AU-20000 Skeletons - Repair Tips

[Geek]

When it comes to Sansui, or any vintage amplifier, recapping the electrolytics will solve a plethora of problems. However, there are times when the obvious isn't so obvious. I'd like to share some tips discovered via experience.

1) Bias wandering. On the driver boards, there's a TO-1 cased transistor with long leads attached to the heatsink for the driver transistors. Replace it, even if it tests good. You can use a 2N2222 no problem and just compression mount it with a piece of metal and a dab of heat sink goop.
This transistor is prone to failure at the most inopportune time, usually when things heat up.
Another failure mode of this transistor gives the same symptoms as motorboating... it will cause the outputs to lock hard to one of the supply rails, putting full V+ or V- to the speakers, tripping the protector relay to chatter, giving hundreds of watts of "machine gun" coming through the speakers.

2) Everything checks out, but protector still stays engaged. The PCB the output transistors are mounted on, sucks. Hairline cracks in the traces will drive one crazy. The best answer is to solder the thermistor leads "P" and "Q" straight to the thermistor.

3) Testing the bias. When doing the test, you have to have the test probes dangerously close to the spring retaining clip for the driver connector. Just temporarily solder some wires to test points "2" and "4". Set to 35mV at start, then re-adjust after 30 minutes.

4) A note on recapping the regulator PCB - I use Panasonic FC caps. Yes, FM or Nichicon PS are lower impedance, but also so low that you're popping things with surge currents. Remember that even todays worst cap is superior to yesterday's best cap.

5) Replacing the giant can caps. If you find you have to, you're in luck, they still make them! Digikey 493-8984-ND or Mouser 647-LNT2A223MSE. They aren't cheap. But even better news! If you're capable of dispensing with the screw terminals and using a smaller bracket, Digikey 338-1986-ND (no Mouser equivalent) out-specs the original by a good margin and is 1/4 the price!

6) The 1uF polarized electrolytic capacitor positions on the meter board (F-2496), the silkscreen is wrong! The polarity of those caps is reversed to the silkscreen.

I hope this information is of some help for someone.

Cheers!

 

[Hyperion]

I hope this information is of some help for someone.

Cheers!

Should be - thanks for posting.

 

[JoshHendi]

I am working on an AU20k right now... The TO-1 transistor on the two driver boards you are referring to, is it the one with the long legs folded over? Seen in the picture here

 

[Hyperion]

is it the one with the long legs folded over? Seen in the picture here

Yes, I would say that is the one.

 

[roger2]

When it comes to Sansui, or any vintage amplifier.......

4) A note on recapping the regulator PCB - I use Panasonic FC caps. Yes, FM or Nichicon PS are lower impedance, but also so low that you're popping things with surge currents. Remember that even todays worst cap is superior to yesterday's best cap.

................................

Could you point to further information on this point? I have wondered whether some of the ultra low impedance caps could perhaps cause issues in some circuits, and have occasionally (rarely) come across warnings but nothing concrete. I would be interested to look at any links you may have on this topic. Thanks.


Oops...just realized this is an old post. Not sure if the OP is around currently...

 

[Hyperion]

I don't buy that, I don't know why I didn't react to that point originally, the other points are valid IMO.

I use low ESR capacitors (Panasonic FM / Nichicon HE etc) in all PSU and filtering locations and I have never experienced anything 'popping' as a result of their use. I wouldn't describe the above as 'ultra low ESR' either - just low ESR.

 

[roger2]

And I won't "buy" anything without additional information sources, so I asked for it.

Whatever word one chooses to describe it, HE and FM are lower impedance than FC, which (I assume) is lower impedance than original vintage era caps. In one of the cap comparison charts, Nichicon refers to HE as "extremely low impedance". In the FM spec sheet, that cap's impedance is referred to as "40% to 70% lower than FC series".

I don't know how the impedance of the vintage era caps compares to HE/FM, but it is possible that the difference could be significant. I also do not know whether the original designers factored capacitor impedance into their designs and whether it may be possible that in some circuits, changing that impedance could potentially be detrimental. Use of the phrase "popping things" is most likely overstating things. But if the OP has read something relating to issues with use of modern low impedance caps, I would like to read it too.

 

[Hyperion]

Fair enough regarding references, but I have never heard anyone stating, or even mentioning, that something 'popped' or might 'pop' because of the use of very low ESR capacitors.

 

[roger2]

Agreed totally. "Popping" due to use of low impedance caps is not something that I have heard of, or believe to be credible. But perhaps that reference came from somewhere, albeit re-worded. Perhaps something else was being described that was summarized in the OP. I have, a couple times, come across references to low impedance caps that were contrary to the generally accepted knowledge here at AK. Mind you, I read a lot over at diyaudio.com as well. And it is a fact that everything one reads online must be taken with a grain of salt. I am simply probing for information, and if something comes it can be evaluated. Not expecting anything, but you never know. Knowledge is never complete, but evolves continually.

 

[Hipocrates]

 

[JoshHendi]

Going back to the bias transistor... I purchased the recommended 2N2222 replacement, but the legs on the 2N2222 are very short... Does the bias transistor NEED to be mounted to the original heatsync to 'detect heat' ? (I've read about the original bias transistor on other sites and it is mounted to detect heat? Which doesn't make sense to me, but I am not that familiar with bias transistors) Or can it be mounted like a standard transistor? Not sure what the OP means by 'compression mount it'

 

[Hyperion]

Yes, the bias control transistor MUST be mounted as it was originally to ensure bias is under thermal control. The term 'compression mount' sounds like a flowery way of saying 'mounted securely and in physical contact with' - which of course it must be.

It is usual to use thermal compound (the white stuff) in between the bias control transistor and the heatsink to make absolutely sure the temperature of the bias control transistor is as close as possible to the temperature of the heatsink to which it is attached.

As for short leads - Extend Them!

 

[JoshHendi]

As for short leads - Extend Them!

Just solder 'leg extensions' to the 2N2222s ? Then I'd assume to use the plastic leg covers like the original Hitachi's have

 

[JoshHendi]

Just solder 'leg extensions' to the 2N2222s ? Then I'd assume to use the plastic leg covers like the original Hitachi's have

I went ahead and did this, and used the original metal transistor wrap on the 2N222... Got the amp up and running, bias adjusted, so far so good. I did notice after being on for an hour or so that the two driver board heatsyncs got fairly warm, but assume that is normal

 

[Hyperion]

the two driver board heatsyncs got fairly warm, but assume that is normal

How warm is 'fairly warm' - can you touch them? - hold them for several seconds? - and yes it is normal for them to get a little warm, that's why they have heatsinks right?

 

[JoshHendi]

When I touch the top of the heatsyncs on the driver boards, they are warm to the touch, but I can keep my fingers on them for 10+ seconds (but the heat is noticeable).

 

[Hyperion]

When I touch the top of the heatsyncs on the driver boards, they are warm to the touch, but I can keep my fingers on them for 10+ seconds (but the heat is noticeable).

From your description it sounds completely normal, as I said, that's why they have heatsinks in the first place - to get rid of heat. If they were both boiling hot, or at different temperatures then you would have an issue or two to resolve.

What temperature would you like them to be?

 

[JoshHendi]

From your description it sounds completely normal, as I said, that's why they have heatsinks in the first place - to get rid of heat. If they were both boiling hot, or at different temperatures then you would have an issue or two to resolve.

Same temperature on both and amp has been running good for hours now. Heatsyncs connected to outputs are barely above room temp. With the 20k being the monster it is, I know it will generate heat... I am a computer tech and heat always gets my attention and makes me a little nervous. If it runs for another few hours at it's current stability, I think I'm set

 

[Rotoflex4]

The 1uF polarized electrolytic capacitor positions on the meter board (F-2496), the silkscreen is wrong! The polarity of those caps is reversed to the silkscreen.

Just for the record: Today I'm working on my F-2496 board. It's a fairly early model AU-20000. The 4 x 1uF caps on my board are NOT oriented in reverse. In other words, mine agree with the silkscreening. They are the original grey Elnas and the solder looks original too. I'm thinking that maybe somewhere over the course of production they decided to reverse them? From one photo I've seen of a board with reversed caps there are a couple of resistors added to the rear of the board. Mine has no additional components on the backside. I'm tempted to pull my caps and turn them around, but I think I'll leave well enough alone.

 

[Falconet]

Hi guys, I have a chance to buy an AU 20000 with 4 output transistors missing for 300 usd. aside from this defect the seller claims the amp is untouched. not sure if it is a chance or that it's worth the hassle. so what do you guys think?