Revisiting the Sansui AU-999 - repair, rebuild and mods

[smurfer77]

I've not been quiet about saying my honest feelings about the AU-999 in the past. My earlier Sansui experiences were along the lines of AU-666, AU-555A, AU-505, AU-222. I was already in love but when I got my hands on an AU-777 and AU-777A I really fell deeply. I knew I had to have the TOTL from that era, the AU-999, that many seemed to like on these forums at that time. I somehow acquired 3 clean units within a month, tried all three, and was disappointed with the sound. Yes it was clear and solid, but I was gutted by the lack of bass and overall bright (for that era of sansui at least) flavour. The poor preamp gain was more of a mental issue... the main amp was clipping around the same time as the preamp.

I sold all three units immediately.

A few years later I see all of the mods and wonder if I was a bit hasty. I never even rebuilt one to see if the issues were just all three happened to sound the same with decay over time. Anyway, I have had a growing feeling for a while I need to give this beauty another chance and acquired a clean AU-999 recently.


UNPACKING
Doubled boxes - a good sign!

And here she is before cleaning up. Looks good no? And behind the AU-999 you see my wall of shame... the wall of unfinished projects. Actually it's just a part of the unfinished project collection. Well, actually I just did two of those last week (rebuilt a pair of HF-V60) and more will be underway over the xmas break!

REPAIR
Switches needed some cleaning, but she sounded to be working ok, as advertised. DC offset and bias adjustments went ok but check out the output from the speaker outlets. One after another i get amps that sound more or less ok but have an issue. I decided to fix this issue before rebuild... trying to engage my brain instead of hoping for the best with a recap.

Well, I pulled out the schematic and after some probing and thinking I figured it was likely C820 (or C819) which is a green mylar on the speaker output/feedback shunting VHF to ground. I pulled them and indeed one had gone very very low value, and wasn't doing anything at relevant frequencies. I replaced C820/C819 - the blue cap below is the new one.

and above you see it fixed the output. That is 18 Vrms out of both channels into 8Ohms, i.e. 40.5 W. That was about the limit before rebuild.

I did also find some DC getting around the amp where is shouldn't and found 2 other bad green mylar. Don't remember which ones exactly, maybe C715 and one other. Replaced those couple too. Anyone else had a bad time with green mylar in general or this is a bad batch??? They weren't short, just kind of low value (i.e. effectively open circuit for relevant frequencies).

I listened to the AU-999 for a few days in stock form after making sure it was working properly. Well, honestly, not as bad as I remember. I think I was a bit harsh. Definitely not the mid-range magic of the AU-777, and the bass was a bit weak, but not the total disaster I remember from before. I think I over-reacted a bit, even if the AU-999 in stock form isn't perfect. Okay, onto the rebuild and mods to see what the limits are.


REBUILD
Well I don't want to go into my usual detail for the rebuild. There are several threads already. In summary, I went heavy on the rebuild. I replaced all transistors except the drivers and outputs (for now), replaced ALL caps including small value. Those temp stable COG ceramics for smallest value, polypro wima for next smallest, electrolytics for large value. I used polyester WIMA caps for intermediate values in the signal path. Polypro was used up to about 1uF and then polyester up to 10uF. Usually people use electrolytic in that range but polyester is great, smaller than polypro and gets a bad wrap IMO. There is one nichicon muse BP on the driver boards... size was a bit of a problem for that one so I stuck with the electrolytic in that particular case... otherwise it is pretty much film all the way in signal path.

Here are the 10uF (50V) polyesters - smaller than you expected right?!

How does it sound after? Well, let me hold on a second before getting into that, because I actually did some mods before rebuilding (but after making sure the stock voltages and output were fine). So let me describe the mods now. Sterefun and others have outlined the summary of mods here and there so I wont go too much into it but here we go.

 

[smurfer77]

SUBSONIC FILTER REMVOAL
I bypassed the subsonic filters on the underside of the amp (just jumpered over the 47k resistors so I could go back and forther between having filter in or out for testing).

Bypassing the subsonic filter gave a big increase in bass from 20 - 50Hz. For example if I had 4.2V with filter in, I had 5.6V at 20Hz. The graph below paints the picture. Again, this data was taken before rebuild. But look at it. We go from -3.5dB at 20Hz to -0.9dB at 20Hz. While it's only a couple of dB and there is not much music using 20Hz the overall effect on bass was not subtle. Removing this subsonic filter is IMO a must do. Even the wife could A/B this one. (If one really needed a subsonic filter, I would mod to make it steeper and turning over as a slightly lower frequency but I don't need one so the subsonic filter is gone!)

KEVZEP PREAMP MOD
This one gives you more headroom on the preamp and gives you more voltage swing. The mod is simple to implement. As you can see, I also do this mod before full rebuild.

I was definitely able to feed higher input voltage into the preamp before overloading it but I wasn't able to get the 7.2Vrms that Kevzep reported. My max output went from around 2Vrms before mod, to about 2.5Vrms (and i note that this is 7.2Vp-p...). I did check again after replacing transistors though and it made no difference and I tried both lowish gain KSA/KSC as well as the high gain ZTX. A popular (on facebook at least, btw he hate AK) tech also seems to have only about 2.5-3Vrms out of his AU-999 after mod (note that graph is for voltage peak-to-peak, not Vrms).

Anyway, for now I notice only a subtle difference in pereived sound quality (IF you adjust the volume knob to give the same volume with and without mod).

The preamp definitely doesn't clip before the main amp, but i also saw no evidence of this before the mod, with only 1V preamp output required to get full output from the main amp

Also.... I would love to get more output from main amp as report by kevzep (72W) but i see same main amp output before and after mod (which makes sense as the main amp and voltage is the limit, not preamp clipping which happens well past main amp input sensitivity in stock form. The power output i achieved is same as others have reported (less than spec claimed output). .

Also, there might still be more to learn from the AU-888 here, as I noticed my tone boards had the same model number but they clearly had some differences to those in my AU-999, which surprised me (i'm not talking about just different values of components, but actually different board layout and different components. The differences are few, but still I thought it's interesting!).

 

[smurfer77]

STEREOFUN DRIVER BOARD MOD

The original thread is for this mod is here. Basically here we match up transistors/resistors and current in the diff pair on the driver board.

One interesting thing I noticed while doing this mod is that KSA1015 are pretty much identical from the same batch but when I used ZTX they were all over the place and it was much harder to get matches. Later when doing the rebuild I found the ZTX to result in unacceptable noise in the head (phono) amp but moving to the KSA1015 solved the problem (btw that might have been a gain issue, not just inherent noise of ZTX). Anyway i noticed no difference in noise when using either ZTX or KSA for the driver board. I just found the KSA better matched out of the box. I believe I have the ZTX installed now as that was the last thing I tested and it sounds fine.

As a side note i noticed that the AU-888 used 82 Ohm instead of 56 Ohm pulling the current into the diff pair.

The improvement to me was a more solid image; I felt I could pin-point vocals for example. It could be all in my mind as it is a bit of a subtle thing, but the mod is little effort so why the heck not!

Thanks to Stereofun for sharing his findings.



KEVZEP BASS/EQ MOD
This mod gives you more range of use on the tone controls. Before mod turning bass all the way up does what you would expect from 1-2 notches. After mod it does more or less what you would expect it to. I do find optimum setting for tone to my ears to be +1 bass and nothing on the others. However, I still run flat/defeat because I find the +1 bass setting muddies the bass slightly; that could totally be just my combined setup and interaction with my speakers. I have acoustic suspension speakers which are pretty tight in the bass but I do have 4x 10" woofers and the amp couldn't quite keep them under control with + bass settings. (however, I may have fixed that.... read on to NFB later).

Now, I tend not to use the tone controls and did all other comparisons with flat/defeat setting. So for me this mod might be a bit slightly optional, but I actually lean towards 'must do' because it is nice to have a functional tone control, even if I don't use it much.

Again, thanks Kevzep for sharing those findings.



REMOVE SCR SPEAKER PROTECTION

http://audiokarma.org/forums/index.php?threads/sansui-au-999-help-needed.749333/#post-10167466



In the pic you can see the two empty holes with white around them, next to the big caps. Just clip the wires going into there and store away safely. I know this is a bit controversial... I mean it is designed to stop your speakers getting friend. My experience though, when an output transistors or something goes short circuit from rail is that the fuse or something blows a lot quicker than my speakers. But you have been warned! remove at your own risk. I prefer it out, but I don't really hear any negative effect from leaving it in although others say different.

LISTENING AFTER MODS

I did listen at each step after each mod but now I have done all mods, and the full rebuild, I enjoyed the amp for the last few weeks, and compared, using an amp selector box (only way to do it! trust me on this), to my rebuilt AU-777A, a stock AU-888 and a rebuilt AU-70. Flicking between them and you know what I still prefer the AU-777A/AU-70 over the others. But the bass on the AU-999 is now absolutely fine - nothing wrong with it after mods and rebuild, at least in the flat/defeat setting. The highs and mids are also fine, just slightly different to the AU-777A, but very very good. In fact, the best way to describe my hot-rodded AU-999, is ...... an AU-888. The AU-999 sounds very very similar to the AU-888 now, as it probably was intended too. You know some folk are calling the AU-888 the AU-999A as a joke and I think it's close to the truth.

So basically I think the AU-999 after rebuild and mods is a top tier amp, worthy of being someone's only amp. I still prefer the AU-777A and AU-70. I do think the pin-point imaging and clarity of the AU-999 is a bit better and that is noticeable in certain music or movie listening.

Now if I turn it up a notch on the bass settings I have noted that the bass gets a bit muddy with my particular speaker/amp combo. This bothered me a bit, especially since the AU-888 is tight with the same speakers.


NEGATIVE FEEDBACK MOD
The high point of the AU-888 compared to the AU-777A is the sense of clarity that seems to come with doing away the output cap coupled scheme. The AU-999 has that clarity in the top, but in the bass it was a bit muddy as I have mentioned, especially when I turned the bass up. Turning the bass up on the AU-888 and the bass stays tight until much much later.

You might be fine with your speakers in combo with the AU-999 in this respect so maybe just be happy in that case.

In my case, I wanted to tweak. On my AU-999 I have 15 pF ceramic parallel, on back of driver board, over the 33kOhm feedback resistor R819. The other half of the voltage divider that determines the feedback level is a 1.2 kOhm resistor to ground. I.e. the ratio of the voltage divider is 33/1.2 = 27.5. This is quite different to the AU-888:

Model ------ R1 (kOhm)----R2 (kOhm)------R1/R2
AU-999 --- 33 ---------------1.2---------------- 27.5
AU-888 ---- 47 ---------------2.2---------------- 21.3
AU-666 ---- 33 ---------------1.5---------------- 22.0
AU-777A--- 47 ---------------1.2---------------- 39.2

I didn't note the caps, as they are not at values that are audio EQ shaping, but keep in mind values are also different for each amp and may still be important for high frequency and phase stability etc.

So what I found was that changing to 47/1.2=39.2 (less NFB) the amp did remind me more of the AU-777A, but that the bass was a bit out of control. Still, I though it was an improvement. Mids not quite at AU-777A levels still.

Then going to 47/2.2 = 21.4 like the AU-888 and it sounded great across the spectrum, with tighter bass much like the AU-888. This was my preferred NFB setting.

Note that with more NFB you may need to turn the volume knob up to get the same volume (and opposite was true when I decreased NFB). Makes sense, but might give you the false impression based on volume knob psychology that one setting is better or worse than another if you aren't careful. The good news is that with the kevzep preamp mod you can just turn up the volume knob more and get to max power output, no problem. And clipping is at around the same power. Actually, I got slightly closer to spec power with the new NFB setting. (I've previously never gotten spec power at spec distortion out of an AU-999). Before rebuild, mods and adjustments I got about 40 W at spec distortion out of the AU-999, or even just under. After everything I now get about 45-46 W. She will swing enough voltage for what would be about 72 W into 8 Ohms but when you hook up 8Ohms I couldn't beat around 45 W. Honestly, maybe closer to 40W if I'm strict on the distortion and not just looking at clipping.

Looking around frequency response of the different NFB settings I could see the lower (AU-777A style) NFB setting rolled off extreme high frequency a bit... just 0.5dB at 100kHz. The frequency response of the middle (AU-999 original) and higher (AU-888 style) NFB were quite similar. Not only in amplitude but also phase:

I don't really have good enough gear to really get into the high frequency stability. I did adjust the cap value slightly, to keep the cut-off frequency about the same and haven't run into any stability problems so far. I've also popper the original 15pF back in and all seems well for now even though that changes the cut-off frequency a bit... the phase isn't too different up to 100kHz. I think the AU-888 NFB settings are likely very safe in the AU-999... it's basically the same setup. But keep in mind, NFB is not child's play and use this mod with the understanding that I haven't analyzed all aspects of stability fully! Do I recommend it? I don't know; it's very speaker dependent IMO. In my case I wanted the extra damping factor of the increased NFB to keep my woofers under control.

OTHER COMMENTS

- XA495 > ZTX795A (higher gain and current handling, but matching pinout as XA495) [KSA1015 closer gain but collector centre not convenient]

- exception to above was that I used KSA in head amp (ztx gave a lot of noise!) I tried in otherboards too but heard no difference from ZTX.

- I also did the diff pair on the driver board with KSA because they seem to be better matched out of the packet.

- I measured the bias current vs distortion. Going from 15mA to zero slowly it is surprisingly good until you get close to 5mA. Also, pretty much nothing to be gained in distortion moving from 15mA to 100mA. The distortion does go lower as you increase bias from 100mA to 300mA but I think nobody really wants to cook eggs on their AU-999. All of the variations from ~5mA - 100mA are hardly worth mentioning above the noise floor. So I think spec of ~30mA is good (well I do 30mA with hood off in winter taking into account it will only go up). My distortion meters are HP8903A... very nice gear but if someone with more sensitive gear could confirm I would be grateful. Anyway, the decrease in distortion when moving from 15mA to 100mA is trivial and I think the 30mA spec is probably a good sweetspot between distortion and running cool and reliable.

- I've yet to change the TO-66 drivers and the TO3 outputs. I will probably change over to some modern devices when I get bored one day and see what happens.


- C807 (47uF, 16V) on driver board is bi-polar signal path cap. I tried changing this to 100uF (50V) and saw zero measurable difference either bass response or distortion over a modern bipolar nichicon muse 47uF (50V). i did notice that the AU-888 used only 22uF in that position, so probably the 47uF is decent enough for bass response.

- Original 4,700uF PS caps measured ~8000 uF!!! I still put new caps in since I already had them on hand. The new caps I installed were also 8000 uF. Since the old caps measured 8000uF I think we can stop worrying about the original rectifier in the AU-999 for these 8k-10k PS cap 'upgrades'.


- 999 is rated at 50W (into 8Ohms at 0.4%). I could only get 42 W before rebuild and now 46W. Interestingly the AU-888 is listed as 45W for same load/distortion and it gives measures about the same as my final version of AU-999. So I think the AU-999 is not 5 W more powerful than the AU-888; it's just a 5W overestimation error on the AU-999 spec sheet. Note that i could not get more than 46W out of the AU-999 for any NFB setup.

 

[kevzep]

Nice write up!!
The voltage swing I reported is achievable by the pre-amp post mod, it will do this if you inject enough voltage into the input. I probably wasn't clear about this when I did my post. I cannot remember what voltage was being injected to achieve this, I might do some measurements while I have this last AU999 I did here...
The gain of the pre-amp remains the same post mod.

On this last AU999 I restored, I was getting about 46 watts RMS both channels driven. My own one I did had some other mods to the driver board which I think may have been responsible for the 78 watts I got from that unit.
I neglected to take that into account when I reported my findings.

 

[kevzep]

I changed the TO66 drivers out on the last unit I did, I don't think it influences the power output at all as I was getting the same power as you were getting.

 

[smurfer77]

Nice write up!!
The voltage swing I reported is achievable by the pre-amp post mod, it will do this if you inject enough voltage into the input. I probably wasn't clear about this when I did my post. I cannot remember what voltage was being injected to achieve this, I might do some measurements while I have this last AU999 I did here...
The gain of the pre-amp remains the same post mod.

On this last AU999 I restored, I was getting about 46 watts RMS both channels driven. My own one I did had some other mods to the driver board which I think may have been responsible for the 78 watts I got from that unit.
I neglected to take that into account when I reported my findings.

Kev, yes I increased the input voltage to achieve my slightly higher preamp output. I can't get it to go any higher. I will look into it. Would love to know about the other driver board changes you hinted at

 

[kevzep]

Kev, yes I increased the input voltage to achieve my slightly higher preamp output. I can't get it to go any higher. I will look into it. Would love to know about the other driver board changes you hinted at

Hmmm, okay, let me run this 999 I have here and I'll see what it does......
The driver board mod was a constant current modification to the differential pair, involves a CC diode, which are now becoming very difficult to obtain.
You just need to measure the current draw of the diff pair through R845, then get the appropriate CC diode to replace that resistor. Then replace R811 and 813 with 1% metal films.
Then put it all together and measure the current across R807 and 809, and change the value slightly to balance out the current so they are both as identical as possible.
That's it. Makes quite a bit of difference to punch, bass clarity and top end linearity....

 

[smurfer77]

Hmmm, okay, let me run this 999 I have here and I'll see what it does......
The driver board mod was a constant current modification to the differential pair, involves a CC diode, which are now becoming very difficult to obtain.
You just need to measure the current draw of the diff pair through R845, then get the appropriate CC diode to replace that resistor. Then replace R811 and 813 with 1% metal films.
Then put it all together and measure the current across R807 and 809, and change the value slightly to balance out the current so they are both as identical as possible.
That's it. Makes quite a bit of difference to punch, bass clarity and top end linearity....

If you get a chance to look at your 999 preamp that would be amazing, but I know your busy so no worries if you can't... i'm sure I just need to look into what I've done wrong, but i will have more motivation to dive back in if you confirm your preamp output once again.

RE the constant current source for the diff pair, I bet there is a bit of flexibility on the current (I don't mean it isnt' worth matching, I mean the matched value might not be that critical within some range). While R845 was the same on the AU-888, R811 and 813 changed from 56 to 82 Ohm.

PS. thank you for taking my comments constructively. I was worried it might come off a bit of a smart ass move on my part, but really I just want to understand what is going on.

 

[kevzep]

If you get a chance to look at your 999 preamp that would be amazing, but I know your busy so no worries if you can't... i'm sure I just need to look into what I've done wrong, but i will have more motivation to dive back in if you confirm your preamp output once again.

RE the constant current source for the diff pair, I bet there is a bit of flexibility on the current (I don't mean it isnt' worth matching, I mean the matched value might not be that critical within some range). While R845 was the same on the AU-888, R811 and 813 changed from 56 to 82 Ohm.

PS. thank you for taking my comments constructively. I was worried it might come off a bit of a smart ass move on my part, but really I just want to understand what is going on.

Yeah I do get rather busy at times so hard for me to find time to do these things when it gets like that...

Nah I didn't think you were being a smart ass.....all good!!

 

[stereofun]

Thank you for this extensive write up and in- depth analysis. Very impressive work. I am thinking the 999 must be one of the most analyzed, documented and modded vintage amps out there. It is such fine build inside out that it is hard not to wish for it's sonics to fully match it's ambition.

It is interesting how Sansui engineers made subtle re-designs to the near identical but later 888 - hence the 999a reference. Perhaps the 888 NFB config might be a worthwhile and simple tweak to carry out and see what it does on your rig since it is "sanctioned" by the original engineers in a near identical design.

I would be very interested in your opinion and bias recommendation if you do change the drivers and outputs to onsemi mje150's / mj21194's. The last three amps I've worked on all had 1-2 output subs from older repairs. Finding units with all original sets seems to get harder and harder, so the question is do we keep the old drivers assuming they work well with the new outputs or will new drivers from same manufacturer let alone century create more coherence in the final phase when the image projects unto the big wall....

..... controversial subject, but those wimas might need some break in. There are several references online about wimas going from a high-mid /low- tweeter edgy'ness gradually followed a by a finer and more detailed sound.

Great work David, thanks for the effort

 

[Robisme]

Yes, very nice write up. Interesting read.

Now, for fair comparisons, are you going to rebuild the AU-888?

Rob

 

[slimecity]

I havent done the driver board mod yet. Im keen to get into that next.

Rather annoyingly - I cant see voltages on the schematic. For the voltages around TRs 801 & 803 - could KSA1815 be used instead of 1015? or would the voltages here require the 1015? I have heaps of 1815 spare and it would be nice to do this b4 xmas if possible.....

All those WIMA caps look fantastic in those boards BTW.

cheers!

 

[gslikker]

Thanks guys for the writups

 

[SanthoshA]

Wow, nice work and interesting piece of information. !
The subsonic filter removal is interesting , i didn't know that removing the section completely improves bass response, i was under the assumption that the switch off, is same as the mod that was performed.

 

[smurfer77]

Thank you for this extensive write up and in- depth analysis. Very impressive work. I am thinking the 999 must be one of the most analyzed, documented and modded vintage amps out there. It is such fine build inside out that it is hard not to wish for it's sonics to fully match it's ambition.

It is interesting how Sansui engineers made subtle re-designs to the near identical but later 888 - hence the 999a reference. Perhaps the 888 NFB config might be a worthwhile and simple tweak to carry out and see what it does on your rig since it is "sanctioned" by the original engineers in a near identical design.

I would be very interested in your opinion and bias recommendation if you do change the drivers and outputs to onsemi mje150's / mj21194's. The last three amps I've worked on all had 1-2 output subs from older repairs. Finding units with all original sets seems to get harder and harder, so the question is do we keep the old drivers assuming they work well with the new outputs or will new drivers from same manufacturer let alone century create more coherence in the final phase when the image projects unto the big wall....

..... controversial subject, but those wimas might need some break in. There are several references online about wimas going from a high-mid /low- tweeter edgy'ness gradually followed a by a finer and more detailed sound.

Great work David, thanks for the effort

Thanks Tom. Will let you know how it goes when I do the drivers and outputs. I guess I will get around to it eventually.

Yes, very nice write up. Interesting read.

Now, for fair comparisons, are you going to rebuild the AU-888?

Rob

Rob, yeah I will likely rebuild an AU-888. I have a pair of them so I may compare a rebuilt to stock one too.

I havent done the driver board mod yet. Im keen to get into that next.

Rather annoyingly - I cant see voltages on the schematic. For the voltages around TRs 801 & 803 - could KSA1815 be used instead of 1015? or would the voltages here require the 1015? I have heaps of 1815 spare and it would be nice to do this b4 xmas if possible.....

All those WIMA caps look fantastic in those boards BTW.

cheers!

No. You can't replace the 1015 with the 1815. The 1015 is PNP while the 1815 is NPN.

Wow, nice work and interesting piece of information. !
The subsonic filter removal is interesting , i didn't know that removing the section completely improves bass response, i was under the assumption that the switch off, is same as the mod that was performed.

The subsonic filter does not have a switch... it is always in. The "lo filter' is something else.

 

[slimecity]

No. You can't replace the 1015 with the 1815. The 1015 is PNP while the 1815 is NPN.

Oops, hence why several of my rebuild projects end up in the dustbin.....

I got confused between the 1815 and 1845.....

Not this one though! Only one mod to go.

 

[Declivita]

Great this is the best I have seen... a perfect reference ... I certainly would do better now... after seeing this....
my only concerns are the Fine Gold nichicons that I notice have below average.. ripple ratings..! But really like and feel comfortable with the nichicon BT's you have employed that come with above average ripple, E.S.R. and a rated reliable long life of up to 10,000 hour's...!
I replaced half the Mylars on mine without engaging my brain as its not been used in the Technician area for more than a couple decades now..!
Didn't like doing the Mylar's with the thick radial's bent on the track's, It has resulted in a cleaner performance and thinking now I should also do the Mylars on the Minus and Plus Filter block as well...!

 

[Declivita]

I've not been quite about saying my honest feelings about the AU-999 in the past. My earlier Sansui experiences were along the lines of AU-666, AU-555A, AU-505, AU-222. I was already in love but when I got my hands on an AU-777 and AU-777A I really fell deeply. I knew I had to have the TOTL from that era, the AU-999, that many seemed to like on these forums at that time. I somehow acquired 3 clean units within a month, tried all three, and was disappointed with the sound. Yes it was clear and solid, but I was gutted by the lack of bass and overall bright (for that era of sansui at least) flavour. The poor preamp gain was more of a mental issue... the main amp was clipping around the same time as the preamp.

I sold all three units immediately.

A few years later I see all of the mods and wonder if I was a bit hasty. I never even rebuilt one to see if the issues were just all three happened to sound the same with decay over time. Anyway, I have had a growing feeling for a while I need to give this beauty another chance and acquired a clean AU-999 recently.


UNPACKING
Doubled boxes - a good sign!

And here she is before cleaning up. Looks good no? And behind the AU-999 you see my wall of shame... the wall of unfinished projects. Actually it's just a part of the unfinished project collection. Well, actually I just did two of those last week (rebuilt a pair of HF-V60) and more will be underway over the xmas break!

REPAIR
Switches needed some cleaning, but she sounded to be working ok, as advertised. DC offset and bias adjustments went ok but check out the output from the speaker outlets. One after another i get amps that sound more or less ok but have an issue. I decided to fix this issue before rebuild... trying to engage my brain instead of hoping for the best with a recap.

Well, I pulled out the schematic and after some probing and thinking I figured it was likely C820 (or C819) which is a green mylar on the speaker output/feedback shunting VHF to ground. I pulled them and indeed one had gone very very low value, and wasn't doing anything at relevant frequencies. I replaced C820/C819 - the blue cap below is the new one.

and above you see it fixed the output. That is 18 Vrms out of both channels into 8Ohms, i.e. 40.5 W. That was about the limit before rebuild.

I did also find some DC getting around the amp where is shouldn't and found 2 other bad green mylar. Don't remember which ones exactly, maybe C715 and one other. Replaced those couple too. Anyone else had a bad time with green mylar in general or this is a bad batch??? They weren't short, just kind of low value (i.e. effectively open circuit for relevant frequencies).

I listened to the AU-999 for a few days in stock form after making sure it was working properly. Well, honestly, not as bad as I remember. I think I was a bit harsh. Definitely not the mid-range magic of the AU-777, and the bass was a bit weak, but not the total disaster I remember from before. I think I over-reacted a bit, even if the AU-999 in stock form isn't perfect. Okay, onto the rebuild and mods to see what the limits are.


REBUILD
Well I don't want to go into my usual detail for the rebuild. There are several threads already. In summary, I went heavy on the rebuild. I replaced all transistors except the drivers and outputs (for now), replaced ALL caps including small value. Those temp stable COG ceramics for smallest value, polypro wima for next smallest, electrolytics for large value. I used polyester WIMA caps for intermediate values in the signal path. Polypro was used up to about 1uF and then polyester up to 10uF. Usually people use electrolytic in that range but polyester is great, smaller than polypro and gets a bad wrap IMO. There is one nichicon muse BP on the driver boards... size was a bit of a problem for that one so I stuck with the electrolytic in that particular case... otherwise it is pretty much film all the way in signal path.

Here are the 10uF (50V) polyesters - smaller than you expected right?!

How does it sound after? Well, let me hold on a second before getting into that, because I actually did some mods before rebuilding (but after making sure the stock voltages and output were fine). So let me describe the mods now. Sterefun and others have outlined the summary of mods here and there so I wont go too much into it but here we go.

Pulled out C819 and C820 on the minus and plus filter block F1277 & F1231 on mine and noticed one going low slightly down by 5%, replaced the pair with Polyester MKT and the sansui sound's much cleaner in the mid's but the bass has dropped of slightly..! The AU-999 sounds much better detailed though...its very noticeable

 

[BLAH BLAH]

Yes a more noticeble increase in detail, ear friendly without being overly bright! Sounds great!

 

[smurfer77]

Pulled out C819 and C820 on the minus and plus filter block F1277 & F1231 on mine and noticed one going low slightly down by 5%, replaced the pair with Polyester MKT and the sansui sound's much cleaner in the mid's but the bass has dropped of slightly..! The AU-999 sounds much better detailed though...its very noticeable

That's great you are happy with the result. To be clear though, 5% down on that type of cap is within tolerance of the original spec and within uncertainty of your measurement. Anyway a good idea to get some film caps in - they will last a lifetime hopefully.