AU888 restore discussion

[gslikker]

Some testing was done. The lack of power issue was less severe than initially thought. Continuous power appears to be over 39 watts (17.7 VAC) and assumed-just-below 3% worst channel is 43 wats both channels sine wave drive 1kHz. This was after the modified board above (which may have added one watt or so....).

It was found, the distortion meter is UNRELIABLE since it gives different figures left and right. Left gives highest and fluctuating. At higher levels of distortion the difference is much less. Later, it was found swapping the loudspeaker cables made no difference.
also later, I measured the Distortion meter internal loudspeaker shunts from its loudspeaker inputs. They appeared 7.7 ohms instead of 8. This modifies the power measurements a bit, power will be slightly higher than measured (see it as margin....)

Above 1030 millivolts input the distortion suddenly rises ate both channels.
Meter watt output (200W scale), just low of 40 Watts, 17.67 Volts AC, LH channel THD says 0.13% on 0.3% scale.

Meter output Right channel (lowest value 0.06%) at 1030 mV input

Meter output Right channel at 1040 mV input, observe the range change LED from 0.1% to 1% full scale

Figures for Left channel are two to three times as high and fluctuating!, so above is not an indication of the true thing, although the left channel measurement seems to vary a lot (swapping meter inputs problem stays with meter)

Picture of a newly populated amp board, which was swapped over to other side as left/right difference was thought to be the amp at first....

Picture of the (negative) main smoothing capacitor (22000 µF) ripple at 39 Watts both channels. It is some 800 mV tt. 1 kHz is "dancing" on it and it looks a bit like a triangular wave.
Note, at this moment capacitor DC voltage is about 32 Volts which is a loss of about 3 Volts (some 8%) due to the fact the transformer apparently can not keep up (it is not that big.....)

Picture of the (positive) main smoothing capacitor ripple at no input. It is some 60 mV tt


To be continued....

 

[stereofun]

Very interesting write up. Does the 888 and 999 use the same power transformer.

 

[AnalogueBen]

Great to see you going in deep with this build, although much of it is going way way over my head...keep up the good work, looking forward to seeing more.

 

[smurfer77]

I have two AU-888 and both (barely) make spec of 45W at 0.4%THD (as measured by a pair of HP8903A), in stock form.

The AU-999 is one the one that doesn't make spec, coming in at around 45W or so, instead of 50W in stock form (and this is simply why the AU-888 is rated at 45W, instead of 50W; it doesn't really have less power than the AU-999 - the specification on the AU-999 was an error).

I would make sure everything on the AU-888 is up to spec before modding get more output, unless modding is the goal.

 

[kevzep]

Only way to get that amp to produce more power is to increase the rail voltage and current capability..

 

[gslikker]

Only way to get that amp to produce more power is to increase the rail voltage and current capability..

Yes that would imply a comparable form fit bigger transformer. Or putting an extra small toroid somewhere, maybe asking for trouble....

Very interesting write up. Does the 888 and 999 use the same power transformer.

No, 888 has different windings and lower voltage for the supply for the current output stages (54 VAC versus 60 VAC)

 

[gslikker]

I have two AU-888 and both (barely) make spec of 45W at 0.4%THD (as measured by a pair of HP8903A), in stock form.

The AU-999 is one the one that doesn't make spec, coming in at around 45W or so, instead of 50W in stock form (and this is simply why the AU-888 is rated at 45W, instead of 50W; it doesn't really have less power than the AU-999 - the specification on the AU-999 was an error).

I would make sure everything on the AU-888 is up to spec before modding get more output, unless modding is the goal.

Modding WAS not the goal, but it starts looking like it, going along.

 

[gslikker]

The used transistors and small mods in the main amp.
The current drivers and output transistors have not been changed.
The pre-drivers I wanted to use KSP42 and KSP92, but as HFE of KSP92 was twice that of KSP42, I decided put in one more in line (KSC1009).

 

[gslikker]

Deviating thoughts....
Do I want a relay DC speaker protection?
How to do it?
I have an idea not taking another 100 milliamps from the main transformer windings but use the preamp ones instead. As those are not ground-referenced, this MAY give some problems due to the weird way they designed preamp power. The relay circuit must be built having no ground reference and it also must not deteriorate the preamp power supply.

As a start I need a detection circuit, which will not compromise this, so only logical way is use opto coupling.

I modified the F1334 protection board to accept main amp power supply, changed some resistors originally suited for SCR trigger using arbitrary guessed values, and mount 4N24 optocouplers instead of the thyristors.
Although not ideal, I had the optocouplers already and they can be mounted, while optocoupler through hole IC's would be more of a hassle for wiring up.

board is ready for test ======> edit:

The test did not work, I should have checked.....also here the design is questionable, since the sensitivity is over 9 Volts DC.

I want lower.

The circuit needs some 1.3 Volts on the transistor base. Now, if ONE channel is out of business and it has a high DC, it goes through 10 kilo-ohms. However, this is divided by half due to the other 10k to ground, and divided further by the series diode and 3k9 to ground.
So nothing much is left.
This is an unsuitable arrangement, since if you would have 9 Volts DC for any reason (broken null trimmer for example) nothing happens and a low watt loudspeaker could fry its coil.....

So, the emitter resistor are not to provide 0.6 Volts anymore and replaced by 120 ohms. (guess I could even put a wire in the holes instead.)
The 3.9 k resistors became 39K.
The 22nF was taken out and it is 4.7µF metal can tantalum now. (it is an arbitrary value anyway, it could be left out but let's just fill the holes with parts I have excess NOS....)
The opto couplers were moved into the collector line of the transistors.

Now, the circuit triggers at 2.5 Volts DC. How it reacts for AC, will be checked later in the amp.

 

[gslikker]

All the circuits were put back in the amp. So far, it at least works on the aux input.

A real disappointment is the sensitivity. It is much too sensitive, using a compact disc player.
So sensitive, even using the muting switch it does not give a good volume pot swing.

I think, as nothing precedes the volume control but the already modified tone board, this still needs a closer look.



The F1334 board gets back its original position, whether the safety relay project will be successful soon, or not.




The relay will be getting here close to the loudspeaker fuses, if everything gets working.
It is scavenged from some old aircraft testing box, it is called "balanced force" relay, when new it was probably over 100 dollars/euros.....It is made by a German company called LRE, now owned by Leach. (google leach balanced force relay M300 for specs)
I had some small aluminium bar having holes already, which I used for mounting the relay.





The relay circuit prototype. It has to be adapted to use the floating preamp 30V AC windings, the goal is take as little amount of power as possible, adapted for the relay.
I have some power saving idea, by preregulating the smoothing capacitor charging and deplete (not charge) during the AC sine wave peaks, I still have to find out if it works well with the transformer and if it has no effect on the preamp supply.
So, instead of a few components and a relay, this is 35 components already, excluding the F1334 board....

Of course, it will be smaller than this one.....

 

[kevzep]

You didn't do the AU999 pre-amp mod did you? Its not necessary on the AU888 as it seems Sansui got the 888 pre-amp right, as far as I know nobody has tried it because when we had a look at the schematics you can see the designs are different...

 

[gslikker]

Hi Kevzep, thanks for chiming in!

I did your mod indeed (board was same as au999) and I will look into it, again, maybe I did something wrong!
(the stage with the 1M resistor right after all those tone filter stuff, according your mod.).

There are some things going on with these schematics as there are some deviations.

I will try to understand the circuit, also the (extent of the) effect of the emitter capacitor c723 which enhances amplification. Maybe take it out or use two series resistors and attach in the middle, but I need to understand calculations for that....
If I fail to understand, I will try to measure.

The second stage on this board already had a different feedback resistor versus the AU999, so your AU999 was probably even more sensitive?

Anyway, the thing being over sensitive should not be a real surprise, just reading it to be 180 mV for full output, the CD player probably provides way too much of a voltage, also....

(going to sleep now, it is 1:00 here)

regards, Gerard

 

[gslikker]

I tried some other resistors in the TR705 area on the tone F1209 board, after consulting several other one-transistor tone control circuits.
R761: 1.2 k
R757: 180k
R755: 820k
R759, 4.7 k
I leave this to be measured later, assuming it works OK, as it is more in line with resistors normally relating to each other in this kind of circuit

For the sensitivity, having R769 at 2k, I lowered R775 to 4,7k (from 8.2 k). This circuit provides half of the previous amplification and now volume control works quite nice.

Next will be tinkering the relay circuit again....although not much progress to be expected (7 nightshifts ahead)

 

[smurfer77]

I did your mod indeed (board was same as au999) and I will look into it, again, maybe I did something wrong!

The second stage on this board already had a different feedback resistor versus the AU999, so your AU999 was probably even more sensitive?

The AU-888 fixes most of the issues of the AU-999. Think of the AU-888 as the AU-999-II (as Stereofun I think first commented). Most of the (correctional) mods for the AU-999 are logically aimed at making it more like the AU-888, rather than vice versa.

Regarding the feedback, yes, they are different. The AU-888 uses more negative feedback than the AU-999, meaning that it has slightly less gain, not more gain, than the AU-999, at least in this section. Also, note that the negative feedback is determined not by one resistor value, but by the ratio of two resistors forming a voltage divider. see this post here for more info.

 

[gslikker]

Yes, I have seen all those nice topics
As I am satisfied with the volume range now, I leave any further volume tweaking.

 

[gslikker]

I made the relay circuit on a hobby board, schematic to follow later when time avail. Size is small enough to put on a piece of empty space after using the saw.
Some mounting and wiring tidying to be done.

 

[gslikker]

The schematic. Note time cap is 100 µF which I forgot to write in. this gets something like 7 seconds or so delay. Note the input is not smoothed. Smoothing is a big 2200 microfarad cap after the dissipating BDT95 transistor. Original idea to cut cap charging into two pieces (no charging when input voltage above 28) did not really work out because of too many variables. It would have saved half a watt of dissipation, now it is much less. Would half a watt be important? Yes, if paranoid and knowing the transformer is not really up to the task it DOES take away over 1% or so of this amps power capability.
The KSA1220 with the two 1n4148 is a current source circuit not just a switch, it supplies some 75 milliamps to the relay.
The optocouplers coming from the crowbar circuit card are in parallel. They each can take away base current from the BC547B and deplete the 100 uF cap if active. Except for the 10 ohms resistor for the current source, which was choosen to have a reasonable measured voltage on the relay but below official voltage, all component values are very "arbitrary" ( read:rough guesswork and not validated by calculations).

A major property is the fact this circuit can not be measured with respect to grounding.
The transformer windings are not ground referenced and DC voltage against ground is compromised by preamp power supply operation. When I am done I just put it in and hope for the best (which would be no impact on the preamp power supply)

 

[gslikker]

Next will be the power indicator, as the lamp rubber was burnt, it is to become a LED indicator.
It has a rectifier bridge and a potentiometer mounted on the lamp holder itself, although this does not look that neat it is good enough for now.

 

[gslikker]

Some more pictures....

A cup of tea




Adjustable LED power indicator, it is white enough to lit the yellow plastic piece looking yellow




How the main amps look like, now




The realy steering circuit mount




The Sansui Secret Sauce capacitor, which is to stay.




And a death cap, prior to reflow, I used a scavenged piece which actually is a snubber network, hopefully it will cure possible sparks over the mains power microswitch.

 

[Hipocrates]

The realy steering circuit mount

That looka pretty good, very nice!