I plan on getting a Pi 4 with a Topping E30 DAC. I want to add DSP to my setup so I can use REW for room correction. Is this possible with an E30?

[Alesana95]

Hi everyone!

I have a Pi 4 on the way and I have been looking at different DACs to use with it. Originally I was going to go with one of the HiFiBerry HAT DAC chips, but after looking into it I found that the external USB DACs are actually better quality for less money, and that with the Pi 4 the USB is just as good as HAT.

So after reading more into it I've decided on the Topping E30 DAC.

When I was looking at the HiFiBerry DACs I came across the DAC + DSP combo. There was an article on their website about implementing room acoustics correction using the DSP. They also offered a DAC2 HD chip, and you could get a DSP add-on board for it. I was going to get this combo before opting for the E30.

So now my question is, can I still use some DSP board to do room acoustics correction? Does the DSP have to be attached to the DAC, or is the room acoustics correction separate from the DAC to where I could attach it directly to the Pi 4 without any issues?

The only DSP I could find was MiniDSP 2x4 DSP for $104, but it's an external DSP. Would I be able to use that with my Pi 4 / Topping E30 combo? Would using that affect sound quality all (other than the room correction)?

I would prefer to find some DSP chip/board similar to the add-on board, but I can't seem to find anything.

If anyone has any suggestions or pointers that would be greatly appreciated!

 

[HammerSandwich]

Why not run DSP on the Pi? Look into CamillaDSP. It's included in recent Moode versions, if you're looking for the easiest setup.

 

[Alesana95]

Why not run DSP on the Pi? Look into CamillaDSP. It's included in recent Moode versions, if you're looking for the easiest setup.

I saw CamillaDSP (and I think a couple others). I was thinking it required some DSP board in order to be able to do that. Would I just plug a USB calibration microphone into the pi and it would be able to process everything? What would be the point in getting a HiFiBerry DSP board or a MiniDSP at that point?

I haven't tested out Moose vs Volumio vs others yet so I haven't quite decided which one I want to use but if Moode has room correction built in thats a huge plus.

 

[mdsimon2]

I saw CamillaDSP (and I think a couple others). I was thinking it required some DSP board in order to be able to do that. Would I just plug a USB calibration microphone into the pi and it would be able to process everything? What would be the point in getting a HiFiBerry DSP board or a MiniDSP at that point?

I haven't tested out Moose vs Volumio vs others yet so I haven't quite decided which one I want to use but if Moode has room correction built in thats a huge plus.

CamillaDSP is very flexible and only requires a computer (can easily be a RPi 4), no other external DSP board is needed.

As a long time miniDSP user who has recently switched over to RPi 4s running CamillaDSP the "what would be the point in getting a HiFiBerry DSP board or a MiniDSP" question is rather interesting.

MiniDSP and hifiberry are much more plug-and-play than software DSP on a RPi 4. You do not sound too experienced with DSP in general so a more plug-and-play product may better suited to learning about DSP and room correction.

The other big advantage to miniDSP / hifiberry board's is that they can provide additional input/output functionality when using software DSP. For example all of my systems using software DSP still have miniDSP board's involved to provide additional input (TOSLINK, SPDIF, AES, etc) functionality, if your only input is a streaming OS (Volumio, Moode, etc) then this does not matter but if you want software DSP to act like a more traditional hardware DSP you will want additional inputs.

Michael

 

[Alesana95]

That makes sense, the extra inputs would be useful. I didn't realize but it looks like the miniDSP also has a DAC in it. I'm not sure how great it is though. So, the miniDSP board is the initial input, and then it feeds processed input from TOSLINK / etc... to the Pi?

Another thing I wondered (which isn't very relevant anymore since I don't need an external DSP) is if the initial source were the Pi, and if it were to use an external DSP with REW, would it have to first feed the audio to the miniDSP and then back to the Pi before sending it to the DAC? If so would that cause any side effects with the sound? I read somewhere (can't remember where) that someone was concerned that if they used a miniDSP streamed from a Pi 4 that it could be converted from digital to audio, then back to digital, then back to audio again.

 

[mdsimon2]

That makes sense, the extra inputs would be useful. I didn't realize but it looks like the miniDSP also has a DAC in it. I'm not sure how great it is though. So, the miniDSP board is the initial input, and then it feeds processed input from TOSLINK / etc... to the Pi?

Another thing I wondered (which isn't very relevant anymore since I don't need an external DSP) is if the initial source were the Pi, and if it were to use an external DSP with REW, would it have to first feed the audio to the miniDSP and then back to the Pi before sending it to the DAC? If so would that cause any side effects with the sound? I read somewhere (can't remember where) that someone was concerned that if they used a miniDSP streamed from a Pi 4 that it could be converted from digital to audio, then back to digital, then back to audio again.

Probably best to describe a few example setups.

1) Use miniDSP 2X4HD (definitely recommend this over the normal 2X4) as your DSP / DAC. You can use your RPi 4 as an input via USB but also have TOSLINK and analog inputs. This setup obviously uses the 2X4HD DAC which has reasonable performance but is not SOTA (~95 dB SINAD, dominated by noise not distortion). I have run REW on RPi 4s running Raspberry Pi OS and Ubuntu Server and it works but it is not a great experience (rather slow), I imagine it would also require a bit of work to get it to work on an audio OS like Volumio / Moode. With this setup it would be very easy to run REW from a laptop or other computer (using 2X4HD USB input) and then switch back to the RPi 4 once you are done measuring / designing your correction. The other nice thing about this setup is if you want to try Dirac down the line you can upgrade the 2X4HD to a DDRC-24.

2) Use E30 as your DAC with CamillaDSP on your RPi 4. This will definitely work with a streaming OS on your RPi but I am unsure of if you will be able to use the other inputs on your E30. Depending on how they present themselves to the RPi you may be able to capture those inputs in CamillaDSP, apply processing and then output back to the DAC via USB from the RPi. Running REW may be a bit of an issue with this setup, you can definitely play white / pink noise files from your streaming OS and use another computer with the UMIK to measure via RTA but frequency sweeps will be a challenge. You could also run REW another computer directly connected to the E30, design corrections based on that and then switch back to the RPi but I do not see a direct path for how you can measure the results of those corrections.

In your use case I do not see much of reason to run both a 2X4HD and an E30.

Michael

 

[MCH]

You could also run REW another computer directly connected to the E30, design corrections based on that and then switch back to the RPi but I do not see a direct path for how you can measure the results of those corrections.


Michael

Hi Michael, thank you very much for all your input, in this and other occasions, very much appreciated.
I don't get your last sentence, why could be a problem to do measurements+correction with a computer and then switch back to the pi if the speakers room etc are still the same?
Sorry if too stupid just trying to learn something

 

[mdsimon2]

Hi Michael, thank you very much for all your input, in this and other occasions, very much appreciated.
I don't get your last sentence, why could be a problem to do measurements+correction with a computer and then switch back to the pi if the speakers room etc are still the same?
Sorry if too stupid just trying to learn something

It should be fine for the reasons you describe, it is more about verifying the filters you applied actually did what you expected. As mentioned you can still do some verification with white noise and moving mic measurements (which are probably more accurate than frequency sweep measurements anyways). I think I would probably just leave the E30 connected to the RPi and make measurements using noise / MMM method as you don’t need to switch between the RPi and the other computer.

Michael

 

[MCH]

Got it, thanks!

 

[Alesana95]

Probably best to describe a few example setups.

1) Use miniDSP 2X4HD (definitely recommend this over the normal 2X4) as your DSP / DAC. You can use your RPi 4 as an input via USB but also have TOSLINK and analog inputs. This setup obviously uses the 2X4HD DAC which has reasonable performance but is not SOTA (~95 dB SINAD, dominated by noise not distortion). I have run REW on RPi 4s running Raspberry Pi OS and Ubuntu Server and it works but it is not a great experience (rather slow), I imagine it would also require a bit of work to get it to work on an audio OS like Volumio / Moode. With this setup it would be very easy to run REW from a laptop or other computer (using 2X4HD USB input) and then switch back to the RPi 4 once you are done measuring / designing your correction. The other nice thing about this setup is if you want to try Dirac down the line you can upgrade the 2X4HD to a DDRC-24.

2) Use E30 as your DAC with CamillaDSP on your RPi 4. This will definitely work with a streaming OS on your RPi but I am unsure of if you will be able to use the other inputs on your E30. Depending on how they present themselves to the RPi you may be able to capture those inputs in CamillaDSP, apply processing and then output back to the DAC via USB from the RPi. Running REW may be a bit of an issue with this setup, you can definitely play white / pink noise files from your streaming OS and use another computer with the UMIK to measure via RTA but frequency sweeps will be a challenge. You could also run REW another computer directly connected to the E30, design corrections based on that and then switch back to the RPi but I do not see a direct path for how you can measure the results of those corrections.

In your use case I do not see much of reason to run both a 2X4HD and an E30.

Michael

Thanks for the in depth response! I think I understood most of it, I'll have to reread it more thoroughly once I have everything set up.

I'll just skip the external DSP though since it sounds like it would just cause hassle trying to use the MiniDSP along with the DAC (I ended up with the D10s).

 

[abdo123]

dominated by noise not distortion).

I would double check that, from the tea leaf reading i completed on the review thread it has a S/N of 107 dB (15 dB higher noise floor than Topping D50).

Boy reviews back then were rough.

 

[mdsimon2]

I would double check that, from the tea leaf reading i completed on the review thread it has a S/N of 107 dB (15 dB higher noise floor than Topping D50).

Boy reviews back then were rough.

Pretty sure, 107 dB would be better than miniDSP spec of 103 dB (unclear whether this is A-weighted) which seems unlikely. The AKM datasheet for the AK4626 shows typical dynamic range of 100 dB with an 88 dB min spec when running at 96 kHz, unweighted.

I had made measurements in the past and THD was around -100 dB and noise was around -96 dB. That was with the MOTU M4 which at 2V input has a dynamic range of ~102 dB which implies a dynamic range of 97 dB for the 2X4HD. I can try again with the Ultralite Mk5 which is a few dB better from noise perspective than the M4 but doubt we are seeing anything better than 100 dB on DR.

Michael

 

[mdsimon2]

Here are the measurements with the Ultralite Mk5. USB and TOSLINK inputs resulted in the same performance. The Ultralite Mk5 has two kinds of analog inputs, line in and microphone. The microphone inputs have slightly better noise performance (as well as a gain knob that can be used to dial in the signal level to achieve best noise performance) but slightly worse performance from a distortion standpoint.

Line in, RCA to TRS cable. -96 dB N, -101 THD, -95 THD+N.

Microphone in, RCA to TRS cable, +12 dB gain. -97 N, -98 THD, -95 THD+N.

So that -107 dB DR seems a bit optimistic to me. To be honest eyeballing Amir's measurements and considering gain for a 32K FFT I would have thought noise would have been in the -90 dB region at best. Although that review is a bit suspect to me as it was done with the old analyzer and frequency response measurement is clearly wrong.

Michael

 

[mdsimon2]

Also as a point of calibration here is a measurement of the Ultralite Mk5 DAC using the microphone input at 2V output voltage, per Amir's review this should have around -109 dB DR. I measure -104 dB N which implies the ADC is adding around -105 dB N. Based on this I would put the 2X4HD DR at 97-98 dB.

Michael

 

[abdo123]

So that -107 dB DR seems a bit optimistic to me. To be honest eyeballing Amir's measurements and considering gain for a 32K FFT I would have thought noise would have been in the -90 dB region at best. Although that review is a bit suspect to me as it was done with the old analyzer and frequency response measurement is clearly wrong.

Michael

Thanks for the masurements, I was going off by Amir saying it has 15 dB higher noise floor than the Topping D50 which has 122 S/N. i'm not sure if FFT windows are comparable like that but it made sense to me at the time.