advertisement


Off grid rpi dac

sq225917

sq225917

Bit of this, bit of that
This one has been a long time coming, not as long as the Mdac2 though....

So the off grid rpi, dac, fifo board and output stage are finally assembled. Took a bit of dicking around with tiny smt parts to finalise the set up but its sounding very sweet. Ill give it a few days to run in before comparing against the Brooklyn. It runs from a 10 battery psu providing five isolated rails, 2 for the output stage, 1 for the fifo, 1 for the dac and 1 for the rpi.


47728300812_9dab2b0f55_b.jpg


Very pleased with this so far, props to iancanada from diyaudio for the GB.
 
Wooo cracking job there Si, wonder how I missed that GB o_O
I noticed that Misterdog also had one of these and meant to ask about it?
Please tell us more about it as I have a RPI/DAC/Amp so was wanting to see if it can challenge my PC/Jriver Media/M2Tech Young DAC system which sounds excellent ATM.

So is this just a battery power supply for a Rpi, DAC, FiFo, 12s?. Amp maybe?
What is the FiFo board doing, organizing data packets maybe for asynchronous USB?

Alan
 
I have this dim memory that a battery is not a "perfect" PSU, and that (super?) regulators can be better, but I don't recall where.

I will try to find it.

BugBear
 
There's batteries and batteries. Theseare A123 lifepo, lowest impedance of any you can buy commercially and capable of 120amp peak current. So about as near as dammit a perfect voltage source. They have negligible voltage drop with charge state and incredibly low chemical noise. Building with them requires some care as any stray contact could end in rather nasty vapourisation.

Maybe a particularly dialled in supereg could beat them, an lt3045 based reg doesn't, ime.

The pi runs volumio, gets files from airplay or direct connected hd storage. Data hits the fifopi running in locked sync master mode with mclk to the dac board. Dac board is dual ess 9038qm feeding an opamp based iv stage that's just chugging along with basic jelly bean dials at the minute. Opa6122 and adaptors are on their way. I also have some jfet input ad7389 to try. Not expecting much from opamp swaps.

I'll bugger about with clocks as well, ndk parts and high spec crysteks. Not sure what if anything they'll bring.

Well see how it sounds vs the Brooklyn, which is single chip 9028qm.

Mrdogs has defo changed in sound from new, was a little hazy and grainy at the top when box fresh, tightened up and smoothed out. Stereo separation is very very good, really wide and precise soundstage.
 
sq225917 said:
There's batteries and batteries. The are a123 lifepo, lowest impedance of any you can buy commercially and capable of 120amp peak current. So about as near as dammit a perfect voltage source. They have negligible voltage drop with charge state and incredibly low chemical noise. Building with them requires some care as any stray contact could end in rather nasty vapourisation.

Maybe a particularly dialled in superdrug could beat them, an lt3045 based reg doesn't, ime.

The pi runs volumio, gets files from airplay or direct connected hd storage. Data hits the fifopi running in locked sync master mode with mclk to the dac board. Dac board is dual ess 9038qm feeding an opamp based iv stage that's just chugging along with basic jelly bean dials at the minute. Opa6122 and adaptors are on their way. I also have some jfet input ad7389 to try. Not expecting much from opamp swaps.

I'll bugger about with clocks as well, ndk parts and high spec crysteks. Not sure what if anything they'll bring.

Well see how it sounds vs the Brooklyn, which is single chip 9028qm.

Mrdogs has defo changed in sound from new, was a little hazy and grainy at the top when box fresh, tightened up and smoothed out
Click to expand...

Nothing will beat a super reg for low impedance. Noise can be as low as the thermal noise of a 50R resistor as well!

Of course there are other characteristics which may or may not have more influence on sound quality.
 
- two 'other characteristics' that come to mind:
1. total galvanic independance from mains, and from other supplies. May or may not help depending on what you are powering, how well-isolated (say, the dac's) usb input really is, what it is connected between, lots of tiny such things that just may be helpful to simply avoid. This replaces 3 PSUs in this case with an essentially -independant. lump. Quite elegant if the runtime is adequate!
2. While superregs can do better, getting them to perform like that well is quite sensitive to precise implementation, or a power oscillator can result.

(I say these things as a fan of ludicrous-feedback, carefully-stabilised diy regs: and keep my mind/options open; and hey, I use a Mojo often enough...)
 
I've not hit the run time buffer yet. It has a user adjustable countdown feature where can set a switch off in hour increments and it powers the rails down in order with no thumps. The default is five hours so I assume it's good for that at least.

Isolation from the mains should be good as the batteries are isolated from the front end while in use. The front end needs power all the time though, it's not a mobile solution
 
It was all going great guns, till it went silent five minutes ago. Giving it ten minutes switched off before I look at it on the bench.
 
Had a wobble with it today, but it's back working again. The IV stage needs some extra filtering, space for a few extra film caps on the pcb.
 
I've been bodging, added some extra decoupling to the rails on the IV, they've tamed some hash. Couple of hundred uF and 1uf films. Got around to replacing the stock opamps tonight with OPA1612. Things have taken a huge step in the right direction. Much better all round. The treble has come into focus quite markedly. Still a ways to go yet, clocks next then supercaps on the three rails on the DAC board. As it stands it has better soundstage than the Brooklyn but the actual sound isn't quite as good. Not as delicate or as insightful in the upper frequencies. baby steps though.
 
Alan, the guts of it are the fifopi board. Ess Dacs usually run a 100kHz clock and resync everything internally. With the dual clocks on the fifopi you can run in true sync mode using a local master clock to provide the timing. The idea is that it lets you run a crappy source like the rpi without issue, that side of things certainly seems to work. The Iv stage does seem to suffer a bit with hf noise from the dac board which was coupling into the opamps, the extra filtering and different opamps have really cleaned that up.

There's still some way to go, supercaps on the rails on the dac board and possibly adding a reg on the avec as well as the better clocks.
 
Glad it's working and going in the right direction.
Might be worth trying MC33078 for the opamp (take a look at its o/l gain v freq characteristic compared to the OPA1612).
 
I'll take a look at that down the line. I'm making my way down the stack. DAC board next.
 
You must log in or register to reply here.


advertisement


Back
Top