Sansui 907 series X-Balanced/Alpha - a bit of chronology [technical]

fojica

Active Member
Having now almost all 907 series service manuals available thanks to @JimEGR kindness, I decided to address a bit some chronology regarding the 907 series.
I will start again by saying that from a technical perspective "X-Balanced" and "Alpha" amps are just a different marketing name, all of them being balanced amplifiers. Alpha series benefited indeed from some more qualitative parts compared to previous X-Balanced series. But the topology and the way how they work are identical.

This time I won't cover any changes regarding power-amp section, as I mentioned (I believe) all of them in my earlier postings. So there is no reason to repeat the same thing. But, if anybody wants to have all the information in a single thread, no problem - feel free to add my relevant earlier postings also to this thread.
Instead, this time I will focus on an area usually overlooked by many people - the power supply unit (PSU).

Believe it or not, the main filtering caps (being for the power amp or another section of the amp) are in the circuit and they inevitably influence the sound. And before jumping to say that this is a nonsense or whatever, just ask yourself the question "how does the current flows in the circuit when rectifying diodes are not conducting ?". Rectifying diodes are conducting only during the charging pulses, otherwise the circuit can be seen as "open" from their perspective.
After the main caps, any bypass cap and eventually any following voltage regulator has also "something to say", as it will play its part on the final sound.
This is why the PSU is important - because whatever bias/idle and dynamic current is required in the circuit to drive the "load", it will only come from the PSU.
And the general golden rule is "the better the PSU, the better the sound".

First 907 X-Balanced model was AU-D907X, which enjoyed a dual-mono topology for power-amp section and a "shared" closed-loop (CL, aka with feedback) voltage regulator for phono and tone/flat-amp modules. When I say shared it means it's only one voltage regulator which supplies power to both L and R channels. This CL-type regulator was somehow in-line with the older models, as the older AU-919 for example also had a feedback voltage regulator (albeit a more complex one).
d907x.jpg
The topology was a "centralized" one, this being the only voltage regulator and then having just decoupling caps on phono and tone/flat-amp modules.

The same "centralized" topology is to be found also on first Alpha 907, AU-α907 from 1986. But this time instead of using CRD's they used JFET's with resistors for current limiting - doing the exact same thing as the previous CRD's. For anyone wondering., CRD = Current Regulation Diode

All Alpha 907 models don't feature anymore the older AU-D907X's dual-mono topology for power-amp section, instead the main caps are connected in parallel - even if there are separate secondary windings and rectifiers. for each bank of caps The effect would likely be lower PSU impedance as seen by the amp sections, but the price to be paid for that is the existence of some cross-talk between R and L channels. This was also discussed on Hyperion's thread related to his AU-α907MR, so more details can be found there (including a tweaking/mod).
As a general rule, for all 907 Alpha models the main open-loop voltage regulator is shared between R and L channels.

a907.jpg

First change on PSU came with AU-α907i, where the main voltage regulator was changed to an open-loop type (no feedback, classic zener + series pass transistor/darlington). But this time the "centralized" topology was dropped and individual voltage regulators were added for phono and tone/flat-amp modules.
a907lex.jpg
a907i.jpg

The exact same topology was maintained also for next AU-α907L Extra.

Later series AU-α907DR, AU-α907KX and AU-α907XR simplified the regulator for tone/flat-amp, having a single transistor instead of a compounded darlington. The main open-loop voltage regulator was preserved like before, and that goes until late AU-07 Anniversary (most likely also NRA series).
a907dr_kx_xr_.jpg

Then on AU-α907MR and AU-07 Anniversary the voltage regulator for tone/flat-amp module was dropped entirely and replaced with a simple RC filter. Some will maybe say that on MR and later series there are current mirrors, and so there was no need for a local regulator anymore. That could be true or not, at least partly. Because no current mirror or CCS is "immune" to the ripple present on the supply lines. So having a local regulator for sure cannot hurt and can only make things better.

907mr_anv.jpg

For AU-α907NRA is not clear (no SM), but I have no reason to suspect that anything was changed compared to MR series from PSU perspective.

If anybody wants to address in a similar way other sections of the amps, or feel that I missed something - please feel free to add (to this thread).

Regards
 
Register to hide this ad
Really nicely done! Thanks.
I do have some questions. As you may remember we once discussed the floating ground of the AU-D907G. I see some similarities to this in the regulated power supply from the AU-D907X Decade. You can see that there is no connection from the power supply ground to the pre amplifier boards. And there is a 10 Ohm resistor to the centre tab of the transformer. If you look at the connection symbols in the schematic, some of the ground connections of the pre amplifier boards go to the unregulated supply for the input stages of the driver board, some i can not find. Why is this interesting. In the alphas all pre amplifier and input stage supplies are not ground free or ground floating.
Secondly, what is your opinion on the balanced configuration of the alpha 907/907i/907i MOSFET control stages which then connect there signal to both jfets of the differential pair of the power amplifier?
This was used simplified in the AU-GxxX models which was called X-balanced II.

Screenshot_20241023-010057.png
 
Having now almost all 907 series service manuals available thanks to @JimEGR kindness, I decided to address a bit some chronology regarding the 907 series.
I will start again by saying that from a technical perspective "X-Balanced" and "Alpha" amps are just a different marketing name, all of them being balanced amplifiers. Alpha series benefited indeed from some more qualitative parts compared to previous X-Balanced series. But the topology and the way how they work are identical.

This time I won't cover any changes regarding power-amp section, as I mentioned (I believe) all of them in my earlier postings. So there is no reason to repeat the same thing. But, if anybody wants to have all the information in a single thread, no problem - feel free to add my relevant earlier postings also to this thread.
Instead, this time I will focus on an area usually overlooked by many people - the power supply unit (PSU).

Believe it or not, the main filtering caps (being for the power amp or another section of the amp) are in the circuit and they inevitably influence the sound. And before jumping to say that this is a nonsense or whatever, just ask yourself the question "how does the current flows in the circuit when rectifying diodes are not conducting ?". Rectifying diodes are conducting only during the charging pulses, otherwise the circuit can be seen as "open" from their perspective.
After the main caps, any bypass cap and eventually any following voltage regulator has also "something to say", as it will play its part on the final sound.
This is why the PSU is important - because whatever bias/idle and dynamic current is required in the circuit to drive the "load", it will only come from the PSU.
And the general golden rule is "the better the PSU, the better the sound".

First 907 X-Balanced model was AU-D907X, which enjoyed a dual-mono topology for power-amp section and a "shared" closed-loop (CL, aka with feedback) voltage regulator for phono and tone/flat-amp modules. When I say shared it means it's only one voltage regulator which supplies power to both L and R channels. This CL-type regulator was somehow in-line with the older models, as the older AU-919 for example also had a feedback voltage regulator (albeit a more complex one).
View attachment 3326797
The topology was a "centralized" one, this being the only voltage regulator and then having just decoupling caps on phono and tone/flat-amp modules.

The same "centralized" topology is to be found also on first Alpha 907, AU-α907 from 1986. But this time instead of using CRD's they used JFET's with resistors for current limiting - doing the exact same thing as the previous CRD's. For anyone wondering., CRD = Current Regulation Diode

All Alpha 907 models don't feature anymore the older AU-D907X's dual-mono topology for power-amp section, instead the main caps are connected in parallel - even if there are separate secondary windings and rectifiers. for each bank of caps The effect would likely be lower PSU impedance as seen by the amp sections, but the price to be paid for that is the existence of some cross-talk between R and L channels. This was also discussed on Hyperion's thread related to his AU-α907MR, so more details can be found there (including a tweaking/mod).
As a general rule, for all 907 Alpha models the main open-loop voltage regulator is shared between R and L channels.

View attachment 3326799

First change on PSU came with AU-α907i, where the main voltage regulator was changed to an open-loop type (no feedback, classic zener + series pass transistor/darlington). But this time the "centralized" topology was dropped and individual voltage regulators were added for phono and tone/flat-amp modules.
View attachment 3326803
View attachment 3326806

The exact same topology was maintained also for next AU-α907L Extra.

Later series AU-α907DR, AU-α907KX and AU-α907XR simplified the regulator for tone/flat-amp, having a single transistor instead of a compounded darlington. The main open-loop voltage regulator was preserved like before, and that goes until late AU-07 Anniversary (most likely also NRA series).
View attachment 3326812

Then on AU-α907MR and AU-07 Anniversary the voltage regulator for tone/flat-amp module was dropped entirely and replaced with a simple RC filter. Some will maybe say that on MR and later series there are current mirrors, and so there was no need for a local regulator anymore. That could be true or not, at least partly. Because no current mirror or CCS is "immune" to the ripple present on the supply lines. So having a local regulator for sure cannot hurt and can only make things better.

View attachment 3326820

For AU-α907NRA is not clear (no SM), but I have no reason to suspect that anything was changed compared to MR series from PSU perspective.

If anybody wants to address in a similar way other sections of the amps, or feel that I missed something - please feel free to add (to this thread).

Regards

Just when it looked like this forum couldn't get any better - then this brilliant post!
 
Yes, way above my head but fascinating to read about. I think I get the concepts though.

About PSU: how did they evolve re: a pair of large filter caps vs more smaller value caps, for faster response carrying the same ( or close to equilavant ) capacitance? Or did they?
 
The dual-mono to "centralized" Alpha power amp shift is an interesting one, and indeed the justification seems like a mystery considering the benefits of "de-centralizing". And with the dual-mono being deployed in the B-2301 which has been considered to be the "best" power amp Sansui made, the dual-mono omission for the Alpha integrated's is peculiar.

The related conversation in @Hyperion AU-a907MR thread starts here for anyone interested. I've read it before but was glad to read it again! As I look over at my AU-a907Limited wondering how it might sound in a "de-centralized" configuration...

I believe I interpreted all of this correctly. Open to adjust my statements if anything is not technically accurate. Thanks to all of you technically accomplished Sui folks for demystifying a lot of this. Interesting stuff!
 
Last edited:
@GrisPato - You probably refer to the 10R resistors that were the subject in this post -> https://audiokarma.org/forums/index.php?threads/sansui-jpd-adventure.963332/post-15239802, and then on another following it.
If that's the case, then I don't know the answer - for now to me it's not clear why those 10R resistors were there.
If I missed the point, then please post a photo where those resistors are located.

Secondly, what is your opinion on the balanced configuration of the alpha 907/907i/907i MOSFET control stages which then connect there signal to both jfets of the differential pair of the power amplifier?
This was used simplified in the AU-GxxX models which was called X-balanced II.
You probably refer to the fact (which I mentioned previously) that in the first α907 series the flat-amp is balanced, contrary to "normal" topology which was used previously and later. I don't know why Sansui have chosen to do this and then dropped it later. And I won't speculate just for the sake of talking.
But you can find mode details about that configuration (or the idea itself), provided by more knowledgeable people than me, like here: https://www.hairballaudio.com/blog/resources/diy-resources/balanced-and-differential
Hope that answers your question.
 
About PSU: how did they evolve re: a pair of large filter caps vs more smaller value caps, for faster response carrying the same ( or close to equilavant ) capacitance? Or did they?
If one looks through the schematics of older X-Balanced series like AU-D11-II or AU-G90X/99X it would be obvious that it was more capacitance per channel and also usually that was made of 2x caps of the same size, like 2 x 10.000uF.
On later Alpha series (excluding B-2201/2301/2302) the approach was changed and the overall capacity was lower and made from non-equal caps, like 5600uF+8200uF or similar.
But those caps are not necessary the same "type"/manufacturer - usually they are marked as "Gold Supply" and "Pure Focus". About the same size even if capacitance is different..
Why they mixed those - I don't know, probably they have different characteristics and mixing them gets the best of both worlds, sort of that.
But I don't think anybody can explain it from a technical perspective, unless somebody with more advanced measuring tools can check and find details about their parameters and then get an idea. No chance for me to do that as I'm just a poor diy-er, so on my part it will remain a mistery :dunno:
 
Why they mixed those - I don't know, probably they have different characteristics and mixing them gets the best of both worlds, sort of that.
The only reason I can think of is that large capacitors can have inductance** which causes undesirable effects, maybe (as you say) they mixed the sizes to disrupt the inductance effect caused by this apparently difficult to avoid characteristic. Some remediation of this is provided by the use of film capacitors placed in parallel with the large electrolytics.

(** caused I think, because the capacitor foil is effectively a 'wound conductor' i.e. a coil, and therefore naturally has 'inductance')
 
Last edited:
Reading & understanding theses subtle things about the Sansui's Alpha amp design slowly is like reading a classic Agatha Christie suspense novel on a rainy day.

Just when you start to think that the fog is clearing up, it thickens. Wow.....

I have also wondered many a times that -

Why same size but different capacity capacitors used ?

Why used many capacitors of different capacities in filtering ?

and since I've just started learning about amplifier designs and their applications, it's a great great read.

Thanks to all enthusiasts and Gurus to bring their hard earned and acquired knowledge to the forum.

I have complete set of all boards of a Sansui AU 607KX Mos Premium except power stage and transformer and then a power stage pair of AU 607XR coming to me from Japan.

Don't have any chassis and transformer, filtering caps, front plact and knobs.

Will open a new thread once receive them and learn how to start the project.
 
As I look over at my AU-a907Limited wondering how it might sound in a "de-centralized" configuration.
Why don't you try it, Jim. That's likely the easiest possible mod that can be done on a Alpha 907 or on X-1111 + B-2102/2103/2105 MOS versions. Also easy enough to "undo", in case you don't like the result - which frankly I doubt, but that's another page of the story.
With all due respect (as I am fully aware that most of the folks here are not quite on the technical side), but if that one is already bordering "the complicated side of the things", then clearly there's no point to even think starting a topic on modding Sansui balanced amps.
 
Why don't you try it, Jim. That's likely the easiest possible mod that can be done on a Alpha 907 or on X-1111 + B-2102/2103/2105 MOS versions. Also easy enough to "undo", in case you don't like the result - which frankly I doubt, but that's another page of the story.
With all due respect (as I am fully aware that most of the folks here are not quite on the technical side), but if that one is already bordering "the complicated side of the things", then clearly there's no point to even think starting a topic on modding Sansui balanced amps.

That amplifier is currently "the most perfect" in my collection - condition and SQ. So I want to just enjoy it as is for a while - why mess with perfection? :)

I have a fair understanding of balanced tech, and I have been through a half dozen or more traditional Sansui restorations thanks to the help of folks here. But I do know where I rank on the technical side compared to most of y'all so I have no delusions about my abilities. After reading yours and @Hyperion conversation about the mod on his AU-α907MR, I do think that work is in my wheelhouse - and it sounds like the mod will indeed improve the performance of my AU-α907 Limited. I will try it at some point, but will definitely check with all of you Sαnsui Alphα Sαge's before I do!
 
Last edited:
If one looks through the schematics of older X-Balanced series like AU-D11-II or AU-G90X/99X it would be obvious that it was more capacitance per channel and also usually that was made of 2x caps of the same size, like 2 x 10.000uF.
On later Alpha series (excluding B-2201/2301/2302) the approach was changed and the overall capacity was lower and made from non-equal caps, like 5600uF+8200uF or similar.
But those caps are not necessary the same "type"/manufacturer - usually they are marked as "Gold Supply" and "Pure Focus". About the same size even if capacitance is different..
Why they mixed those - I don't know, probably they have different characteristics and mixing them gets the best of both worlds, sort of that.
But I don't think anybody can explain it from a technical perspective, unless somebody with more advanced measuring tools can check and find details about their parameters and then get an idea. No chance for me to do that as I'm just a poor diy-er, so on my part it will remain a mistery :dunno:

Some may recall that Sansui mixed non-equal large capacitors in the late model AU-X11
Early models are 8x10,000
Late models are 4×6800 & 4x10,000

I have both versions and the later model has a much fuller sound to it imo.
 
why mess with perfection
I see your point, definitely nothing wrong with going "original". And I totally understand the "value" of the investment, as these amp don't quite cost 20-50 bucks.
But...in the same time...why not trying to best the perfection ? :biggrin: Accordingly to Hyperion, he found it to be an improvement. I always walked the dual-mono path in my diy amps in the last 15 years or so, for me this is the "default" (whenever possible/achievable).
Yeah, I know, it's the diy-er inside of me who's screaming "but at least try it and see what you get" :D:D
 
Some may recall that Sansui mixed non-equal large capacitors in the late model AU-X11
Early models are 8x10,000
Late models are 4×6800 & 4x10,000

I have both versions and the later model has a much fuller sound to it imo.
I didn't check any non-X-Balanced/Alpha models regarding this (out of scope so to say), so I take your word for it.
But it could be an "older recipe" that was probably re-tried and eventually re-applied at a later point....definitely possible.
 
I didn't check any non-X-Balanced/Alpha models regarding this (out of scope so to say), so I take your word for it.
But it could be an "older recipe" that was probably re-tried and eventually re-applied at a later point....definitely possible.
That is what I was suggesting with my post.
Sansui may have looked into the past to results that worked.

Could you imagine being that engineer at Sansui.
Boss comes to you and says "This is pefect! Now make it better."
That job would be a blessing and a curse at the same time.
 
@GrisPato - You probably refer to the 10R resistors that were the subject in this post -> https://audiokarma.org/forums/index.php?threads/sansui-jpd-adventure.963332/post-15239802, and then on another following it.
If that's the case, then I don't know the answer - for now to me it's not clear why those 10R resistors were there.
If I missed the point, then please post a photo where those resistors are located.


You probably refer to the fact (which I mentioned previously) that in the first α907 series the flat-amp is balanced, contrary to "normal" topology which was used previously and later. I don't know why Sansui have chosen to do this and then dropped it later. And I won't speculate just for the sake of talking.
But you can find mode details about that configuration (or the idea itself), provided by more knowledgeable people than me, like here: https://www.hairballaudio.com/blog/resources/diy-resources/balanced-and-differential
Hope that answers your question.
Screenshot_20241023-135119.png

There they are. Just noticed it, this is something Sansui also used on the C-2301. If you have no idea what the purpose is then I for sure do not know.
The same for the differential pre amp stages. I do think they did it to have something similar to the output stage but it is clearly not, accept, what I read in the article you linked to, maybe a little higher s/n ratio.
Thanks anyway for taking the time to answer. I just love to study schematics and detect differences and wonder what the purpose is.
One thing I find really interesting for example is the relatively small filter capacitance Sansui uses in the alpha models. In the patent that Sansui issued they talk about that the 2 capacitors in a normal power supply ( +-+- ) are effectively in series. With my limited knowledge I can not really see that to be true. Fact is that Sansui designed an output stage power supply with just
21000 uF, while in the AU-X1/11 they use 80.000 uF for the same power output.
 
That job would be a blessing and a curse at the same time.
Well, I've been there :biggrin: Not as my job, but as my personal hobby and altogether quest/goal to see if I can improve a Sansui Alpha topology beyond the level of my modded AU-D77X.
I'm not done yet, but so far I have all the reasons to believe that there are good chances to have achieved that.
But it was quite a journey to reach that point. All the "unknown"/not public schematics that I posted on AK were part of that jorney, as most of it was the understanding part. And when I thought "well, I'm almost there" it proved to be nothing close in reality....so back to the drawing board and have another go later.
Yeah, DIY is far from easy if you're into R&D universe. Having some boards from a group-buy and soldering the components is easy-peasy and a totally different story.

why mess with perfection
Now on a serious note and leaving joking aside - I really don't want to see or hear about broken amps (or another device) around here due to people trying some modding/tweaking.
So I will say this, to be 100% clear: if anyone is ABSOLUTELY confident that he/she knows WHAT is doing and WHY is doing that, then I say "go for it". Otherwise leave it alone. Who knows, maybe some day will come its time. Or ask somebody with proven expertise to do that. But the caveat here is that - if you don't like the result - you will have to ask for services again, in order to have it reverted to the original state.
 
There they are. Just noticed it, this is something Sansui also used on the C-2301.
I just love to study schematics and detect differences and wonder what the purpose is.
Then my suggestion would be to have a look at Onkyo preamp P-3030. Or Onkyo patent US4555751, where the idea is explained.
I think the purpose is similar or maybe "smoothing" the charging pulses, to lower the content of resulted HF from diodes switching.
But I may be barking the wrong tree on this.

Edit: I remembered that we discussed this before, patent pdf in question is also attached to that post:
 
Last edited:
Back
Top Bottom