GALA DAC - New Product, Coming Soon

[pos]

1kHz fundamental to see the distortion spectrum, similar to the SINAD screenshot Amir does in all his recent reviews: https://www.audiosciencereview.com/...dio-gala-dac-dashboard-measurement-png.16314/

0dBFS and -20dBFS would be interesting there, and with a dB scale instead of the Vrms one

Regarding the jitter measurement, as the FFT is not the same as the one used in other published measurement here I cannot tell if the "skirt" we see real or if it would be hidden in the noise floor with other settings. As this is supposedly LF noise I think it would raise with the noise floor with a shorter FFT?

 

[bunkbail]

I had send two GALA boards to the manufacturer for repair and they screwed up both repairs. One board the left channel is broken and the other board left channel is broken. Needless to say I am not going to be using that manufacturer any more.

In any case I was able to do some modifications at home to get everything to work, unfortunately it is not something I can send to Amir, I will have to order new boards.

I will be launching the project and all items will be addressed on the final rev of the board.
Based on all of the work that I did I know, what circuitry needs to be adjusted, see images below.

View attachment 17446

Here are the measurements:

SNR (A-Wt.) -- As per above post when analyzer's residual noise is subtracted out you get 130dB.
View attachment 17447

Dynamic Range (A-wt):
View attachment 17449

Dynamic Range (no-weighting):
View attachment 17450

THD+N vs. Level:
View attachment 17452

Frequency Response:
View attachment 17453

Jitter (256k FFT w/ 16 averages):
(I have a solution for the jitter also, here you can see a significant reduction in the noise floor of the jitter measurement
View attachment 17456

Impressive stuff. Looks at least on par with a Benchmark DAC3. Great job!

 

[Grave]

Impressive stuff. Looks at least on par with a Benchmark DAC3. Great job!

The Benchamrk DAC3 and the Khadas DAC have a SINAD of 110 dB which is slightly better than this though. This looks like the THD+N goes down to 105 dB.

 

[orchardaudio]

Regarding the jitter measurement, as the FFT is not the same as the one used in other published measurement here I cannot tell if the "skirt" we see real or if it would be hidden in the noise floor with other settings. As this is supposedly LF noise I think it would raise with the noise floor with a shorter FFT?

The skirt is below audible levels, it will be hidden in all other measurements.

I will post 1kHz fft later today when I get home.

 

[orchardaudio]

I will post 1kHz fft later today when I get home.

In the final design the SINAD will be slightly better because there will not be floating wires running all over the place as you can see in my setup here:
https://www.audiosciencereview.com/...-dac-new-product-coming-soon.4316/post-114722

 

[rajapruk]

This is probably a stupid question, but how come SINAD is like 106-107dB when 1KHz distortion is below -110dB? Is it higher distortion in another frequency range that shows up in SINAD?

 

[orchardaudio]

This is probably a stupid question, but how come SINAD is like 106-107dB when 1KHz distortion is below -110dB? Is it higher distortion in another frequency range that shows up in SINAD?

I am not sure what you are asking but SINAD is a sum of all harmonics and noise combined.

 

[pos]

This this coherent with the THD vs level curve, and should stay that way down to -20dBFS.
Nice to see only low order HD products

IMD vs level would also be interesting to see by the way!

 

[orchardaudio]

IMD vs level would also be interesting to see by the way!

IMD vs. Level:

 

[orchardaudio]

The GALA project will be launching on Tuesday Nov 13th, 2018.

If you singed up for the mailing list, please be on the lookout for the special link around that time to receive your discount.

The discount will only be available for 24 hours from launch.

Please share this project with your friends, family, coworkers and on social media to help it succeed.

 

[pos]

I must say I am pumped, measurements look really good now, kudos

I would love to be able to wait for Amir measurements before committing though...
What power supply are you using here?

I would be curious so see the same THD and IMD vs level curves with the internal volume of the DAC set to say -10dBFS or -20dBFS, just to see if it simply translates to the right, or if there is more to it.
Regarding those curves "stalling" past -20dBFS, is this due to the analog output stage, or is it intersect to the DAC chip?

Now that I think of it, THD vs frequency measurement is also missing.

 

[orchardaudio]

I use a wall-wart 12V 6W power supply. It does not really matter, since it is not directly used by anything on the board. The board generates all power supplies locally and uses ultra low noise LT3042 linear regulators for analog circuitry.

Also I was messing around and learned that the PCM1794A has slightly better performance when MCLK is 256FS vs 128FS. I had been using 128FS all along. Here are some plots with 256FS MCLK. The reason I had tough about this is because the ApplePi DAC uses 256FS MCLK as there is really no other options available, but with the S/PDIF receiver you can choose between 128FS and 256FS.

THD+N vs Frequency (256FS MCLK):

THD+N vs Level (with 256FS MCLK):

64k FFT @ 0dBFS (256FS MCLK):

64k FFT @ -20dBFS (256FS MCLK):

I will now be concentrating on building a new revision of the board with all of the changes that I discovered from testing. Once I have the new board with all fixes. I will post measurements from that board.

Also all products that go to customers will be fully tested to ensure that they meet published specs.

 

[pos]

Comparing the new THD vs level (256FS MCLK ) to the old one (presumably 128FS MCLK), it looks like things get better in the -15dBFS to 0dBFS range, but everything gets a ~5dB hit for levels lower than -15dBFS. Does this modification increase noise level?

Also when comparing the 0dBFS and -20dBFS 1kHz plots, it looks like 2nd and 4th harmonics do decrease faster than the signal (eg for 2nd order around 130dB down relative to the fundamental in the -20dBFS measurement, compared to around 120dBFS down in the 0dBFS one), whereas the 3rd order one stays constant relative to the fundamental.
Could this (very relative) rise in even order harmonics towards 0dBFS be caused by the analog output section?

 

[orchardaudio]

GALA is now launched:
https://www.indiegogo.com/projects/gala-ultra-high-performance-stereo-dac/x/19247162#/

 

[SchwarzeWolke]

#1

 

[orchardaudio]

#1

Thank you.

 

[rajapruk]

I have lost confidence in the GALA dac, for various reasons.
Only new measurements from Amir of a production unit later might restore my confidence. But, I am not holding my breath anymore, for this one.

 

[studio19atx]

I have lost confidence in the GALA dac, for various reasons.
Only new measurements from Amir of a production unit later might restore my confidence. But, I am not holding my breath anymore, for this one.

I am looking forward to Amir's measurements. Hopefully they will be close to what Leonid measured.

I have no affiliation with Orchard Audio, but I have been following the threads around the DACs. It is a startup and Leonid seem to be making an honest effort to be open about the state of development and willingness to address any concerns/shortcomings of the product.

Even some very large established companies at times ship some very expensive units that are dead arrival or have fallen through the cracks of quality control.

I am in market for DAC and the choice is down to GALA and DAC DAC. Both have same features and components. It will be interesting to compare the measurements of both. May be DAC (Digital Audio Company) can send a DAC for review to Amir.

 

[Veri]

@studio19atx no offense but that DAC DAC looks like a lot of mumbo jumbo BS :/ he claims SNR specs but no real measurement of proof anywhere. Pricing is really high, too.

 

[studio19atx]

I am not too worried about the looks. I guess they are in line with their other products offerings.
Like I said, DAC should send a unit to Amir for testing .