AU-α907Limited - overheating

Hyperion

Roobarb & Custard
Subscriber
I've been having a look at this AU-α907Limited, and the overheating it seems to have suffered. I discovered the probable reason, i.e. that the bias was set at something around 40mV (should be 28.6mV) for both channels, the somewhat higher setting (45mV) was for the Driver/OP module pictured below.

Here are some pictures of one Driver/Output assembly

IMG_2494a.jpg

Here is a close up - showing the shrunken sleeves of the 4 capacitors (2 pairs of 2) in the foreground centre.
IMG_2495a.jpg

I have removed the worst looking two (they are 2.2µF 50V) and measured them, they have different ESR's at 0.49Ω & 2.0Ω although the values seem to have held well. The combination, they are joined + to + to form a bipolar, measures 5.0Ω ESR.

I think they have obviously suffered some heat stress, and really should be replaced - however 'Elna Silmic α Super Gold' capacitors don't seem to be around any more from a reliable source. So, I am thinking of replacing them with some nice films - one film capacitor per original pair, which will fit very well in the space available. Elna Silmic II, or Nichicon Muse ES would seem to be the only other choices.

My thinking here is, 'why not take the opportunity to get rid of 2 electrolytics' in favour of films.

Any thoughts?
 
Last edited:
Toasty!!......The old shrinking plastic sleeve!! Those caps are done, even if they test good.....
One of the G6700's I was working on was like this...
 
John,

How long did you wait for the bias to settle down (and/or did you 'preheat' to speed things up)?

The reason I ask is because I wonder if it's coincidence that you mention ~40mV, which is is what I measured in my 907limited ...before it warms up. Without preheating it takes a full 60 minutes to reach steady-state. So on my amp 40mV at start up (well 38 - 39mV actually) gives about 28mV after my unit warms up.
 
I recently replaced the same capacitors (which were Elna Silmic IIs with around one year of use) in my AU-x711 with a polypropylene film cap - I found a Panasonic with the correct lead spacing to suit.

I do think it added a little more transparency to the sound, but I have no way of confirming 100%.
 
How long did you wait for the bias to settle down (and/or did you 'preheat' to speed things up)?
Ah ! - no preheating used, those OP transistors got plenty hot all by themselves it seemed to me, but 60 minutes! - that's crazy. In the time I had the unit on the bias zoomed up to ~40mV (x2) for one channel, and over 45mV (x2) - for the other channel. I didn't actually wait to see where the higher one ended up to be honest, I just thought it was far too high. I think I had it powered on for about 20 minutes, I adjusted the bias settings down to the correct value during that time - but they showed no signs of reducing that I could see. The bias tracking transistors were not mounted as efficiently as they could have been, rather carelessly done in my opinion for such an important part, I'll be improving on that.
I found a Panasonic with the correct lead spacing to suit.
Thanks Pete, I think I may have some of the same ones, ;) Panasonic ECW 1µF, 250V Polypropylene.

Ok, thanks for your pointers guys, time to order some trimmers I think, and get the amplifier re-assembled for another test...
 
Ah ! - no preheating used, those OP transistors got plenty hot all by themselves it seemed to me, but 60 minutes! - that's crazy. In the time I had the unit on the bias zoomed up to ~40mV (x2) for one channel, and over 45mV (x2) - for the other channel. I didn't actually wait to see where the higher one ended up to be honest, I just thought it was far too high. I think I had it powered on for about 20 minutes, I adjusted the bias settings down to the correct value during that time - but they showed no signs of reducing that I could see. The bias tracking transistors were not mounted as efficiently as they could have been, rather carelessly done in my opinion for such an important part, I'll be improving on that.

Ok. If the bias showed no signs of moving after 20 minutes then yes that is kind of weird and I think you lost the bias servo feedback (as I think you already suspect).

You may find that after fixing those poorly mounted servo transistors that you may end up turning the bias back up. My 907LTD drops from 40 mV to 28mV over an hour (without 'preheating') and in fact the first time I tried biasing I set to 28mV after 10 minutes, and then at 1hr I had close to 0mV! Have fun :).
 
I have removed both power amp modules, changed the trimmers and heat affected capacitors in both channels. I have remounted the OP transistors using new compound after thoroughly cleaning all the mounting hardware, and paid attention to the mounting of the bias tracking transistors. Initial tests suggested the bias drift was very much reduced, but now I am not so sure, overall I am much happier with it now compared to how it was. ;)

I have reset the bias and given it a 1½ hour warm up, and I now have the Left channel running at ~36°C and the Right channel running at ~42°C (covers off) - with the bias at the same level for each side.

I am a bit surprised by the temperature difference, (not what I am used to), but I am thinking it may well be just a characteristic of this amplifier and the OP devices used.

I would welcome any advice or comments on the above. ;)

@smurfer77 ;)
 
Last edited:
I have removed both power amp modules, changed the trimmers and heat affected capacitors in both channels. I have remounted the OP transistors using new compound after thoroughly cleaning the mounting hardware, and paid attention to the mounting of the bias tracking transistors. Initial tests suggested the bias drift was very much reduced, but now I am not so sure, overall I am much happier with it now compared to how it was. ;)

I have reset the bias and given it a 1½ hour warm up, and I now have the Left channel running at ~36°C and the Right channel running at ~42°C (covers off) - with the bias at the same level for each side.

I am a bit surprised by the temperature difference, (not what I am used to), but I am thinking it may well be just a characteristic of this amplifier and the OP devices used.

I would welcome any advice or comments on the above. ;)

@smurfer77 ;)

Great work! I should change the VRs on mine.

Can I ask what was your approximate ambient temperature? I got high 40s at steady-state but if I recall my ambient was mid 20s.

I think I've mentioned elsewhere that on all of my dual power amp board alphas the right heat sinks run hotter. I attributed this to the left heat sink being directly mounted to the chassis but the right heat sink being less efficiently coupled to the chassis as its coupling to chassis is largely intermediated by phono-section mounting. On my amps I've seen temperature spread of 2-4C. 6C is a surprisingly big difference but I guess it could be ok. I would have been more worried if you told me right was 6C cooler than left.

Lastly, if you happen to have the data, when you amp is set to ~28.6mV warm, what does it read intially when you turn it on cold? As I mentioned on mine it's about 40mV. I'm curious to know because this is a bit of a time-saving fact, when rebiasing if you know roughly where it should be at cold.
 
I need to recheck the bias again after a 1 hour warm up, I did tweak it a couple of times to flatten out the warm up behaviour. Eventually I had it powered on for 2 hours monitoring every 15 minutes or so. The early stages (10 mins) of the warm up was with the amplifier standing on its left side, the next time I run it up, I want to minimise this to firm up on the results so far obtained. For those people who have yet to encounter one of these - the bias monitoring points are underneath the amplifier, and there are 4 of them, thus potentially needing 4 meters to see them all at the same time.
Lastly, if you happen to have the data, when you amp is set to ~28.6mV warm, what does it read intially when you turn it on cold? As I mentioned on mine it's about 40mV. I'm curious to know because this is a bit of a time-saving fact, when rebiasing if you know roughly where it should be at cold
Yes, this is next ;) - I want to know this too, not primarily from a time saving point of view - but more from a reliability and consistency point of view. ;)

PS: ambient was about 20-22°C

Thank you so much for your words of wisdom and re-assurance, feeling much more comfortable about this now. :thumbsup:
 
Last edited:
Ok, the results.

Ambient is 21°C

Cold start
Left = 22 & 22 - peaking at 35mV in 5 minutes - eventual settle reading 28.6mV after 1 hour. heatsink temperature = ~36°C

Right = 20 & 20 - peaking at 34mV in 5 minutes - eventual settle reading 28.6mV after 1¼ hours. heatsink temperature = ~42°C

I am still somewhat concerned about the difference in heatsink temperature, but very pleased that the bias for both channels zeroes in on the set point very nicely now - and is very stable too. As for warm up time, '30 minutes good, 60 minutes better' for this amplifier. :)
 
Last edited:
Ok, the results.

Ambient is 21°C

Cold start
Left = 22 & 22 - peaking at 35mV in 5 minutes - eventual settle reading 28.6mV after 1 hour. heatsink temperature = ~36°C

Right = 20 & 20 - peaking at 34mV in 5 minutes - eventual settle reading 28.6mV after 1¼ hours. heatsink temperature = ~42°C

I am still somewhat concerned about the difference in heatsink temperature, but very pleased that the bias for both channels zeroes in on the set point very nicely now - and is very stable too. As for warm up time, '30 minutes good, 60 minutes better' for this amplifier. :)

The right side amplifier gerbils may be running a fever... Tylenol should help
 
Ok, the results.

Ambient is 21°C

Cold start
Left = 22 & 22 - peaking at 35mV in 5 minutes - eventual settle reading 28.6mV after 1 hour. heatsink temperature = ~36°C

Right = 20 & 20 - peaking at 34mV in 5 minutes - eventual settle reading 28.6mV after 1¼ hours. heatsink temperature = ~42°C

I am still somewhat concerned about the difference in heatsink temperature, but very pleased that the bias for both channels zeroes in on the set point very nicely now - and is very stable too. As for warm up time, '30 minutes good, 60 minutes better' for this amplifier. :)

I will rebias mine shortly and follow you timing for measurements and see what I have for precise comparison.
 
I will rebias mine shortly and follow you timing for measurements and see what I have for precise comparison.
That will be good to know. :thumbsup:

I am listening to it now, the precision of the imaging is gorgeous, it has a relaxing presentation, tone controls flat, no loudness needed, but sounds very full and accurate. I've tried the 'Source Direct' setting, it makes a difference (slightly lower level), but turning up the volume to the same level yields no difference in SQ that I can detect. ;)

I listened to it from cold, and couldn't detect any change in SQ as it warmed up, I'll leave it on for a while longer, and try it with some different music.

I need to take the front panel and knobs off for a good clean, as they are filthy! - then I think I'll have to give this beauty back to its owner. :( :thumbsup:
 
Hi John,

I have to ask these questions while you still have the amplifier,prefacing them with the admission that most of the TOTL Sansuis sound the same (and by this I mean very good) to me.

Any sound differences that you can attribute to MOS FETs?

How does this Alpha series model stack up to your AU 919, D11 II, and X1 or any others on which you care to comment?

Any idea as why the Sansui MOS FETs have lower rated power? Is this because they need to be biased to run hot with temperature being a limiting consideration?

Any opinions or speculations would be welcome.

Cheers.

Marc
 
most of the TOTL Sansuis sound the same (and by this I mean very good) to me.
This is true, they are very difficult to separate.
Any sound differences that you can attribute to MOS FETs?
None at all, their use in this amplifier at least, seems completely transparent to me. Although I have a Technics SU-A900 MkII, which is a MOSFET design which I find seems less 'bright' - smoother, but losing some clarity, if that makes any sense.
How does this Alpha series model stack up to your AU 919, D11 II, and X1 or any others on which you care to comment?
Just a hair better than the AU-919 or the AU-D11 II, with a bit more of a gap (two hairs) to the X1 - which although it has its strengths (i.e. phono sections) has very slightly less delicacy than the '919 or the 'D11 II in my opinion. I wasn't sure about the AU-919 until I had done (my version of) the 'Leestereo upgrades' to it, now it stands right up there with the best.
Any idea as why the Sansui MOS FETs have lower rated power? Is this because they need to be biased to run hot with temperature being a limiting consideration?
Not sure on that one, that's one for the MOSFET experts. ;)
 
Last edited:
Back
Top Bottom