Swedish Chef said:Magura,
excellent of you to share your findings!That was exactly the hard data I was looking for here. Have you any secret bag of tricks for making the alu/copper junction as good as possible, or do you just sand them down and apply thermal grease? How about screw tightening arrangements, do you just run the M3's for the TO-247's all the way through or do you use lots of complementary screws?
Cheers
/Magnus
The tests were made with 6*M6 or M5 for 100*50*10.
Thermal grease and mica was used between the copper and the heatsink, just thermal grease between the device and the copper.
Whenever I make something with heatsinks and/or heatspreaders, I allways machine the contact faces. Sanding just makes matters worse, as as it will make the ithem turn convex.
Any machine shop should be interested in cake for the afternoon coffee
Magura
Good point. Something like that was my worry about the thermal expansion thing, i.e. the copper plate slightly bending thus loosing contact with the heatsink somewhat. We're only talking a temp rise of 50-60 degrees here so maybe it's not much of an issue. But I have seen my share of thermal expansion problems so I don't take anything for granted anymore. Most people just don't realize the tremendous mechanical forces involved here.
Pretty much in the same direction as copper spreaders is the issue of using heat pipes (the convecting/phase changing stuff used in most laptop computers). Has anyone here had a look whether it might be worth the trouble for an audio amp?
/M
I used 5mm thick copper plate in my class A follower. I have to dissipate 40W with TO-220 device. The biggest problem was the thermal resistance of the insulation plate. So I mount the device on the center of the 6*6cm copper plate, without insulation. Then I insulate the plate from the heatsink. To fix the copper I used 8pcs. of M4 screws. It still works without any failure.
sajti
sajti
I see some facts that demonstrate that the use of copper in that design is just cosmetical :
- They use copper also in the case of the amplifier, look at the pieces that fix the walls to the floor and the capacitors to the floor. Copper is obviously not expensive for them
- The copper plate appears to be 1mm thich so it spreads essentially nothing. Even a TO-220 output device has a copper plate slighty thicker than 1mm [and featuring 0,5ºK/W Rth between the junction and the external side of the copper plate
] and the MT-200 Sanken devices feature about 2,5mm thick copper plates
- The cooper plate does not extend above the output devices so half of its heat-spreading capabilities are lost!
- There are no screws nor any means of fixing this copper plate to the heatsink. The copper plate is only guaranteed to make good thermal contact to the heatsink under the transistors. Also think about copper dilatation versus aluminum dilatation. the copper plate must not be continuous, it has to be divided in several pieces with dilatation joints between them since dillatation coefficient for aluminum is 26% higher than for copper
- The ouput devices are placed on the upper side of the heatsink. Shame! the upper side of any heatsink working with natural convection cooling is allways hotter than the lower side. For such a heatsink I would expect 10ºC difference
- The worst case dissipation for 1400W into 1 ohm resistive with +-65V rails is 1000W and this means about 40W per MT-200 device. Not only such a thin copper plate is ridiculous but also these heatsinks with such a thin heat storage/spreading base without fans are ridiculous for 1400W continuous. If they seriously wanted to reduce operating temperature they would have placed a good quality air blower with speed control. In the other hand, in normal music operation with 4 ohm speakers these MT-200 devices are not supposed to dissipate more than 5W each
Does it make any sense to use copper plates with MT-200 devices dissipating between 5W and 40W ??? How much would junction temperature decrease? Maybe 1ºC?
I think that kind of gear and bookleets are intended to look attractive to all those audiophiles ready to spend $20.000 or more in an amplifier and knowing nothing about thermal resistances of materials and their junctions and thermal expansion coefficients. They just like to see copper because copper looks great and belive all the pseudo-science statemens included on the booklets
Copper plates built into all transistors and ICs are intended to integrate the thermal energy pulses generated in the die [like a capacitor] and reduce the peak temperature reached by the die during transient dissipation conditions [thermal circuits are modelled like electrical circuits]. Placing extra plates only makes sense if they are of the optimum thickness to increase the 'thermal capacitance' without increasing the DC thermal resistance to the heatsink
In the other hand, audio amplifiers work at audio frequencies so there is no point in extending the 'heat-integration' capabilty below 30Hz, and MT-200 devices already provide quite decent 'heat-integration' due to its massive built-in copper plate of 35x20x2.5mm
- They use copper also in the case of the amplifier, look at the pieces that fix the walls to the floor and the capacitors to the floor
. Copper is obviously not expensive for them- The copper plate appears to be 1mm thich so it spreads essentially nothing. Even a TO-220 output device has a copper plate slighty thicker than 1mm [and featuring 0,5ºK/W Rth between the junction and the external side of the copper plate
] and the MT-200 Sanken devices feature about 2,5mm thick copper plates- The cooper plate does not extend above the output devices so half of its heat-spreading capabilities are lost!
- There are no screws nor any means of fixing this copper plate to the heatsink. The copper plate is only guaranteed to make good thermal contact to the heatsink under the transistors. Also think about copper dilatation versus aluminum dilatation. the copper plate must not be continuous, it has to be divided in several pieces with dilatation joints between them since dillatation coefficient for aluminum is 26% higher than for copper
- The ouput devices are placed on the upper side of the heatsink. Shame! the upper side of any heatsink working with natural convection cooling is allways hotter than the lower side. For such a heatsink I would expect 10ºC difference
- The worst case dissipation for 1400W into 1 ohm resistive with +-65V rails is 1000W and this means about 40W per MT-200 device. Not only such a thin copper plate is ridiculous but also these heatsinks with such a thin heat storage/spreading base without fans are ridiculous for 1400W continuous. If they seriously wanted to reduce operating temperature they would have placed a good quality air blower with speed control. In the other hand, in normal music operation with 4 ohm speakers these MT-200 devices are not supposed to dissipate more than 5W each
Does it make any sense to use copper plates with MT-200 devices dissipating between 5W and 40W ??? How much would junction temperature decrease? Maybe 1ºC?
I think that kind of gear and bookleets are intended to look attractive to all those audiophiles ready to spend $20.000 or more in an amplifier and knowing nothing about thermal resistances of materials and their junctions and thermal expansion coefficients. They just like to see copper because copper looks great and belive all the pseudo-science statemens included on the booklets
Copper plates built into all transistors and ICs are intended to integrate the thermal energy pulses generated in the die [like a capacitor] and reduce the peak temperature reached by the die during transient dissipation conditions [thermal circuits are modelled like electrical circuits]. Placing extra plates only makes sense if they are of the optimum thickness to increase the 'thermal capacitance' without increasing the DC thermal resistance to the heatsink
In the other hand, audio amplifiers work at audio frequencies so there is no point in extending the 'heat-integration' capabilty below 30Hz, and MT-200 devices already provide quite decent 'heat-integration' due to its massive built-in copper plate of 35x20x2.5mm
I am using copper too, but in rather different configuration. I have all output transistors screwed at massive copper prism ( 20 * 10 mm or 30 * 10 mm ) without insulation, 'cos I'm using this prism as power rail ( compare it with diameter of copper on PCB ) . Insulated from heatsink is only this prism. I mean that this is the best solution, which have all advantages .
) . Insulated from heatsink is only this prism. I mean that this is the best solution, which have all advantages .
Upupa Epops said:I am using copper too, but in rather different configuration. I have all output transistors screwed at massive copper prism ( 20 * 10 mm or 30 * 10 mm ) without insulation, 'cos I'm using this prism as power rail ( compare it with diameter of copper on PCB
This is same application I use for my hybrid amplifier. This configuration reduce the internal resistance of the power supply, and also reduce the thermal resistance between the device, and the heatsink.
sajti
Eva said:I see some facts that demonstrate that the use of copper in that design is just cosmetical :
- They use copper also in the case of the amplifier, look at the pieces that fix the walls to the floor and the capacitors to the floor
- The copper plate appears to be 1mm thich so it spreads essentially nothing. Even a TO-220 output device has a copper plate slighty thicker than 1mm [and featuring 0,5ºK/W Rth between the junction and the external side of the copper plate
- The cooper plate does not extend above the output devices so half of its heat-spreading capabilities are lost!
- There are no screws nor any means of fixing this copper plate to the heatsink. The copper plate is only guaranteed to make good thermal contact to the heatsink under the transistors. Also think about copper dilatation versus aluminum dilatation. the copper plate must not be continuous, it has to be divided in several pieces with dilatation joints between them since dillatation coefficient for aluminum is 26% higher than for copper
- The ouput devices are placed on the upper side of the heatsink. Shame! the upper side of any heatsink working with natural convection cooling is allways hotter than the lower side. For such a heatsink I would expect 10ºC difference
- The worst case dissipation for 1400W into 1 ohm resistive with +-65V rails is 1000W and this means about 40W per MT-200 device. Not only such a thin copper plate is ridiculous but also these heatsinks with such a thin heat storage/spreading base without fans are ridiculous for 1400W continuous. If they seriously wanted to reduce operating temperature they would have placed a good quality air blower with speed control. In the other hand, in normal music operation with 4 ohm speakers these MT-200 devices are not supposed to dissipate more than 5W each
Does it make any sense to use copper plates with MT-200 devices dissipating between 5W and 40W ??? How much would junction temperature decrease? Maybe 1ºC?
I think that kind of gear and bookleets are intended to look attractive to all those audiophiles ready to spend $20.000 or more in an amplifier and knowing nothing about thermal resistances .......
Obviously you know everything about thermal behaviour and putting successful amp design together. At least that's how it seems from your write up. Well, here are few pointers:
1. I can bet that none of the capacitor's clamps is made out of copper, nor the angles and brackets supporting PCB assembly. Clamps are chromated (or whatever the process is called) while brackets are made out of steel an plated with what looks like copper. That's a common practice with Japanese makers.
2 The copper plate is more like 1.5mm thick and it only extents as much as heatsink size allows.
3 I'm not sure how you did come to conclusion that there are no screws holding the plate to the sink. Whatever we can see, there are screws attaching devices (I see at least 22 of them, some of them with 2 screws), the board is attched with 6 screws and each bracket with 3 screws. How many screws all together? And we still don't know if they used any under the PCB.
4 The output devices are placed not only on the upper side of the heatsink, but also on the bottom side, and the same assembly is placed on the other side. I still didn't notice where the actual UHC-MOS devices are placed.
5 The MT200 devices are probably placed on the same heatsink for thermal stability. I did the same when building my A75 amp ( I placed input devices on the main heatsink) as this provided for much less DC offset drift and more steady work conditions (after initial warm up period).
6 I think this type of comments (as yours here) are good for someone who never actually built a high power amp, but again, they bring no positive input and only critism for no actual purpose, unless $20,000 price tag really annoys you.
What about the new SA1 amplifier which has just been introduced?
I recently bought the PMA-S10 Integrated amplifier which is a 50 watt/channel desigh. An amazing amplifier I must say, it also has these uhc mosfets, I tested four different type of speakers on it from 4 Ohms to 15 Ohms and it admirably drove everyone of those without batting an eyelid.
I recently bought the PMA-S10 Integrated amplifier which is a 50 watt/channel desigh. An amazing amplifier I must say, it also has these uhc mosfets, I tested four different type of speakers on it from 4 Ohms to 15 Ohms and it admirably drove everyone of those without batting an eyelid.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.