Its funny you say that. I had the exact experience recently with the IXFN BAF2015 after listening mono for a few days, and it led me down to this...still need to finish the other channel and tart the chassis up a bit. I think the Singing Bush will be worth all the effort.
Attachments
Last edited:
@TungstenAudio. Nice build so far. Curious what you're using for transformers. Antek? The 500 and 600VA models have additional 12V and 18V secondaries. I was hoping to adapt the 18V secondaries for the bias circuit but I'm unsure of the calcs for the Radj on the LM317 or whether that is even advisable or not. Trying to avoid having to buy multiple transformers for each channel.
The extra 18V secondaries on the Antek transformers would work well for this. For this configuration, however, I'm reusing some older Hafler DH-220 transformers that don't have the extra outputs. (I have several lying around.) I'm trying to stuff everything into a single chassis. Should be entertaining.
Unlike the standard TO-247 package that's typically used for outputs,
which allows for different mounting/clamping methods, the SOT-227
package used here really only allows one method to mounting it, and
thats with a pair of screws. I guess it would depend on the depth of
metal/material on a typical CPU cooler face it has before you breach
the water jacket.
which allows for different mounting/clamping methods, the SOT-227
package used here really only allows one method to mounting it, and
thats with a pair of screws. I guess it would depend on the depth of
metal/material on a typical CPU cooler face it has before you breach
the water jacket.
I have finished (well, almost) my build, listened to it for a bit. Imaging is great, everything else remains to be evaluated with longer listening. I think I will like it
One thing I wished I had thought through better is PSU. I went all passive, CRCRC, with 165'000uF per channel. While this approach seems to work well sonically, the energy stored in caps is C*Vsquared/2
These 60V squared need to be dealt with. Regulated slowly ramping up PSU must be a much much calmer beast
One thing I wished I had thought through better is PSU. I went all passive, CRCRC, with 165'000uF per channel. While this approach seems to work well sonically, the energy stored in caps is C*Vsquared/2
These 60V squared need to be dealt with. Regulated slowly ramping up PSU must be a much much calmer beast
Yes, next time surely something like this. Here a big 21R NTC and 5A slow fuse seem to work. Will see how the fuse holds over time
Worth considering is a CLC supply. I used CLC, as I mentioned in post #487 (The Singing Bush). I ended up with 44mF - Hammond 195ZJ - 44mF per channel. A single thermistor slows the charge to the capacitors.
The cost was probably no more than CRCRC with 165mF per channel as 80V electrolytic capacitors are not inexpensive, and the choke takes up no more room than the capacitors.
My CLC supply is very quiet - no noise from my 103dB speakers.
Papa Pass suggested CRC of 47mF - 0R5 - 47mF:
BAF 2015 Coverage
The cost was probably no more than CRCRC with 165mF per channel as 80V electrolytic capacitors are not inexpensive, and the choke takes up no more room than the capacitors.
My CLC supply is very quiet - no noise from my 103dB speakers.
Papa Pass suggested CRC of 47mF - 0R5 - 47mF:
BAF 2015 Coverage
do not forget that we can use common mode choke ( damn, is that right term), properly phased for DC , thus relaxing tasks for heavy iron
well, not that I'm objecting about heavy iron
example, glorious one : https://www.diyaudio.com/forums/solid-state/161663-nem-ai-50-a.html#post2091636
observe that this is the case of LC smoothing supply......... which can be tricky to dial right away for final values, but Duncan's PSUD is always being great help there, if one is not having drekload of parts for prototyping
well, not that I'm objecting about heavy iron
example, glorious one : https://www.diyaudio.com/forums/solid-state/161663-nem-ai-50-a.html#post2091636
observe that this is the case of LC smoothing supply......... which can be tricky to dial right away for final values, but Duncan's PSUD is always being great help there, if one is not having drekload of parts for prototyping
Last edited:
Looking at the SIT datasheets, it seems that the gain of the SIT is affected a lot more by the VDS instead of Id. If so, then cascoding the SITs could significantly improve its linearity...
I'm pretty sure that one of the clever chaps here will have tried it - any luck ? Did it work well ?
I'm pretty sure that one of the clever chaps here will have tried it - any luck ? Did it work well ?
