How much "worse" does having a properly made bracket (out of aluminium alloy) with a large contact area (as in half the back of the heatsink, one half for one set of TO-3's and one half for the other set of TO-3's....) and very very flat surfaces and goop and lots of bolts/screws/whatever holding it down, in reality, would it add another 0.5c/w making it totally useless?
Thanks
Aaron
Thanks
Aaron
NUTTTR said:
why not measure it and see -- it isn't preposterous to suggest this -- all you need is an Igloo cooler filled with water and a thermometer.
Make two devices -- one consisting of a TO-3 device connected via bracket to the heat sink, the other with the device directly mounted onto the heat sink. Immerse the fins of the heat sink portion into the water and measure the temperature rise over at precise point in time. You can convert the change in temperature back to calories -- (but you will have to know the volume of water.)
time to get a TIG welder -- or find an aircraft mechanic who will weld the aluminum brackets to the heat sink.
Re: Keld, in you environment...hehe....do not need so much heat sinking as here.
Maybe in theory it would work. But what if you want to move the amp or the window is in the wrong end of the room or --aso.
And in the winter you want to keep the heat inside your house. (BTW this winter is really warm, now (19:00PM) its 0 degrees Celsius thats freezing point of water for all you who use other mysterius kind of scales)
I think the best way to transfer lot of heat is watercooling.
Fans are for PA, disco and partys, or to show that you cannot afford the right size heatsinks.
destroyer X said:
Maybe in theory it would work. But what if you want to move the amp or the window is in the wrong end of the room or --aso.
And in the winter you want to keep the heat inside your house. (BTW this winter is really warm, now (19:00PM) its 0 degrees Celsius thats freezing point of water for all you who use other mysterius kind of scales)
I think the best way to transfer lot of heat is watercooling.
Fans are for PA, disco and partys, or to show that you cannot afford the right size heatsinks.
I was actually wondering last night - what if i use the MF30's (with the channels for TO-3's) and fan cool them, i.e. have them mounted inside the amp and have 3 fans blowing from the base up through the top? How well would that work? I dont mind the higher price, the MF18 makes it a squeeze to fit 10 TO-3P devices on it (manageable but messy and tightly packed!!)....
However the other ones being 30cm long would be EASY to fit the to-3's on it (and yes, i'd still prefer to-3's!!) What do you think? Realistically it's a 0.25c/w - i only want 1 heatsink per channel though, i don't mind 2 slow running fans per channel.... possible? Because i'd LOVE to do that!
Aaron
However the other ones being 30cm long would be EASY to fit the to-3's on it (and yes, i'd still prefer to-3's!!) What do you think? Realistically it's a 0.25c/w - i only want 1 heatsink per channel though, i don't mind 2 slow running fans per channel.... possible? Because i'd LOVE to do that!
Aaron
Nuttr,
A 0.25C/W sink with slow (but forced) air cooling will give you roughly 2.5 times the thermal capacity, viz 0.1C/W. This would mean that with proper device connections you could easily dissipate 200W evenly along the length with only a 20C rise above ambient. On a day of 35C, this would be 55C sink temp, and 75C case temp and likely about 90C junction temp. These are all estimates, based on experience, and this is acceptable, just. But long heatsinks don't easily transfer heat lengthwise, so you'd need to spread the TO3 devices out nicely - and moreover, I'd suggest you need to use the MF30-1F-100, with a mounting ledge of 8mm thickness. Don't use a flatback for this; no amount of careful joining of the device mount against the flatback will be entirely satisfactory and there is always quite a large temperature gradient at the join which does not happen with the ledged heatsinks.
You might find it tricky to direct vertical air flow over a 12" length; here, I'd suggest you either use a tube/manifold approach with a 120x120mm fan, or even use a tangential fan at low speed. You must direct air in at the base for best cooling, directing it downwards will increase the air speed required.
In high heat Class A, it is very, very difficult to use conventional, free air sinks and achieve satisfactory junction temperatures. It's too bulky, too expensive, and highly dependent on amp positioning. Better to use low fan speeds and forced air.......
Hope this helps,
Hugh
A 0.25C/W sink with slow (but forced) air cooling will give you roughly 2.5 times the thermal capacity, viz 0.1C/W. This would mean that with proper device connections you could easily dissipate 200W evenly along the length with only a 20C rise above ambient. On a day of 35C, this would be 55C sink temp, and 75C case temp and likely about 90C junction temp. These are all estimates, based on experience, and this is acceptable, just. But long heatsinks don't easily transfer heat lengthwise, so you'd need to spread the TO3 devices out nicely - and moreover, I'd suggest you need to use the MF30-1F-100, with a mounting ledge of 8mm thickness. Don't use a flatback for this; no amount of careful joining of the device mount against the flatback will be entirely satisfactory and there is always quite a large temperature gradient at the join which does not happen with the ledged heatsinks.
You might find it tricky to direct vertical air flow over a 12" length; here, I'd suggest you either use a tube/manifold approach with a 120x120mm fan, or even use a tangential fan at low speed. You must direct air in at the base for best cooling, directing it downwards will increase the air speed required.
In high heat Class A, it is very, very difficult to use conventional, free air sinks and achieve satisfactory junction temperatures. It's too bulky, too expensive, and highly dependent on amp positioning. Better to use low fan speeds and forced air.......
Hope this helps,
Hugh
Hugh - thanks for the help you've given me so far! Been excellent!
I don't mind using 3 fans per sink if i have to - or i can go the manifold route, which was something i was thinking of, just getting it made up might be a little tricky, but i can work around that...
All the MF30's have one shelf or two, i was looking at the MF30-2F-151.5 but it only has 6mm shelfs, but 2 of them, this is perfect for my setup that i wanted, it allows me to add voltage "rails" directly to the back of the heatsink, etc to set it up and also allows one ledge to have the -ve side and the other the +ve side...
I am using 5 pairs per channel (this really could be upped to 6, but i'm happy using 5 - should distribute fairly evenly) - so 3 of these sinks with some fans hanging off them! That would mean - per "ledge" i will have about 75-80w disspaited heat - around about 15w/device..
Those calcs seem about correct to me and if i get the cooling right should work a treat!
Another question - would it be better to have the fans and heatsinks on the sides of the case? Or internally ? i would have thought that internally would make the entire casing substantially hotter.... However i guess i could have "external" side panels covering from the bottom of the amp up to the bottom of the heatsinks, then have open air sinks (with fans hidden behind the panel) and let it suck air from the floor?
Thanks for the help guys
Aaron
I don't mind using 3 fans per sink if i have to - or i can go the manifold route, which was something i was thinking of, just getting it made up might be a little tricky, but i can work around that...
All the MF30's have one shelf or two, i was looking at the MF30-2F-151.5 but it only has 6mm shelfs, but 2 of them, this is perfect for my setup that i wanted, it allows me to add voltage "rails" directly to the back of the heatsink, etc to set it up and also allows one ledge to have the -ve side and the other the +ve side...
I am using 5 pairs per channel (this really could be upped to 6, but i'm happy using 5 - should distribute fairly evenly) - so 3 of these sinks with some fans hanging off them! That would mean - per "ledge" i will have about 75-80w disspaited heat - around about 15w/device..
Those calcs seem about correct to me and if i get the cooling right should work a treat!
Another question - would it be better to have the fans and heatsinks on the sides of the case? Or internally ? i would have thought that internally would make the entire casing substantially hotter.... However i guess i could have "external" side panels covering from the bottom of the amp up to the bottom of the heatsinks, then have open air sinks (with fans hidden behind the panel) and let it suck air from the floor?
Thanks for the help guys
Aaron
krell heatsinking...
Hi Guys,
For comparison we should probably take a look at the amps in question, they used a tunnel style heatsink, the ledge being large enough for a pair of to3s, 4 ledges around the perimeter, allowing for a total of 8 output devices...The length of the tunnel can't have been much more than 4 inches, and as far as I know no-one complained about the fan, so it can't have been blasting air too fast. For the ksa50 that solution had to work for both channels and dissipated ~320 watts at idle, the outputs idling at ~2.2A per channel with ~37v rails.
The KSA100 used one such tunnel for each channel, basically because it had to dissipate twice as much power from twice as many devices. I think using a 12" tunnel for each channel with 3 outputs per ledge, giving 12 to3's or to3p's per tunnel will be able to cope. If the extrusion were identical it should have 2 or 3 times the capacity, though the airspeed may need to be higher...Are we really talking about dissipating 600-1000watts per channel?
Steve at ApexJr has 12" tunnel extrusions drilled for a bunch of to3s, for $25usd a pair, about as cheap as you are going to find the solution...
The nominal power rating of the mj15003/4 is 250w, so even applying the normal derating curve still offers huge margin at krells original 40w per device. The plastic transistors are smaller, but not by much at 200w. Increasing the count to lower the per device dissipation to 30w sounds like a sound plan, and as far as I know is being followed by all those people cloning the ksa50...
Ciao
Stuart
Hi Guys,
For comparison we should probably take a look at the amps in question, they used a tunnel style heatsink, the ledge being large enough for a pair of to3s, 4 ledges around the perimeter, allowing for a total of 8 output devices...The length of the tunnel can't have been much more than 4 inches, and as far as I know no-one complained about the fan, so it can't have been blasting air too fast. For the ksa50 that solution had to work for both channels and dissipated ~320 watts at idle, the outputs idling at ~2.2A per channel with ~37v rails.
The KSA100 used one such tunnel for each channel, basically because it had to dissipate twice as much power from twice as many devices. I think using a 12" tunnel for each channel with 3 outputs per ledge, giving 12 to3's or to3p's per tunnel will be able to cope. If the extrusion were identical it should have 2 or 3 times the capacity, though the airspeed may need to be higher...Are we really talking about dissipating 600-1000watts per channel?
Steve at ApexJr has 12" tunnel extrusions drilled for a bunch of to3s, for $25usd a pair, about as cheap as you are going to find the solution...
The nominal power rating of the mj15003/4 is 250w, so even applying the normal derating curve still offers huge margin at krells original 40w per device. The plastic transistors are smaller, but not by much at 200w. Increasing the count to lower the per device dissipation to 30w sounds like a sound plan, and as far as I know is being followed by all those people cloning the ksa50...
Ciao
Stuart
I designed the attached circuit to drive the cooling fan in class A amplifier. It sets the speed as necessary, and avoid the noise generated by the full speed fan.
The T1 is the sensor device. I mounted it into the heatsink, with glue in 5mm diameter hole. When the temperature is low, the fan runs on 5V. This is enough for all the fan I ever seen, to run with very low speed and generating no noise.
If the heatsink reach the temperature, where the output of the sensor is lower than the voltage set by the pot, the fan starts to incerase the speed. It reach the maximum speed in 2-3degre.
I used this circuit in many different amplifier and it works very well. Easy to set, and You can find that full speed is not necessary in most of the time.
sajti
The T1 is the sensor device. I mounted it into the heatsink, with glue in 5mm diameter hole. When the temperature is low, the fan runs on 5V. This is enough for all the fan I ever seen, to run with very low speed and generating no noise.
If the heatsink reach the temperature, where the output of the sensor is lower than the voltage set by the pot, the fan starts to incerase the speed. It reach the maximum speed in 2-3degre.
I used this circuit in many different amplifier and it works very well. Easy to set, and You can find that full speed is not necessary in most of the time.
sajti
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.