FootFungus
Super Member
I just got a Nikko Alpha 450 back up and alive from terminal fuse-itis. 5 of the 6 outputs were toast in the left channel. The bias transistor and drivers survived, as this looks like it was collateral damage from a liquid spill incident.
Given Toshiba 2SA1095/2SC2565 outputs weren't cheap when they were plentiful and real, I went looking for substitutes. I came up with the Sanken 2SA1215/2SC2911 and the On-Semi NJW1302/NWJ3281. Both sets are slower ft-wise - about 60/50 mhz versus the 80/60 mhz for the original Toshibas. The Sankens are 150w and the On-Semi's 200w. I looked at the NJW0302/0281, which are a bit faster and closer in cob. However, the SOA at higher voltages did not match well.
The Sankens use the same screw holes as the Toshibas, but have a larger MT-220 footprint. So new thermal interfaces would be needed. The NJW1302/3281s are TO-3P so they require a new hole in the heatsink or a piece of bar stock to press them to the heatsink. The Sankens are about $13 a pair from Bdent and the On-Semi's $3.50 a pair from Mouser. So naturally I dug out my 4-40 tap for the On-Semi's.
The 2SD882 bias transistor (still widely produced) is screwed to the top of the center PNP output. Given the packaging change on the NJWs, I had to extend the legs a bit for the 2SD882. A counterfeit 2sa1306 provided some extra legging - it wouldn't be needing them....
I replaced the drivers for completeness. The Alpha 450 uses the 2SC2336a/2SA1006a combo. Bdent carries these for $6/pair if you get the 2SC2336a and the 2SA1006B. The Alpha 440 is almost exactly the same design, but with 2SC2275/2SA985 drivers.
About $18 dollars later I had a working amp. I still need to accurately set the bias and adjust the clipping, meters, etc. I may redo the good channel for symmetry and opportunity to liberate 3x pairs of A1095/C2565s for future purposes.
In the spirit of show and tell, I attached pictures of the output sections.
Given Toshiba 2SA1095/2SC2565 outputs weren't cheap when they were plentiful and real, I went looking for substitutes. I came up with the Sanken 2SA1215/2SC2911 and the On-Semi NJW1302/NWJ3281. Both sets are slower ft-wise - about 60/50 mhz versus the 80/60 mhz for the original Toshibas. The Sankens are 150w and the On-Semi's 200w. I looked at the NJW0302/0281, which are a bit faster and closer in cob. However, the SOA at higher voltages did not match well.
The Sankens use the same screw holes as the Toshibas, but have a larger MT-220 footprint. So new thermal interfaces would be needed. The NJW1302/3281s are TO-3P so they require a new hole in the heatsink or a piece of bar stock to press them to the heatsink. The Sankens are about $13 a pair from Bdent and the On-Semi's $3.50 a pair from Mouser. So naturally I dug out my 4-40 tap for the On-Semi's.
The 2SD882 bias transistor (still widely produced) is screwed to the top of the center PNP output. Given the packaging change on the NJWs, I had to extend the legs a bit for the 2SD882. A counterfeit 2sa1306 provided some extra legging - it wouldn't be needing them....
I replaced the drivers for completeness. The Alpha 450 uses the 2SC2336a/2SA1006a combo. Bdent carries these for $6/pair if you get the 2SC2336a and the 2SA1006B. The Alpha 440 is almost exactly the same design, but with 2SC2275/2SA985 drivers.
About $18 dollars later I had a working amp. I still need to accurately set the bias and adjust the clipping, meters, etc. I may redo the good channel for symmetry and opportunity to liberate 3x pairs of A1095/C2565s for future purposes.
In the spirit of show and tell, I attached pictures of the output sections.
