The MONGREL (supersym II)

[ostripper]

Layout options....

This is very important , more than just the thermal considerations .. the practical construction issues (heatsinks , available space) have to be carefully
explored.

(Picture 1) The flat opposed layout. It does work quite well on my PB250 , allows direct thermal contact , lends itself well to a good NFB takeoff and general circuit symmetry. The rails can be run close together to cancel current pulses and is the nicest looking (subjective opinion). I have seen this on the genesis and many other esoteric amps , it's only drawback is that you need a tall heatsink 125mm (5") this style would be hard to implement on a 2U setup.

(picture 2 ) Flat aluminum angle mounting. Very common here at DIYA. The oblivious drawback is the additional thermal interface of the aluminum angle. It also requires that you source said angle. It does look nice and is easy to layout and service. With my proposed 100 X 150mm boards you would lose 200mm of interior chassis space which just might leave you with 200+ mm in the center to mount your power supply components.
PS ... you ALSO could mount the angle/amp ABOVE the power supply components , freeing up space in the chassis center ! I personally do not like this method because of the thermal issue , but this would be minimal. some heatsinks have the Extrusion for this layout method , but are expensive.

(picture 3) Edge mounted layout. This is nice for low profile cases , it still takes up that 200mm of chassis space but a 4-600VA http://www.antekinc.com/pdf/AN-6440.pdf or this :
Antek - AS-4440 will work out well.
This layout also allows direct thermal OPS contact and can be laid out impeccably.

COME ON GUYS !!! You PM and email me ... tell me what you want !!!

OS

 

[Lazy Cat]

Something like this, of course directly to heatsink, no bracket.

 

[ostripper]

Something like this, of course directly to heatsink, no bracket.

Littlefish ... good amp. That is design 2 (pix2- above , my last post) , flatly mounted without the L bracket would be better . In that case , design 1 (flat opposed) would be preferable. You could layout the rails and star ground better with "opposed" as well. Of course , one could implement any of the 3 layouts into any case with a little imagination. Thanks for feedback.

OS

 

[Rudi_Ratlos]

Lazy Cat's idea implies PTH PCBs.
I vote for ALU-brackets, but in the standard way:

From top-to-bottom: Transistors - ALU-bracket - PCB.

Use some thermal grease on the back of the ALU bracket, before you fix it to the heatsink.

Best regards - Rudi_Ratlos

 

[meanman1964]

I think those special CPU metal pads are very usefull for that

 

[AndrewT]

Something like this, of course directly to heatsink, no bracket.

half way down

The PCB is designed with all the power transistors mounted underneath. Bolts pass through the PCB, then the transistors, then into a heatsink, as shown in the diagram:

Don't do it like the diagram.
The FETs are mounted too far away from the heatsink.
The angle should be turned the other way with the leg pointing to the opposite side of the FETs.
The PCB does not need to overhang the FETs.
The Fets can be moved until they almost touch the heatsink.
The angle mounting face can be machined/ground/filed/polished for good/best heat transfer.


When the FETs are mounted on the same side as the angle leg, you lose all these advantages.

 

[ostripper]

Something to consider , in the CPU world the bracket method would be both unacceptable and possibly destructive. CPU's will dissipate 65-100W these days. Overclockers will polish the HS surface's to gain every last .01 C/W , use silver compound , heatpipes. EACH of our OPS devices will possibly transfer 50W into whatever thermal solution we use (200W+ total) , so ... should we not at least "keep up" with the PC people ?

I am not saying the "bracket" method is wrong , in fact, this method can be mounted flat with the devices underneath as Lazy cat suggested for max transfer. However , I must agree with Andrew that it is NOT ideal. The heat transfer on my PB250's is awesome ... the device bodies are just warm even as the Genesis heatsinks are nuclear hot (nearly full transfer).

The new Aspen amps are like the littlefish w/ onboard supply caps , by seeing the photo's , they can be mounted either way (bracket or direct).
This could be a good idea , as this is the only layout scheme that allows for both options.
OS

 

[MJL21193]

One very convenient advantage of the angle bracket is the ability to use through bolts and nuts to mount the outputs, as opposed to drilling and tapping. Most here seem to ignore this though and still run the risk of breaking off taps doing it the 'right' way.
I use course sheet metal screws that cut their own thread in the right size hole.
The opposed layout can be more compact if the board overhangs the devices and has the majority of the trace routing (that would normally be in the middle). Potentially have a board 2" tall.

 

[ostripper]

Thanks for your input , John. You like the angle bracket , as do most others.
So , I am leaning toward example 2 , as it can also be direct mounted. Some might not have a Homo-depot around the corner or might just like direct mounting (like me). I like example 1 and your new amp , John (I saw it ) , but the "kit" is for DIYA.

To allow for both topside or bottomside OP device mounting , I guess onboard main supply caps are out of the question. I will give the amp multiple mid-sized (220 -470uF per device) local OPS decoupling caps with a separate ground return. This seems to be strictly adhered to on the nikko's layout.

Is 4 X 6 " (100 X 150 mm) too big ?

PS ... yeah , some might not be into tapping or the possibility of a snapped off tap.
OS

 

[AndrewT]

The opposed layout can be more compact if the board overhangs the devices and has the majority of the trace routing (that would normally be in the middle). Potentially have a board 2" tall.

someone recently posted a compact, opposed layout that did exactly that.
Anybody got a better memory than me and could provide a link?

100*150 is too big for me. That's almost as wide as the Leach clones which are enormous @ 107 wide.

 

[sajti]

I allways design my pcb to align with the fins of the heatsink, I plan to use. For diy it's not very complicated. This means that I mount my output devices directly to the heatsink.

Sajti

 

[AndrewT]

I allways design my pcb to align with the fins of the heatsink,

or is it the gaps between the fins that are to be used for through bolting?
This only works if you can machine a spotface large enough to accommodate the bolt head and/or nut between or into the flanks of the fins.

 

[sajti]

or is it the gaps between the fins that are to be used for through bolting?
This only works if you can machine a spotface large enough to accommodate the bolt head and/or nut between or into the flanks of the fins.

OK, my mistake: aligned with the gaps between the fins. I use M3 imbus screw, which are small enough for most of the heatsink. You can se one of my amplifier in the "Post your solid state picture" topic.

Sajti

 

[ostripper]

I guess the design will have to accommodate the skill level of the end user. I overlooked this aspect. Most will not like to tap , even as I find it quite easy (NEVER snapped one) , some may not. Fishing the screws through the fins is also too much to ask of the typical DIY'er.

Andrew , 100 X 150mm is too big ??? Have you seen the Siliconchip (UDL) blameless (150mm X 150mm - at least) or Dr. selfs "signal transfer corp." blameless... HUGE (100 X 140mm - for just 2 main OPS devices ?) No cap mults. , either. The "leach clone" that Ed Lafontain personally showed me was 100mm X 200mm , but had 8 or 12 OPS device capability. There also is the end user consideration of eyesight. I mean... 100 x 150mm fits in the palm of my hand , will easily fit in a standard chassis ... I also just happen to have several dozen precut 100 X 150 HQ FR-4's.

PS - If I was to go fully commercial (like the Aussieamps) , I would go smd ... 65 X 100mm ... all pre-assembled . At this point , it is no longer discrete DIY in my opinion , just a glorified chipamp.

OS

 

[Bill_P]

Mounting devides directly to the heatsink is the way to go. Tapping is no problem at all and like you I have never broken a tap.

The kit might be best offered in several forms: PC boards only, PC boards with semiconductors, and a full kit of parts with semiconductors and passives.

 

[sajti]

This is the output buffer of my hybrid.

Sajti

 

[ostripper]

This is the output buffer of my hybrid.

Sajti

With your heatsinks, "fishing" the screws through the fins would be no problem , but not everybody has YOUR heatsinks. What if the fins were spaced different ?? A "tappin'" you would have to do !

OS

 

[sajti]

With your heatsinks, "fishing" the screws through the fins would be no problem , but not everybody has YOUR heatsinks. What if the fins were spaced different ?? A "tappin'" you would have to do !

OS

You have some space to move the output devices, say 8-10mm is possible, due the lenght of the legs. So this design can adopt for another heatsink.
I like this arrangement, because it takes less space from the box.

Sajti

 

[MJL21193]

Thanks for your input , John. You like the angle bracket , as do most others.
Is 4 X 6 " (100 X 150 mm) too big ?

PS ... yeah , some might not be into tapping or the possibility of a snapped off tap.
OS

I'm not particularly attached to anything (just ask my ex's ). I would only use the most practical or convenient way. In my 6 channel, it made the most sense to use the angles, as the heatsinks were made up of pieces.

Taps should have no place in our world, not when you can buy thread cutting screws from places like McMaster Carr. 100 stainless steel tapping screws for the cost of one tap? A no-brainer.

I like this arrangement, because it takes less space from the box.

 

[ostripper]

Taps should have no place in our world

TAP HATERS UNITE !! I guess idiots don't either ( I am not referring to you).
Self- tapping , nuts/bolts through angle , or "anal" 4-40/6-32 tap action , whatever gets it done !

PS ..to-220's with their insulated thin-walled "shoulder" washers might have a issue with self tapping screws.

The OEM's reflect this in direct proportion to price/ value.

(photo 1) dirt cheap sony uses self tapping hardware.
(photo 2) $400 Nikko uses 4/40 tapped.
(photo 3) $7000 genesis uses hardened 6/32.

All options would work , threaded /tapped could most likely be torqued harder. I suppose this falls into the same category as people that paint their speaker cones...

OS