Audigy upgrade project for quality USB DAC

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I will be upgrading a Creative Labs Audigy USB sound card. It is 6 channels and has built in Dolby digital decoding. This is useful if you want to create a digital crossover for your speakers in software on your PC, or if you want a cheap surround sound decoder. Since yesterday I have established the following:

1. Current DAC chip is useless
2. Access to MCLK, BCLK and 3 times SDATA for 3 stereo channels is easy.
3. Possibly inflexibility with sampling rates, possibly only 48KHz???
4. Looking for schematics or anything else which might be useful, who can help? Schematics not required for simple conversion planned, though.

Plan to create a better power supply, perhaps a dedicated one for the DAC section.

Also plan to build separate or built in DAC's of the new breed. Currently I am eyeballing the TI PCM1730 -- although not as highly specced as the new AD unit it does not require all the support circuitry typical of AD units, and possibly even quartets of them per stereo channel.

Also plan on upgrading crystal and possibly changing frequency to allow for different sampling rate. Certainly the oscillator can be positioned close to MCLK on the DAC's themselves.

Upgrade connectors so that we get true balanced XLR out.

This is just too easy -- anybody want to participate?

Agree on the power issue, but with the incredibly poor quality of the DAC chip itself it is kind of hard to tell whether it is the power supply that is the big problem or the DAC chip itself ;) Thanks for the input, though!

Off topic: I have almost completed the (not X topology) Balanced 2 stage DC couple stage with 75dB gain. At this point I am battling with myself as to whether I should use a passive RIAA which requires series capacitors to avoid very high gain at zero Hz (high pass filter) or whether I should do a split RIAA (active and passive) type setup to roll off low freq. gain through single stage feedback instead of the also dreaded series cap.

Findings from my measurements of a real Extigy

TGhe DAC chip is Philips UDA 1328T. Sound pipeline 48KHz (not yet sure how many bits, my scope is not fast enough). The 48KHz rate is a disappointment.

Status of measurements for transplant

Static pin on DAC chip is high. This means that internal volume control etc. is not used. This is incredibly good news!

In standby mode without connection to computer:

SF1 (L3MODE) Low
SF0 (L3DATA) Low

So what is going on here is that we have I2S bus input. Since PCM1730 (chosen DAC chip in my case) requires knowledge of how many bits are presented, it might be necessary to make som form of manual switching arrangement, but my guess is that the sound processing pipeline always provides the same number of bits so this is hardware programmable. Unless I get access to a faster scope, I might well have to try this out by ear.

Note: WS==L/RCLK in normal terminology ….
Frequency of this is about 48,044KHz which is our actual sampling rate. This is not very pleasing, but it is possible that this can be changed with software to 44.1 or if we must change rate – at least go to 96KHz immediately. This however does not seem to work. Perhaps I can do it by changing the crystal. MCLK is 12,287020 which is fine – 128fs. Now this is kind of strange – the master crystal oscillator in the system is 11,0592MHz which is not a multiple at all. I suspect this is simply the cheapest crystal available. The master clock is received by DAC chip from Creative CA0181-DCT chip through R104 (zero ohm resistor) should be about pin 48 or something in that region :) (counting such pins is not fun!). The good news is that this means that Creative can possibly change this clock in software. I briefly thought about changing the main crystal to 10,16064, 15,24096 or 20,32128 if one really wanted to give it to them. If everything worked which there is no guarantee, it would be possible to change the bit rate to 44,1KHz and do it with 256FS or 512FS. I really have no idea if this will work, but hey look, I have an oscillator here :) (but unfortunately it was not a MHz oscillator so I was unable to try it :(, and I only found that out after removing the board from the case :() Now that I think about it, it might not be necessary after all since there will be resampling going on anyway so if anything, the route to go is up – ie potentially doubling to 22,1184MHz to get 96KHz playback -- i.e really give it to Creative ;). It is quite possible that the chip designers at Creative have designed the chip so that it understands the different frequencies coming in/out etc, so it would be extremely interesting to try this out!

If you have an oscillator and an Extigy, the voltage across the crystal is 3.5V peak to peak. Please consider trying it out.

Volume control:
So since Creative obviously are not using the internal volume control of the DAC chip, how do they then control volume??? My guess is that they use a digital volume control and play with the data word (divide by whatever). Such division is most likely to be a part of the box since the box can be standalone. Now, it is not obvious from the volume control to what it is connected. This volume control is a rotary encoder. Minimal component count near it is probably just debouncing caps or more likely pull-up resistors. They could possibly have been messing with the DAC reference voltage (Vref) but this is neither recommended (DAC performance is usually optimized for one level, at least when we are talking about real DACs and not this Philips crap where it might not be so important) nor is this actually done according to my measurements.

What I think is going on is that Creative takes in whatever they get, divide to correct level and output this in 24 bit format over I2S protocol. This needs to be verified, though (in particular number of bits transmitted).

So all in all, it appears easy to graft in new high quality DAC chips. Now it is just a matter of doing it :)

In case you wanted to know what I want to use this DAC for, the application is Digital PC based crossover ...

In case you wanted to know what I want to use this DAC for, the application is Digital PC based crossover ...

That sounds pretty interesting all by itself - I'm considering using a PC based home theater setup. I'm currently using an SB Live! but will upgrade to an Audigy for full 5.1. I know about the hack to do 5.1 with an SBLive, but I believe the S/N ratio is better with an Audigy.

Your digital crossover sounds very interesting, let us know more!
Digital crossover

The plan is for a PC based digital crossover that will take a stereo signal sourced by the PC, re-equalize for "correct" in-room response + do the crossover job digitally inside the PC. The Extigy will be the output device, or at least that is the current plan.

I am not doing the software myself and am thus waiting for others. If successful (and if distribution rights to commercial stuff used is obtained) you will all get access to it.

Right now, all I can do is focus on the hardware improvements.

More measurements

Crystal frequency is 11,0592MHz
CLKA = 12,28701MHz
CLKB= 11,288691MHz
CPUCLKC – 11,999434
CLKD 24,574036MHz (quality issues?)

So Creative are making it incredibly easy to try out 24?/96. I am now seriously considering overclock the CA0181. The current original DAC chip is able to handle 128fs so I must patch to this chip also. Assuming this works, it should be possible to use 128FS since other DAC chips seem capable of 128FS.

Heat. If the CA0181 is overclocked to 200%, it might need some cooling. This should be easy to implement, and I am not even sure that such cooling is needed given that the chip does not run particularly hot anyway (warm to the touch unlike certain other chips). Landing Clock D to R104 would be just about all that was needed .... Nice one, Creative!

Clock B is particularly interesting in that it is 256fs at 44100. That probably means that Creative at some stage intended to do true 44100 playback – I just don’t see where else they might need such a clock – recording perhaps but that is also unlikely.

So, if the CA0181 has a sample rate converter included this will likely work. Now if only there as some minor technical documentation about these Creative chips free product improvements would be much easier to donate to Creative Labs.

Digital Remark

Conversion from 44.1k to 48k is not as straightforward as it seems.

Notably the DSP must use good dither in order not to destroy the sound.

I'd be very suspicious of the algorithms in this DSP.

Consider an EGO-SYS WaveTerminal for instance. It has 8 channels of 24*96 out on an USB wire. And it comes with the good AKM dacs.

Cheers !
I dont belive 24/96 is possible due to limitation on the USB port.

6 Channels @24/96 requires approx. 13 Mbit/s. The current USB standard is fixed at 12 Mb/s.

I have not been able to find any other USB device that supports multiple channels at 24/96.

It would be more fruitfull to look at cards like M-audio Delta 410 which is priced comparably to the audigy. Other options are RME, ESI, etc. Some of them provides the D/A as a sepearte physical box ( though more costly)

What kind of software are you going to use for the Digital crossover and room compensation ?


This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.