Pioneer SX-1980, AWR-154/155 power supply re-design

[rcs16]

Hello Ak'er,

I have a SX-1980 monster in for a re-build, it was working for a while when I received it. The owner wants it re-capped, new or re-built PS etc. The PS had repairs done to it already, what else the +80V reg blew, but the owner was wanting a new PS pcb installed because we can see the old one has been hacked at. I am still waiting to get the new PS, months have passed, I want to get this moving and done, so I will go ahead and design the pcb.

This is a low volume item, so not much return on investment for time and resources, none the less I continue on this quest in interest of science. Wife thinks I am nuts as I should be working on the garage electrical, insulation, walls, but I will use the its cold story and run with that I can call it my new job for now.

While I was adjusting the wonky tuner and testing out the unit, I heard a loud POP, the PS blew, the top of Q209 blew its lid, the -ve 80 smoked, with it sitting on its side, the bottom plate off.
So now I have a dead SX-1980 PS. The original PS is a bit of a PITA to service since you have to remove the transformer secondary wires to get access to the back side. I can see the remnants of the last job I'll take a pic.

We had been discussing this in another thread, so I suggested I'd take a stab at a new PS design, thus this thread to discuss the design and review what I have come up with.

We all know that this original PS design has issues, based on failures and user reports, heat is #1, no ventilation and under sized HS's. jfet constant current sources (CCS) fail regularly, so I have decided to use much larger heatsinks, EW's CCS ckt, the Pioneer mod to double up the zeners, terminal blocks for easy install, service, etc. The pcb will be spec'd to use 2 oz copper and be double sided and lots of copper to strengthen it up. I want to keep it as low cost as possible.

I am not convinced that one has to remote mount the hot regs, if the PS heatsinks are sufficient in size, so I will go that route. It is a fair amount of work to remote the regs, drill, tapping of the PA HS's, wire harnesses etc hopefully can be avoided with this new design. If someone wants to remote mount the regs then they can always do so, but my first attempt is to use much larger HS to start with.

The plan for me is to do the re-design, build, test, then if all goes well, I can offer the pcb's blank or stuffed/tested as required.

I have the first pass at the schematic done (see attached) and am presently in layout to do component placement etc. I still have to remove the old PS and measure the pcb size, mounting holes, determine the max comp heights etc. Pls let me know if you see anything wrong with the schematic. I will also include the preliminary BOM, so you know about the component choices I have made.

One issue that I have a concern, is continuing to use the PA3004 speaker protection device. Can we rely on getting them? There is the option of re-designing the protection ckt, but that is another task. For now I'll use the PA3004 as is.

At one time I did a PS re-design in ltspice simulation, making it a dual tracking +/- regulator with current limiting, but for this rev, I think I will keep it as close to original as possible except for the newer CCS, much larger HS's, terminal blocks etc..

All input is welcome.

Time for me to march on ...

Cheers
Rick

 

[rcs16]

I guess you did not understand my meaning, where I said

I am still waiting to get the new PS, months have passed

So knowing this, I would also like to cost reduce it for myself and change the design to meet my specifications

 

[kevzep]

Is there any sense in going full metal jacket and using some modern voltage regulators? This is what I was thinking...I haven't explored it at all but was just thinking about it a while back...maybe its easier to work with the Pioneer design with mods for now?
Do you think there's any sense in making those 80V supplies adjustable, I'm not sure its necessary? I know the TX turbo boards are....

Looking good though, I agree on the protection, a lot of these chips like that PA3004 are going to become scarce I suspect...

I also agree on the remote mounting, it is a lot of work to make that happen, totally recommend if you are using the original board, but since this is a new venture, would be good to have it all on the same board, it would save a lot of time.

 

[rcs16]

full metal package = TO-3,TO-5,TO-39, hermetic sealed = military app = expensive and cumbersome
Not going to make the +/-80V adjustable, there is no good reason to do so. If you make +/-80V adj, why not the +/-34V as well?
In my simulation of the upgraded design, I made the +/-80V a tracking type, one zener reference, like they did in the Sansui AU-X1.
using 1% R's will tighten up the differences. Could spec tighter tol zeners as well.
also no axial ecap option, way more selection and better specs on radial types.
Thinking about a LM317 for the 13.5V but they have the 13.5V controlled by the +80V supply, so to keep that feature, I'd need a reg with an enable or a bjt switch etc, so what I gain, I'd also loose. it is all about trade offs.
I will think about using the more readily avail TA7318 or uPC1237, but space is limited so it may not fit, it would require stuffing options since they are not pin compatible, the uPC1237 is close to the same.

 

[restorer-john]

The uPC-1237 is the go. Whether Nippon Precision Components (NEC) still makes them is the issue. My Japanese supplier no longer lists them however, as currently available in quantity. There's a stack of Chinese clones, but whether they are any good is anyone's guess.

 

[kevzep]

full metal package = TO-3,TO-5,TO-39, hermetic sealed = military app = expensive and cumbersome
Not going to make the +/-80V adjustable, there is no good reason to do so. If you make +/-80V adj, why not the +/-34V as well?
In my simulation of the upgraded design, I made the +/-80V a tracking type, one zener reference, like they did in the Sansui AU-X1.
using 1% R's will tighten up the differences. Could spec tighter tol zeners as well.
also no axial ecap option, way more selection and better specs on radial types.
Thinking about a LM317 for the 13.5V but they have the 13.5V controlled by the +80V supply, so to keep that feature, I'd need a reg with an enable or a bjt switch etc, so what I gain, I'd also loose. it is all about trade offs.
I will think about using the more readily avail TA7318 or uPC1237, but space is limited so it may not fit, it would require stuffing options since they are not pin compatible, the uPC1237 is close to the same.

All good reasoning there, I agree there's no good reason to make the voltages adjustable.
Yes I noticed you had that zener in there too. Sansui use that in a lot of their power supply designs, seems like it was their "go to". I think its a good way to do it.
1% R's is a good idea too, all my resistors in stock are 1%, makes more sense to me..
I was wondering about the LM317 too but you raise a good point there too.
Oh well, looks like you have it pretty much nailed.

 

[markthefixer]

I just wish to say "Good Luck" and genuinely wish you well in this endeavor..

My investigations of the PSU left me convinced that there is insufficient airflow in that physical area of the chassis and that no matter how much heat sinking is in that volume that the heat transfer to the air will be the limiting factor. I had looked into heat transfer to the chassis itself. The extra wattage of remotely mounting the regulators to the audio heat sinks and the increased heat effects thereof were checked by myself and found to be negligible. I acknowledge that remote mounting the regulators IS a PITA.

As for the "full metal jacket" of using modern voltage regulators, IMO there isn't a lot of chips that can handle those voltages (up to 110v DC in and 80v DC out) , and handle them quietly. I was looking to make some testing power supplies years ago, to use as quick substitutes. There isn't a lot of equipment that needs power supply voltages like that anymore, so the chip makers don't have a lot of incentive.

 

[rcs16]

You know I might not even use the PA3004, I see that uPC1237 is still in production, http://www.unisonic.com.tw/datasheet/UPC1237.pdf

Thx Mark for your encouragement. right now I am looking at a rats nest and a dead 1980.
I want to make it simple for myself, I do not want to invest the time, in a new thermal design, which means to me mechanical work
Going to have to leave it as a Japanese cooker, so goal is too try to reduce device op temps in effort to improve reliability and serviceability

You are correct that there is limited IC parts for the task. The +/-80V regs pass devices each dissipate about 2W, there is a lot of wasted power in converting voltages down.
They should have used 2 transformers or 2 more taps, choices, choices.
I have some options
1) try to fix the old burnt out pcb eek
2) Same design with mods but no advance features like current limiting
3) IC design using TL783,LM317HV,LM337 etc
All these types can float above ground, the TL783 will current limit at supply V of up to 125VDC, LM317HV is good to 57V, whereas the regular LM3x7 will fry if shorted if above ~40V (same issue as original design)
TL783 as it is designed for the task, but it is only a +ve reg and you can't use it for the negative, because of the transformer center tap.
4) Same sort of design using all discrete, the one I simulated, it is a lot of components.

To be honest I think I am now leaning towards option #3 now that I see this design in layout, TL783 alone, makes life so much easier for me and anyone having to build &/or service

 

[onwardjames]

Tech stuff rules, tho I don't understand it.

 

[kevzep]

I just wish to say "Good Luck" and genuinely wish you well in this endeavor..

My investigations of the PSU left me convinced that there is insufficient airflow in that physical area of the chassis and that no matter how much heat sinking is in that volume that the heat transfer to the air will be the limiting factor. I had looked into heat transfer to the chassis itself. The extra wattage of remotely mounting the regulators to the audio heat sinks and the increased heat effects thereof were checked by myself and found to be negligible. I acknowledge that remote mounting the regulators IS a PITA.

As for the "full metal jacket" of using modern voltage regulators, IMO there isn't a lot of chips that can handle those voltages (up to 110v DC in and 80v DC out) , and handle them quietly. I was looking to make some testing power supplies years ago, to use as quick substitutes. There isn't a lot of equipment that needs power supply voltages like that anymore, so the chip makers don't have a lot of incentive.

You got that right, I did it on the one I restored, but not sure i would do it again to be honest, so much fiddling around...

I had a quick look yesterday and came to the same conclusion, the +/-80V are the difficult ones..

You know I might not even use the PA3004, I see that uPC1237 is still in production, http://www.unisonic.com.tw/datasheet/UPC1237.pdf

Thx Mark for your encouragement. right now I am looking at a rats nest and a dead 1980.
I want to make it simple for myself, I do not want to invest the time, in a new thermal design, which means to me mechanical work
Going to have to leave it as a Japanese cooker, so goal is too try to reduce device op temps in effort to improve reliability and serviceability

You are correct that there is limited IC parts for the task. The +/-80V regs pass devices each dissipate about 2W, there is a lot of wasted power in converting voltages down.
They should have used 2 transformers or 2 more taps, choices, choices.
I have some options
1) try to fix the old burnt out pcb eek
2) Same design with mods but no advance features like current limiting
3) IC design using TL783,LM317HV,LM337 etc
All these types can float above ground, the TL783 will current limit at supply V of up to 125VDC, LM317HV is good to 57V, whereas the regular LM3x7 will fry if shorted if above ~40V (same issue as original design)
TL783 as it is designed for the task, but it is only a +ve reg and you can't use it for the negative, because of the transformer center tap.
4) Same sort of design using all discrete, the one I simulated, it is a lot of components.

To be honest I think I am now leaning towards option #3 now that I see this design in layout, TL783 alone, makes life so much easier for me and anyone having to build &/or service

option 3 you think? I'll be interested to see what you come up with.

I was even thinking a separate auxiliary transformer, but thats getting a little extreme.... Jeez, its a real conundrum this power supply, right?

I think its pretty well jammed in that power supply area, but I wonder if you could mount a miniature fan somewhere and duct some cool air to that board? Hmmmmm...I wonder? I was thinking about doing that to my SX1010, but then I sold it...

 

[zebulon1]

I had a laser cutting shop quote on a bottom cover that had extra cooling holes cut and located in a better location. The forum had discussed the mod and decided the cost was on the high side. Around $80 for a complete modified bottom cover. I liked the idea of keeping the original in tact.
I could follow up on the quote and see if any savings can be had over the last one and if they still have the file.
I also used separate heatsinks for the regulators which after testing showed improvement.
Pages three and four has some relevant info and MTF shares some wisdom about the cooling improvements throughout the thread.
http://audiokarma.org/forums/index....d-heatsink-preliminary-drawings.615197/page-3

 

[rcs16]

Look'n good
Done the preliminary layout of option #2, using the schematic that was posted earlier. I have a feel for the flow of the original design.
Now that the bulk of the layout is sorted out, there is enough room to add the uPC1237 option, in the Q303 area.
The +/-80,+/-34V series pass bjts get there own HS, The 13.5V,8V share a HS.
Next task, take out the old PS, measure it up and complete the layout process.
Anyone have one in hand?, know the pcb dimensions?
Will add some extra holes on the pcb, under the heatsink fins, allows for the heat to circulate, if it is possible.

Sorry, fan option will not be incorporated in this version of the Japanese cooker PS design

 

[zebulon1]

I like the HS's you selected and the small package relays.

 

[kevzep]

I think thats really good Rick, I think those HS packages are really good and I think your idea of the holes in the board under the HS's is a great idea too.
Yep shaping up nicely....

 

[Verde]

Rick, were you waiting on a board from TXTurbo?

 

[rcs16]

That was the original plan by the owner, but I gave up waiting, so decided to design my own version, I found ways to cost reduce the design.
Finished manually routing the pcb connections, still have to do some finishing touches and more checking.
Plan is to order the parts ASAP, have in hand, this allows me to double check the pcb comp footprints before I commit to fab the pcb.

Fan option, once I got back into the unit to remove the old PS pcb, I can see how you could mount a small muffin fan on the back plate to circulate/suck hot air out, thus cool the PS. You could run the fan off the switched AC.

 

[kevzep]

Fan option, once I got back into the unit to remove the old PS pcb, I can see how you could mount a small muffin fan on the back plate to circulate/suck hot air out, thus cool the PS. You could run the fan off the switched AC.

I like this idea.....best way to get rid of that heat buildup...

 

[zebulon1]

The heat is an issue but the OEM is for operation without a fan. Granted, the PS runs hot regardless of the volume but it runs for years like that.

 

[kevzep]

The heat is an issue but the OEM is for operation without a fan. Granted, the PS runs hot regardless of the volume but it runs for years like that.

Yes agreed, but look what happens to them, I don't think it would hurt to reduce the ambient temp in there.

 

[rcs16]

Up and running using a old SX-1250 transformer as my test jig supply. Seeing LEDs light up and relays clicking is a relief.

Enjoy
Rick