Sansui JPD adventure

[GrisPato]

In light of my JDP adventure and because in the end I bought a few more amplifiers then that I planned I bought some stepdowntransformers in Japan, Yahoo auction, Kashimura.





One 1000VA for the AU-D707 and one 1500VA for the AU-D907 Limited / AU-D907 G Extra.
Still have one 2000VA and a Nissho 1100VA EL Stepdown for the bench on the way.
This 1000VA is brand new, not used. I know nothing important but always nice.

 

[GrisPato]

@fojica
Something interesting coming soon.
My AU-D907G Extra has arrived, at least to the country.
Sadly our postal services went private some years ago and these days they are as unreliable as the weather here in the Netherlands.
The stepdowntransformers are from the same shipment but have not heard anything about the amplifier yet.
From previous experiences it can take up to two weeks before they know where they are and get them to my house. ( Funny details: they come by ship in 7 weeks from Japan but can easily take two weeks for the last 100 km.)
I am really curious how Sansui implemented there ground free circuit!
Hope you can chime in to help.
Just a little patience.

 

[Hyperion]

Have a look here:-
https://audio-heritage.jp/SANSUI/amp/au-d907gextra.html

 

[GrisPato]

Have a look here:-
https://audio-heritage.jp/SANSUI/amp/au-d907gextra.html

thanks John.
It's a beautiful amplifier isn't it!
I was revering to what was discussed in the thread I started about the Diamond Differential circuit.

Think I found something....this is a real AU-D907G And, yes, follow the red arrows, looks like 4 resistors there - at least 2 of them have 99% sure 10ohm...and those are connected to the midpoint (floating ground) of the main caps if I'm not mistaken
View attachment 2286395
Source: here

P.S. Looks like you are right, John - neither these ones are on a PCB

So now i am going to be able (soon I hope) to check the values of the resistors and see if there are more changes visable.

 

[fojica]

@GrisPato - with some good photos I'm pretty confident we can figure it out
For now I think you refer here to those mR99 10R resistors that are mentioned in that "very brief" service manual (or better said just a supplement) for AU-D907G Extra. It's possible to be like B2102, as @skippy124 mentioned in another thread, but let's see when you'll have it in front of your eyes

Regards

 

[Hyperion]

The 'G Extra' is a Super-Feed Forward design, with MOSFET's in the feed forward output correction circuit. (so it's actually an AU-D11 on steroids).

In other respects I think the B-2102 is closer in design to the AU-D11 II. (which isn't an SFF design), or in JD speak the AU-D907X , but still with some significant differences.

 

[GrisPato]

May be some confusion with the AU-G99X , which is the integrated most similar, in electronic and visual design as the C/B2101/2.
Clearly "cheaped" down versions, not so much in power, but build to a certain pricepoint needed for export.
AU-Dx07G Extra are full DD SFF designs with the first version of groundfree, floating ground, as @fojica was so kind to explain me the difference between the two as good as possibly I can understand.
What I can find is that the AU-Dx07 F series already had the upgrade to MOSFETs in the error correction part (FeedForword) in the Extra serie
So not extra power, extra speed.
On Ishinolab is the story told about the development of the G series. Here the boss of the department where Mr. Ishino works, ( it never becomes clear to me what exactly is his position in the company) asked for a AU-D707F to be brought home to him so he can perform some modifications he has been playing with in his head.
Later he invites his staff to come look and listen to what he did.
He cut the center tab of the transformer, added some resistors between places and they found it sounded better and went ahead to engineer the G series.
PS
I remember now, there was mentioned that indeed there were some similarities between the G's ground floating and the X's ground free, that had to do with resistors between ground and the circuit at places.

 

[fojica]

Yep, I know it's SFF topology.
My assumption is that those resistors somehow in the ground path could be related with the DF (dumping factor)
But anyway...let's see how it unfolds

 

[GrisPato]

Yep, I know it's SFF topology.
My assumption is that those resistors somehow in the ground path could be related with the DF (dumping factor)
But anyway...let's see how it unfolds

Yes I realized that after I wrote the post. For that the PS

 

[fojica]

@Hyperion - John, I mentioned B-2102 here not because it has the same topology with AU-D907X, but because of the presence of those resistors that are interposed in the ground path. However, they are there probably for a different reason than in case of AU-D907X, because the topology are different between them.



Will see how it goes

 

[Hyperion]

because of the presence of those resistors that are interposed in the ground path.

Ah, yes - those

On the B-2102 those ground points 11/1, 13/1 & 12/1, 14/1 go to the cold side of the Protection Board Zobel Networks and also to the same spots for the Driver Board Zobel Networks. So those 10Ω resistors (ringed in red), elevate that virtual ground point just above chassis ground, why? - I don't know.

 

[GrisPato]

I can't wait.
Just got a response from the Dutch Royal Post service on there IM ( instant messaging ) customer service after 48 hours on my request for information on the whereabouts of my package.
I am right, they have now idea where my package is, they hope to find it soon and will notify me when / if they do.

 

[fojica]

On the B-2102 those ground points 11/1, 13/1 & 12/1, 14/1 go to the cold side of the Protection Board Zobel Networks and also to the same spots for the Driver Board Zobel Networks. So those 10Ω resistors (ringed in red), elevate that virtual ground point just above chassis ground, why? - I don't know.

John - it's like below drawing, tried to do a high-level one to be easier to understand. However, that "virtual GND" is not used for any high-current path, nor for any biasing as far as I can figure it out. So why those resistors are there, what and why are they trying to "isolate" from GND.....no clue




And sorry to @GrisPato for the off-topic . Bring the AU-D907G
Regards

 

[Hyperion]

John - it's like below drawing, tried to do a high-level one to be easier to understand. However, that "virtual GND" is not used for any high-current path, nor for any biasing as far as I can figure it out. So why those resistors are there, what and why are they trying to "isolate" from GND.....no clue

Thank you for this - much appreciated, one day we'll find out what those resistors are for.

 

[GrisPato]

John - it's like below drawing, tried to do a high-level one to be easier to understand. However, that "virtual GND" is not used for any high-current path, nor for any biasing as far as I can figure it out. So why those resistors are there, what and why are they trying to "isolate" from GND.....no clue

View attachment 2413901


And sorry to @GrisPato for the off-topic . Bring the AU-D907G
Regards

No excuse needed. The contrary, I always value your participation. Not many people interested in this subjects.
About the 10ohm resistors. In my naivety I thought they were use to give some reference for the center voltage for the groundfree supply trough this connection with the other powersupply that has a centertap. Especially because these come together at the driverboard. ( The two powersupplys )

 

[fojica]

You're welcome guys, glad to be of some help around here
I agree, I'm more on the technical side of the things - but I think this is the challenging part in electronics after all
Normally, those mR3 10R resistors would not be there ( I mean it's simply a direct connection to chassis or global GND) - so it looks like they are trying to isolate that virtual ground point from the global GND.But in the vast majority of other Alpha's / X-Balanced amps they are not there, this is the reason why myself and John were debated them.
Putting some 10R in the ground path (on input side) was done sometimes to break the possible ground loops formed by input cables, but in our case it just doesn't have any sense from this point of view. I tried to imagine various logical scenarios to justify their presence - nothing obvious or just something to make any sense That's why I called them "mysterious resistors" in that drawing

And, in the end, some eye-candy my AU-D77X modded - now with dual-mono design, cascoded 2SK389 and MOSFET outputs on power modules - sort of a AU-D77X MOS Limited

 

[GrisPato]

And, in the end, some eye-candy my AU-D77X modded - now with dual-mono design, cascoded 2SK389 and MOSFET outputs on power modules - sort of a AU-D77X MOS Limited

Really nice to see this. Do I see correctly that you made it with a dual bridge rectifier?
Did you have to modify a lot to get the MOSFETs working, driver transistors, bias change?

 

[fojica]

Yep, there are 2 separate bridges for power stage, one for each channel and then continuing with the separate filtration per channel to make it like a dual-mono. Both bridges are connected to the same secondary - being that the trafo has only one wiring
This helps for a better separation, since the most of crosstalk happens at higher currents - at least in theory.
PSU for input & drive stage was also changed to a better voltage regulator instead of default basic R + Zener.

To exchange output BJT's to MOSFET's you have to modify:
1 - Vbe multiplier or bias spreader circuit, to make for higher voltage on it due to higher Vgs for MOSFET's compared to Vbe of BJT's. Basically there's a single resistor to be increased, you can see this one on any MOS amp
2 - modify the resistor between drivers emitter's (or outputs gates / bases), depends how much current you want to give them. From my perspective, Sansui are "cooking" the drivers in MOS amps a bit - which I don't like. So I increased that resistor a bit
3 - change the resistors from BJT's bases because they are too small (4.7-10R), you need larger values for MOSFET's there (27+47R)
4 - obviously replace BJT's with the MOSFET's
And kind of that's it, nothing outstanding if you ask me
There is no need to change the pre-drivers or the drivers - also Sansui themselves used the same ones between "normal" amps (with BJT) and MOS versions. Didn't tried yet, but the only (I think minor anyway) gain would be to change the drivers for a higher Ft ones, like 2SC5171 and 2SA1930. But I don't expect "miracles" due to this - and it's also possible to need adjusting the compensation caps due to higher Ft, so this should be done with care and with proper equipment to check afterwards.
Hope I answered your questions

 

[fojica]

A word of caution, however, regarding putting 2SC5171 / 2SA1930 as drivers: due to their higher Ft one may gain a bit of speed, but their SOA is lower compared with 2SC4793/2SA1937. In other words, for the same supply voltage you can't push 5171/1930 as much as 4793/1937. Win some, loose some So, depending on the voltage on the rails, it may be not the wisest move in the end. This (SOA) is something that should be accounted for, in case you plan doing so.
You have been warned....as the word is

 

[GrisPato]

That's why I started this. To do something like that. Although not adjusting an amplifier but to build one for myself.
I have an AU-D77X. Could be an option to try something like you did. Later on when I understand more of it myself.
Will take some years, have to learn a lot still, I am not in a hurry.
Thanks for sharing. Would love to see a thread about it, maybe in DIY or solid state, although it's lik a selfmade allpha 607 Mo's limited, it could fit in exclusively Sansui.