You mean equilibrium LM3886 AND heatsink?
I was thinking of how fast the metall in the heatsink closest to LM3886 gets to about the same temperature as the LM3886.
Guess the heats transfers with radius^3 so 1 mm takes 1/1000 of the time of 10mm?
So the temperature changes rapidly early and slowly in the end. At some point the change by the heatsink getting to equilibrium is less than that heatchange made by the music in the lm3886.
Haven't made experiments though.
Yes.
Most calculations I've seen assume that the heat sink is isothermic. That's a pretty good assumption in my observation. Aluminum is after all very thermally conductive.
I have. You can get an idea of the die temperature by monitoring the forward drop of the 3-diode stack in the MUTE circuit.
Tom
THAT is GENIUS!!
i didn't think of that,
excuse me for hijacking this thread for a moment, but do you have a nice mothood of monitoring temperature thatway ?
the lm3886 is used around here within small PA applications in some local discos, and this would be a perfect way for me to add some nice temperature controlled fans to the thing. those darn rack things they like to use are a nghtmare for me. and i prefer to not have fans running full blast anyhow. i know tis not a problem as these amps get the hell driven out of them by idiots, but that would be a nother topic of epic stupidity among "users" and so called "laptop DJs".
That´s genius [2]
care to share how you do it so we don´t reinvent the wheel multiple times?
Just the sensing circuit; actual fan speed or anything else´s use will probably be different in each case.
Thanks
thats genius solution #3 .
but i have question.
if one wants to track heatsink temperature , isnt it better to use good ol thermistor bolted near or on ic than using mute diodes as sensor? because temperature difference?
or to better say same thing , internal temperature is higher from external ,so to have useful fan speed regulation there should be some sort of lag circuit for fan to wait for heatsink to get to target temperature,and wich would make fan spin longer than it should when chip doesnt need more cooling .
but i have question.
if one wants to track heatsink temperature , isnt it better to use good ol thermistor bolted near or on ic than using mute diodes as sensor? because temperature difference?
or to better say same thing , internal temperature is higher from external ,so to have useful fan speed regulation there should be some sort of lag circuit for fan to wait for heatsink to get to target temperature,and wich would make fan spin longer than it should when chip doesnt need more cooling .
actually the heatsink temperature is not important, as counterintuitive it may sound.
but we are interested in not overheating the internals of the ic.
i know there is a relation between heatsink and ic temp (hoooollly cow, who would have thought of that?) but heatsink temps lag behind the internals of the ic.
so if we have the internal temp, and adjust our fans according to that, then the internals o the ic will less likely overshoot the desired temperature.
and the bolted on thermistor, i mean come on.. that would be just far too easy.
bad guys could just shoot the good guys in the movies , yet they HAVE to have sharks with lasers mounted on them to deal with the problem.
otherwise the movie would end 5 secs after starting.
but we are interested in not overheating the internals of the ic.
i know there is a relation between heatsink and ic temp (hoooollly cow, who would have thought of that?) but heatsink temps lag behind the internals of the ic.
so if we have the internal temp, and adjust our fans according to that, then the internals o the ic will less likely overshoot the desired temperature.
and the bolted on thermistor, i mean come on.. that would be just far too easy.
bad guys could just shoot the good guys in the movies , yet they HAVE to have sharks with lasers mounted on them to deal with the problem.
otherwise the movie would end 5 secs after starting.
The information on nonuniform distribution of heat in heatsink with lm3886 package I have from National:
http://www.diyaudio.com/forums/atta...96d1473374772-20-years-ago-today-img_1911.jpg
They said 1.5 is lowest practical thermal resistance of the heatsink. Lower than that the heat distribution gets nonuniform.
And the temperature gradients in aluminium is fast, but how fast. Clearly under 100ms figure 11 shows that the heat disipation is not linear with 1/time (W = J/s) so chip and heatsink can't operate as one unit. (The whole heatsink don't get to 250 C in 100 ms)
Still I do not understand why the time to equlibrium takes 15 minutes in, for example, this case:
Total termal resistance 3 C/W
Total Vcc 50 V
Po max 30 W
8 ohm
5 watts electrical output is about 15 thermal watts (figure 36)
Then the heatsink gets to 45 degrees above ambient.
I really do not understand that it will take 15 minutes to reach 45 degrees above ambient for the whole heatsink to air interface?
http://www.diyaudio.com/forums/atta...96d1473374772-20-years-ago-today-img_1911.jpg
They said 1.5 is lowest practical thermal resistance of the heatsink. Lower than that the heat distribution gets nonuniform.
And the temperature gradients in aluminium is fast, but how fast. Clearly under 100ms figure 11 shows that the heat disipation is not linear with 1/time (W = J/s) so chip and heatsink can't operate as one unit. (The whole heatsink don't get to 250 C in 100 ms)
Still I do not understand why the time to equlibrium takes 15 minutes in, for example, this case:
Total termal resistance 3 C/W
Total Vcc 50 V
Po max 30 W
8 ohm
5 watts electrical output is about 15 thermal watts (figure 36)
Then the heatsink gets to 45 degrees above ambient.
I really do not understand that it will take 15 minutes to reach 45 degrees above ambient for the whole heatsink to air interface?
the bigger the difference is between the heatsink and ambient, the more it will be able to radiate, so heating up the heatsin takes more and more power.
you have to make up with the heat dissipated by the stuff, the excess is left there to raise the temps more. and as it gets hotter, the slower will it be able to rise its own temperature.
so yess, it can take that long with ease.
you have to make up with the heat dissipated by the stuff, the excess is left there to raise the temps more. and as it gets hotter, the slower will it be able to rise its own temperature.
so yess, it can take that long with ease.
i see where you are comming from, but even so , if nothing else than the "cool factor" it self would worth to read chip internal temperatures.
let alone make swiftly acting measures to combat thermal runaways.
who knows, i might have to make a nother amplifier for a nother retard who without thinking twice, will instantly fire it up full blast into a load that is on the real limits of the chip it self.
let alone make swiftly acting measures to combat thermal runaways.
who knows, i might have to make a nother amplifier for a nother retard who without thinking twice, will instantly fire it up full blast into a load that is on the real limits of the chip it self.
Thank you. Yeah, I thought it was pretty neat.
Sure. Attach your Audio Precision APx525 input, DC coupled, to the mute pin. Use the recorder mode to track voltage vs time.
I considered something similar. The tempco of the PN junctions is determined by the semiconductor process (doping concentrations, etc.). It should be pretty reliable from part to part. Sadly, the absolute value of the voltage varies a lot from part to part, so to get an accurate voltage-to-temperature translation, you'll have to calibrate the system at a known temperature. You'll also find that it varies with supply voltage and a slew of other parameters. In addition, it is not a parameter that is controlled for in the fabrication (at least not directly), so it isn't the most reliable temperature sensor in a production environment.
For a one-off measurement of the die temperature, it can be made to work. If you're willing to perform some sort of calibration, you might be able to make it work in a production environment as well.
Tom
these are usually just "dumb poweramps" witch means a basic configuration without any circuitboards or controlls. to p2p wiring, quick and dirty job. i know its not the best, but money is short, time is short, and idiots blow up everything anyways so makes no difference.
however, i don't mind doing a measurement for every one of them i make, since i don't make more than 2-5 a year. they do the job well enough.
so how would you hyjack the mute pin voltage drift so it can be used ?
however, i don't mind doing a measurement for every one of them i make, since i don't make more than 2-5 a year. they do the job well enough.
so how would you hyjack the mute pin voltage drift so it can be used ?
heat the heatsink
You can heat the heatsink by a dissipating resistor with maximum IC 's dissipation, regulated by a thermo-switch. See picture bellow as well as new thread ,hybrid lm3886+triode ,Andrea, a democratic hybrid amplifier for luxury sound.http://www.diyaudio.com/forums/tubes-valves/297206-andrea-democratic-hybrid-amp-luxury-sound.html
You can heat the heatsink by a dissipating resistor with maximum IC 's dissipation, regulated by a thermo-switch. See picture bellow as well as new thread ,hybrid lm3886+triode ,Andrea, a democratic hybrid amplifier for luxury sound.http://www.diyaudio.com/forums/tubes-valves/297206-andrea-democratic-hybrid-amp-luxury-sound.html
Attachments
- Status
- Not open for further replies.