[scootsit]

Hello all!
 
A few months back, I picked up an old SA-100B off of ebay. There was a LOT of static on it. The static is on both channels and does not change with volume level.
 
I started trying to trouble shoot it, I was working with someone on audiokarma, but he's had to take some time away and I could use a bit more help. I tried AK first because it's not headphone related (though at ~1WPC, it would be a good electrostatic amp).
 
Anyway, it's a very low-power (maybe 1WPC) positive ground amplifier, it is also always on.
 
Schematic:

 
 
What I've done so far:
I've replaced the PS cap to no avail.
 
I noticed that the secondaries of the transformer were not equal when the transformer was connected to the board, I disconnected the transformer and now everything appears to be okay. The two red wires are V+
The black wire is V-
The white wire is the V- off the lamp. The voltages are:
Red-Red: 38VAC
Red1-White: 25.6VAC
Red2-White: 12.3VAC
Red(both 1 and 2)-Black: 19VAC
White-Black: 6.7VAC
 
Point is, the transformer is fine when it's off the board.
 
I tested the diodes, too. In reverse, both show that they are open, forward, one is 0.511, the other 0.509, which should be about right. So, if it's not the ps cap, it's not the diodes, and it's not the transformer, what could it be? I'm considering just replacing the diodes to see what happens, is that a good next step?
 
The original AK thread is here: http://www.audiokarma.org/forums/showthread.php?t=481150

 

[scootsit]

I replaced both rectifier diodes with 1N4004 (or 1N4005... either way) and changed the PS cap to a brand new 2200uF cap. The loud hum is pretty well gone, but the static is still unacceptably loud. The static is independent of the volume knob, and equal in both channels. Any thoughts?
 
Thanks!

 

[duncan1]

Sounds like an earthing problem.I cant make out the circuit too well as it is so small.Check also the input capacitors if electrolytic then change for quality types or if High impedance input then input capacitor will be low value then change for poly-prop types.Check for multiple earth loops /only need one for whole audio set up/ connected via audio inputs
 

 

[scootsit]

I already replaced all caps in the power supply. Is that what you mean, or the actual input caps?
 
Also, the schematic gets bigger if you click it.
Once there, you can click the "original" button in the lower right, and it will get bigger still.

 

[duncan1]

You're right. Yes it has a high impedance input so the series input capacitor will be a small value worth replacing both with high quality -polyprop types The input I mean by the input to the actual transistors not the components in front of it.
AS I said you should only have one earth return in the whole set-up . If 2 or more from other equipment is connected to the mains with earth returtrns then that is wrong  Earths travel along the shielding of the co-ax to p[provide an earth to each piece of equipment.I still think it is an earthing problem only other thing would be if the amp is wide-band and is picking up your cordless house phone/electrical equipment that is causing it. But it still comes back to earthing  or if not the shielding of your audio equipment is not too good

 

[scootsit]

The input caps are polypropylene, which I thought had a pretty long shelf life. You really think they could be the source of the noise?

 

[duncan1]

I cant tell what they are as I dont  have a component list . If they are poly-prop -your right  not likely to be faulty.I have come across DECT house-cordless phones causing noise in audio and computer equipment. The other area to check is the two smoothing capacitors-C23/C24 which are  500uf in value . The circuit is pretty old and goes back to the 70s at least. Notice the 2 "interstage transformers. Not used in the 80s onwards.So if it is as old as that .And I have worked on a lot from that era. The capacitors of the 70s were Really bad and could "dry up" if electrolytic. 
Try changing them .Make sure the polarity is correct when soldering.If that doesnt sort it Then the change the output capacitors-C20/C21-get the best you can afford as they have enormous bearing on the fidelity-Just look at the Quad 303 PA-a big capacitor that does the sound quality no good.[so smooth you would fall asleep listening] People think electrolytics last forever not after 40YRS . Also change C25 a main smoothing capacitor.Let me know how you get on and make sure the replacements  are of good quality its possible they will be a smaller size as well .

 

[scootsit]

When I first started, the fact that the noise is consistent across the volume and between the two channels led me to think it was in the PS. C25 and the two rectifying diodes were the first things to go. I actually increased the size of C25 to either 1200 or 1800 uF (I forget now), that had no impact. Changing the diodes seemed to help slightly, but not really much. I'm aware that it's an old design, and that old caps go bad. My thinking is that if this isn't something in the PS, what could it be? If I can eliminate the noise, I'll probably do a full recap, but if I can't solve the noise, I'll probably stop before investing in a full new set of caps. I do see the 500uF smoothing caps as a potential place to go, but aren't they likely to have a minor effect if C25 is in? It would seem that C23/24 are just quieting little transients, not really dealing with major noise.
 
Regarding the household circuit, I've got plenty of dead-silent audio gear on the same outlet, and I've tried this amp on multiple outlets. Also, I do not own a cordless phone.
 
There's something in the amp that's noisy as can be, but it's not the rectifiers, the transformer, or the PS cap.
 
I've thought about the ground channel, but it's this weird active ground and I've had a hard time figuring it out. Between the ground and the negative on the transformer, there are ~15 V. I also think there's a short somewhere because the voltages between V+ and V- are not equal from ground, I believe the ground is closer to V+ than V-, but I could be remembering wrong.

 

[scootsit]

I do appreciate your help, but I've tried basically putting other caps in parallel with some of the caps in there, figuring that their total capacitance would be additive, and if it changed, that it would be a place to look. The result has been pretty much nothing.

 

[KimLaroux]

There are 0.01µf caps between the gate and collector of the output transistors. I have a feeling these are there to filter out high frequencies. Though since the noise is audible, it's not like you could just replace these with larger values... it'll just lower the high frequency dropout. (just thinking out loud)
 
R43 and C23 create an RC filter for the input stage. Raising the values of both of these will raise the attenuation of the filter. You may be able to raise the resistor to 1k or so, but it'll lower the voltage on the stage. It may be worth a try though.
 
You could also add an RC network directly to the PSU. There's a 1000 µf cap for power smoothing (C25), but nothing between it and C24, the power decoupling cap of the output stage. It may be worth a try to add a second 1000 µf cap and a resistor between both or them. I'm guessing you only have ~27 V on C25, so you don't have much to spare. You may get away with dropping a volt or two across a resistor. It's another cheap test.
 
If you feel motivated, you could also add a linear regulator in there. Though be careful what design you use since this is a "positive ground". You'd need a negative regulator, I think.

 

[duncan1]

Kim. Looking at C23 it is a power supply smoothing capacitor of 500uf terminated between the supply voltage to the first 2 input transistors and system ground not a filter as such. R43 of 270K sets the input impedance.
R15 and C910 in the local neg-feed-back are used as a filter.and are in standard series connection for that use.   old interstage transformers used in early transistorized audio amps were very liable to have shorted turns[low quality] and as the power goes through them your intuition could be right .
In that case scrap it not nice to say and negative but if the mods dont cure it then cut your losses as you are not going to get another transformer. And buy from the 80s onward unless it is very top of the line equipment.
 

 

[scootsit]

So given that I've basically rebuilt the p's, new caps and diodes, and there's still very loud white noise, what would you replace next?

 

[duncan1]

I admire your determination to get this working right. The first  2 transistors are low noise so it shouldnt be coming from them. You have to appreciate that I am working "long distance" from the amp instead of having it in front of me. The next test is to S/C the input that is  from the input of C5 --  1/C6--1 to earth[bypassing the input components]. This is a standard method to make sure that the noise is actually coming from the circuit itself and not being induced by the high-impedance of the input-270K.
In other words to see if it is picking up radiated interference.If not the case then that leaves the resistors 1960s/70s resistors werent low noise and if the equipment has been lying a long while not being used they can change value as well.Do any look overheated Carbon resistors of the period gave out a good bit of noise.
Check-R9/R10-R17/R18 voltage feeds to the input transistors for correct value and if possible change them for low noise -metal 1% ones If you have measured the voltage to both transformers and one is different from the other that could point to shorted turns in the transformer  but you would then have one good and one bad output.         .Other than that without it being in front of me to work on I am running out of ideas.Remember old equipment isnt up to modern standards and if you have modern high resolution listening equipment Faults in old equipment will show up easier. AS listening transducers of that era just dont have the same resolution.Best of luck.!

 

[KimLaroux]

Quote:

Kim. Looking at C23 it is a power supply smoothing capacitor of 500uf terminated between the supply voltage to the first 2 input transistors and system ground not a filter as such. R43 of 270K sets the input impedance.
R15 and C910 in the local neg-feed-back are used as a filter.and are in standard series connection for that use.   old interstage transformers used in early transistorized audio amps were very liable to have shorted turns[low quality] and as the power goes through them your intuition could be right .
In that case scrap it not nice to say and negative but if the mods dont cure it then cut your losses as you are not going to get another transformer. And buy from the 80s onward unless it is very top of the line equipment.
 

 
LOL! There are two R43 in the schematic! I was referring to the 470R resistor at the top of the schematic between the output and input stage supplies.

 

[duncan1]

Yes there are 2 R43-just shows you what Japanese early circuit layout was like -As far as R43[at the top] this is standard means of lowering both current and voltage to the input pair of transistors to isolate noise from the output circuit as the output active devices in most amps are not low noise.
This doesnt matter so much at the output because the ratio of signal to noise is very low. But it does matter at the input as even the tiniest will be amplified and heard at the output.
Look at Hi-Fi power amps and voltage circuit feeds between input and output .You will find this is the simplest approach a more intelligent approach is to have separate power supplies to each half.[stabilized] The more money you spend the higher quality of circuit[usually]
So even R43-2 isnt a filter of any sort that we are talking about -only a cheap way of stopping noise from the output.