Sansui AU 5900 dead channel - Page 4 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 7th November 2016, 07:12 PM   #31
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Depaj View Post
I used the +23v from the power supply of a sansui au 2900 to feed the preamp of the au 5900 and it gets very quite and sounds perfect so it's definitely the secondary power supply that is faulty. Yet I replaced the 4 transistors, the zener diode and the resistors but the hiss is still there
What about the electrolytics?

I would estimate the AU-2900 +23V supply to be a fair bit less noisy - it's only a cap multiplier that's cleaned up passively on top of that. The AU-5900 reg would have zener noise amplified by a few dB at least.

Where exactly did you inject the signal when you wrote
Quote:
-Than I took the signal from F-2603 and directly injected it into the tone control board (bypassing the preamp section) and the hiss was gone.
R19/20-ish?

That tone amp arguably is not the last word in PSRR. Depending on transistor type, I'm seeing power supply noise on the output reduced by about 10-12 dB only when simulating the preamp. Not great. Distortion performance is OK with BC547C or 2SC4102 models, the 2N5088 fared 20 dB worse for some reason. Anyway, doing something about that would seem to be a good idea.

Ah yes, cleaning up the supply for the two 68k resistors (R03/04) with 1k + 100 to ground in series brings this up to almost 36 dB. Still not great but kinda more like it.

It's interesting that power supply noise is the limiting factor here. I would have expected the highish-impedance surroundings (e.g. 250k volume pot) to be the culprit.

Now let's tackle the regulator... Hmm, they're running a fair bit of current through that zener, that's more than 10 mA plus almost 4 mA from the transistor. So 14-15 mA total. That won't be easy to clean up. You could cheat a bit and add some resistance in series with the zener. Like 33-100 ohms max, at which point a few hundred F in parallel should clean things up quite nicely. Who cares about DC stability anyway. I'm also wondering whether the extra current provided by 820 ohm R43 is even needed any more then, I think they mainly did that to reduce impedance and improve regulation, since TR11 current depends on the input-output voltage differential.

If DC stability turns out to be an issue, you can always reconfigure the whole shebang for another (like 5.6 V) zener, all you have to adjust is the three gain-setting resistors R45+47-R49 and preferably give R45+R47 a little parallel cap like R59 has. Vout ~= (V(D11) + Vbe(TR11)) * (R45+R47+R49)/R49; actually it's a bit higher because of TR11 bias current. Ideally you pick a zener type such that zener and Vbe tempco just cancel at the given current.

The negative regulator just tracks the positive one in this arrangement - the R59/65 midpoint voltage must end up being Vbe(TR15) below ground.
Attached Files
File Type: asc AU5900-tone-amp.asc (3.4 KB, 14 views)

Last edited by sgrossklass; 7th November 2016 at 07:26 PM.
  Reply With Quote
Old 7th November 2016, 08:41 PM   #32
Depaj is offline Depaj  France
diyAudio Member
 
Join Date: Aug 2015
Location: lyon
Quote:
Originally Posted by sgrossklass View Post
What about the electrolytics?
They have all been replaced by nichicon FG or wima mks2/Nichicon ES in signal path.

Quote:
Originally Posted by sgrossklass View Post
Where exactly did you inject the signal when you wrote, R19/20-ish?
It was a while ago but I think it might even have been after the tone control circuit.

Quote:
Originally Posted by sgrossklass View Post
That tone amp arguably is not the last word in PSRR. Depending on transistor type, I'm seeing power supply noise on the output reduced by about 10-12 dB only when simulating the preamp. Not great. Distortion performance is OK with BC547C or 2SC4102 models, the 2N5088 fared 20 dB worse for some reason. Anyway, doing something about that would seem to be a good idea.
After recommendation from some other people on the forum here I replaced all the small signal transistors by some fairchild KSA992/KSC1845, as replacing those in a previously also noisy AU-3900 did the trick.

Quote:
Originally Posted by sgrossklass View Post
Ah yes, cleaning up the supply for the two 68k resistors (R03/04) with 1k + 100 to ground in series brings this up to almost 36 dB. Still not great but kinda more like it.
Just before R03/R04 there is a .022F ceramic cap going to ground, you suggest replacing this by a 100F +1k in series or adding this in parallel with c31/c32 ?

Quote:
Originally Posted by sgrossklass View Post
Now let's tackle the regulator... Hmm, they're running a fair bit of current through that zener, that's more than 10 mA plus almost 4 mA from the transistor. So 14-15 mA total. That won't be easy to clean up. You could cheat a bit and add some resistance in series with the zener. Like 33-100 ohms max, at which point a few hundred F in parallel should clean things up quite nicely. Who cares about DC stability anyway. I'm also wondering whether the extra current provided by 820 ohm R43 is even needed any more then, I think they mainly did that to reduce impedance and improve regulation, since TR11 current depends on the input-output voltage differential.
I'm still learning a lot, how exactly is adding that resistor and cap going to clean things up ? Should I just take R43 out after installing those and see what happens or leave it be as I don't expect it can do any harm ?

I will start with this and see where this leaves us.

I suppose the file you added was the simulation, I can't open it, would jut a simple screen shot help me understand your process ?
  Reply With Quote
Old 8th November 2016, 01:33 AM   #33
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Depaj View Post
It was a while ago but I think it might even have been after the tone control circuit.
Which is kind of apples to oranges since you've taken out like 14-16 dB of gain as well.
Quote:
Originally Posted by Depaj View Post
Just before R03/R04 there is a .022F ceramic cap going to ground, you suggest replacing this by a 100F +1k in series or adding this in parallel with c31/c32 ?
The problem is, it still is the supply for the entire tone amp at this point, which would be drawing about 280 A + 3.2 mA + 2.2 mA for TR01, 03 and 05, respectively. That's about 6 mA total, a lot more than TR01 alone. If you put in a 1k there, you'll lose 6 V out of your 22, which is a bit much.

By contrast, in series with R03, you'll only lose an insignificant and easily afforded 0.3 V.

It may still not be a bad idea to clean up the entire supply, but I'd say about 150R series max.

I don't like the tone amp much either, but if in doubt you can always bypass that by activating TONE DEFEAT to see how much difference it makes.
Quote:
Originally Posted by Depaj View Post
I'm still learning a lot, how exactly is adding that resistor and cap going to clean things up ?
Zeners are quite noisy (not as noisy as they used to be, but still), and the noise is amplified like the zener's DC voltage by the regulator. If you're just slapping in an electrolytic in parallel, however, you get an RC filter where the R is the dynamic impedance of the zener. The higher the current, the lower that is. I haven't looked up the datasheet, but I'd guess low-mid double-digit ohms, maybe even single-digit territory. So you'd have to use a really fat electrolytic for it to be effective, like a 1000 F.

As you can see, filtering gets much easier with some extra series resistance and reduced current. So we add some resistance in series with the zener and then add the parallel capacitor, which should be less unwieldy at this point.
Quote:
Originally Posted by Depaj View Post
I suppose the file you added was the simulation, I can't open it, would jut a simple screen shot help me understand your process ?
Oh, sorry, that's an LTspice schematic. Just the preamp section.
Attached Images
File Type: png au5900-tone-pre.png (28.5 KB, 58 views)

Last edited by sgrossklass; 8th November 2016 at 01:42 AM.
  Reply With Quote
Old 8th November 2016, 02:09 AM   #34
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Now with tone amp added, plus two outputs (out = tone defeat / out-2 = no defeat). Tone controls tweaked for linear frequency response (set AC component of V1 to "0." and that of V3 to "1." to plot FR). Needed a bit more bass and a hair more mids.

PSRR on tone amp is somewhat better than for the unmodified preamp section but still not exactly great. Seems its basic bias network (the 220k/33k) is the main problem as I originally suspected, RC filtering the supply of that (e.g. 2k2/22u before R23) gives much better results. I normally prefer low-noise bias networks la Grundig - R/RC between supply and ground, and an R to transistor base tapped from midpoint. That may not be as easy to hack into the existing circuit though. You might just as well split R23 into 100k + 120k, then you can pretty much get away with a 1 filter cap...
Attached Images
File Type: png au5900-tone-pre-2.png (38.9 KB, 57 views)
Attached Files
File Type: asc AU5900-tone-amp-2.asc (8.7 KB, 2 views)

Last edited by sgrossklass; 8th November 2016 at 02:37 AM.
  Reply With Quote
Old 8th November 2016, 11:31 AM   #35
Depaj is offline Depaj  France
diyAudio Member
 
Join Date: Aug 2015
Location: lyon
I will take it slow !!
So far I installed the 1K resistor and the 100F cap before R3/R4 and indeed the hiss dimmed a bit but was still clearly present. Than I went for a 91R resistors in series with the zener diode (I had this value laying around and as you said between 33 and a 100 I thought this would do it) and added 220F in parallel. This got rid of all the hiss and it sounds very good with headphones but I now have +24v instead of 22 and when I turn up the volume the hiss comes back quite quickly and I now also have a fairly loud hum (also when turning up he volume).

I will tackle the tone amp next but wanted to get these first steps right before going any further.
  Reply With Quote
Old 8th November 2016, 02:58 PM   #36
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Depaj View Post
I will take it slow !!
So far I installed the 1K resistor and the 100F cap before R3/R4 and indeed the hiss dimmed a bit but was still clearly present.
At this point it would have been interesting to turn TONE DEFEAT on.

Quote:
Originally Posted by Depaj View Post
Than I went for a 91R resistors in series with the zener diode (I had this value laying around and as you said between 33 and a 100 I thought this would do it) and added 220F in parallel. This got rid of all the hiss and it sounds very good with headphones but I now have +24v instead of 22 and when I turn up the volume the hiss comes back quite quickly and I now also have a fairly loud hum (also when turning up he volume).
Higher voltage is expected as Vzener increased to Vzener + 91R * Izener, i.e. about 1.4 V more. That would be a good 2 V more at regulator output.

That could also be the reason for the hum. You didn't use the phono input to test by any chance? This one uses the -29V, which now may have gone up high enough for the negative reg to temporarily drop out of regulation (input is only -33.5 V nominal, computed output over -31 V). Could you measure input and output voltages? If the differential seems suspiciously low, try temporarily soldering about 220k in parallel to R59 to get voltage back down. Or maybe you've got one more 91R that you could add in parallel to the first one? Or plan C, remove R43 to reduce zener current.
In any case I don't think basic ripple rejection should be any worse than before now, the 220 should sink remaining hum quite effectively. If I'm wrong I'd have to whip out LTspice again and simulate the whole shebang.

Increased noise when turning up the volume is expected - you've got a 250k volume pot there, that's never going to be a low noise miracle. In the long term you may be better off modifying the headphone resistor dropper, adding about 56-100 ohms (1 W) from L and R output to ground after the 220R series resistors, e.g. at the headphone jack. Stock output level is way more than needed anyway.

Last edited by sgrossklass; 8th November 2016 at 03:25 PM.
  Reply With Quote
Old 8th November 2016, 03:11 PM   #37
Depaj is offline Depaj  France
diyAudio Member
 
Join Date: Aug 2015
Location: lyon
I actually did before touching the power supply : when all the settings are at 0 (bass, treble and mid) turning the tone defeat on doesn't affect things all that much, the hiss doesn't change in volume, maybe a little in pitch but only very slightly.

I used the Aux input for testing. at the input I have -40v after R51 (where I should have only -33.5) and +35.6v after R35 and at the output I get +24v/-31.5v.

Concerning the volume pot, several of the AU series amps have 250k pots and I definitely don't have that much noise on them. Before the modifications the volume pot didn't have any influence on the hiss, how come it does now (half way up the hiss is a lot louder than it was before).

Also it's very sensitive too touching, it makes little poping noises when I touch the chassis which it didn't before. The hum gets better when I touch any part of the chassis and worse again when stop touching it.

Last edited by Depaj; 8th November 2016 at 03:23 PM.
  Reply With Quote
Old 8th November 2016, 08:22 PM   #38
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
All quite odd, this.

If the amp is "touchy", maybe you have accidentally disconnected the chassis from circuit ground?

The hiss definitely shouldn't be louder than before. That points towards higher current noise. Possibly you've picked kind of a bad grounding spot for your 100, and now it's inserting power supply noise into that. When it comes to old single-sided layouts with thin traces, ground definitely does not equal ground.

Hmm, it looks like F-2604 gets its combined signal and power ground (grrr) from the star ground on F-2598 via F-2602 and F-2603. That's quite the way off. Power supply ground connects to F-2598 via RCA jack ground. There is no dedicated power ground available on F-2604. I don't like that. It's not an issue for the stock circuit, but try to add filtering like we're doing here, and you may be in for a nasty surprise. (The noise current from your 100 has to travel all the way back along the signal ground connection, which has a decidedly finite, non-zero resistance.) Like I said, ground != ground. Hrmpf.

You may just have to run your own power ground connection from F-2604 back to the regulator (F-2599 post 23, +/-) - ideally along the power lines, keeping a bit of a distance from the real high-impedance sections. Multiple wires of a decent thickness would be ideal, or sort of a flat braid, so you get a certain width (keeps inductance down). Then you can use that to attach your 100 + 1 supply filter capacitors. At this point noise should finally turn out as low as predicted.

So much for some quick minor mods, huh?

At this point just brute-force RC filtering the regulator output doesn't sound quite so unattractive any more. Still, I'd always prefer the circuit itself to have some decent local PSRR, being sensitive to supplies that are run through half the device could result in some funny issues.

If you want to rule out the regulator for the time being, consider temporarily shorting your series 91R to bring DC conditions back to where they should be. I don't think it's that though.

Last edited by sgrossklass; 8th November 2016 at 08:45 PM.
  Reply With Quote
Old 9th November 2016, 03:18 PM   #39
Depaj is offline Depaj  France
diyAudio Member
 
Join Date: Aug 2015
Location: lyon
I checked and didn't disconnect any ground wires by accident. I did notice though that the signal inputs for F-2604 were very close to the ground wire going to the regulator circuit. I moved those two appart a bit and the hum was already a bit better. Than I decided to install a power cord with an earthing and now I have almost no hum until the volume button is all the way up which seems very acceptable.

I wired the ground for the filtering on F-2604 directly to the ground on the regulator board along the +22v line and the hiss is down as well even when I turn up the volume.

There still is some audible noise compared to other amps but that again I haven't done the filtering for the biasing circuit yet so that will be my next step whenever I'm able to go and buy the right value resistors.

We're getting somewhere thanks to your help so minor or not I'm very grateful. I will let you know how the rest goes when I get there.

Last edited by Depaj; 9th November 2016 at 03:20 PM.
  Reply With Quote
Old 9th November 2016, 08:09 PM   #40
diyAudio Member
 
Join Date: Nov 2004
Location: Germany
Quote:
Originally Posted by Depaj View Post
Than I decided to install a power cord with an earthing and now I have almost no hum until the volume button is all the way up which seems very acceptable.
Eww. You just turned an IEC Class II device into a Class I one. Make sure you remember this in case you ever have to hunt down ground loops when using e.g. a PC as source.

Manufacturers put in extra effort so they can make these as Class II - better insulation, and there's a shield winding in the transformer that just about eliminates primary-secondary electrostatic coupling. This allows device ground to float very well.

If anything, the connection of ground to PE should be relatively weak. Maybe 1 kOhm, possibly in parallel with a pair of antiparallel rectifier diodes (1N54xx or a bridge rectifier wired up accordingly) if you also want some kind of extra protective function.
Quote:
Originally Posted by Depaj View Post
I wired the ground for the filtering on F-2604 directly to the ground on the regulator board along the +22v line and the hiss is down as well even when I turn up the volume.
Phew. So that one worked as predicted. We're getting somewhere.

You could still try running the power ground to F-2598 (star ground) rather than directly to the regulator. Might be marginally less good for hiss but better for hum. You could also choose your RC filter resistor somewhat bigger than 1k (it doesn't have to be quite that small - 2k2-4k7 would still be OK) and reduce noise insertion that way.
Quote:
Originally Posted by Depaj View Post
There still is some audible noise compared to other amps but that again I haven't done the filtering for the biasing circuit yet so that will be my next step whenever I'm able to go and buy the right value resistors.
It doesn't have to be exactly 100k + 120k. 68k + 150k would also work, or 39k + 180k with a marginally bigger capacitor (22).

Does Tone Defeat make a difference now?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Sansui AU-555A Dead channel argonrepublic Solid State 14 22nd December 2016 09:31 AM
Sansui AU-70 hum in one channel jeffrey88 Tubes / Valves 29 28th November 2015 06:25 PM
Sansui AU 555a One dead channel ? No power ? Evilution Solid State 59 17th March 2014 03:45 PM
Sansui AU-217 - Weak Left Channel mortron Solid State 0 30th September 2012 06:51 PM
one chan dead on sansui au-4400 amp. dzikamai Solid State 3 17th November 2009 12:44 PM


New To Site? Need Help?

All times are GMT. The time now is 03:20 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2
Wiki