Article: Berkeley Audio Design Alpha USB Review

[easternlethal]

I don't understand. CA had this unit for four months and instead of comparing against other similar products, he uses that time to compare between different sources and DACs. <br />

<br />

Then, when it comes to comparing against other converters he uses units that cost half as much at most before concluding a) another unit which costs $1000 actually comes close, and b) he knows of no converter better than the Alpha USB. <br />

<br />

Firstly I think a lot of users would actually not mind paying nearly half the price of the Alpha USB for another product that comes close (despite what he says about close not being 'good enough'), secondly I am not surprised he thinks it's the best converter since he only compared it against cheaper products. Did he even compare against any products in the same or higher price range, like the sonicweld diverter? If he did then why didn't he mention it?<br />

<br />

Sorry but this is as faulty a review process as I ever seen.

[easternlethal]

Okay you want us to treat your review as subjective opinion. That is fine. <br />

<br />

But you dissed the practice of comparing components and said they are a disservice on the one hand then on the other you say in the article "[Alpha DAC + USB] is possibly the best digital I've heard in my listening room". So what... proper comparisons are wrong, but incomplete impressions based on subjective opinions are okay. Is that the kind of style you want us to get accustomed to?<br />

<br />

Secondly you said to Mark Powell regarding avoiding spdif conversion that this means he is satisfied with mediocrity and not pushing the boundaries of engineering. Does this mean that developments in converting usb - spdif - i2s = pushing the boundaries of engineering, wheras developments in converting usb - i2s directly are not? I think XMOS and all the other DAC manufacturers and manufacturers of USB - I2s converters (such as Emprical Audio) would disagree. <br />

<br />

Here's another bit of honesty. BA developed the USB because their DAC has a SPDIF-I2S converter and they just wanted to make something else to sell that can improve the sound without overhauling this architecture . It has nothing to do with advanced engineering or even whether usb - spdif - i2s is the best method of getting data into the dac circuitry (I doubt whether BA themselves would know).

[easternlethal]

I'm not sure why you're interpreting my comments in this way. Yes I have personal feelings about this site and you in general, but I have not expressed them here or anywhere else (and I am not about to discuss them with anyone either).<br />

<br />

I have personal feelings about the review as well, but I did not express them here either (only on the Phasure site in passing, which has nothing to do with my comments here). <br />

<br />

Here, I only made observations about the review process and the conclusions you drew. These points are still valid no matter what your opinion is about me or my motivation (I am not going to comment about this either).<br />

<br />

The fact is, quite a few people have come up against the question of whether to upgrade their dac or improve its connections by getting a SPDIF converter like the BA USB. There are DAC manufacturers who think this is not the optimum approach if you don't want to keep the DAC (and I think Peter is one of them - which is why I wanted his views so I could understand why). Your review, suggests you prefer BA's DAC+ USB approach. But it does not provide any rationale behind it and I was really just pointing that out. <br />

<br />

At the end of the day the question still remains over whether USB-SPDIF-I2S is better (implementation issues aside). Perhaps for the BA USB+DAC it is, because the USB/SPDIF conversion can be isolated more easily than USB - I2S. Perhaps its DAC clock deals better with SPDIF jitter than USB jitter. Either way I think it would have been better to explain this in the review (especially if you have discussed with BA engineers which I have not) as opposed to just reiterating that it is better than the other alternatives you had at the time and that therefore it must be the best solution. The BA guys might be very well meaning, but I am still skeptical over whether or not they would have come up with this solution if they could reinvent their whole DAC all over again. If you disagree, then it would be great if you could tell us why.<br />

<br />

And sorry for hurting your feelings. I thought this forum was set up to allow us to express our thoughts. Let me know if that is incorrect and I'll express them elsewhere which does.<br />

[easternlethal]

I have a weiss dac2 and was thinking the same thing. From my experience the firewire driver of this device leaves quite a bit to be desired and one would need to decide whether to upgrade the connection or just sell the dac and use that money (plus the money you were going to spend on the BA USB) to upgrade to a better<br />

Dac. This is why it's important to understand the merits of the different data path options available (usb-spdif-i2s, soundcard-aes-i2s, usb-i2s or firewire-i2s) as well as other reasons that may have nothing to do with spdif such as cleaner power or better mechanical isolation.<br />

<br />

So a more helpful test might involve comparing first with a pure spdif converter like the hiface or offramp (no batteries) to see whether there is an improvement, then if so, comparing that result against solutions that offer isolation as well as spdif such as the BA USB and sonicweld diverter and then finally against an isolation solution that does not offer spdif but does offer isolation like the vertex hirez (using the original firewire). This type of comparison should help you decide on technology as opposed to just what 'brand' is better. And if you do it for spdif + power solutions then you'll be able to rank your preferred solutions and assess audiophile components that way instead of just taking cues from the subjective opinions of other people. Who knows... After all that you might even find out it's the weiss's burr-brown dac chip you don't like and end up upgrading the dac.

[easternlethal]

If only one component in a system at a time is substituted there should be no difficulty in determining the effect that component has on the system no matter how holistic your view is of the total system. There are several technologies deployed within the BA USB and my belief is if you want to understand the overall impact of that unit you should understand the technologies first and foremost by testing against alternative products including only that particular technology (sort of equivalent to scenario analysis). Then having an approximate appreciation of alternative products and the effects of each technology, it's possible to hypothesize on how your system might sound once if those technologies were adopted. Then, with that hypothesized idea, compare against the actual result achieved by the unit. If the sound is different from what you expected then you will know it is not the technologies tested that made the difference, and if such difference is an improvement over the original hypothesis then this residual difference could be implementation or some other technology that was not compared (such as superior quality or construction etc.). Then at least you'll start to approach some ability to quantify the difference instead of just knowing something is different without understanding why. Either way it's a scientific process and something manufacturers themselves do when performing r&d. To get an idea take a look at the program on sean olive's blog, which is used by harman kardan to train listeners to evaluate components.<br />

<br />

But not many here seem to care much about the process and just want recommendations from chris or anyone else - who incidentally would not even be able to tell what effect something will have on anyone's system (only his). <br />

<br />

This observation i am making on the lack of a scientific process (which applies to the way most people and even manufacturers evaluate their systems - including the specific review above) will no doubt be taken as a personal attack and people might even say they are not interested in it or whatever, but so what. It is still a valid point.

[easternlethal]

I tested the BA USB + Dac on lamm ML3s and Magico Q5s and i have to agree - an improvement over the original dac only solution but still sounding too compressed for what it is.

[easternlethal]

@chris<br />

<br />

Most people who are not reviewers would perform their 'tests' in the following way - by going into a shop and hearing it in the shop's system. They are not familiar with the overall setup, but by swapping out the particular unit and listening to the differences they start to build a picture of how it sounds. Unless they are lucky enough to be in a shop that has all components they want, they usually have to repeat this process in several demo rooms using different components and then mentally regressing the results and hypothesizing on how the components they actually want sound together. And when they buy the component(s) and assemble them at home quite a few people actually achieve a very good result. Equally it's also possible for someone to take a component home to test against his own system, decide they like the new component more, buy it, then realise afterwards they could have achieved a similar or better result by tweaking their existing system in a different way at a much lower cost. Faulty conclusion too. So it is not the equipment but the tester.<br />

<br />

I did my testing in a room at a local store where I do a lot of testing (as much as I can when new components come in) so I am familiar with the sources and components. I also tested it against a Ray Samuels Raptor and Audeze LCD2 which I own (and brought in). The setup (with the lamms and magico) was recommended by the owner. It did not sound bad but it turned out I did not get a different result using my audeze compared to the magicos. Transients and detail was good as was the frequency response in the mid-range and bottom end. And since the source was a laptop running cplay (usually noisy and bright as hell) I thought the electrical isolation in BA USB would have helped with the dynamic response - but it did not do much. Maybe it will perform better in another system which needs the isolation but it did not do much in these ones. It seems bobbha came to the same conclusion as well with the Offramp (which does not adopt electrical isolation but I think has better oscillators).<br />

<br />

<br />

[easternlethal]

I am not angry. I just like being direct. As I said before I have not made any comments about the general usefulness of Chris's reviews or this site.<br />

<br />

My biggest concern in getting a new BA DAC would be the limitation on its sampling rate (at 192hz). If I were in the market for a new DAC I would at least make sure it can play DSD or 384hz (after all this is the one advantage digital music has over analogue - whatever your feelings about that might be), if I wanted to keep the dac for more than 3 years.<br />

<br />

Anyway in answer to your question about SQ I think BA DAC employed the same technology (or at least general technical approach) in the USB unit than with its DAC (i.e. electrical isolation + clocking). I said in my post above, the system did not benefit from this but what I really meant was it did not benefit it in the way I liked (because of course, more isolation always benefits).<br />

<br />

I feel its dynamic range and response could have been better and it was not as full in the upper mid range (surprising given the help it had from the valve setup from the lamms). But it improved transients and has good frequency response - so it is up to you which characteristic you are going for (and type of music you like). I personally felt the extra isolation was not required because the characteristics in the dac remains and will never be improved without overhauling the dac.

[easternlethal]

@accwai - sorry it was the Q1. I got confused with the magico model numbers (it's hard when one can't even tell what they look like from their site). Yes ML3 wouldn't have been able to drive those big units. The general characteristic... hmmm lets just say I think I would have preferred something with higher sensitivity like horns :-) but that shop does not sell horns so... <br />

<br />

@chris - i said I did not make any comments about the GENERAL usefulness of your review or site. This was a response to Paul's comment about your previous recommendations being helpful as well as your own earlier comments about me not liking your 'service'. Why do you insist on broadening the scope of what I am saying? I criticised your specific review of the BA USB unit and that was it. So if you think I'm trolling, then fine. I'm trolling. My comments are STILL valid.

[easternlethal]

I didn't say the Berkeley sounded dull (that was somebody else). I said it lacked dynamic range and response (not quite the same) and the USB solution did not really improve on that. I'm not sure what you mean by 'lit' but if you mean frequency range and response and transients then I agree (the USB did improve on it). I'm not saying DAC+USB is no good at all. Just not good enough for what it charges.<br />

<br />

I just think low wattage SETs benefit high sensitivity drivers because they are extremely detailed and the 'greater acoustic output' of horn topology gives all that fineness and sensitivity the 'oomph' without destroying too much of it. Something loudspeaker cabinets just don't achieve as well (even for the very best designs).<br />

<br />

What do you mean by 'front-end'? Do you mean source or other converter? If converter I think sonicweld is better and I am very curious about the off-ramp. If you mean source then that'll be for another discussion. But if I had to change something in the ML3 - Q1 combo I would not change the 'front-end'. I would ditch the Q1 and get something better (like a Cessaro).

[easternlethal]

Interesting to hear about the off-ramp. Would you care to speculate on why it is better? It is the clocking or power do you think? Based on comparisons with the hiface board inside the pdx (which means data path = usb-i2s?) I suspect it is clocking, which supports my earlier statement about better oscillators (further because both converters use two oscillators). It also supports my belief that Berkeley's approach to spend all that time on electrical isolation was not the most optimal way to improve usb conversion (off-ramp employs none i don't think). I haven't been able to find out about what the 'turbo clocks' are.

[easternlethal]

I think you can try upgrading the converter first with something cheaper like a hiface to see what kind of improvement it brings. Maybe that will help with understanding the 'trajectory' of going down the converter route vs upgrading the dac. <br />

<br />

In our hi-fi world, most of the current engineering breakthroughs are happening in the dac-related space, which when compared to amps and speakers have not reached the same perceived level of importance (similar to amplification circuits in the early days). That, and the rapid pace of development in our understanding of digital data transfer means - short product life cycles + components selling for relatively less compared to other components + rapid improvements in performance (small diy manufactured dacs of today can outperform the majority of the best and most expensive from large manufacturers made 1 or maybe only 2 years ago).

[easternlethal]

One would expect betters result using Alpha USB than without (just Debussy) because of more isolation + separate clocks.<br />

<br />

Using the USB, between the alpha dac and dcs the difference in technologies I can surmise are: dsp filters + mechanical isolation (dcs) vs electrical isolation + separate clocks (BA). Again one would expect BA to come out on top because it has a superior clocking solution, which I think the best way to address audio transfer, and dsps are a big no-no.<br />

<br />

It would be great to compare the results against this hypothesis. If dcs comes out on top we might have to hand it to them for superior implementation or the better quality components.

[easternlethal]

sorry - I forgot another big difference. custom dac chip (dcs) vs. delta-sigma chip(BA). <br />

<br />

hmm... <br />

<br />

@mark powell - reputedly. because dcs makes their own chip. but the technology is not fundamentally different (they still use clocks, FPGA, circuit boards etc...) only the design and layout is different.

[easternlethal]

I didn't think using an FPGA to implement a dac was that unique. Xilink, Entegra have done it as well as diy'ers in the past (even incorporating PLL and DSPs).<br />

<br />

But I dcs have a patent so I suppose something must be different. Question is.. what? and why would changing the topology of a board (something that can be done inside a single chip) improve anything? would like to look into this somemore..

[easternlethal]

ha ha.. that is too cynical. <br />

<br />

This is a temporary solution for BA. I bet anything they will produce an Alpha III in the future incorporating USB, that will cost more than Alpha DAC II + USB. They will not be able to not produce non USB dacs. Even Weiss (longtime firewire proponent) has started incorporating USB into their 202. Even Phasure! Is this an 'honest' strategy? Well... it does benefit existing users.<br />

<br />

Anyway I don't think it's whether it's inside or out that is the issue. I think it's whether USB is implemented properly. If so then a USB converter would not do much. And I guessed dcs would benefit from the BA USB because I didn't think dcs implemented USB as well (their efforts were spent on the FPGA). speaking of which... doesn't NOS1 also use FPGAs?

[easternlethal]

All the site really reveals is the magic's actually in the code (implying it can also used to manage sample rates). Feel free to share if you have anything more.

[easternlethal]

thanks for posting the result. seems my prediction was correct and dsc was not saved by its FPGA after all.

[easternlethal]

to be accurate, i called it a hypothesis. but who cares. <br />

<br />

I mainly just want to demonstrate the process of understanding the technology, hypothesizing and then comparing against actual results, which I don't think is done enough amongst audiophiles. And I think I've shown that as far as can be done in a forum.<br />

<br />