• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Measurement and Review of Berkeley Alpha DAC

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,955
Location
Seattle Area
#1
This is a detailed measurement and review of the Berkeley Alpha DAC. The DAC retails for $4995 through dealers. It is an older design and this review is based on a prototype unit prior to manufacturing. There is now a version 2 with better clocking, etc.

Berkeley Alpha DAC.jpg
NOTE: My company (Madrona Digital) is a dealer for Berkeley products. And I led the acquisition of their previous company, Pacific Microsonics (makers of HDCD format) into Microsoft. So I consider the founders professional colleagues. Feel free to read as much bias as you see fit in this review.

The Berkeley DAC as you see above has very good connectivity sans the important USB input. They believe the digital noise from that interface must be kept outside of a DAC and make a USB to serial digital box. Outside of that omission, the box comes with remote control and adjustable volume (controlled in software). Everything internally is asynchronously resampled, putting the DAC clock chip in charge (rather than tracking the input).

Let's see how the device performs.

For explanation of what these measurements mean and the setup, please see: https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/

Measurements
Frequency response is essentially rule flat:
Frequency Response.png


There is a tiny rise however due to design of the digital filter. Note that due to heritage, the Berkeley Alpha DAC supports HDCD and with it, reduces the output level by 6 dB at 44.1 KHz sampling. It does not do this at higher sample rates. So if you are doing AB comparisons against another DAC, you better pay attention and level match. Otherwise the Berkeley will sound more anemic due to potentially lower output at 44.1 Khz.

Next let's look at noise and jitter and compare it to my other pricey DAC, the Exasound E32 (retail: $3,400):

Berkeley Alpha DAC Jitter Measurement.png


The Berkeley produces a very clean noise floor whereas the Exasound has those low frequency wiggles plus a pair of tiny sidebands which indicate jitter. The Exasound noise floor seems better but that is due to its output also being lower. Compensating for that, they are both equal. So here the Berkeley wins.

Next let's look at harmonics and noise from a 1 Khz tone (with the tone filtered out):

Berkeley Alpha DAC distortion and noise Measurement.png


The Berkeley has a single third harmonic and then remarkably clean until we get to higher frequencies where a few peaks show up. The Exasound E32 (in green) has much lower noise floor which then allows its harmonic distortions to be more visible. It also suffers from the same closely situated spikes as we saw in J-Test.

Note that it is possible the Berkeley level and Exasound were not matched so the difference in noise floor may be different.

Hard for me to judge which one wins here. Neither produce audible concerns as the distortion spikes are at extremely low level (-120 dB and lower).

Let's now look at my favorite, linearity test:

Berkeley Alpha DAC linearity Measurements.png


The Exasound E32 pulls ahead here with one more bit of resolution and error that seems to be noise dominated. The Berkeley on the other hand has a chewed up deviation that rises higher than Exasound. So the Exasound wins here.

Let's see the same response in time domain and see how well a very low amplitude -90 dB sine wave is recreated:

Berkeley Alpha DAC low level linearity Measurement.png


The higher noise floor seen in Berkeley output earlier shows up here making our sine wave look jagged. The Exasound on the other hand creates a beautiful facsimile of a sine wave. So it wins easily here.

Now let's look at intermodulation distortion:
Berkeley Alpha DAC SMPTE IMD Measurement.png


Here both Exasound E32 and the much cheaper Topping DX7 pull ahead of Berkeley with lower distortion at all output levels.

Conclusions
The Berkeley is clearly a competently designed DAC. It is showing its age though in lacking USB input (which today is a mandatory input) and somewhat lower performance in objective measurements. So it is hard to recommend it to purchase now over the newer entrants.

P.S. Yes, you see references to Schiit Yggdrasil in the above graphs. :) When I show the performance of that DAC, I will include comparisons to Berkeley Alpha DAC.

As always, all feedback, questions, corrections, etc. welcome.

If you like this review, please consider donating funds for these types of hardware purchase using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
 

Attachments

Last edited:

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
476
#2
Another job off the to do list. As my teachers used to say, "keep up the fair work", as opposed to the good work others, apparently, were doing.

Seriously, another superb job by the King!

One thing, though. Was the Exasound measured via its spdif or USB inputs?

Since I own an Exasound e28, I am also concerned by the (small amounts of ) mains and related noise consistently in your E32 measurements. Exasound now ships its top DACs with a TeddyPardo 12/2 LPS. Do you use that, their stock SMPS wall wart, or your own PS?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,955
Location
Seattle Area
#3
I tried to match the tests by using S/PDIF on Exasound but I am not sure I was consistent in all the tests.

On Exasound, mine is with the little switcher supply. For this price range, they should have never shipped it with this unit. I will re-test with a linear supply and report back.
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
476
#4
I tried to match the tests by using S/PDIF on Exasound but I am not sure I was consistent in all the tests.

On Exasound, mine is with the little switcher supply. For this price range, they should have never shipped it with this unit. I will re-test with a linear supply and report back.
Thanks, Amir. Seems like you got the SMPS wall wart, same as I use.

Ok, back to our regular program, the Berkeley.
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,462
Likes
1,401
Location
Riverview, Fl
#5

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,023
Likes
1,478
Location
Monument, CO
#7
@amirm: "The Berkeley has a single second harmonic..." Looks like third harmonic?
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,289
Likes
2,500
#9
Do you think the nastier -90 db sine wave of the BADA is due to its low level linearity?

Also, off topic, you have any connections to borrow a Pacific Microsonics DAC and/or ADC to test? The model One or Two I think they were called. It has something of a legendary status in the pro recording and audiophile world. Would be very nice to have that in your list of measurements.
 

stunta

Senior Member
Patreon Donor
Joined
Jan 1, 2018
Messages
437
Likes
232
Location
Boston, MA
#10
I am absolutely loving this combination of budget and expensive DACs being measured SxS here. Outstanding work, Amir.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,955
Location
Seattle Area
#11
Do you think the nastier -90 db sine wave of the BADA is due to its low level linearity?
I think the indication for that is the higher noise floor we see in the 1 Khz noise+distortion residual.

Also, off topic, you have any connections to borrow a Pacific Microsonics DAC and/or ADC to test? The model One or Two I think they were called. It has something of a legendary status in the pro recording and audiophile world. Would be very nice to have that in your list of measurements.
I kick myself for not getting one after we acquired the company. :) I remember thinking about it but then thought it was too big and fancy for home use.

Our newest member Rene was part of the company so maybe he knows?
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,023
Likes
1,478
Location
Monument, CO
#12
I wonder how much if the difference in low-level sine waves is affected by noise decorrelation (dither) intentionally added to the signal? I have used various dither schemes over the years but most give up one to several lsb's to dither. However, I agree completely that a higher noise floor will do the same thing. Might be interesting to see the spectral content of the noise just for grins (don't expect it would really show anything). Some of the more interesting noise decorrelation work used colored noise, e.g. narrowband noise of higher magnitude outside the frequency band of interest. Easier to do in RF DACs where there is generally a lot of unused bandwidth. One design I did placed dither from roughly 1 to 10 MHz for a converter designed to handle 100 to 1000 MHz signals. In that case the dither was fairly large but did not impact the band of interest.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,955
Location
Seattle Area
#13
Unfortunately that was a loaned unit and I only had access to it for a few hours. I have one myself for a few months but it was sold. So I can't do any further analysis.
 

rebbiputzmaker

Addicted to Fun and Learning
Joined
Jan 28, 2018
Messages
542
Likes
72
#14
Do you think the nastier -90 db sine wave of the BADA is due to its low level linearity?

Also, off topic, you have any connections to borrow a Pacific Microsonics DAC and/or ADC to test? The model One or Two I think they were called. It has something of a legendary status in the pro recording and audiophile world. Would be very nice to have that in your list of measurements.
Below is some on the description of the PM model one. Notice the second entry, rather interesting. "Discrete, multibit A/D and D/A converters designed by Pacific Microsonics"

Eight Motorola 56009 DSPs providing 400 millions of instructions per second (MIPS) in processing power
Discrete, multibit A/D and D/A converters designed by Pacific Microsonics
Ultra-low jitter; precision A/D and D/A reclocking circuitry
Pacific Microsonics-designed 2-MHz bandwidth discrete operational amplifiers used throughout analog circuitry
High-isolation discrete shunt power regulation used throughout
All analog and digital circuitry shielded and electrically isolated
Balanced, direct-coupled high common mode rejection ratio (CMRR) analog inputs and balanced +24 dBu direct-coupled analog outputs
Isolated, independent linear power supplies
Vibration-isolated, thermally controlled cooling fan
Copper-plated chassis
 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
2,685
Likes
1,041
Location
Australia
#15
Below is some on the description of the PM model one. Notice the second entry, rather interesting. "Discrete, multibit A/D and D/A converters designed by Pacific Microsonics"

Eight Motorola 56009 DSPs providing 400 millions of instructions per second (MIPS) in processing power
Discrete, multibit A/D and D/A converters designed by Pacific Microsonics
Ultra-low jitter; precision A/D and D/A reclocking circuitry
Pacific Microsonics-designed 2-MHz bandwidth discrete operational amplifiers used throughout analog circuitry
High-isolation discrete shunt power regulation used throughout
All analog and digital circuitry shielded and electrically isolated
Balanced, direct-coupled high common mode rejection ratio (CMRR) analog inputs and balanced +24 dBu direct-coupled analog outputs
Isolated, independent linear power supplies
Vibration-isolated, thermally controlled cooling fan
Copper-plated chassis

And, your point is?

Enough of the riddles. o_O
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,023
Likes
1,478
Location
Monument, CO
#16
Unfortunately that was a loaned unit and I only had access to it for a few hours. I have one myself for a few months but it was sold. So I can't do any further analysis.
Sorry, I meant in general, since you have seen fairly significant differences among low-level sinusoids generated by these DACs.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,289
Likes
2,500
#17
Below is some on the description of the PM model one. Notice the second entry, rather interesting. "Discrete, multibit A/D and D/A converters designed by Pacific Microsonics"

Eight Motorola 56009 DSPs providing 400 millions of instructions per second (MIPS) in processing power
Discrete, multibit A/D and D/A converters designed by Pacific Microsonics
Ultra-low jitter; precision A/D and D/A reclocking circuitry
Pacific Microsonics-designed 2-MHz bandwidth discrete operational amplifiers used throughout analog circuitry
High-isolation discrete shunt power regulation used throughout
All analog and digital circuitry shielded and electrically isolated
Balanced, direct-coupled high common mode rejection ratio (CMRR) analog inputs and balanced +24 dBu direct-coupled analog outputs
Isolated, independent linear power supplies
Vibration-isolated, thermally controlled cooling fan
Copper-plated chassis
Multi-bit doesn't mean 16 or 24 bit necessarily. What were they doing with 400 MIPS processing if it was a straight multi-bit design?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
14,768
Likes
7,955
Location
Seattle Area
#19
Multi-bit doesn't mean 16 or 24 bit necessarily. What were they doing with 400 MIPS processing if it was a straight multi-bit design?
They had to perform the HDCD encoding/decoding and digital filtering.

BTW, Keith Johnson designed a clever glitch removal using dual DACs.
 
Top Bottom