Mooly, there would be no sense in putting a linear regulator in front, you missed the connection of winding 7 from the chopper transformer, and the purpose of the inductors around the pre-regulator.
This setup is similar to all Sony PLPS amps, in that it has a buck pre-regulator in front of a power oscillator. So, it rectifies the input, pre-regulates it to a stable voltage, and the second stage self-oscillating chopper provides sonversion to voltages as needed by the amp as well as galvanic isolation. All the regulation is done on the primary side by the pre-regulator, after that the whole thing is unregulated to the output. Winding 7 is used to synch the pre-regulator and the chopper so that you don't get beat frequency problems with a separately oscillating pre-regulator. The pre-regulator starts in a pseudo-linear mode so the chopper gets power to start oscillating, at which point the pre-reg goes into buck regulator mode. The weak point in this setup is the possibility the chopper does not oscillate properly, at which point not only does the chopper dissipate much more than designed for, but also the pre-regulator might fail to exit linear mode, resulting in overheating of it's pass transistors and failure.
Having a higher regulated primary side voltage requires more windings on the chopper as the output voltage is directly determined by the winding ratio - the chopper transformer is really just a HF transformer.
Regarding two different models, it seems illogical but is down to rules and regulations. Certification is (was?) quite different for US only and international models. Seemingly inconsequent these days when everything seems to have IEC plugs and sockets, the power wire and connection tot he amp was different, for instance the german model had an IEC cable, while this required completely different certificates for the US where such devices were not supposed to have a ground line. The UK again used some different rules, japan had it's own, etc.
This setup is similar to all Sony PLPS amps, in that it has a buck pre-regulator in front of a power oscillator. So, it rectifies the input, pre-regulates it to a stable voltage, and the second stage self-oscillating chopper provides sonversion to voltages as needed by the amp as well as galvanic isolation. All the regulation is done on the primary side by the pre-regulator, after that the whole thing is unregulated to the output. Winding 7 is used to synch the pre-regulator and the chopper so that you don't get beat frequency problems with a separately oscillating pre-regulator. The pre-regulator starts in a pseudo-linear mode so the chopper gets power to start oscillating, at which point the pre-reg goes into buck regulator mode. The weak point in this setup is the possibility the chopper does not oscillate properly, at which point not only does the chopper dissipate much more than designed for, but also the pre-regulator might fail to exit linear mode, resulting in overheating of it's pass transistors and failure.
Having a higher regulated primary side voltage requires more windings on the chopper as the output voltage is directly determined by the winding ratio - the chopper transformer is really just a HF transformer.
Regarding two different models, it seems illogical but is down to rules and regulations. Certification is (was?) quite different for US only and international models. Seemingly inconsequent these days when everything seems to have IEC plugs and sockets, the power wire and connection tot he amp was different, for instance the german model had an IEC cable, while this required completely different certificates for the US where such devices were not supposed to have a ground line. The UK again used some different rules, japan had it's own, etc.
Last edited:
Hi ilimzn, thanks for the explanation... I hadn't fully appreciated that aspect of the design actually, that it slipped from linear to switching in the pre stages, it just didn't register if I'm being honest. Thanks.
So that brings another "critical" component into play, the flywheel diode D601. I'm not familiar with the type number listed but know from experience that the old rectangular "Semikron" brand SKE2G type and similar used to fail by intermitently going open circuit.
Must be losing it this morning lol
So that brings another "critical" component into play, the flywheel diode D601. I'm not familiar with the type number listed but know from experience that the old rectangular "Semikron" brand SKE2G type and similar used to fail by intermitently going open circuit.
Must be losing it this morning lol
Wow - all that's way over my head unfortunately 
. But does any of it have any bearing on whether to stick to the factory installed 100K for R607 (which should be for the US/Canadian units only), or to change that to the 56K which is mentioned 3 times in the manual for the UK/AEP E versions of the amp? (Mines the 'AEP, E' version). The only difference I can see between the UK and the 'AEP, E' version is the UK consumes 550W, and the AEP, E model consumes 450W...
Jaycee is of the opinion it makes little difference at the end of the day, but I really would like to fit the correct part if possible.
Just checked diode D601 and it tests fine thank goodness!
My spare 2SC1810s arrived today - exactly the same as the originals and still in their 70s style Sony packaging - niiice!
Ta,
- John
. But does any of it have any bearing on whether to stick to the factory installed 100K for R607 (which should be for the US/Canadian units only), or to change that to the 56K which is mentioned 3 times in the manual for the UK/AEP E versions of the amp? (Mines the 'AEP, E' version). The only difference I can see between the UK and the 'AEP, E' version is the UK consumes 550W, and the AEP, E model consumes 450W...Jaycee is of the opinion it makes little difference at the end of the day, but I really would like to fit the correct part if possible.
Just checked diode D601 and it tests fine thank goodness!
My spare 2SC1810s arrived today - exactly the same as the originals and still in their 70s style Sony packaging - niiice!
Ta,
- John
Last edited:
Hi John...
I'd stick with the 100K.
Why... because it's what was fitted originally for one thing.
Second is that it is the value that makes sense, but that goes against the parts in the manual. My reasoning is this,
The resistor has 6.3 volts on one end. That's an absolute as it's set by the zener diode and the B-E volt drop of Q605. The other end goes to the unregulated side of things with around 340 volts DC on it. So the resistor passes around 3.3ma, and that is the current available for the error amp to work with. All seems reasonable.
If the resistor is dropped to 56 k there is nearly twice as much current available. Not a problem in itself, and it will not alter the operation of the circuit as such with regard to voltages. It would allow for "cleaner" switching as more current would be available to charge and discharge stray capacitance and switch the transistors cleanly. At the frequencies involved here that's not really such an issue. It will increase power dissipation though.
Thinking aloud now... if you fitted 56K then that would give a dissipation of W=V2/R which is 2 watts. And the manual says 1 watt rating...hmmm.
The 100K dissipates 1.1 watts... which is still over it's rating.
Thoughts anyone ? Is this where commercial ratings of parts come in. A 1 watt metal oxide type probably would be OK in practice and considered acceptable.
Getting back on track, the 56K makes sense for the lower unregulated supply of the US model as that keeps the current available similar to the 100K for UK voltages.
So keep it same as fitted
I'd stick with the 100K.
Why... because it's what was fitted originally for one thing.
Second is that it is the value that makes sense, but that goes against the parts in the manual. My reasoning is this,
The resistor has 6.3 volts on one end. That's an absolute as it's set by the zener diode and the B-E volt drop of Q605. The other end goes to the unregulated side of things with around 340 volts DC on it. So the resistor passes around 3.3ma, and that is the current available for the error amp to work with. All seems reasonable.
If the resistor is dropped to 56 k there is nearly twice as much current available. Not a problem in itself, and it will not alter the operation of the circuit as such with regard to voltages. It would allow for "cleaner" switching as more current would be available to charge and discharge stray capacitance and switch the transistors cleanly. At the frequencies involved here that's not really such an issue. It will increase power dissipation though.
Thinking aloud now... if you fitted 56K then that would give a dissipation of W=V2/R which is 2 watts. And the manual says 1 watt rating...hmmm.
The 100K dissipates 1.1 watts... which is still over it's rating.
Thoughts anyone ? Is this where commercial ratings of parts come in. A 1 watt metal oxide type probably would be OK in practice and considered acceptable.
Getting back on track, the 56K makes sense for the lower unregulated supply of the US model as that keeps the current available similar to the 100K for UK voltages.
So keep it same as fitted
And yet according to the manual the US/Canadian model uses 100K hahaha! I'm thinking this must be a mistake in the manual, and the resistor values have evidently been transposed between the UK/US models by mistake. As you say the 100K is fitted to my 'AEP, E' model. The US *should* be the 56K. I'll correct that (or add a note) in the service manual as I'll PDF it at some point incase others need it.
I'll forget about it now - I'm getting way too OCD about this
As always, mucho thanks for all the help & advice
Hope everyone's having a good weekend!- John
I'll forget about it now - I'm getting way too OCD about this
As always, mucho thanks for all the help & advice
- John
You're welcome. It's quite an interesting PSU actually. I think 100K would work in all variants tbh particularly if the transistors in the error amp are high gain, particularly Q604.
The power consumption will come down to the efficiency of the PSU at different line voltages I suspect.
That insulating gasket is on a 2 week back order with RS - arghhhhh! I guess that'll force me to continue to take it slowly with this project however. Very slowly - there's not much I can do with the board now until it arrives except solder in the new caps. The transistors will have to wait until the die-cast heatsink is back in place. I always seem to choose the out of stock items - it's already happened with 4 of the items I ordered from Farnell for this project
Oh well - the amps nice to look at anyways
Oh well - the amps nice to look at anyways
That's the insulation for the four switching transistors ?
We need to test as we go really. Are Q601 and 613 fitted ? as this section is the first to power up and test.
Bit hard to visualise exactly what it all looks like and how it all fits together. I can imagine but it's the details that count
We need to test as we go really. Are Q601 and 613 fitted ? as this section is the first to power up and test.
Bit hard to visualise exactly what it all looks like and how it all fits together. I can imagine but it's the details that count
Hi Mooly.
Yes it is, but the insulation is sandwiched between the die-cast aluminium part (which is fitted to the top of the PCB), and the PCB itself. Until that is all back in place none of the BUT11A transistors can be soldered into place as all 6 use that die-cast part as the heatsink.
I'll aim to get some pictures up later today.
Yes it is, but the insulation is sandwiched between the die-cast aluminium part (which is fitted to the top of the PCB), and the PCB itself. Until that is all back in place none of the BUT11A transistors can be soldered into place as all 6 use that die-cast part as the heatsink.
I'll aim to get some pictures up later today.
Here's a picture which shows the die-cast part which acts as both a heatsink, plus an anchor to which L604 and T3 is fitted. The insulating silpad fits underneath that.
Note the parts which press Q601 Q613 against the heatsink have been removed in this pic. I'll be using BUT11As here, and for the 4 transistors underneath the PCB. Those 4 transistors press against 4 mini heatsinks, which in turn press against the large silpad, and then the die-cast part through cut-outs underneath in the PCB. You can see the mini heatsinks plus the large silpad towards the top-left of the picture below:
These were all taken last week before I'd removed the caps & bad transistors.
An externally hosted image should be here but it was not working when we last tested it.
Note the parts which press Q601 Q613 against the heatsink have been removed in this pic. I'll be using BUT11As here, and for the 4 transistors underneath the PCB. Those 4 transistors press against 4 mini heatsinks, which in turn press against the large silpad, and then the die-cast part through cut-outs underneath in the PCB. You can see the mini heatsinks plus the large silpad towards the top-left of the picture below:
An externally hosted image should be here but it was not working when we last tested it.
These were all taken last week before I'd removed the caps & bad transistors.
Last edited:
Ahh now I see. Hm, they only used one big silpad for ease of assembly to be honest. You could have just as easily used 4 TO-220 sized pads, or TO-3P ones if you wanted to be sure. Similarly for Q601 Q613, two pads would work just as well, and they used one for ease of assembly.
Not sure why they bothered to insulate underneath the switching transistors. They could have used the fact that the tab is connected to the Collector in the PCB design. Pity they didn't as that would've improved thermal conductivity too by not needing any insulation. Again personally, I would have used TO-3P pads and just cut them down a bit with a sharp scalpel. Cutting the notches in them doesnt really seem to be neccesary.
Not sure why they bothered to insulate underneath the switching transistors. They could have used the fact that the tab is connected to the Collector in the PCB design. Pity they didn't as that would've improved thermal conductivity too by not needing any insulation. Again personally, I would have used TO-3P pads and just cut them down a bit with a sharp scalpel. Cutting the notches in them doesnt really seem to be neccesary.
I think using T03P pads like Jaycee and myself wondered about might just be do-able if you had too. It's still not the same seeing a piccy as handing the thing somehow. The bottom line is all the transistors of course have to be insulated.
Looking at your pictures in post#1
If you can fit the two reg transistors as they were in that piccy then we should be able to test the first part of the circuit.
Looking at your pictures in post#1
If you can fit the two reg transistors as they were in that piccy then we should be able to test the first part of the circuit.
Thanks Jaycee/Mooly - yes looking at it I cannot see why one couldn't use TO-220 insulators under each of the four switching transistors, pressing against the die-cast attachment with the transistor clamp (which I will make this week). This would do away with the large insulating pad, plus the 4 mini heatsinks too.
I'm stumped why Sony didn't do this - it would have been cheaper & easier to produce surely? I'm going to take another look & more pics and see if I can see any reason at all not to go that route. The back of that die-cast block actually is the bottom of the SMPS - it screws into the bottom of the black aluminium box shown in the picture a few pages back. I would have thought as Jaycee says that having the transistors 'closer' coupled to the die-cast block would improve heat dissipation too.
As it stands you have the transistor clamp, the transistor, then some heatsink paste (no insulation incidentally), then the miniheatsink (1mm thick aluminium rectangle), then more paste, the large silpad, then the die-cast block...
Hmmmmmm!
I'm stumped why Sony didn't do this - it would have been cheaper & easier to produce surely? I'm going to take another look & more pics and see if I can see any reason at all not to go that route. The back of that die-cast block actually is the bottom of the SMPS - it screws into the bottom of the black aluminium box shown in the picture a few pages back. I would have thought as Jaycee says that having the transistors 'closer' coupled to the die-cast block would improve heat dissipation too.
As it stands you have the transistor clamp, the transistor, then some heatsink paste (no insulation incidentally), then the miniheatsink (1mm thick aluminium rectangle), then more paste, the large silpad, then the die-cast block...
Hmmmmmm!
Last edited:
The picture below should give a slightly more obvious look at how it all fits together - the only bits not in the shot are the 4 mini heatsinks, but you can see from the remains of the heatsink paste where they fit on the silpad. I just do not understand the logic of having a mini heatsink, then silpad, then the large die-cast block/heatsink!
And a look at what the topside of the board looks like without the die-cast block fitted (whoops sorry should have rotated this before uploading!):
- John
An externally hosted image should be here but it was not working when we last tested it.
And a look at what the topside of the board looks like without the die-cast block fitted (whoops sorry should have rotated this before uploading!):
An externally hosted image should be here but it was not working when we last tested it.
- John
Last edited:
Ahh now it is clear what the silver pads on the underside of the PCB are. I thought they were just tinned sections.
They used paste AND silpad ? Hmm... modern silpads do not need paste, and in fact if you use it you hinder performance. I guess the materials have changed since. I'd honestly just use TO-220 or TO-3P pads. As long as everything's still insulated as before, it'll be fine.
They used paste AND silpad ? Hmm... modern silpads do not need paste, and in fact if you use it you hinder performance. I guess the materials have changed since. I'd honestly just use TO-220 or TO-3P pads. As long as everything's still insulated as before, it'll be fine.
Yeah I thought the same, but in the service manual it says to put a little heatsink paste on the visible rectangles of silpad, as well as some on the mini heatsinks. It's almost like they were deliberately decreasing the efficiency of the heatsink process perhaps? Or as you say the silpads weren't very efficient back then, and used more for insulation than heat transferring abilities.
As to the two transistors on the side, the service manual does not mention needing to use any paste. The last guy who worked on this smothered it on (you can see it on an earlier pic of mine - post 111 - behind those two red transistors) to the extent I'm not surprised those two blew!
I think I'll cancel my order for that large silpad (it came to £18 from RS once VAT and delivery were included!!!) and get four small ones instead.
- John
P.S. would you also do away with the four rectangular mini-heatsinks too? I can see no reason for those now. More efficient to couple them (via the TO-220 insulators) directly to the die-cast block, which is itself coupled to that black aluminium box which acts as a heatsink itself.
Thanks,
- John