Back into town this afternoon, and my parts were waiting for me. Dug into the amp, and decided to first play with the Zobel filter before trying the 47 ohm resistors on the drivers...
I also had ordered 10 ohm 3W resistors, so had to decide which to use...the 10 ohm or the 5.1 ohm. Also had to decide if I wanted to move the Zobel to before the R||L filter.
Because the feedback for the opamp is taken before the R||L filter, I thought it best to also have the Zobel there to minimise any instability. Also, the 5.1 ohm power resistors I have are nice non-inductive types, and I'm not so sure about the 10 ohm ones I have.
So I moved the Zobel so it is before the R||L filter, using the 5.1 ohm resistor, and replaced the cap with a very good metal polypro .1µf Matsushita.
Guess what? Output is perfect!!! Load, no load, square wave test, anything I throw at it looks beautiful. I'm so happy I'm beside myself!!!
I was sooo sick of messing with this I was about ready to throw it in the river, but now I have reason to celebrate!!!
To me, it simply appears to be a slew of poor design decisions. Who the hell says that just because you have a job designing amplifiers for a big company that you know what the hell you're doing? More and more I'm realizing that a lot of guys doing this might have been better off selling shoes....
Thanks for you guidance Eva...even if I didn't use your ideas, you made me think, and I learned something.... appreciated.
I also had ordered 10 ohm 3W resistors, so had to decide which to use...the 10 ohm or the 5.1 ohm. Also had to decide if I wanted to move the Zobel to before the R||L filter.
Because the feedback for the opamp is taken before the R||L filter, I thought it best to also have the Zobel there to minimise any instability. Also, the 5.1 ohm power resistors I have are nice non-inductive types, and I'm not so sure about the 10 ohm ones I have.
So I moved the Zobel so it is before the R||L filter, using the 5.1 ohm resistor, and replaced the cap with a very good metal polypro .1µf Matsushita.
Guess what? Output is perfect!!! Load, no load, square wave test, anything I throw at it looks beautiful. I'm so happy I'm beside myself!!!
I was sooo sick of messing with this I was about ready to throw it in the river, but now I have reason to celebrate!!!To me, it simply appears to be a slew of poor design decisions. Who the hell says that just because you have a job designing amplifiers for a big company that you know what the hell you're doing? More and more I'm realizing that a lot of guys doing this might have been better off selling shoes....
Thanks for you guidance Eva...even if I didn't use your ideas, you made me think, and I learned something.... appreciated.
Hehe Nice to hear that everything's ok but that means no more tinkering which is baaad
Maybe adding 0.1uf met polyprop as power supply bypass? You may solder them on the back of the PCB on the rail connectors and if you're real picky (or your friend is) you can physically clean up the PS rails connectors real good (soldering here helps a lot!
) and I think that the input cap may benefit a lot from bypassing with a .1 or even removal if everything upstream has no DC. Not so if the signal passes through a lousy preamp, though
Nice to hear that everything's ok but that means no more tinkering which is baaad Maybe adding 0.1uf met polyprop as power supply bypass? You may solder them on the back of the PCB on the rail connectors and if you're real picky (or your friend is) you can physically clean up the PS rails connectors real good (soldering here helps a lot!
) and I think that the input cap may benefit a lot from bypassing with a .1 or even removal if everything upstream has no DC. Not so if the signal passes through a lousy preamp, though
When I got the nice .1µf cap for the Zobel, I also got several for bypassing the power supply...nice and close to the output devices.
Yeah, there are probably a few other thngs I could do to it, but I'm doing this as a favor for a friend, and I've probably got 100 hours invested in it, plus better than $160 US in parts.
At this point, it works...and it works pretty well. For a free fix, I think he got a hellova deal.
I'm quitting while I'm ahead. After all, the enemy of 'good' is 'better'.
Yeah, there are probably a few other thngs I could do to it, but I'm doing this as a favor for a friend, and I've probably got 100 hours invested in it, plus better than $160 US in parts.
At this point, it works...and it works pretty well. For a free fix, I think he got a hellova deal.
I'm quitting while I'm ahead. After all, the enemy of 'good' is 'better'.
I know its a very old and almost dead thread, but I have question to EchoWars. Before fixing the oscillation problem his comment on this AMP is not to take even if some one gave it for free. I would like to ask him whether he changed his opinion after he fixed the oscillation problem. Why I am asking is I got this amp from some one in exchange for my Kyocera R-861 receiver + Denon DRA-1000 preamp. Currently the amp works but I would like to replace the caps and speaker connectors and zobel filter mod etc before using is as my main amp. Is this amp worth the effort? If not I would like tear down the driver boards and build an high power amp based on LME49810 chip kit. I like the cool meters and hefty heat sinks of the amp. I agree with EchoWars Philips tried to use the cheapest (except CA3100) available part to build this amp. It's amazing how these amps are working even after 30+ years.
Thanks,
Routhun
Thanks,
Routhun
echo wars .....
very nice effort really !!!!!
i know these machines are strange to play with .....
there is a couple of things id like to add
A) my aprroach will always be that if you have any form of ocilation there is no point to try to treat it with cortizone ..... meaning there is no point on trying to supress it in the out put .....you have to find where is it coming from .
B ) changing all these drivers and so on i would play with other values of miller caps ( and other quality) in driver or vas stage ...
C) and finally and most funny i will tell you that a mosfet amplifier that due to simple and poor design was very horny for any ocilation prooved to me a huge mistake that was taking place under my bench .....
the test socet, signal genarator, and scope installed in my bench were actually creating i huge EARTH LOOP !!!!!!!!!!!!
BUT !!!!! all the amplifiers i ve repaired or constructed in the past with the same configuration never show any problems related to that ..... it prooved that this particular amplifier that was very easy to ocilate made me rearange all that ....so i was able to find out if ocilation was coming from my instruments interconection or the amplifier !!!!!
very nice effort really !!!!!
i know these machines are strange to play with .....
there is a couple of things id like to add
A) my aprroach will always be that if you have any form of ocilation there is no point to try to treat it with cortizone ..... meaning there is no point on trying to supress it in the out put .....you have to find where is it coming from .
B ) changing all these drivers and so on i would play with other values of miller caps ( and other quality) in driver or vas stage ...
C) and finally and most funny i will tell you that a mosfet amplifier that due to simple and poor design was very horny for any ocilation prooved to me a huge mistake that was taking place under my bench .....
the test socet, signal genarator, and scope installed in my bench were actually creating i huge EARTH LOOP !!!!!!!!!!!!
BUT !!!!! all the amplifiers i ve repaired or constructed in the past with the same configuration never show any problems related to that ..... it prooved that this particular amplifier that was very easy to ocilate made me rearange all that ....so i was able to find out if ocilation was coming from my instruments interconection or the amplifier !!!!!
Just recently I've had my hands under the hood of an old Sansui 7070 receiver. And guess what, it has the same zobel error - the zobel is at the output terminal, after the output inductor. This precisely defeats the purpose of the zobel network - it is supposed to be a 'default' load at HF, which it certainly cannot be through a coil - except at the resonance frequency of the coil and cap. Further, the resistor in the zobel is a wirewound, which creates another resonant circuit (fortunately at very high frequency) at which the zobel becomes ineffective. Needless to say I will modify that...
OK, I just found the Sansui 7070 service manual on the web. The output network is as you describe, but with a 4.7 ohm resistor in parallel with the coil. The coil is labelled as 2.5mH, but I’ll assume that is an error and the coil is 2.5uH.
The Zobel R is also 4.7 ohms and the Zobel C is 100nF.
This network does indeed provide a smooth, low impedance high frequency load to the amplifier output.
Unloaded at HF beyond the resonant frequency it will look like a purely resistive 9.4 ohm load (the sum of the two 4.7 ohm resistors in series). Note that the most favoured Zobel resistance is 10 ohms.
I have attached a sim of the network below. R3 is added to give the source a 9.4 ohm output impedance and the frequency response at “OUT” is plotted.
As can be seen the resonant frequency is at about 300kHz, at which the network provides the lowest impedance and after that the attenuation rises to –6dB and the phaseshift reverts to zero degrees, indicating the network is 9.4 ohms resistive.
This of course assumes that the resistors are not inductive
The Zobel R is also 4.7 ohms and the Zobel C is 100nF.
This network does indeed provide a smooth, low impedance high frequency load to the amplifier output.
Unloaded at HF beyond the resonant frequency it will look like a purely resistive 9.4 ohm load (the sum of the two 4.7 ohm resistors in series). Note that the most favoured Zobel resistance is 10 ohms.
I have attached a sim of the network below. R3 is added to give the source a 9.4 ohm output impedance and the frequency response at “OUT” is plotted.
As can be seen the resonant frequency is at about 300kHz, at which the network provides the lowest impedance and after that the attenuation rises to –6dB and the phaseshift reverts to zero degrees, indicating the network is 9.4 ohms resistive.
This of course assumes that the resistors are not inductive
Attachments
I probably said that out of frustration. Once the thing is sorted out, it works and sounds pretty good. It's just a bear to work on, and an oddball design.routhun said:
Move the output Zobel, and replace the poly caps with high-quality non-inductive types. The Zobel worked really well with a .1µf and a 5.1 ohm 3W resistor. You can also replace the input opamp with whatever you like, but when I did the above amp 4+ years ago, I had a bunch of opamps here, and I thought the 5532 sounded best for this amp.
Edit: ...and yes, in that drawing in post #2 has an error...that cap in the Zobel filter is 0.1µf, not 1µf.
Oh, and here is where I got the complete service manual for the 7070 from:
http://www.eserviceinfo.com/downloadsm/23258/Sansui_7070.html
http://www.eserviceinfo.com/downloadsm/23258/Sansui_7070.html
ilimzn said:
Thiele was not daft.G.Kleinschmidt said:OK, I just found the Sansui 7070 service manual on the web. The output network is as you describe, but with a 4.7 ohm resistor in parallel with the coil. The coil is labelled as 2.5mH, but I’ll assume that is an error and the coil is 2.5uH.
The Zobel R is also 4.7 ohms and the Zobel C is 100nF.
This network does indeed provide a smooth, low impedance high frequency load to the amplifier output.
Unloaded at HF beyond the resonant frequency it will look like a purely resistive 9.4 ohm load (the sum of the two 4.7 ohm resistors in series). Note that the most favoured Zobel resistance is 10 ohms.
I have attached a sim of the network below. R3 is added to give the source a 9.4 ohm output impedance and the frequency response at “OUT” is plotted.
As can be seen the resonant frequency is at about 300kHz, at which the network provides the lowest impedance and after that the attenuation rises to –6dB and the phaseshift reverts to zero degrees, indicating the network is 9.4 ohms resistive.
This of course assumes that the resistors are not inductive
I think that putting a wirewound resistor in series with a cap to form a zobel damping network is wrong. The wirewound has inductance and creates a resonant circuit in a place you really don't want it.
Carbon composition resistors are the right part to use there. Unfortunately, I don't think they are made anymore, although you can sometimes find them through surplus and nos dealers.
Also it is debatable as to whether the output terminal of the amp should be the feedback point or not...depends on the circuit, really... although in this case it helped fix your problem.
Carbon composition resistors are the right part to use there. Unfortunately, I don't think they are made anymore, although you can sometimes find them through surplus and nos dealers.
Also it is debatable as to whether the output terminal of the amp should be the feedback point or not...depends on the circuit, really... although in this case it helped fix your problem.
Thanks EchoWars!!!
I took out the driver boards from this amp, they need good cleaning, capacitor replacement and Zobel filter mod. This amp is pretty heavy and philips used cheap and thin metal chasis. Only the heatsinks holding this amp together. If heatsinks are removed the front, back and bottom parts are not connected, they were held together by the heat sinks. I wish they used littlebit strong metal bars to held the chasis. If anyone wants to remove the heatsinks to remove driver boards just remove one heatsink at a time and put back this hint sink before removing the other. Otherwise it is pretty nasty to put together because of this nasty heavy power transformer.
Thanks,
routhun
I took out the driver boards from this amp, they need good cleaning, capacitor replacement and Zobel filter mod. This amp is pretty heavy and philips used cheap and thin metal chasis. Only the heatsinks holding this amp together. If heatsinks are removed the front, back and bottom parts are not connected, they were held together by the heat sinks. I wish they used littlebit strong metal bars to held the chasis. If anyone wants to remove the heatsinks to remove driver boards just remove one heatsink at a time and put back this hint sink before removing the other. Otherwise it is pretty nasty to put together because of this nasty heavy power transformer.
Thanks,
routhun
he he he
once i had to fix a double deck cassete player from philips ....
of course you need torx and some other special tools to remove things from it
then of course you need three working hands to do that since two are not enough
and of course there is point when clik sound appears and your shop is full of coils springs and plastics .....
philips R&D ha ha ha
once i had to fix a double deck cassete player from philips ....
of course you need torx and some other special tools to remove things from it
then of course you need three working hands to do that since two are not enough
and of course there is point when clik sound appears and your shop is full of coils springs and plastics .....
philips R&D ha ha ha
I'm sure glad I found this thread. I've been sob-storied into working on one of these oddball things. One channel is completely shorted, and the other has plenty of audible distortion, the filter caps need to be replaced, and that's all I've discerned so far. It's certain that after the amp channels and unregulated supplies are repaired, there will be more. I've already told the owner to be prepared to spend a fortune.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.