As a potential DA10 buyer, I'm interested to know how the unit does sample rate conversion when playing a CD standard 16/44.1 input.
Is the conversion synchronuous or asynchronous? Does it depend on the setting of the "wide/narrow/crystal" switch?

Does asynchronous sample rate conversion (by some called "upsampling") introduce aliasing noise or is that a "solved problem" or a myth alltogether?

I've looked for this information in various forums and the lavryengineering site without success.

Thanks in advance,

The DA itself is always operating in an up sampled mode which is synchronous. This is needed because it allows a real world design of an anti imaging filter. Prior to the concept of up sampling, the analog anti imaging filters were made of a lot of parts, but never enough to yield a reasonable performance. It would take dozens of opamps and precision resistors and caps to have the proper filtering, and while doing so, other things fall apart...

Also, there is another reason for up sampling:
a DA with no up sampling has very non flat amplitude vs. frequency response. In theory, a DA is perfect, because the samples are “zero width”, each sample with proper amplitude. But in practice, zero width samples, or very narrow pulses, carry very little energy, so the outcome will be very weak. A weak signal calls for a lot of amplification, which raises the noise, and that is undesirable.

So instead of narrow pulses, we go for a “stair case” waveform, where each value is held steady until the next sample. That practice (we call it NRZ for “not return to zero). We do so instead of the theoretical narrow pulses (we call them RZ because with a narrow pulse the signal between samples is zero most of the time).

Now, doing NRZ (stair case) solves the noise problem, but it brings on another problem – it causes some attenuation when you get to higher audio frequencies – nearly a dB at 20KHz if I recall (see my paper on Sampling, Over sampling, Aliasing, imaging under the support section). That roll off curve (sinX/X shape) is NOT something you can fix with an analog EQ (poles and zeros). When you up sample, that problem goes away. See the graph in the paper I recommended.

So the question regarding sample rate conversion belongs somewhere else - in the circuitry leading to the DA conversion.

The widely used way to accommodate low jitter clocks is to build crystals oscillator circuits for the desired frequencies (such as 44.1,48, 88.2, 96KH). Another way, is to use an internal fixed crystal, and convert whatever rate comes in to the internal crystal rate. That is done via a sample rate converter, asynchronously.
The jitter is good, but there is a tradeoff – the data itself has to be recomputed for the “new” internal rate.

The DA10 provides BOTH modes. Why? I prefer the older and more costly way of individual crystals for known frequencies, but including the SRC mode (sample rate converter mode) enables the unit to be used in applications other then pro and high end audio (such as broadcasting). SRC’s are getting better all the time, and as a rule, I am not against the use of SRC’s. In the DA10 case, you have both.

In either case, the DA itself is using up sampling, which is a synchronous SRC. Wide stands for asynchronus SRC.

Regards
Dan Lavry

Thankyou very much for taking the time to enlighten me with such a thoroughly reply to all my questions!

Any plans for a remote controlled version?

Wow - Mr. Lavry - I think I actually understood that - you should be a professor as well as an entrepreneur. I am looking forward to my DA10 which should be arriving in a week or so.

Let me see if I have this right.

The DA-10 has the multiple crystals to take various inputed smaple rates but can also do SRC like a computer, onboard, to "upsample" say a 44kHz input to 96kHz.

I do SRC at the PC and upsample 44kHz to 96kHz because I also have a bunch of native 96kHz content.

It's a quality SRC built-in to J. River Media Center, and they some very competent coders/ audio engineers onboard.

I'm sort of unclear as to why the DA10 then operates "always in upsample mode". What does it do with a 96kHz input before passing it to the DA?

JC

Ji JC

Almost all of the modern DA's operate at a very high up sampling ratios. In fact, the majority of modern DA's are sigma delta type, and those operate at X64 to X1024 upsampling ratios.

A DA that operatas at say X64 has to compute 63 out of 64 samples. A DA operasting at say X512 uses only around %0.2 of the "original data" and the rest of it is done by upsampling. There are a number of reasons why we upsample and as always it is a matter of architecture and other considerations with thier associated tradoffs...

There is a lot of concern regarding sample rate conversion, and much of it is "misplaced". The sample rate does take place, so the question is about the quality of implementation. If done well, the advatages are great, and if done poorly the resuts will reflect it.

I recently saw a post on another forum where someone stated that he gets much better audio when he upsamples the data to 88.2KHz before it is fed to the DA. His conclusion was that upsampling is a good thing.

But the fact is, the data is being upsampled in BOTH cases. When feeding 44.1KHz to the DA, the first upsampling stage to 88.2KHz is done by the DA. When first upsampling the data up to 88.2KHz, the DA can skip that first stage. In BOTH cases, the DA continues the upsampling to the final rate...

So the differenaces are not conceptual, they are about implementation. Indeed, the most critical (difficult) stage in the upsampling is the first stage (44.1-88.2KHz), so it is very easy to accept that some external upsampler can outdo some given DA. But that is not always the case.

It is correct to say that sigma delta type DA's tend to have higher upsampling ratios then resistor based architecture. There is no reason to fear upsampling or other sample rate conversions. One should go for good implementations, both in terms of computation issues and clock timing issues.

Regards
Dan Lavry

Dan,

What setting on the DA10 = individual synchronous crystals and what setting = async SRC??

Thanks!

aerial wrote:

The Crystal and Narrow settings operate at the fixed frequencies of 44.1, 48, 88.2 and 96 kHz.

In "Wide" setting, the DA10 uses asynchronous SRC and will accept valid input signals with sample rates between 30 and 200 kHz. This is indicated by all four of the sample rate indicators illuminating on the front panel of the DA10. A valid signal is one that conforms to the AES3 or IEC 60958 stereo digital audio format.

Brad Johnson
Lavry Tech Support

Thanks, Brad. All the best for 2009 and beyond!