In another thread, the question of cranking up the bias of an amplifier
to the absolute maximum has come up. As this is a tutorial opportunity,
I have started a new thread:
Attached is the portion of the IRF244 datasheet which lists the maximum
ratings. We see that the dissipation figure is 125 watts at 25 deg C, and
you must derate that by 1 watt per degree above that temperature.
The maximum dissipation calculation for this is predicated on the notion
that the maximum junction temperature is 150 deg C. This is pretty
universal for Silicon transistors.
The formula is:
Tj = Watts X (Rj + Rc + Rh) + Ta , where:
Tj = the temperature of the junction (deg C)
Ta = ambient temperature
Rj = thermal resistance of junction to transistor case (deg/watt)
Rc = thermal resistance of transistor case to heat sink
Rh = thermal resistance of heat sink to ambient (usually 25 deg)
Keep in mind that Rh is the amount of heat sink allocated to one transistor,
so if there are 2 transistors on the sink, the Rh figure is doubled.
From the 125 watt rating, we can infer that Rj = 1 deg/watt, and from
experience we know that the Rc value is also about 1 deg/watt.
If we simply know the temperature of the heat sink or the case, the
calculation is more simple:
Tj = Watts X (Rj + Rc ) + Th (Th is temperature of heat sink)
or
Tj = Watts X (Rj) + Tc (Tc is temperature of transistor case)
Let's take the case of a 65 deg heat sink. If the part is dissipating 50 watts,
then
Tj = 50 X (Rj + Rc) + 65
Tj = 50 X (1 + 1) + 65
Tj = 165
This a lot more than the manufacturer thinks is reliable. The rule of thumb
is that the lifespan doubles for every decrease by 10 deg C of the junction.
And contrariwise.
Lets say an Aleph 2 has 12 devices dissipating about 200 watts, or about
16 watts each. The temperature on the heat sink at the devices is about
60 deg C. This means that the Rh (experienced by each transistor) is
(60 - 25) / 16 or about 2 deg/watt
The maximum dissipation per device here can be estimated by
150 = W * (1 + 1 + 2) + 25
125 = W * 4
w = 125 / 4
w = 31 watts.
In our Aleph 2 example
Tj = 16 * (1 + 1 + 2) -25
Tj = 116 deg C.
And this is why Aleph 2's don't break.
😎
to the absolute maximum has come up. As this is a tutorial opportunity,
I have started a new thread:
Attached is the portion of the IRF244 datasheet which lists the maximum
ratings. We see that the dissipation figure is 125 watts at 25 deg C, and
you must derate that by 1 watt per degree above that temperature.
The maximum dissipation calculation for this is predicated on the notion
that the maximum junction temperature is 150 deg C. This is pretty
universal for Silicon transistors.
The formula is:
Tj = Watts X (Rj + Rc + Rh) + Ta , where:
Tj = the temperature of the junction (deg C)
Ta = ambient temperature
Rj = thermal resistance of junction to transistor case (deg/watt)
Rc = thermal resistance of transistor case to heat sink
Rh = thermal resistance of heat sink to ambient (usually 25 deg)
Keep in mind that Rh is the amount of heat sink allocated to one transistor,
so if there are 2 transistors on the sink, the Rh figure is doubled.
From the 125 watt rating, we can infer that Rj = 1 deg/watt, and from
experience we know that the Rc value is also about 1 deg/watt.
If we simply know the temperature of the heat sink or the case, the
calculation is more simple:
Tj = Watts X (Rj + Rc ) + Th (Th is temperature of heat sink)
or
Tj = Watts X (Rj) + Tc (Tc is temperature of transistor case)
Let's take the case of a 65 deg heat sink. If the part is dissipating 50 watts,
then
Tj = 50 X (Rj + Rc) + 65
Tj = 50 X (1 + 1) + 65
Tj = 165
This a lot more than the manufacturer thinks is reliable. The rule of thumb
is that the lifespan doubles for every decrease by 10 deg C of the junction.
And contrariwise.
Lets say an Aleph 2 has 12 devices dissipating about 200 watts, or about
16 watts each. The temperature on the heat sink at the devices is about
60 deg C. This means that the Rh (experienced by each transistor) is
(60 - 25) / 16 or about 2 deg/watt
The maximum dissipation per device here can be estimated by
150 = W * (1 + 1 + 2) + 25
125 = W * 4
w = 125 / 4
w = 31 watts.
In our Aleph 2 example
Tj = 16 * (1 + 1 + 2) -25
Tj = 116 deg C.
And this is why Aleph 2's don't break.
😎
Attachments
Actually, what's crazier?
You crank up your amp bias a little bit, or
you are traveling with your car, you know it though, the brake is wrong?
You crank up your amp bias a little bit, or
you are traveling with your car, you know it though, the brake is wrong?
Killer quote!!!!
let's not talk numbers, but actual failures.. Like I said im dumb, well sorta, for some, we just need to push it and find out! Crank it up till it blows up and back it off was great!!
let's not talk numbers, but actual failures.. Like I said im dumb, well sorta, for some, we just need to push it and find out! Crank it up till it blows up and back it off was great!!
Burn it?
maybe im a crazy diyer, the articile was appricated, thank you np. Let's go where no man has gone before. Sometimes the only way to find out is try.. I like the smoke thing, rock ON!
maybe im a crazy diyer, the articile was appricated, thank you np. Let's go where no man has gone before. Sometimes the only way to find out is try.. I like the smoke thing, rock ON!
No, the latter is not insane, but rather irresponsible.
But this crazy world forced to do so.
I think I have finally will learn to say no.
"This a lot more than the manufacturer thinks is reliable. The rule of thumb
is that the lifespan doubles for every decrease by 10 deg C of the junction."
sounds like what a tranny cooler would do for a th 350 gm. If the amp will benifit from more bias why not try, not much to loose... Exept the power supply...
is that the lifespan doubles for every decrease by 10 deg C of the junction."
sounds like what a tranny cooler would do for a th 350 gm. If the amp will benifit from more bias why not try, not much to loose... Exept the power supply...
Awesome! So we can project maximum useable bias from manufacturer's data and heatsink temp. Is it reasonable to use "1" to represent Rc? Won't our "Mica and Goop" influence this?
I remember reading Gray's thread about his water cooled aleph a while back. In it, he mentioned that he 'had trouble' with the outputs biasing properly. He figured that this was due to the outputs being 'overcooled'. Lets say I can keep a water cooled plate at a consistent 10C, that would give me Tj of 110 for 50 watt dissipation. I guess the question is, should the bias per device be set on a particular junction temp?
That is the Mica and goop. Actually, I think the real figure is
abut 1.1, and the Silicon pads a bit higher.
😎
Doesn't this put the aleph 3 on the edge? By numbers I mean. There has to be bigger fet's that can take a lot more, but that topic might be over my head right now.
The Aleph 3 is about 25 watts per device, and runs slightly cooler than
the Aleph 3 (more effective sink per device).
It doesn't break either, though.
😎
the Aleph 3 (more effective sink per device).
It doesn't break either, though.
😎
Mine hasn't broken in 12 years, running pretty hot. I accidentally left it on for a weekend idling, my room was warm... I can put more bias and cool it better, even bring the temp down some. I still think that there might be an application for soft start bias.
I am sorry, but it is late and I am lost.
Tj = W * (Rj + Rc + Rh) + Ta
Tj = 16 * (1 + 1 + 2) + 25
Tj = 16 * 4 + 25
Tj = 64 + 25
Tj = 89
Where did 116 come from?
Thanks for a great post. A nice equation makes things so much easier to understand!
- Status
- Not open for further replies.