Measurements of MiniDSP 2x4HD

[Blumlein 88]

Yes pretty clear you don't want the miniDSP doing your SRC.

Nice and clean without aliasing or harmonics in the JRiver version.

BTW, your video in post #39 says not available.

 

[pos]

My understanding is that the minidsp uses the in built SRC block of the 21483 DSP.

I think they are using a TI SRC4382
http://www.ti.com/product/SRC4382

That is what they are using in pretty much all their designs (openDRC-DI, UDIO, etc.).

It is pin compatible with the SRC4392 which has better specs:
http://www.ti.com/product/SRC4392

 

[oivavoi]

This is highly interesting. Thanks, @BE718 !

Would anybody mind explaining more about how minidsp does upsampling? So they don't only change the bit depth, they also change the sample rate?

Any chance of seeing additional measurements from Amir? (edit: or it might be redundant if everything has been covered here, I'm not up to date with the different testing procedures)

 

[March Audio]

I think they are using a TI SRC4382
http://www.ti.com/product/SRC4382

That is what they are using in pretty much all their designs (openDRC-DI, UDIO, etc.).

It is pin compatible with the SRC4392 which has better specs:
http://www.ti.com/product/SRC4392

Are you sure? That would be odd considering the shark dsp adsp21483 they have in this unit has it as built in functionality.

 

[Frank Dernie]

The DSPeaker Antimode 2 uses 24/48 too. I wonder if there is some available DSP programme they both use internally which only works on data of this file type?
Just wild speculation.

 

[pos]

Are you sure? That would be odd considering the shark dsp adsp21483 they have in this unit has it as built in functionality.

You might well be right here, I don't see an SRC chip on the board here : https://www.minidsp.com/images/documents/Product Brief-nanoSHARC.pdf

In the minishark (the one found in the openDRC) the DSP also has builtin ASRC capabilities, but as the in/out cards are external the DIGI-FP (the one used in the openDRC-DI) includes the ASRC using the Ti chip:
https://www.minidsp.com/images/documents/Product Brief - DIGI-FP.pdf

I would be curious to see a comparison of the nanoshark (2x4HD) and minishark (openDRC-DI) with your test procedure.

 

[March Audio]

The DSPeaker Antimode 2 uses 24/48 too. I wonder if there is some available DSP programme they both use internally which only works on data of this file type?
Just wild speculation.

No, I think it is purely to simplify the dsp filtering processes. One set of coefficients regardless of input signal.

 

[pos]

Here a description pof the ASRC block in the ADSP21483, page 9: http://www.analog.com/media/en/tech...sheets/ADSP-21483_21486_21487_21488_21489.pdf

The sample rate converter (SRC) contains four SRC blocks and is the same core as that used in the AD1896 192 kHz stereo asynchronous sample rate converter and provides up to 128 dB SNR.

Although the "real" AD1896 is supposed to have a 142dB SNR:
http://www.analog.com/en/products/audio-video/sample-rate-converters/ad1896.html

 

[March Audio]

This is highly interesting. Thanks, @BE718 !

Would anybody mind explaining more about how minidsp does upsampling? So they don't only change the bit depth, they also change the sample rate?

Any chance of seeing additional measurements from Amir? (edit: or it might be redundant if everything has been covered here, I'm not up to date with the different testing procedures)

The operation is described in the datasheets linked in posts above. In terms of measurements of the SRC functionality I have simply used the usb connection in loopback so it is purely the function of the dsp being looked at. The adc and dac sections are bypassed, so nothing to gain from further measurements in that respect, however more dac adc measurements are good for comparison.

One major difference between amirs AP system and my QA401 is that the AP has a notch filter functionality which can be used to filter out high level fundamental tones in test signals. ADCs / input amps used in test equipment generate their own noise and distortion. By filtering out the high level fundamental you can minimise the noise and distortion contribution of the analyser. You will see Amir now includes a 1kHz high level signal (notched) test to look at distortion.

You can listen to the adc conversion and compare the tracks found in this post

https://audiosciencereview.com/foru...ments-of-minidsp-2x4hd.2459/page-2#post-72278

Best to do it with something like foobar abx

 

[March Audio]

Here are some measurements of the 2x4HD A/D converter. Overall its not very impressive.

0dBFS - It has a 2V RMS FS input and shows quite a lot of harmonic distortion

-60dBFS - Still showing a surprising amount of HD

IMD SMPTE - Pretty messy in line with the HD

Jitter 12kHz (not Jtest) - Bunch of sidebands

Noise floor - Appears to be 8&16kHz tones, presumably USB packet noise.

 

[March Audio]

These are measurements of the DAC output using the same analogue input signals as above, so the overall input to output performance.

odB FS

-60dB FS

IMD SMTPE

Jitter 12kHz (not Jtest)

Noise Floor

I really like the miniDSP products, their concept and ease of use, just wish they would make a truly high performance version.

 

[March Audio]

Sweep analogue in. Just noticed watching the "stereo" meter that there appears to be some phase difference between the channels.

Video

 

[restorer-john]

Here are some measurements of the 2x4HD A/D converter. Overall its not very impressive.

0dBFS - It has a 2V RMS FS input and shows quite a lot of harmonic distortion

So, it is specified at 2.0v input via the analogue input, but did you wind up and down the input level slowly to confirm that 2.0v wasn't overloading it (aliasing) and causing the harmonic spray?

 

[March Audio]

So, it is specified at 2.0v input via the analogue input, but did you wind up and down the input level slowly to confirm that 2.0v wasn't overloading it (aliasing) and causing the harmonic spray?

I did indeed, and as you allude to you suddenly get a massive increase in hd when you hit FS. The difference isnt subtle . The actual 0dB test level was about 0.5dB below input FS at 1.94 V RMS. So makes FS about 2.05V RMS. If you look closely at the ARTA measurements you will see the input level in dB ref FS. The high HD is also there at lower levels ie -6dBFS way below FS.

 

[March Audio]

My view on this unit overall is thst the DAC section is pretty average, the ADC is not great and the internal resampling, if using digital input, is definitely something to be avoided.

However, in the context of price and functionalty it still is good value for money. I still think it would probably beat a passive crossover.

So come on miniDSP, time to build a truly high performance unit

 

[Blumlein 88]

Really dig the video of the sweeps in Musicscope by the way.

 

[March Audio]

Really dig the video of the sweeps in Musicscope by the way.

Yes, thanks to Amir for finding that software. Its great for showing just how pointless and crap most of my collection of remastered 24/96 music is

I think I will include a sweep with this software in any new measurement tests.

 

[Blumlein 88]

Is the video capture capability built into the Musicscope or if not what are you using for that?

 

[Blumlein 88]

Also a question I have had and I know Amir was interested in at one time recently. They have the bit meter to show which bits are encoding only noise. What is typical and what is the best you have seen on recordings for how many bits are not noise? My guess would be you have a hard time finding recordings that have non-noise info above 14 bits.

 

[restorer-john]

They have the bit meter to show which bits are encoding only noise.

I think I need glasses. I saw "They have the bit meter to show which bits are exploding only noise."