jerdobi
New Member
I wanted to say a big thank you to the entire audio community both on AudioKarma and YouTube for all the help and guidance in restoring my Sony TA-5650. As we all know, these amps are no small project—the output FETs are extremely difficult (if not impossible) to find and match properly.
I picked up this unit on eBay back in 2022 for $250 as a parts-only piece. I started the restoration by pulling the FETs and immediately found two failed SK60s on the left channel along with blown fuses. I searched Yahoo Auctions Japan and luckily sourced two matched pairs of SK60/SJ18 FETs in the proper rank 54.
I began with a full recap and replaced the infamous “death diodes.” Next I started testing the power supply and protection board, but I managed to short out several transistors during testing thanks to a classic fat-finger moment. Discouraged, I lost interest and tossed a towel over the whole thing… until just a couple of months ago. This past February I discovered I could actually have a relatively meaningful conversations with Grok. That restarted everything. I asked Grok how to safely get back into the project and how to test the fault I caused to the power supply board. I bought a dim-bulb tester on eBay and re-watched all my reference YouTube videos.
The PrimeHiFi TA-5650 restoration became my main overall guide, and I made sure to follow Sony Service Bulletin 49 for the updated capacitors and resistors. I also learned proper flux-cleaning techniques that made the boards look brand new. Another lifesaver was the “whatcouldgowrong” Sony TA-4650 restoration video—especially the first video on testing the FET source and gate pin voltages and setting initial bias.
I also made sure to clean every pot and switch while the boards were out. One critical lesson: you absolutely must use 25-turn Bourns pots. The bias is incredibly sensitive, and temperature regulation is everything. Using the lid to stabilize the amp’s temperature is essential for getting a rock-solid 50 mV bias, and the dual-diode replacements must be properly thermally coupled to the bias transistors. Another thing I did which will get negative feedback is I used 4, 6 & 10 pin plastic connector blocks for the amplifier and power supply boards. They were great as I mounted and removed these two boards many times and didn't need to hassle with any soldering.
I started testing the right channel with just two pairs of FETs to see if I could even set bias and no shorts or voltage spikes. It was nerve-wracking watching the bulb light up brightly on first power-up before slowly dimming. Once that checked out, I installed the full complement of FETs. My first attempt with a 60-watt bulb that came with the DBT and 1-amp fast-blow fuses resulted in blown fuses and a bright bulb again. (a rookie misstake for sure on my part). I switched to a 150-watt bulb and 1-amp slow-blow fuses, and everything behaved perfectly ultimately putting back the 3.2 amp fuses.
Confident now there were no shorts, I removed the dim-bulb tester and powered the amp up with the full FET set. No smoke, no drama—and I was able to dial in a stable 20 mV bias to start off with my sound testing and break-in.
When I initially tried playing music there was no sound. I found that the speaker relay simply wasn’t clicking. After some research on the protection circuit and double-checking the service manual, I discovered I had accidentally installed a 100 Ω resistor where a 100 kΩ resistor belonged on the protection board stopping it from closing the relay. Once I corrected that, the relay clicked in and the amplifier came alive with absolutely crystal-clear, gorgeous sound.
The final lesson was being patience and not making unnecessary mistakes and trusting your work. I found the bias tends to drift where I attributed initially to break-in or the diode pairs not well thermally bonded to the bias transistor. But, it really was temperature of the room —even breathing on the amplifier board can change the readings. My workshop is in a cool basement (around 70 °F), and this amplifier is extremely sensitive to both room temperature and whether the lid is on or off. I let it burn in for more than six hours over two days, checking and adjusting the bias with the lid on after making numerous tweaks with lid off and then finding the idle different with the lid on. Finally, after a few tweaks I called it good and bolted the lid to let it bake a while.
Overall, I’m incredibly happy and proud I didn’t brick this beautiful amplifier. It sounds angelic—warm and rich with sparkling treble. A true classic brought back to life. If anybody undertaking this feat for the 1st time has questions, send them my way.
I picked up this unit on eBay back in 2022 for $250 as a parts-only piece. I started the restoration by pulling the FETs and immediately found two failed SK60s on the left channel along with blown fuses. I searched Yahoo Auctions Japan and luckily sourced two matched pairs of SK60/SJ18 FETs in the proper rank 54.
I began with a full recap and replaced the infamous “death diodes.” Next I started testing the power supply and protection board, but I managed to short out several transistors during testing thanks to a classic fat-finger moment. Discouraged, I lost interest and tossed a towel over the whole thing… until just a couple of months ago. This past February I discovered I could actually have a relatively meaningful conversations with Grok. That restarted everything. I asked Grok how to safely get back into the project and how to test the fault I caused to the power supply board. I bought a dim-bulb tester on eBay and re-watched all my reference YouTube videos.
The PrimeHiFi TA-5650 restoration became my main overall guide, and I made sure to follow Sony Service Bulletin 49 for the updated capacitors and resistors. I also learned proper flux-cleaning techniques that made the boards look brand new. Another lifesaver was the “whatcouldgowrong” Sony TA-4650 restoration video—especially the first video on testing the FET source and gate pin voltages and setting initial bias.
I also made sure to clean every pot and switch while the boards were out. One critical lesson: you absolutely must use 25-turn Bourns pots. The bias is incredibly sensitive, and temperature regulation is everything. Using the lid to stabilize the amp’s temperature is essential for getting a rock-solid 50 mV bias, and the dual-diode replacements must be properly thermally coupled to the bias transistors. Another thing I did which will get negative feedback is I used 4, 6 & 10 pin plastic connector blocks for the amplifier and power supply boards. They were great as I mounted and removed these two boards many times and didn't need to hassle with any soldering.
I started testing the right channel with just two pairs of FETs to see if I could even set bias and no shorts or voltage spikes. It was nerve-wracking watching the bulb light up brightly on first power-up before slowly dimming. Once that checked out, I installed the full complement of FETs. My first attempt with a 60-watt bulb that came with the DBT and 1-amp fast-blow fuses resulted in blown fuses and a bright bulb again. (a rookie misstake for sure on my part). I switched to a 150-watt bulb and 1-amp slow-blow fuses, and everything behaved perfectly ultimately putting back the 3.2 amp fuses.
Confident now there were no shorts, I removed the dim-bulb tester and powered the amp up with the full FET set. No smoke, no drama—and I was able to dial in a stable 20 mV bias to start off with my sound testing and break-in.
When I initially tried playing music there was no sound. I found that the speaker relay simply wasn’t clicking. After some research on the protection circuit and double-checking the service manual, I discovered I had accidentally installed a 100 Ω resistor where a 100 kΩ resistor belonged on the protection board stopping it from closing the relay. Once I corrected that, the relay clicked in and the amplifier came alive with absolutely crystal-clear, gorgeous sound.
The final lesson was being patience and not making unnecessary mistakes and trusting your work. I found the bias tends to drift where I attributed initially to break-in or the diode pairs not well thermally bonded to the bias transistor. But, it really was temperature of the room —even breathing on the amplifier board can change the readings. My workshop is in a cool basement (around 70 °F), and this amplifier is extremely sensitive to both room temperature and whether the lid is on or off. I let it burn in for more than six hours over two days, checking and adjusting the bias with the lid on after making numerous tweaks with lid off and then finding the idle different with the lid on. Finally, after a few tweaks I called it good and bolted the lid to let it bake a while.
Overall, I’m incredibly happy and proud I didn’t brick this beautiful amplifier. It sounds angelic—warm and rich with sparkling treble. A true classic brought back to life. If anybody undertaking this feat for the 1st time has questions, send them my way.
