Wow! Excellent plots Michael. You're right about the filter not having an effect on the noise down in the audible spectrum. Something else is going on there. Part of it might be the impedance of the polypropylene caps being higher at certain frequencies compared to the impedance of the types used on the stock SI and the Fenice (ceramic X7R on the sonic and I don't know what on the Fenice). The Ref-T also spits out more RF than I thought it would! The multiples of the switching frequency are clearly defined and you can see the spread spectrum result of the variable switching frequency. Different inductors might reduce that a bit but certainly won't knock it down to Fenice levels. You said the tests were done with the outputs unloaded, correct? If that was the case, then the output filter and zobel network aren't playing too nicely with each other. I just used the values recommended by Tripath in their datasheets.
From your measurements the Fenice board is clearly well designed in terms of emitted radiation and noise. Those Italians did some nice work! I think one important thing to pull away from these graphs and listening tests is the importance of a good power supply. The Ref-T has a lot more noise on the output but sure doesn't sound like it when hooked up to speakers because the supply rail is clean and at a higher voltage. I can't see a much better (and reasonably compact) way to lay out the board with only 2 layers... a 4 layer board would be great to help reduce emitted radiation and reduce coupling between traces but then the cost goes up quite a bit. Now I wish I had taken some RF classes in college
Oh, before it slips my mind again, is there any chance you could do some channel separation measurements? I also tried to keep the interaction between the channels to a minimum with the layout, but there's only so much you can do with such minimal space.
From your measurements the Fenice board is clearly well designed in terms of emitted radiation and noise. Those Italians did some nice work! I think one important thing to pull away from these graphs and listening tests is the importance of a good power supply. The Ref-T has a lot more noise on the output but sure doesn't sound like it when hooked up to speakers because the supply rail is clean and at a higher voltage. I can't see a much better (and reasonably compact) way to lay out the board with only 2 layers... a 4 layer board would be great to help reduce emitted radiation and reduce coupling between traces but then the cost goes up quite a bit. Now I wish I had taken some RF classes in college
Oh, before it slips my mind again, is there any chance you could do some channel separation measurements? I also tried to keep the interaction between the channels to a minimum with the layout, but there's only so much you can do with such minimal space.
Hey Brian,
The amps all had 8 ohm noninductive resistors on the outputs.
By unloaded, I meant the power supply. As the amps in these tests have no signal, the power supply is hardly working. Once the amp starts to ump out some sound, thing look different.
I did some testing with a speaker as a real world load, but the results were pretty much the same.
All 3 amps were powered from the same source, a home made linear PSU with large transformer. The Ref-T got its power before reguation. So about 20V of filtered DC.
If you look at this graph:
You will see that the Ref-T has lower noise than the Sonic thru most of the audio band. It's just at the top that there is a peak. The Sonic drops nicely at the top, don't know why.
I don't have the PicoScope for much longer, so I will be concentrating on some pulse tests to see how well each amp handles dynamics. A real test of the PSU!
The amps all had 8 ohm noninductive resistors on the outputs.
By unloaded, I meant the power supply. As the amps in these tests have no signal, the power supply is hardly working. Once the amp starts to ump out some sound, thing look different.
I did some testing with a speaker as a real world load, but the results were pretty much the same.
All 3 amps were powered from the same source, a home made linear PSU with large transformer. The Ref-T got its power before reguation. So about 20V of filtered DC.
If you look at this graph:
You will see that the Ref-T has lower noise than the Sonic thru most of the audio band. It's just at the top that there is a peak. The Sonic drops nicely at the top, don't know why.
I don't have the PicoScope for much longer, so I will be concentrating on some pulse tests to see how well each amp handles dynamics. A real test of the PSU!
amt will be sending the amp back this week and then it will be sent out again for another person to demo. Panomaniac was on his honeymoon this past week so I doubt he did any listening! I'll let him keep it for another week or so to allow him to have some more listening time.
I'm in the midst of working on a few electronics projects so my focus has been shifting back and forth lately. Not much free time doesn't help when I want to get things done. Regardless, I'm working on compiling a detailed BOM and some other things that are needed to get accurate price quote from a few assembly houses. In short, I'm working on it
I never really made a schematic because I basically followed the one from the Tripath data sheet and added my modifications directly in eagle. The only differences are the input offset circuitry and the extra power supply components. It's the layout of the Ref-T that is quite different from the other amps, and that's a farily large factor in how the amp performs as well.
I'm in the midst of working on a few electronics projects so my focus has been shifting back and forth lately. Not much free time doesn't help when I want to get things done. Regardless, I'm working on compiling a detailed BOM and some other things that are needed to get accurate price quote from a few assembly houses. In short, I'm working on it
I never really made a schematic because I basically followed the one from the Tripath data sheet and added my modifications directly in eagle. The only differences are the input offset circuitry and the extra power supply components. It's the layout of the Ref-T that is quite different from the other amps, and that's a farily large factor in how the amp performs as well.
I've come to the decision that I will no longer pursue production of this version of the Ref-T. Right now I'm focusing my effort on the D10.1s because they sound every bit as good and can be made for less. I have ideas to bring the cost down and to revise the layout to improve it even further but this will be down the road a little.
As such, I've put one of the Ref-T's up for sale here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=83635
Thanks everyone for following along. If anyone has any questions about anything I'd be happy to try and answer them over email.
As such, I've put one of the Ref-T's up for sale here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=83635
Thanks everyone for following along. If anyone has any questions about anything I'd be happy to try and answer them over email.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.