Leestereo's Restoration and Upgrade of a Sansui AU-777A

Leestereo

Super Member
A little while ago (last summer), a very nice AU-777A was on the workbench for a complete restoration and upgrade.

Part 1: HEAD PRE-AMP BLOCK (F-1194)

The F-1194 board includes the phono stage (600-series components) and the line level stage (700-series components).

index.php


A. Phono Stage
The stock capacitors (C601/C602) in the high-pass filter for the phono input were 1.5µF polarized electrolytic types; these were upgraded to 1.0µF polyester film types. The emitter bypass capacitors at C605/C606 were originally 33µF/10V and were replaced with 47µF Nichicon KZ types; the increased capacity of the replacements extends their effectiveness to 20Hz. Similarly, the 10µF emitter bypass capacitors at C613/C614 were replaced with 100µF Nichicon FG types. The C607/C608 capacitors in the feedback high-pass filter were 33µF polarized types and these were replaced with 100µF Nichicon ES bipolar types. The stock phono stage output capacitors (C611/C612) were originally 10µF polarized electrolytic types and were replaced with 22µF Nichicon ES bipolar types. The failure-prone 2SC458 transistors (TR601-TR604) were replaced with KSC945C(G) transistors (hFE matched pairs).

The capacitors for the RIAA correction, C615-C618, were originally 10% polyester film types and these were upgraded to 5% polypropylene types. The RIAA resistors at R625/626 were originally 22kohm 10% carbon film types and their replacements were 22.6kohm 1% metal film types. Similarly, the RIAA resistors at R627/628 were originally 270kohm 10% carbon film types and the replacements were 1% 267kohm metal film types. Note that changes in the resistor values are deliberate and improve the RIAA accuracy to -0.1dB/+0.2dB. Furthermore, although the stock component values can theoretically yield excellent RIAA accuracy, given the 10% tolerance of the stock components, that is not very likely.

index.php


B. Line Stage
The input high-pass filter capacitors (C721/C722), originally 0.22µF polyester film types, were upgraded to 0.47µF Panasonic polypropylene film types. All other stock polyester signal path capacitors (C731-C738) were upgraded to polypropylene film types of the same capacity. The coupling capacitors at C727/C728 were originally 10µF polarized electrolytic types and these were replaced with 22µF Nichicon ES bipolar types. The 1µF signal path capacitors at C729/C730 and C739/C740 were upgraded to stacked film types. The 47µF emitter bypass capacitors at C725/C726 were replaced with 100µF Nichicon FG polarized types; the increased capacity extends their effectiveness to below 20Hz. The failure-prone 2SC458 transistors (TR705-TR710) were replaced with KSC945C(G) transistors (hFE matched pairs). The local decoupling/filtering capacitor at C743 was originally 470µF/25V and was replaced with a 1200µF/35V low ESR type.

index.php


index.php


index.php
 
Lots of people interested in this model (and its variants), my goodness, that finished board looks gorgeous! :thumbsup:

I see C741 & C742 got changed to WIMA films. ;) :thumbsup: (can't see the value though - guessing 3.3µF as original).
 
...I see C741 & C742 got changed to WIMA films. ;) :thumbsup: (can't see the value though - guessing 3.3µF as original).

Good eye, @Hyperion; yes the replacements for C741/C742 are indeed 3.3µF. The F-1194 board is so densely populated that I sometimes fail to see everything that I replaced when I do these write-ups.
 
Thanks for going the extra mile and documenting this and other restorations so thoroughly. Had you just shown us your amazing work we would be thrilled enough, but now with detailed write-ups and thorough documentation many of us are downright grateful too.

P.s Hyperion sees everything :)
 
Part 2: DRIVER AMP. BLOCK (F-1183-1)

index.php


index.php


The stock capacitors (C801/C802) in the input high-pass filter were 10µF polarized electrolytic types; these were upgraded to 4.7µF WIMA polyester stacked-film types. The 10µF polarized electroltyic capacitors at C805/C806 were upgraded to 22µF Nichicon ES bipolar types. The 3.3µF electrolytic coupling capacitors at C807/C808 were upgraded to same value polyester film types. The stock local decoupling/filtering capacitor at C809 was replaced with a low ESR type (capacity was increased from 47µF to 120µF). The 1000µF/6.3V emitter by-pass capacitors at C811/C812 and C815/C816 were replaced with 1000µF/35V Nichicon FG types. The inter-stage coupling capacitors at C821/C822 were originally 100µF polarized types and these were replaced with same value Nichicon ES bipolar types.

The four 100ohm/3W cement resistors at R819-R822 were replaced with 100ohm/5W low TCR wirewound types. The smaller physical size of the replacement resistors results in greater distance from the surrounding capacitors and the off-board mounting affords better cooling; the detrimental effect of the stock resistors' close proximity to C811/C812 and C815/C816 can be seen in the shrinking of their plastic sleeves.

The stock input transistors at TR801/TR802 were the failure-prone 2SC458LG transistors and these were replaced with KSC945C(G) transistors (hFE matched pair). The open-frame trimmers at VR801/Vr802 were replaced with cermet multi-turn types.

index.php
 
Noticed you are using ksc945 for 2sc458 replacement. Haven't noticed those before, but seems like a better package than my standard 458 replacement, which is ksc1845. The ksc945 has much higher ft and can handle 150ma Ic so a bit more versatile.

Leestereo, are those 945's what you recommend as a go-to for npn signals, just like 1845's and before that 2240's ?
 
The KSC945C were used as replacements for this project since their specs. are a better match to the original 2SC458LG than the KSC1845:
A. KSC945 fT (300MHz) is comparable to that of the 2SC458 (230MHz), whereas the KSC145 fT at 110MHz is quite a bit lower (although that is more than adequate for AF use)
B. KSC945 Ic (150mA) is higher than 2SC458 (100mA), whereas KSC1845 Ic is again lower at 50mA (although not likely to be an issue for signal amplification).
C. KSC945 Noise Figure is 4dB which matches the 2SC458 NF; the KSC1845 is described as a low noise device, but its difficult to compare directly since its spec'd as 25mV (which I think is lower than 4dB)

The lower noise level of the KSC1845s might be advantageous in a phono stage where the gain is very high, but for this 777A project, the decision was to use the closest match to the original 2SC458LG. If the original transistors had been 2SC1312/2SC1313, I probably would have used KSC1845 replacements as was done for the late production 555A in this thread: Repair and Upgrade of a Previously Recapped Sansui AU-555A
 
Should also mention that if the original 2SC458LG is specified as a "B" hFE ranking, it can be replaced with a KSC945C with a "Y" hFE rank; the 2SC458LG with a "C" ranking should be replaced with a KSC945 with a "G" rank.
 
Thanks for the shoutout Lee. Lovely work as always! Truly outstanding effort on the parts labeling too.

Is there a reason you downgrade the capacitance of the input caps on both the phono and driver when you changed to film, or it was just a matter of physical space? Polypro is big as you know but I found 5-10uF polyester WIMA caps will fit if you buy the right ones. So on my AU-777A I managed to upgrade those values (except the 10uF where I kept same value), and go film, rather than electrolytic where you did in several places. I found film (a) it sounds better with same value C film than bigger electrolytic (and low f cut-off is already low enough except for those input coupling caps (which is why I raise the issue of keeping or increasing the value)) and (b) the film caps will hopefully never need replacing again.

Stereofun, hope all is good! I have been ranting that the 1845 is not an appropriate sub for the 458. 50mA vs 100mA is too much unless the role/circuit is well understood by the tech. KSC1815 is the way to go, even if the speed is slower (80MHz) (it's not relevant in these older designs as long as it's fast enough); you will notice the lower noise from the 1815/1845 and both phono and tone boards, compared to the slightly bigger small signal transistors including the ZTX (haven't tried the 945). I did quite a few tests with this... and even made audio recordings.... and this is what led me to start going on about transistors making a bigger difference than recap to noisefloor etc.

Keep up the good work Lee!!!
 
Last edited:
Thanks guys for chiming in - I am almost done with a Sansui 800 resto and thankfully bought a batch of 1815's to replace 281's - There is still time to undo my 1845 subs, its only two in the tone board, but unfortunately four in the Equalizer. (Here the SM lists 2 x 458LG's and 2 x 281's - but actually only had 458LG's on the board....guess that is not uncommon during production runs)

Ok no more interference in this exiting thread, cant wait to see part 3 and on....
 
Thanks guys for chiming in - I am almost done with a Sansui 800 resto and thankfully bought a batch of 1815's to replace 281's - There is still time to undo my 1845 subs, its only two in the tone board, but unfortunately four in the Equalizer. (Here the SM lists 2 x 458LG's and 2 x 281's - but actually only had 458LG's on the board....guess that is not uncommon during production runs)

Ok no more interference in this exiting thread, cant wait to see part 3 and on....
i bet the 1845 will be 'ok'...people have been using them like that without problems for ages. just trying to keep 'best practices' alive :p
 
Thanks guys for chiming in - I am almost done with a Sansui 800 resto and thankfully bought a batch of 1815's to replace 281's - There is still time to undo my 1845 subs, its only two in the tone board, but unfortunately four in the Equalizer. (Here the SM lists 2 x 458LG's and 2 x 281's - but actually only had 458LG's on the board....guess that is not uncommon during production runs)...

Since all of the KSC1845s you mentioned are for AF amplification in low level circuits, where low noise is important and high current is not needed, I suggest that you keep them in place. Further, if the KSC1815 are "Y" rank, the gain is somewhat low for your application, the higher KSC1845 gain is preferred.
 
Since all of the KSC1845s you mentioned are for AF amplification in low level circuits, where low noise is important and high current is not needed, I suggest that you keep them in place. Further, if the KSC1815 are "Y" rank, the gain is somewhat low for your application, the higher KSC1845 gain is preferred.

Sadly, these days it looks like Y rank is about all we can buy of the 1815. That said, the transistors in application are never used close to their open loop gain, and each stage only asks for a gain in the range of 1-10, in Stereofun's application, if I'm not mistake. Y rank gain of 120-230 I think should be fine.

And, the 1815 is, like the 1845, a low noise transistor. (Spec sheet says it only adds 1dB of noise relative to the minuscule thermal noise at the input of the transistor. It's a bit hard to compare spec sheet of 1845 because they used a different noise figure approach but my personal experiments indicate it's in the same league as 1845 in this regard). So I still believe 1815 is the preferred sub if original part was >50mA.

I expect will not agree on this, so I say, enough with this nonsense... .bring on the next part of the AU-777A!!! :)
 
Last edited:
@smurfer77, @stereofun: All is good, the transistor substitution discussion is worthwhile. I do agree that the transistors are not used anywhere near their open loop gain, but for reasons not completely clear to me, Sansui often specifies different gain rankings for the same transistor when used in different positions, e.g., in the 2-transistor phono stages of the 555A and 777A, the second transistor is spec'd for higher hFE.

BTW, will try to post the next instalment of my AU-777A saga tonight.
 
Is there a reason you downgrade the capacitance of the input caps on both the phono and driver when you changed to film, or it was just a matter of physical space?

Yes, since film capacitors are invariably larger than their electrolytic capacitor equivalents, available board space for a replacement film capacitor is often a consideration in their selection. Fortunately, it may not always be necessary to replace the original electrolytic capacitor with a same capacity film capacitor. For input RC high-pass filters, a good rule of thumb is to use a capacitor that gives a F3 that is 2 octaves below the lowest frequency of interest, i.e., typically, the minimum acceptable F3 is 5Hz. Further, if the capacitor is an electrolytic type, to minimize LF distortion (see http://www.diyaudio.com/forums/parts/192811-capacitor-measurement.html#post2643470), it is good practice to use a capacitor that is ~10X the size required to achieve the F3=5Hz, e.g., if a 1µF film capacitor is sufficient to give F3=5Hz, an electrolytic capacitor substitute should be 10µF. However, since one cannot assume that the electrolytic capacitor installed in the filter followed this ~10x size recommendation, it is best to calculate the replacement film capacitor size.

In the 777A phono stage, the input impedance is 50khom and to achieve a F3<5Hz, a 0.68µF capacitor would be sufficient; hence the installed 1.0µF WIMA film is good (F3=3.2Hz). The stock 1.5µF capacitor also appears to be sufficient, giving a F3= 2.1Hz, but since it is an electrolytic type, ideally it should have been at least 6.8µF. Also, since the F3 of serial filters are additive, it is good practice to lower the F3 below 5Hz whenever possible to maintain an overall passband that is flat down to 20Hz. An example of a very low F3, is the input high-pass filter for F-1183-1, where the input impedance is 300kohm and the 4.7µF film capacitor gives a F3 of 0.1Hz!
 
Yes, since film capacitors are invariably larger than their electrolytic capacitor equivalents, available board space for a replacement film capacitor is often a consideration in their selection. Fortunately, it may not always be necessary to replace the original electrolytic capacitor with a same capacity film capacitor. For input RC high-pass filters, a good rule of thumb is to use a capacitor that gives a F3 that is 2 octaves below the lowest frequency of interest, i.e., typically, the minimum acceptable F3 is 5Hz. Further, if the capacitor is an electrolytic type, to minimize LF distortion (see http://www.diyaudio.com/forums/parts/192811-capacitor-measurement.html#post2643470), it is good practice to use a capacitor that is ~10X the size required to achieve the F3=5Hz, e.g., if a 1µF film capacitor is sufficient to give F3=5Hz, an electrolytic capacitor substitute should be 10µF. However, since one cannot assume that the electrolytic capacitor installed in the filter followed this ~10x size recommendation, it is best to calculate the replacement film capacitor size.

In the 777A phono stage, the input impedance is 50khom and to achieve a F3<5Hz, a 0.68µF capacitor would be sufficient; hence the installed 1.0µF WIMA film is good (F3=3.2Hz). The stock 1.5µF capacitor also appears to be sufficient, giving a F3= 2.1Hz, but since it is an electrolytic type, ideally it should have been at least 6.8µF. Also, since the F3 of serial filters are additive, it is good practice to lower the F3 below 5Hz whenever possible to maintain an overall passband that is flat down to 20Hz. An example of a very low F3, is the input high-pass filter for F-1183-1, where the input impedance is 300kohm and the 4.7µF film capacitor gives a F3 of 0.1Hz!

I enjoyed reading that a lot, especially about your thought process on C value for film vs electroytic regarding additional care with LF distortion. I had read such things before but I must admit I tended to not worry about that too much and just look at the cut-off frequency. But, i think that's great to take into account....I may think about that more when forced to use electrolytics. I do appreciate the physical space limitations of using film caps :). Some of my AU-777A boards look stupid with caps hanging of the edges due to limited space.

Thanks for taking the time to respond to my question in detail. Much appreciated.
 
Back
Top Bottom