Life after Yggdrasil?
Mar 16, 2017 at 4:27 PM Post #1,111 of 1,366
True, there are no caps in the direct signal path. But there are caps in nested NFB loops that perform the necessary anti-imaging reconstruction filtering.


What caps did Mike choose to use there and what caps would be better choices and why?
 
Mar 16, 2017 at 4:42 PM Post #1,112 of 1,366
What caps did Mike choose to use there and what caps would be better choices and why?


From the Schiit page, photos looks like WIMAs. More than likely MPK 2 metalised film polypropylene due to their small footprint for PCB mounting. A competent all around good film cap. There are "better" (film and foil PP, or PTFE) but usually much larger physically for the same value. Trying to mount those larger caps with long leads to the PCB would increase the inductance the filter sees and alter the designed response. How much? You would need to measure the result.
 
Mar 16, 2017 at 6:00 PM Post #1,113 of 1,366
From the Schiit page, photos looks like WIMAs. More than likely MPK 2 metalised film polypropylene due to their small footprint for PCB mounting. A competent all around good film cap. There are "better" (film and foil PP, or PTFE) but usually much larger physically for the same value. Trying to mount those larger caps with long leads to the PCB would increase the inductance the filter sees and alter the designed response. How much? You would need to measure the result.

From the web:
 
The Wima caps are just to bypass the electrolytic caps with (decoupling caps) and are fine for doing that. They are not coupling caps and have no music signal passing through them all.
 
Again, why would Mike pick what he did if he could have chosen a better part for similar money??
 
Mar 16, 2017 at 6:17 PM Post #1,114 of 1,366
From the web:

[COLOR=0000CD]The Wima caps are just to bypass the electrolytic caps with (decoupling caps) and are fine for doing that. They are not coupling caps and have no music signal passing through them all.[/COLOR]

Again, why would Mike pick what he did if he could have chosen a better part for similar money??


Sigh... I can't help you.
 
Mar 17, 2017 at 9:48 PM Post #1,121 of 1,366
From the Schiit Yggy specs page:

http://schiit.com/products/yggdrasil


Excellent, thanks!  Interesting that they got MCI to make a dual-choke (seeing all the pins threw me off until reading that they had it wound as a dual choke for their bi-polar supply).
 
I love true DC choke-filtered power supplies (and I don't mean little common-mode chokes).  In fact, the big DC filter choke of our own JS-2 5-7A LPS is the heart of its design.
 
Thanks again for correcting me.  I should not have made that assumption.  Makes me want to get a Yggy!  
rolleyes.gif

 
Best,
 
--Alex C.
 
Mar 17, 2017 at 10:00 PM Post #1,122 of 1,366
 
Excellent, thanks!  Interesting that they got MCI to make a dual-choke (seeing all the pins threw me off until reading that they had it wound as a dual choke for their bi-polar supply).
 
I love true DC choke-filtered power supplies (and I don't mean little common-mode chokes).  In fact, the big DC filter choke of our own JS-2 5-7A LPS is the heart of its design.
 
Thanks again for correcting me.  I should not have made that assumption.  Makes me want to get a Yggy!  
rolleyes.gif

 
Best,
 
--Alex C.

 
Glad to help you.
 
I'm a little busy now, right in the middle of triage on my KGST..... popped a 150V Zener in a 400VDC stack. Arggggh!
 
Mar 19, 2017 at 3:12 PM Post #1,125 of 1,366
BTW Alex, your Uptone LPS-1 is a game changer.

 
Thanks very much.  I can only take credit for the inspiration/idea to start the project (close to two years ago) and to pouring lots of money into it.  But ALL the engineering credit belongs to my brilliant partner John Swenson!  The actual journey--through multiple topologies and approaches to making it a reliable, buildable product--is a story unto itself.  The first pass of all the logic was done all discrete.  The thing would have had about 380 parts and reliability would have been questionable.  But once he had done it all discrete, it was obvious then (at least to him) how to write all the code and do in an FPGA.
 
We eventually will scale it up into high voltage, high current versions, but doing so quickly multiplies size and cost.   A multi-rail 6-7 amp unit would likely end up selling for about $3K. 
eek.gif

 

Users who are viewing this thread

Back
Top