slew rate of 3V/us. All other things being equal, (which they are not)...no wonder people complain about the card having a 'sloppy' and or 'lifeless' sound. Am I correct in this assertion? :confused: Compare this to some of the op-amps of choice recommended by Warren Young's 'Notes on Audio Op-Amps', such as the AD8620, with a slew rate
of 50V/us. The recommended Burr-Browns also have similar slew specifications.

I know that surface mount soldering is difficult, but the little pins on these look just-about doable. Assuming I can find one that operates within the 4556A's specs, would it be worth replacing them with better ones? I guess so. Are there any issues I should be aware of?

Also, would it be worth stacking Op-Amps in the same way that DACs can be stacked? Speaking of which...

The DAC

The Cirrus Logic 4382

Apparently this is a fairly decent DAC, though not perfect. The prospect of DAC-stacking this surface mount beastie is certainly beyond my abilities...unless anyone has a technique for such things. The only thing I can think of doing is building a simple clock generator and tapping it into the master clock. Is this the only pin that needs to have the clock fed into it? Or would the creative controller chip need it too? (I know good clocks aren't 'cheap', but anything's probably better than Creative's effort, judging by the choice of op-amps, one can only wonder at the other high-performance components in there! ;) )

Would I need to desolder the 'master clock' leg? Would this not work happily? Is it feasible?

If it's not, the only thing I can think of is upgrading the crystal itself to a more stable version, though I don't believe that this will result in a lower-jitter signal necessarily because of the oscillator circuit that surrounds it. Am I correct?

Other Ideas

The only other things I can think of is upgrading the regulator ICs to higher-performance ones for cleaner power, or even powering the entire card independently of the PCI bus (and thus, the PC's power supply) using batteries, or simply giving the DAC a clean, battery-fed power supply. It would also be quite simple to 'upgrade' the no doubt awfully cheap Jamicon capapcitors for even non-audiophile high quality caps, if this would help. Any comments?

Thanks in advance for your patience and ideas,

- Toddy.

PS. I have used the 'search' function, but couldn't truly find what I was after, and I'd rather bow to experience than charge out with my own odd experimentings. (At least for now.)
 
I'v been thinking of doing the same for quite some time on my audigy 2, but never got around to it, too much hazle involved.

though i wonder, are the opamps on your board dual?

about the dac soldering, you can use the "flood-and-suck" technique..
read here

oh, and welcome to the forum BTW :)

-Marius
 
Hey, demogorgon.

Thanks for your reply. With regards to the nature of the op-amps, I'd guess that they were. Certainly, the data sheet on the NJM4556A states (the primary output) that it's certainly dual, and I'd imagine the rest were, too.

The SMD soldering guide has been duly bookmarked. :) Cheers. The flood-suck method might well work on a stacked DAC chip, or stacked op-amps, and I suppose one could always cover over the pads that one wanted to tap into or out of with something heat-resistant and non-conductive.

For future reference, I'm hoping that I can turn this thread into a running reference on sound-card modification in general later on. I believe that the modification of sound cards is underrated. For example, if one were to build a lovely CMoy to power headphones with some nice high-slew op-amps, and take output from the soundcard, is there much point if the sound has already been turned sludgy beforehand by an op-amp that is ten times 'slower'?

Oh, and why use a nasty Live! 24-bit anyway? Well, it's pretty ghetto, if nothing else. :)

Am I correct? Anyone have any more advice before I begin my tinkerings, particularly with regard to the clock issues? I'll probably begin with the op-amps anyway, as it's the simplest.

-T.
 
J.R.Todd

Why dont i ever learn?
i should have stopped looking into post i know will eventually make me buy something..

I like the idea of creating a reference thread, andfor what its worth il join in on the action :)

opa2604 and some opa2134 are on their way as we speak.
about the dac, i dont know very much about dac's other than in general theory, i'm gonna have to do something about that soon..
perhaps i'l find one thats pin for pin compatibe with the one allready there? - something to look into.
the clock though looks like to much hazle for me, fideling with such high frequeny signals can very easy get more interesting than planed, i imagine. if someone with a scope would join in the thread, i think perhaps we least could find out a thing or two, and go from there.
though there arn't many people around with 100mz scopes of their own. (considering the pci bus is 66mhz, the scope would probably have to be that?)

-Marius
 
ok, here's the dac on my card atleast,
http://www.cirrus.com/en/pubs/proDatasheet/CS4362_F1.pdf
haven't found any replacement yet, but looking.
doesnt look too hard to do something about it though.

-edit
well, there's a cs4382A available, doesnt look like much of an improvement though. i'l be having to take a look at what's allready on my board.

what that codec thingy does however, i still wonder..
anyone able to afford an explanation?
the codec is the SigmaTEL STAC9721 AC'97 codec.

-Marius
 
Hi,

The 4556 op-amp is rated for 70mA and is used for driving headphones. If you don't intend to use headphones you could simply bypass it. Replacing the 4558s with AD8620s sounds a good idea, but remember that faster op-amps need better HF decoupling, so you should consider making sure their supply pins are bypassed with small value film caps. Also the DC specs of the 8620s are much better than the stock devices, so you should be able to replace any DC blocking caps with a wire link or solder bridge.

As for the clock, simply remove the crystal and associated resistor/caps and apply the new oscillator.

One thing to keep in mind: simple mods like swapping op-amps are worthwhile, but there comes a point with upgrading sound cards where it becomes more practical to vuild an external DAC. Your decision of course.
 
My sincere apologies, chap. :)

The 'on their way' thing sounds familliar - an AD8620 is being delivered courtesy of the Analog Devices samples department. :) Other than that, you might have to take the lead on the first replacement, as I'm woefully underequipped to do any type of electronics work. There's still tools to aquire...and the Electronics department doesn't allow non-students to use their facilities. :(

With regard to the DAC, they're far less industry-standard than op-amps and so finding a drop-in superior replacement will be very difficult, which is why I thought it best to concentrate on reclocking the beast. We do share the same DAC, though, so at least we can collaborate on this a little, I hope. :)

The PCI bus, bog standard, I believe functions on 33Mhz, with various variations. Though I don't think one can seriously 'reclock' the master PCI bus clock!

http://en.wikipedia.org/wiki/PCI

I was hoping to, at some point, simply get a good, clean master clock signal into the DAC...but I don't know if the Creative controller chip will also need to be reclocked.

Best of luck with your endeavours, chap.

-T.

For the curious, here's the components list of my Live! 24-bit, with a diagram.

ST Microelectronics (ST) 33078 - Low noise op-amp, presumably for microphone input. (#1)
Japan Radio Corp (JRC) 4556A - Headphone (primary) line-out. (#2)
ST Microelectronics (ST) 4558C - Other line-out op-amps. (#3)

Cirrus Logic (CS) 4382 - 24-bit 96kHz DAC. (#4)
Wolfson Microelectronics (WM) 24-bit 96kHz ADC. (#7)

24.576Mhz generic crystal. (#5)

Various regulators, bus buffers, etc.

As for #6...yeah...what is this? :)
 

Attachments

  • live24.jpg
    live24.jpg
    91 KB · Views: 691
J.R, offcource the PCI is 33mhz, i was confusing it with AGP.:headbash:
And i was conserned the bus frequency was the same as the pci bus itself, as the card would already have that clock available..

I think i have some 8620's, but they are dip i seem to recall, i'v been in contact with their sample department as well :D

We do have the option of testing the A V the non A cs4382.
i hear there are differences between pieces with different suffixes on other dac's. worth a shot.

and as you say, finding a BETTER dac, is a chalenge in itself, finding a drop-in is impossible it would think.

your electronics department sound like the same kind of bozos the mechanical department here are.
but hey, if you have problems getting it done yourselv, postage isn't much over the pond anyways, so perhaps we can work something out. ;)
i have experience with smd soldering, and good equipment.

good luck to you as well
-Marius
 
'lo all.

Demogorgon - It appears that I might've rushed my post. We don't, indeed, share the same DAC, but it appears to be at least in the right series (43xx) so I'm sure its at least similar. I hope. :)

Spartacus said:
The 4556 op-amp is rated for 70mA and is used for driving headphones. If you don't intend to use headphones you could simply bypass it.

Mmm. Well, I don't know about Demogorgon, but for me, it's all about the headphones, though placing a little bypass switch on the backplate of the card for a 'pure' lineout sounds like an interesting modification project.

Replacing the 4558s with AD8620s sounds a good idea, but remember that faster op-amps need better HF decoupling, so you should consider making sure their supply pins are bypassed with small value film caps. Also the DC specs of the 8620s are much better than the stock devices, so you should be able to replace any DC blocking caps with a wire link or solder bridge.

Mm...I've read up on the need for this...but I'm not sure on how to implement it. And there appears to be a whole, ghastly array of SMD capacitors and resistors in place already around the op-amps, so they may be decopled already. I think it's a 'solder and see' scenario.

As for the clock, simply remove the crystal and associated resistor/caps and apply the new oscillator.

The question is - does the oscillator only generate the clock for the DAC and ADC, or does it drive the Creative on-board chip as well? Or does it run from the PCI bus clock?

If the former is true, it's going to be hard to find it without a 20+Mhz scope and some time. It's hard to trace a signal on such tiny pins (inside an operating computer, mind).

I presume that the DAC runs on a 24.576Mhz master clock signal, as it's a supported frequency in the data sheet, and the crystal runs at the same speed. Can I simply put a new 24.576Mhz (or other supported speed) master clock signal into the MCLK leg of the DAC (see attatchment) and leave the L/R signalling leg (which presumably comes from the creative controller chip?

There are a plethora of other clocks, too.

One thing to keep in mind: simple mods like swapping op-amps are worthwhile, but there comes a point with upgrading sound cards where it becomes more practical to vuild an external DAC. Your decision of course.

Indeed, indeed. Though I think the modification of a sound card is more helpful for neophytes such as myself because you don't have to worry so much about the power suppy, surrounding components, etc. and replacing the existing components, hopefully, have the same support componentry arround them.

-T.
 

Attachments

  • cs4382a-2.png
    cs4382a-2.png
    35.8 KB · Views: 586
demogorgon said:
We do have the option of testing the A V the non A cs4382.
i hear there are differences between pieces with different suffixes on other dac's. worth a shot.

Be careful, see Flecom's post about DAC part numbers. I think that the Cirrus Logic DAC, certainly my CS4382, is rated for SACD decoding (in another mode, it's capable of transforming DSD streams) so I would imagine that the Audigy 2 / Live 24-bit DACs are let down far more by their clocks / output op-amps / power supplies than by their own internal designs.


..but hey, if you have problems getting it done yourselv, postage isn't much over the pond anyways, so perhaps we can work something out. ;) i have experience with smd soldering, and good equipment.

Lucky! :) Whilst this option is tempting, I have to develop these skills for myself, thanks for the offer.

-T.
 
J.R, have you come any further since my last post?

I shure have had my hands full, so nothing have been done on my end, and i need to get some other things done before any more modifying as well.

hehe, I have the cut out mdf for two peerless XLS 12' subs, the parts for a legacy gainclone (lm1875++), with the need for soldering and a chassi, a moped that's completly stripped down and painted before selling it, a bunch of tossed out audio+video stuff from work i have to go through, salvaging parts (there is insanly nice stuf in there)...

my hands full to be honest.
but let me know what your endevours turn into :)

-Marius
 
J. R. Todd said:
Also, would it be worth stacking Op-Amps in the same way that DACs can be stacked? Speaking of which...
I wouldn't do this. You can stack two current-output DACs - the output currents of the two DACs will sum.

However, op-amps are voltage output devices - connect their outputs directly with any voltage difference between their outputs and you'll likely get one amplifier feeding a lot of current into the other, resulting in hot chips, instability and all sorts of other bad things.

And connecting the inputs together doesn't ensure that the output voltage will be the same, unless the offset voltages of the op-amps are exactly matched. And when you actually put audio through the stacked pair, variations in open-loop gain and frequency response between two op-amps will cause the voltage outputs to differ.

I've seen cases where multiple power op-amps (eg, LM3886's) have been put in parallel - usually this involves the outputs being summed through series resistors, each chip being given its own adjustable feedback loop, and each chip having an adjustable DC bias on it to correct for offset voltage.
 
Hello, all.

The first mod is slowly coming together. Analogue Devices has been good enought to deliver a few free samples to my door, an AD8620 dual op-amp for the Sound Blaster, and a couple of AD843JNZs, just in case I ever go for the pure line-out idea and build myself a Chu Moy.

The real delay is in the aquisition of tools - poor students don't get the good tools, generally speaking. :) This should happen in the next few days.

GMarsh - Thanks for the advice...current feeding into another sounds logical, and very, very counter-productive.

Demogorgon - As it happens, I was just thinking of the salvaging
thing. Dead electronics must emerge from this University on a
daily basis...but where does it go? :) I wonder.

Best of luck in all your endeavours.

-T.
 
Having read of the very, very cheap trick of covering the clock crystal with blue-tack, I can confirm that this indeed proves the sound when applied to sound-cards. I simply placed a bigish lump of tack covering the crystal. Now, counting on the amount of vibration in a PC - anything must be an improvement, frankly. One forum member removes them, covers them in lead tape, and (I think) lays them sideways, on tack, on the PCB.

To my ears, it sounds like the upper end of the frequency range has been cleared out quite significantly, and general clarity has improved a little. Bass is much improved. Everything sounds a little 'warmer'. Is this normal?

Maybe it's just psychosomatic, I don't know. :)

Demogorgon, want to try this out? Even one as busy as you should be able to have time for this one. :D

-T.
 
J.R, i must admite, your description sounds much like placebo, sygarpill effect. i recon much of what you "hear" is just extra focus on the sound that was allready there.

been there done that. im my case it was the gainclone and it's snubber. carlosfm is a very convincing person ;)

but.. a drop of blue tac? why not? never heard of it, what is it? glue, silicone?

-Marius
 
Well, the arguement that I read regarding the damping of the crystal was that if the crystal is subject to excess vibration, the sine wave it generates is less consistent than it could be. Thus daming the crystal (covering it with a blob of blue tack) increases the stability of the sine generated.

This would, it follows, (I guess - in my very limited knowledge of these things) improve the consistency of the clock signal generated, and thusly a better sound. I'll try and find some quotes. People do, after all, mount their cd players, amps, DACs and speakers on heavy stuff to dampen vibration...might be worth a try.

Who knows. But blue-tack is simply the putty used to stick things such as posters to walls. :)

-T.
 
well, its my honest oppinion that people do a lot of very strange things in the name of belife ;)

not to mention what they are capable to accept to protect their belife.

I'l use some silicone\rubber glue and test. with a sceptic mind :D
I dont belive this works one bit to be honest.

I'l listen hard now as well. to avoid the most common pitfall, not really remember how it truly sounded before.

-Marius
 
I do not know if you have tried that already, but if not, it is a must:

One of the many things things which are not perfect about the SBlive - is the original driver. It actually completely and utterly sucks!
The SB drivers costed so much nerves and pain in the last years, it almost prevented me of using Creative products at all. But there is a solution:

The free open source kx-project driver is much better in all ways, from the sound (playback and recording, yes it works!) to the inbuild synthesizer and effects processor and to the general usability. Download it from http://kxproject.lugosoft.com/ and read all the help files etc. You will be fully amazed what the humble SBlive is able in reality.
It even (I tried that with an older SB live model recently) has a very smooth overload characteristic, because very short overloads are just softly clipped, unaudily mostly.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.