ESS Sabre Reference DAC (8-channel)

[tritosine]

also the venerable pcm1704 with its 112 dnr must be real bad. For all that money ... what where they thinking...

[rossl]

Originally posted by NicMac
Why would a DVD player producer use a two-channel DAC?

use three of them.

[ismosys]

Originally Posted by InfiniteGain

Hi,

Does anybody know where to get hold of a unit or two of the ESS Sabre Reference DAC? The minimum dynamic range is 128dB!! THD better than -112dB any frequency.
http://www.esstech.com/products/digi...italaudio.shtm

Its the worlds best measuring DAC and supposedly the best sounding. The thing is, the DAC handles PCM, DVD Audio, SACD etc with unheard of quality and almost total lack of jitter. I was wondering if anybody may know how we can get our hands on them?

Thanks,
InfiniteGain

Hi, Can be obtained online through the European stockist www.ismosys.com where you'll also get tech support as necessary.

[OzOnE_2k3]

Hi all,

I finally got around to building the 8-channel Sabre DAC (design courtesy of Nicolas / NeoY2K), but I'm having problems with crackling noise...

I have the DAC hooked up to a Denon decoder via I2S (for DD / DTS decoding etc.), and I keep getting random digital noise?

I just improved the wiring a bit, and I'm now using the TP LCDPS to power the DAC itself, while the DSP board and CPU also have separate supplies. This has improved things a lot, but I still get the odd crackle now and then. This happens more often when the CD track changes (after automute etc.)?

The noise is definitely digital, and is caused by the DAC itself, and before I changed the wiring and power supplies, the DAC was VERY sensitive to interference - If I even touched the ground plane of the PCB, I would get "more snaps and crackles than a Firebreather in a fireworks factory!"

The setup is pretty much exactly the same as I had with the TP Buffalo DAC, but one main thing which bothers me is that this new design uses an 80MHz crystal?...

In the Sabre (ES9008) datasheet, it says that the minimum cycle time for MCLK is 13ns. This would mean that the MAXimum recommended clock would be 76923076.92MHz ??
EDIT: Sorry, I meant 76.923MHz of course!


ie. 1000 / 0.000013 ms

I can't remember exactly what was previously discussed about this stuff, so I searched the forum and noticed that Russ mentioned using a 80MHz clock for the Buffalo...

ESS Sabre Reference DAC (8-channel)

Is there a recommended maximum MCLK for the ES9008, or is 80MHz alright?

(This problem could be further complicated by a soldering issue with this board, but I'll save that for another post.)

OzOnE.

[Russ White]

The 80mhz clock in practice works just fine, and was actually the frequency suggested by Dustin in order to support SPDIF up to 192khz. I would suggest that the clock frequency has not much to do with your issue.

Cheers,
Russ

[rossl]

I agree with Russ, you problem is not the clock.. Most likely a bad solder connection. Are your I2S lines using twisted pairs and terminated to damp the reflections?

[OzOnE_2k3]

After posting, I read some more of the archives and saw that Dustin did indeed recommend an 80MHz clock and that a faster clock also improves the DPLL locking accuracy or something?

Thanks for the confirmation. It looks like the problem is to do with my source (a cheapo DVD player). I'm now using a Sony DVD player, and it's working MUCH better now.

I should have tried another source sooner, but the reason I didn't was because the cheap player works perfectly fine with my Denon amp, so I'm not sure why the DSP + Sabre setup has this "sensitivity" issue? (I tried both optical and coax connections to the DSP board with no improvement.)

The SPDIF receiver on the DSP board is actually a Sanyo LC89057 with it's own crystal (this chip also generates the main BCLK and LRCLK signals for the DSP). The symptoms do seem to suggest a problem with the receiver lock, so I'll have to grab the o'scope back from my brother and give the signals a poke.

btw, the I2S lines are connected via short lengths of mini-coax (between two and four inches each signal), and the shields are only connected at the DSP end. I think the I2S connection is fine now as there are no audible errors during playback with the Sony player.

So, it's working fairly well now, but I have to say that I did have the same type of issue when using the Buffalo DAC with direct S/PDIF input (via optical module or otherwise). The crackling would appear with some digital sources and not others, while these sources all worked fine on commercial amps?

The way I solved that problem before was to use the receiver on the DSP board? Now with the new 8-channel Sabre DAC and prety much the same setup, I'm getting the issue again with certain sources?

Also, even with the "good" source, I still occasionally get noise for the split-second after the Sabre comes out of auto-mute? (As a side note, I don't currently have a way of changing the Sabre's I2C config until my buspirate arrives, so I can't yet change the DPLL bandwidth etc. I am using the Volumite though, but the SPDIF issue exists with or without.)

I'll try separating / upgrading the power supplies even more.

OzOnE.

[pftrvlr]

OzOnE,

My TP ES9008 DAC has the exactly the same issue. It seems the DAC has problem re-locking the SPDIF signal when track changes. It is better when config to I2S.

[OzOnE_2k3]

Hi,

Well, that's the problem really. I'm using the separate SPDIF receiver to convert to I2S for the Sabre.

I've done some more testing, and it's working better, but still crackles when the signal first locks on and the Sabre comes out of auto-mute.

This happens with any type of software (PCM, DD, DTS), and it's the same sort of noise each time, so I don't think it's the receiver chip because the DSP would have started decoding to I2S way before the crackling stops (if that makes sense?)

The crackling noise disappears after around 400ms and playback is then perfect (excellent quality), so it's probably not caused by data corruption as such.

What I will do to confirm if it's the data or not is to power up the original DAC on the DSP board (AD1837) and see if the noise is still present. This will confirm if the I2S stream is good. I just need to build a +/- 15V supply for the DAc's opamp I/V stages (the original PSU parts were scavanged for something else).

I've attached a photo of the DSP board with the Sabre DAC added to it. Yes, I know that it's still quite messy, and yes, I know that Kynar wire can be evil! (this will be replaced by proper wire and mini-coax very soon).

The DAC does not have any I/V stages yet, but the FL / FR outputs are connected directly to some Sympatico modules, so should (in theory) be OK for testing.

Some of you might also have noticed that the Crystek module is hovering over the PCB slightly. This is due to a small PCB layout error (pattern was mirrored), so I had to place the module sideways, and use Kynar wire to reverse the power connections. The clock output is connected directly to the PCB pad via a very short piece of wire (around 2mm), and everything is quite well shielded and insulated.

OzOnE.

[rossl]

Originally Posted by OzOnE_2k3

btw, the I2S lines are connected via short lengths of mini-coax (between two and four inches each signal), and the shields are only connected at the DSP end.

The mini-coax should be connected to signal and ground at both ends, otherwise you will have an impedance mismatch. For the digital signals, anyway. I'm not sure how your analog is connected.