Raspberry Pi I2S to SPDIF Hat

For best performance you will want to build two of them. You could potentially just run with one of them and use the 5V jumper on the HAT, but there will be more noise (I presume) on the 5V of the HAT due to the Raspberry introducing noise onto the 5V. Plan to build two, they are actually not all that expensive compared to some of the other LPS's out there, and should provide the best performance.

Let me know if you need suggestions on cases, parts, etc. I have 3 of them for powering my digital gear and they have been built with some of the best parts I could find to maximize performance. (I can also help you build them as cheaply as possible if that is your goal).

 

thanks for the info, I'll get 2 of them and try to build them myself.

 

Why would you suggest to power the Pi as well from a LPS? You think the noise from a SMPS will make it through to the HAT ?

 

When the 5V jumper is disconnected, noise on that rail should not couple into the HAT from the pi. However, the pi has to generate the i2s signal to feed the HAT, and I'd imagine that clean power to the chips/circuitry used to process music files and generate the i2s signal could be important for overall fidelity. However, I have no empirical proof for this yet and it will have to be tested.

I'll be testing this out when I get my SPD2; I'll try running SMPS to pi + LPS to HAT and compare it to LPS to both pi and HAT (as well as many other comparisons).

 

@Mban . Sure. Please post your findings when you have done some testing.
I did some testing myself over the weekend with and ifi ipower, normal SMPS connected directly too the Digi+ and then via the Pi USB port . The difference between the USB feeding the Pi and the SMPS feeding the Digi+ directly was easily noticeable. There was some difference between the iFi ipower and my DLink SMPS, but it was rather small, so I returned the ifi. I assume for not that much more money I can build a LPS that could feed both PI and Digi+, so that is why I am interested to hear what you find and what is the costs involved.
Please also share some more info about your LPS build. I am interested to know the price

 

@Michael Kelly

Hi, is there still room on the test list for your D1/D2 or your 503/504 Dac's? Thanks!

 

For probably around $150-200 (incl. all the shipping costs) you could get all the parts to build and augment a dual Reflektor-D build (that price also includes the price of a case, connectors, etc. to have a complete dual rail LPS build). When you look on the Reflektor-D schematic you can replace all resistors (except R6) with non-inductive wirewound resistors (e.g. Ohmite WN series resistors). Capacitor C2 can be replaced with an aluminium organic polymer cap (e.g. APSG160ELL102MH20S). The rectifiers can be replaced with C3D03060F SiC rectifiers or equivalent. I would use this (EGPD250ELL123MM40H) or equivalent capacitor as my CM cap. For R6, get the lowest PPM and tolerance resistor you can find (I went with this one: PTF65120R00BYBF). For R1, I would go with something like WNDR50FET. With that 0.5ohm resistor you should be able to push 1amp of output current from the Reflektor-D. Remember: If you are putting this much current through the Reflektor-D you will also have to electrically insulate the rectifiers and mosfets and sink them to something that can take the heat! Don't just use the onboard heatsinks, they are not sufficient for that much current. For the transformer to run this all, I'll leave this up to you, but something such as VPT18-2780 should be sufficient (or VPT18-5560 if you want to have a whole bunch of extra capacity in case you add something to the setup down the road).

P.S. If you use a big enough case, you could put the Rpi and Digi/SPD board, two Reflektor-D builds, and the transformer to run the whole thing into a single case and have an all in one unit for streaming your music for pretty darn cheap. Even if you put a touch display on the unit, it would still be one hech of a lot cheaper than the Bryston BDP-Pi

Edit: Forgot to mention that if you really want to be intense you can feed DC into the Reflektor-D's from another LPS (I have mine fed off a LM338 based LPS that feeds them a highly stable 10V). In the case of feeding them DC, you would need a 0.1uF decoupling polypropylene cap instead of the big CM electrolytic and would not need the rectifiers on the Reflektor-D builds.

 

Think most of us are avoiding the Bryston, I sure don't want to be locked into their design.

 

@Mban, you sure aren't messing around, sweet!

 

If I may ever so politely ask, is it there yet?

 

Finally an update! We built the first protos of the SPD2 and tested them yesterday. All is looking good in default mode (Digi+ compatible, 27Mhz clock and PLL). We will be sorting the code for the clock options the next day or so. The downside will be that you can not use the default Digi+ driver. We need to supply a patch that you will have to apply to your Raspbian Linux. Alternately we will be creating a modified version of Volumio 2 that you can run. Is Volumio 2 OK with my test dummies? Let me know asap.

BTW, we are only able to test BNC in-house. The differential XLR will have to wait until we get something to receive it.

Cheers,
Michael

 

. Yep, volumio 2 works, too.

 

I can work with that

 

Volumio 2 works for me!

 

Too bad we missed the long weekend, that would have worked well.

 

Thank you for the update! Not to be difficult, but can I get the patch instead? The reason is that I will be using HQPlayer which requires a bit of software running on the Pi. It took me a while to get it configured so I would rather not mess with burning Volumio2 then trying to get the HQPlayer software installed and working again. I guess I can be a test dummy for both the SPD2 and the patch! Two dummies for the price of one!

 

Yes, we will make the patch available for everyone who wants it. I will assume you have "some" experience in applying patches though.

Cheers,
Michael

 

Is everyone using Volumio 2 RC1 (github download)? MoodeAudio might be a good choice also, a lot of quick development on it by Tim Curtis (can talk to him on diyaudio.com).

I'd vote for picoreplayer, but they are still on the 4.1.x kernel and just now planning on 4.4.x -- unless it's easy to add the patch.

 

Not to put a line too finely, couldn't people just get another SD card and make another build there? That would be a no brainer, unless I am missing something about fine adjustments to software and repeatability?

Looking forward to the HAT.

 

My issue is that I am not using Volumio. I am using a network audio adapter (NAA) from HQPlayer which took me some time to get right so I would prefer to not have to do through this process again.