[NewBuyer]

For me, the biggest selling points in a USB digital computer-to-hifi solution include galvanic isolation and ASIO output (for bit-perfect transfer). In fact, I was just about to purchase the M-Audio Transit (currently available for under $70) which reportedly does both, using ASIO drivers and Toslink output.

Then I learned about the HagUSB, which does the same, but using transformer-isolated coax output (instead of Toslink), and the native Windows USB driver along with optional simultaneous ASIO4ALL wrapper. It is also roughly twice the price of the M-Audio product.

I assume that a design choice was deliberately made here, to use transformer-isolated coax instead of Toslink. I'm just genuinely curious to learn the reasons for the coax design choice please, as this will help me to understand the benefits and make a more informed purchase decision.

So is there any technical reasoning you could please share, that could help a potential buyer like me to choose your product, instead of the other less-expensive alternative?

Thank you in advance for your help and advice...

[hagtech]

I'm not sure what they use in the m-audio, but the HAGUSB comes with some premium quality parts.  It's the sort of thing we do in the high-end audio world.  No scrimping for minimal cost.  In fact, the output signal transformer costs me nearly $10.  It's a very wideband design with tight coupling for maximal signal transfer, fidelity, and impedance matching.  All important things in order to deliver the transitions to the receiver properly.

The TOSLINK optical connection is nice because of the electrical isolation, but suffers from very poor bandwidth.  It is a super low cost consumer interface.  The coaxial one was way too expensive.  Sure, fiber can have much higher bandwidhths (I've designed boards running optical transceivers at 3Gbps).  But not in the case of TOSLINK.  Hey, even S/PDIF has it's issues for signalling.  All interfaces have their bugaboos.

And I didn't want to design a consumer product, either.  So I chose the high-end audio solution of 75 coax.  The problem then was a loss of galvanic isolation - except the expensive tranformer restores that.  So I get the best performance for both parameters.

Sure, there are things I could do to make the circuit even better, but the costs really start to add up.   

jh

[NewBuyer]

Hey, thanks for the reply! May I please get your thoughts on this follow-up post? I apologize in advance if my question seems too basic, but I'm very grateful for your advice.

You explained that a transformer-isolated coax output is more expensive to make than a Toslink output.

The HagUSB outputs data rates at most 48kHz @ 16-bit. My understanding is that 2-channel 44.1kHz S/PDIF requires less than 3MHz bandwidth, and that standard plastic Toslink cable allows up to 5-6MHz (good quality Toslink cable up to around 10MHz).

This *seems* to imply that in this application Toslink bandwidth is not really an issue, yes?

So since Toslink gives ideal galvanic isolation, is less expensive, and has plenty of bandwidth to spare for this application, I still am confused why the buyer would wish to pay more for the HagUSB.

Please help me understand? I really appreciate your help, and I'm sorry if I'm missing something here...

[hagtech]

The issue is that S/PDIF delivers more than just bits.  It also encodes the timing.  If all you need is bits, the 3MHz is more than enough. 

Now think of it as a clock.  Do you want rounded edges full of noise triggering the receiver?  Timing becomes ill-defined and you end up with a lot of jitter.  So in layman's terms, it's all about jitter.  You can add a lot, or you can add a little.

Plus, keep in mind that just because a toslink cable allows 10MHz signals, it doesn't mean you're getting it from the driver and receiver.  Every link in the chain makes a difference.

jh

[NewBuyer]

Thank you for these replies, I really appreciate it! It's great to receive your help and advice. I'm hoping you don't feel these questions are too low-level.

So if following the USB-to-S/PDIF converter, one uses an upsampling/asynchronousDAC or any subsequent reclocking device (that strips the timing clock and keeps only the bits, then reinserts its own clock), I wonder if this would negate the coax benefits you described?  Also, about electrically noisy data edges mis-triggering a DAC digital receiver: Wouldn't this only apply to a non-isolated wire (coax) connection, but not a light-fiber pulse connection? And since the HagUSB isolates its S/PDIF output (kills groundloops), but does not isolate its USB input, wouldn't computer powersupply/device noise already induce these described signal jitter problems within the HabUSB's internal digital circuitry, before the signal even reaches the HagUSB's transformer-isolated output?

Again, thanks so much for your replies! This is really helping me, I hope it is helping others too...

[hagtech]

would this negate the coax benefits you described?

No.  But it can help a lot.  Once you add in the crap, some of it is really hard to remove.  The chain works better if you never add the noise in the first place.

but not a light-fiber pulse connection?

Why would these be any different?  You can do signaling with electrons or photons.  You can even use dominoes.  Or water or sound or some other medium a signal can propogate through.  A wavefront is a wavefront.  There is nothing sacred about photons.

And don't forget, you have the electrical to optical conversion process, plus the reverse at the other end.  All of these processes and circuits add noise and reduce bandwidth.  Look, there is no reason you couldn't make a superb optical connection.  It's just that toslink ain't it.

computer powersupply/device noise already induce such signal jitter problems

Indeed, you have the same situation in both cases.  Clearly you don't want the raw power line from the computer to directly drive the circuits.  And that's why I use special filtering prior to the linear regulators, to reduce the effects of such noise.  Then I add extra wideband decoupling and layout tricks to optimize performance.  Sure, like I mentioned earlier, you could take it further by throwing more circuitry and cost at it.  It is not hard to make the HAGUSB cost two or three times as much.

jh

[NewBuyer]

Thank you Jim, I really appreciate your time and information. You are very generous with your replies!

To recap, you are saying that for 2-channel, 16-bit, 44.1kHz S/PDIF, Toslink (compared to coax):

1) Is less reliable for acceptable transfer of such data, due to insufficient bandwidth;

2) Has an LED emitter/receiver that produces electrical noise and bandwidth-restriction, corrupting data transfer through induced jitter;

3) These problems remain even when such signal is received by an upsampler, or a reclocking DAC.

I *think* the above summarizes your points accurately - if I am misunderstanding you anywhere,  I trust you will please correct me.

Thanks again Jim! The HagUSB looks really nice.

9/22/06 Edit: Your site says the HagUSB works with Windows XP. Does it also work with Windows 2000?

[hagtech]

Does it also work with Windows 2000?

Yes, but only with updates.  There is a list of operating system compatibility at the bottom of the data sheet.  Search www.ti.com for the PCM2704.

jh