Thanks for all the support, much appreciated.
Hugh, yes the fins are vertical, please see my pictures. I still have one channel mounted on the sink, as you can see I am using a copper spreader, 100mm x 3 mm thick . Tomorrow I will power it up and take more detailed results, but when I had it powered up for an hour or two it was difficult to keep a hand on the sink and the devices could only be touched for an instant.
I do not have a posh reflective type thermometer, but I will post the results I find.
Regards
Alan.
Hugh, yes the fins are vertical, please see my pictures. I still have one channel mounted on the sink, as you can see I am using a copper spreader, 100mm x 3 mm thick . Tomorrow I will power it up and take more detailed results, but when I had it powered up for an hour or two it was difficult to keep a hand on the sink and the devices could only be touched for an instant.
I do not have a posh reflective type thermometer, but I will post the results I find.
Regards
Alan.
Those sinks should be big enough for an Alpha 20. Your spreader may be too thin. Try direct coupling to one of those sinks each?
That was my thoughts about to. You need at least 6mm or more copper.
I know copper is expensive but here you can not save on that.
Before you do anything please measure the temperature after fully warm up at the heatsink the one in the middle and in the last and first one.
If the temperature difference more than 4-5C degree you will need the thicker copper.
I believe that Alan's heatsinks are designed for forced cooling based on their their internal structure. Such heatsinks do not work well with natural convection; you need open fins, not too closer than 10mm apart.
So, Alan, either a fan underneath the heatsinks, OR move to CPU heatpipe/fan cooling. This would be very disappointing to you..... and this is the 20W, imagine the heat issues with the 50W Big Boy?
Hugh
So, Alan, either a fan underneath the heatsinks, OR move to CPU heatpipe/fan cooling. This would be very disappointing to you..... and this is the 20W, imagine the heat issues with the 50W Big Boy?
Hugh
Alibear,
Those fins look widely spaced too me. So probably was designed for natural convection as opposed to forced convection. I think judging from size of those heatsink sub units, even one per MOSFET directly connected should handle 30w no problem. It needs open air to access the currently sealed “pocket” formed when spaces so closely together. Just two widely spaced units should work. That’s my feeling. Might be worth a test. Remove copper spreader. Attach directly and use only two sub assemblies with space in between. Also, change to R131 as 0.47R will help a lot too.
Last edited:
I do not think this arrangement can work with natural convection as only the outer surface is effectively cooling, not the inner fins. Just imagine the heat flow and physics: the huge volume of (very well conducting) and permanently heated aluminium vs. the tiny volume of (poorly conducting) air between the fins that should transport all the heat away? Here you have a burning carbon in a pocket...;-). I agree with X that the sub units need to be wider apart to allow more air to exchange...which is the principle of forced cooling, if you not only allow the air to flow but push it with a fan 😀.
Thanks X. I have just set up again as in the picture, I will let it run for two hours and try to record some more definite results. Then I will do as you suggest, just use two sections, one for each device and see those results. I would really like to use passive sinking if I could.
This is the second channel producing lovely music.
Couple of questions: both channels produce an almost identical offset of minus 15mV. The pot in the emitter of V101 has no effect on this reading on either channel. If I blow on the area of V101-V103 the offset reduces to zero, again both channels. I thought this may be due to matching V101 - V102, but if I hold a pair of long nosed pliers on them the offset does not alter. If I do the same withV103 the offset drops towards zero.Does this make sense, or am I being stupid?
Regards
Alan
This is the second channel producing lovely music.
Couple of questions: both channels produce an almost identical offset of minus 15mV. The pot in the emitter of V101 has no effect on this reading on either channel. If I blow on the area of V101-V103 the offset reduces to zero, again both channels. I thought this may be due to matching V101 - V102, but if I hold a pair of long nosed pliers on them the offset does not alter. If I do the same withV103 the offset drops towards zero.Does this make sense, or am I being stupid?
Regards
Alan
Thank you MBA, looking at it logically I agree. The fins are 90 degrees wrong orientation, this makes sense. I will go through the testing I describes in my previous post just out of interest. I will have learned something and if it is of use to others that will be good.
Regards
Alan.
Regards
Alan.
would kapton be useful here? those are certainly monstrous heatsinks.
two channels with these heatsinks arranged back to back with 2 or 4 fans gently blowing/sucking air (through the tunnel formed), might keep things cool.
the cups are certainly upto the task (to be able to hold the weight😀)
These are the minor variations in voltage drops across V101/V0102/V103 and V104(diode). They are all silicon circa 0.6v and variations up/down should statistically all wash out to about 0mV. But sometimes off by as you see 15mV (but that is quite stable from turn on to turn off). If it bothers you, try substituting a 500R pot for R106 (set nominally to 220R) and tweak that to get 0mV. Then remove it and measure the value and replace with a hard-set exact 1% or 0.1% metal thin film resistor of same value. I think there is about 2mA flowing so a change of 7ohms up or down on that pot would be circa 15mV. Let's ask Hugh if I have this correct, but I would not worry about 15mV offset.
Last edited:
The cooling depends on the surface area and the temp gradient between the heated element (heatsink) and cooling medium (air). With natural convection the warmed (and lighter) air is rising along the heatsink and sucks in the cold air from below. So the fin orientation in your case seems Ok. But I am afraid the tiny air volume between the inner fins gets so hot in the low part of the heatsink that while rising it does not cool in the upper part any more - as there is no or only small temp gradient between hot heatsink and hot air...
Last edited:
~6hours into the second solid day of listening to my new alpha and it's sounding better and better.
Temperature, dc offset are all rock solid.
Now I need to build a real case!
Temperature, dc offset are all rock solid.
Now I need to build a real case!
Nice to hear Gab!
I took voltage out from other rectifier to dc-dc regulator, adjusted 12v -> fan. Now I can adjust fan power down if heat allows that.
Installed thick copper plates under aluminium clamp that holds mosfets against sinks. So coppers against mosfets, and 12mm aluminium bar bolted on to give nice pressure over mosfets.
Started to install 120va 15v trafo and adding speaker delay/protection-boards. I have some turn on/off thumb, don't dare to drive horns/drivers without delay.
I took voltage out from other rectifier to dc-dc regulator, adjusted 12v -> fan. Now I can adjust fan power down if heat allows that.
Installed thick copper plates under aluminium clamp that holds mosfets against sinks. So coppers against mosfets, and 12mm aluminium bar bolted on to give nice pressure over mosfets.
Started to install 120va 15v trafo and adding speaker delay/protection-boards. I have some turn on/off thumb, don't dare to drive horns/drivers without delay.
Offset is completely safe up to +/-100mV. The goal is to get it within +/-30mV.
The benefit of a LTP is that both active devices, situated in one spot, share the same thermal environment.
As I mentioned, an offset within +/-30mV is bagatelle. A voice coil is unaffected by this offset.
All LTP transistor work under this regime. This one is actually a little tighter than most because of the unusual quasi current mirror at the collectors.
Cheers,
Hugh
The benefit of a LTP is that both active devices, situated in one spot, share the same thermal environment.
As I mentioned, an offset within +/-30mV is bagatelle. A voice coil is unaffected by this offset.
All LTP transistor work under this regime. This one is actually a little tighter than most because of the unusual quasi current mirror at the collectors.
Cheers,
Hugh
I checked progress of B.B. Beta boards production. It took them 5hrs to drill the billion via holes in the 10 boards. 
I noticed the on/off thump as well. I think I have some of the DIYAudio Store speaker protection boards, I may assemble one and connect it.
It’s pretty mild and slow (0.2 second long slow motion) . If you have a high pass filter it’s fine. Of course if running direct drive active system then you may need delay delay. Also, if you are running a cap Mx, it has soft start ramp up voltage feature so I don’t think there is thump.
