• Please note that there are a few updates and clarifications made in the Audiokarma Rules, mostly relating to advertising and the addition of the new "Paying it Forward" & "Giving back" forums in the AudioKarma Audio Marketplace section.

Sansui AU-D5 unstable bias current

gascar

New Member
I'm need of technical assistance on this sansui AU-D5 restoration, I've replaced some transistors on the power amp section (F-3221), and bias compensation board (F-3236). I got this unit with a shorted right channel and got a new pair of equivalents for drivers (From 2SD358 to EVVO 2SC4793 and 2SB528 to EVVO 2SA1837) and outputs (2SC2580 to Toshiba 2SC5200 and 2SA1106 to Toshiba 2SA1943) as well as the bias compensation transistors (2SC1845 to ONSEMI KSC1845 and 2SA992 to ONSEMI KSA992).
These transistor changes now have caused me to have what appears to be some bias runaway, on cold start the bias starts at around 10mV and in two-three minutes it all the way up to 50mV, cooling down the output with air does drop the bias current, but it keeps climbing anyways. I'm trying to find a solution to this problem but so far no luck, the board has also been recapped, but so far no changes. The bias current target according to the service manual is 15mV image_2026-03-19_000340692.png
 
Last edited:
Register to hide this ad
Are the temperature sense transistors in the V(be) multiplier (F-3236 board) correctly mounted on the heatsink? I'm not sure if all three transistors need to be thermally coupled to the heatsink as I am not familiar with this three-transistor circuit but they prolly should.
 
Yes, the Vbe multiplier transistors do go mounted on the heatsink, although they're like one and a half inches away from the outputs. The Vbe multiplier consists of a pair of 2SC1845 and a single 2SA992, I believe this kind of Vbe multiplier is only found in the AU-D5 and the more powerful AU-D7 but i wouldn't be sure. The heatsink bracket only has two screws in this image but I've tested it with all three screws and thermal paste.
 

Attachments

  • IMG_20260319_124701.jpg
    IMG_20260319_124701.jpg
    118.7 KB · Views: 5
I must confess, I'm utterly baffled by this bias circuit. I have no idea what problem Sansui's engineers were trying to fix. It looks like it is optimized for the semiconductors used and that the modern replacements have different V(be) behavior. I mean, why the termistors in the base drive of kQ8 & -10? Maybe kQ12 & -5 like to have additional cooling. If that doesn't help just perform a stress test (worst case scenario for any amp) by outputting a square wave of half the supply voltage (in your case 52 Vpp) into your dummyload. Start with a really low bias current, say 10 mV across the test points. Then check it as the heatsink heats up. If it really goes through the roof, say 100-200 mV you should be concerned otherwise just leave it. There is not much else you can do I'm afraid barring measuring V(be) voltages on kQ11-3 (or even kQ11-16) over temperature and see which ones run away fastest.
 
I must confess, I'm utterly baffled by this bias circuit. I have no idea what problem Sansui's engineers were trying to fix. It looks like it is optimized for the semiconductors used and that the modern replacements have different V(be) behavior. I mean, why the termistors in the base drive of kQ8 & -10? Maybe kQ12 & -5 like to have additional cooling. If that doesn't help just perform a stress test (worst case scenario for any amp) by outputting a square wave of half the supply voltage (in your case 52 Vpp) into your dummyload. Start with a really low bias current, say 10 mV across the test points. Then check it as the heatsink heats up. If it really goes through the roof, say 100-200 mV you should be concerned otherwise just leave it. There is not much else you can do I'm afraid barring measuring V(be) voltages on kQ11-3 (or even kQ11-16) over temperature and see which ones run away fastest.
Honestly i cant seem to comprehend it. Testing it with both sine and square 10kHz Amp 200mv at maximum output reveals zero distortion or oscillation in the slightest even after 10-15 minutes warmup, hot heatsink, cold spray on the thermistors does drop bias but even on hot conditions bias appears to settle at around 90-110mV, but then again no distortion or oscillation I think im just gonna leave it as is. I don't plan to use this amp on max volume at all, I think I'll just get a better thermal paste. Any recommendations?
I'll close it up and update on the sound :)
 

Attachments

  • IMG_20260320_145325.jpg
    IMG_20260320_145325.jpg
    97.5 KB · Views: 2
  • IMG_20260320_145645.jpg
    IMG_20260320_145645.jpg
    101 KB · Views: 2
  • IMG_20260320_145929.jpg
    IMG_20260320_145929.jpg
    101.7 KB · Views: 2
  • IMG_20260320_150006.jpg
    IMG_20260320_150006.jpg
    88.6 KB · Views: 2
Back
Top Bottom