A NOS 192/24 DAC with the PCM1794 (and WaveIO USB input)

Hi Palmito,

Assuming you connected to rhe right pins on the Pi; it might be a software issue.

I have also used volumio and raspify on a Pi but went back to my trusted and very stable setup with an Alix running mpdpup.
I had too much issues like, slow connection via iPad, drop outs, software crashes. I am using it with an USB wifi dongle for datastreaming and control; that might be a too heavy load, thus creating package loss.

How do you control the Pi? Is it with a mouse/keyboard or through a client?
 
After reading for about two hours last night there's a host of issues that I need to look at. I was using a local usb drive *and* the network connection, not a good idea in the pi with it's shared usb. The cabling wasn't optimal, although I did the i2s cable to the minimum possible length. I used a usb keyboard to set it up but rebooted without keyboard to have one less thing on the usb bus. Other people pointed out that the pi has a hard time handling large drive sectors and my usb drive is 1 gig so it has big sectors.

There might be a better driver than what I used in /etc/modules and I didn't do anything with the mpd setting except the most basic stuff to preserve bit perfect playback.

I'll try once again tonight using a network share for the music, so the only thing on the usb in the pi is the ethernet. I'll dress up the cables and see what it does. And test with the es9023 to make sure it's not a setup thing rather than problems strictly in the i2s communications with Doede's dac and the pi.

Someone who knows more about this than me can correct me, but I don't think it's bad wiring, as there are only four wires and if you not connect or misconnect *any one* of them, you get no sound. The i2s will obviously not play without ground or data, and the other two are bit and word clocks, which are pretty much needed to get anything out that resembles music.

I did get stuff that sounded ok, but there were muffled parts of the few tracks I played. Like castanets in a track I know very well were completely muffled (they of course sound great in the waveio/DoedeDac combination); the sound was floating in the middle of the speakers, almost as if it was mono, very narrow soundstage; there was a background grunge when music was played, (sort of like comsic microwave radiation :eek:), don't know how to best explain it.

Anyway, I'll report when I do some testing. Sorry for the long post... And Doede, if you think this if off-topic please let me know.
 
It's working. The main thing was finding Russ White's post (http://www.diyaudio.com/forums/twis...en-embedded-audio-applicance.html#post3806457) describing which and in what order to list modules in /etc/modules. I'll listen to it for the next few days before making a judgement, but it sound pretty good.

I did the following:

1) Get volumio and install it

2) Hook up the rpi to the Doede's dac waveio i2s terminal using the four i2s signal wires listed in the link above. Keep the wires as short as possible (I think 10cm is the longest recommended).

3) Turn on i2s output in volumio. I'm not sure how important this is since step 4 below overwrites some of the modules added to /etc/modules. But not knowing the full effect of *not* doing it I did it.

4) As indicated in the link above, /etc/modules should have the following modules in this order:

snd_soc_bcm2708
snd_soc_bcm2708_i2s
bcm2708_dmaengine
snd-soc-pcm1794a
snd_soc_rpi_dac

5) Configure your volumio network and music library.

I cleaned up the wiring, removed the harddisk and usb keyboard from the rpi and used a shared drive as the music library. The only connection the rpi had were power and ethernet. But what really fixed the problem was reading the link above.

If you have a raspberry pi with the P5 header, it's worth trying I would say, as it's very easy to setup and there's room to improve the rpi power supply and doing software tunning.
 

Attachments

  • doede_dac_rpi_i2s_2.jpg
    doede_dac_rpi_i2s_2.jpg
    354 KB · Views: 689
RPI external power

I also cut the micro usb cable and connected it to the 5v ps that was feeding the waveio, so now the rpi is getting clean power. It seems to have a positive effect (as many others have noted), I would say that at least it sounds a bit clearer. Before this I was using a cheap 5v/2a usb wall wart.
 
Hi Palmito,

Cool that you got it working. if I have the time I will try it myself.

But today another job landed on my door step: I recieved the new I2S/SPdif motherboard for the dac.
The board looks very nice, dark blue.

One comment on the board: it looks like you need to choose if you want to install the Tent clock or the ossillator, but do not mount both because they are connected parallel; there is no jumper.

The LF33 in the corner is for feeding the clock, jitter can improve by using a nice Tent shunt regulator here.
The other LF33 is for feeding the WM8804, I think using a shunt would also help on this spot.

Just wondering if using a shunt for the LF50 feeding the shift registers would help sonically.................


@ Doede,

Is the schematic and layout of the new board available?
 
only connect the POS and NEG dac output to the primary, regardless if you do serial or parallel. I would the latter by the way...

so NOT connect the COMMON from the DAC to any thing.... just leave it open

the secondary from the transformer goes to hot cinch and GND.

two connections input, two connections output...

Doede, do I still use the 10k resistor when wired in parallel on the secondaries? Thanks.
 
Hi Palmito,

Cool that you got it working. if I have the time I will try it myself.

But today another job landed on my door step: I recieved the new I2S/SPdif motherboard for the dac.
The board looks very nice, dark blue.

One comment on the board: it looks like you need to choose if you want to install the Tent clock or the ossillator, but do not mount both because they are connected parallel; there is no jumper.

The LF33 in the corner is for feeding the clock, jitter can improve by using a nice Tent shunt regulator here.
The other LF33 is for feeding the WM8804, I think using a shunt would also help on this spot.

Just wondering if using a shunt for the LF50 feeding the shift registers would help sonically.................


@ Doede,

Is the schematic and layout of the new board available?

I've been having the same thoughts Stefan. Had my original green board a while now but still haven't assembled it. Tent regs and clocks are the way I'm going to go with this I think.
 
I've been having the same thoughts Stefan. Had my original green board a while now but still haven't assembled it. Tent regs and clocks are the way I'm going to go with this I think.


I have used a new class d 5v regulator on the old board.(instead of 3.3v) I can not explain it in englisch but the sound was more "solid" / "grounded" / "balanced".
I will also use it on the new spdif board. (Board arrived 2 days ago, indeed very good quality)
Tent clock is on its way and i will upgrade the regulators for 3.3 also only not decided on what brand to buy.tentlabs / new class d / belleson etc.
Also i am not sure how to install the parts on the new board, so i will wait till someone builds one with the tent option.
Found a schematic here of the new board, c4 is a coupling cap for the spdif in am i correct?
So this can be replaced by a "audiophile" type?

p.s. I used super through, silmic ii, mkp1837 and amtrans on dac, ps and m-board . More to my flavor. Did not try new class d regulator with original capacitors that Doede supplied.
I only use spdif. No usb. So i will have a very empty chassis with the new board. Thinking of making some shielded compartements in it for power supply and dac.
 
Last edited:
Big thanks to Doede for a brilliant kit :) I've only got 1xdac board for now, but I've just installed TentLabs shunts this week and the results are wonderful :)

Got a question though... If I'm using a standard non-balanced output and the dc offset across the positive and negative output pins is very low (less than 200mv) is there any reason why I shouldn't just wire my rca to these pins and do away with the coupling caps? Is this dc offset figure likely to change for any reason? A few guys are running their kits like this and reporting good results with only a very slight hum at very high volumes with no music playing. The best cap is no cap at all, right?
Thanks,
James
 
I would think the LF50 replaced with shunt would give an improvement because it also powers the optocoupler of the Wave IO, a fast power supply will lower jitter introduced by the optocoupler.

I need the current draw for the adjustment of the shunt.
Can anybody tell me how much current the LF50 delivers on the new motherboard?

@doede,

The SPdif input is usually build with an input (pulse) transformer. Why did you choose not to use a pulse transformer?
 
Last edited:
Check my sales Site. i offer them as well;)
on the 1543 section to be precise ....

supersurfer, on the spdif transformer, well, most applications do not use them, but with transformers, it is the same with output capacitors, everyone has their own brand of choice.... that is why I do not include them. If some one like to experiment with this than that is the DIY and tweaking we all like so much :)

without, results are normally excellent, specially as your ground system has no loops or otherwise causing issues.
 
I would think the LF50 replaced with shunt would give an improvement because it also powers the optocoupler of the Wave IO, a fast power supply will lower jitter introduced by the optocoupler.

I need the current draw for the adjustment of the shunt.
Can anybody tell me how much current the LF50 delivers on the new motherboard?

@doede,

The SPdif input is usually build with an input (pulse) transformer. Why did you choose not to use a pulse transformer?

This way we have a choice of input, I used the Audio Note input transformer that allows me to use Balanced SPDIF. The next DDDAC I build will have BNC input and no transformer :)