No.
You need to find the manufacturer's data for a sink that is very similar.
That will give you an Rth s-a, when deltaT is their specified value. Sometimes this is hidden in another ap note. But typically deltaT is from 70 to 80Cdegrees. NOTE !!!!!! this is NOT a temperature which would be written as 75 degreesC (°C).
DeltaT is a temperature difference.
I wrote it out for you twice, but you did not take in that crucial difference, deltaT = Ts-Ta, is not the same as Temperature.
"Ts-Ta heatsink temperature rise (deltaT)"
........................"determine the maximum heatsink deltaT (Ts-Ta)"
.
From that Rth s-a and Power you can predict the heat, but you NEED another bit generally hidden an ap note the derating factor when DetltaT is =75Cdegrees.
No.
Do the sink to ambient first.
Then do the case to sink taking account of the thermal resistance of the electrical insulator.
Completely wrong conclusion.
Last edited:
Let me try one more time. first define the terms:
Ta= ambient temperature and let it be 25deg C.
Ts= temperature of the heatsink
Tc= temperature of the case of the transistor.
Lets assume the Rth from transistor casing through the Silpad to the heatsink is 1deg/W.
Important point. The chassis is made in China. I don't even want to try to ask whether it's 200W. Just take it FIRST on face value. Let's assume the heatsink is spec. for 200W at 100deg C. So Ts-Ta=70deg C.
So the rise in temp with 200W heat is 70deg C. therefore it is 2.86W/deg C. That means thermal resistance. 0.35deg/W. But I am going to use 0.4deg/W to give it some margin assume the Chinese lie about 200W.
If I run current so I generate 75W of heat. AND assume Ta=25deg. Ts-Ta=75W X 0.4deg/W=30deg C. So the final temperature of the heatsink is Ta+30deg=55deg C. or Ts=55deg C.
from the Silpad spec, Lets call it from Ts to Tc is 1deg/W. I have 10 transistors(5 pairs) to dissipate 75W. each transistor dissipate 7.5W. So Tc-Ts=7.5deg C. or Tc= 47.5+7.5=62.5deg C.
I think I am correct. Please read first. I don't think casing of 62.5deg C is pushing. If you think it's not correct, please explain.
Ta= ambient temperature and let it be 25deg C.
Ts= temperature of the heatsink
Tc= temperature of the case of the transistor.
Lets assume the Rth from transistor casing through the Silpad to the heatsink is 1deg/W.
Important point. The chassis is made in China. I don't even want to try to ask whether it's 200W. Just take it FIRST on face value. Let's assume the heatsink is spec. for 200W at 100deg C. So Ts-Ta=70deg C.
So the rise in temp with 200W heat is 70deg C. therefore it is 2.86W/deg C. That means thermal resistance. 0.35deg/W. But I am going to use 0.4deg/W to give it some margin assume the Chinese lie about 200W.
If I run current so I generate 75W of heat. AND assume Ta=25deg. Ts-Ta=75W X 0.4deg/W=30deg C. So the final temperature of the heatsink is Ta+30deg=55deg C. or Ts=55deg C.
from the Silpad spec, Lets call it from Ts to Tc is 1deg/W. I have 10 transistors(5 pairs) to dissipate 75W. each transistor dissipate 7.5W. So Tc-Ts=7.5deg C. or Tc= 47.5+7.5=62.5deg C.
I think I am correct. Please read first. I don't think casing of 62.5deg C is pushing. If you think it's not correct, please explain.
Last edited:
I just about spewed on my keyboard, when I saw your other thread , that an "RF guy" asking about coax. LOL
post a picture of the speakers, room and music taste and well tell you.
that will be closer than any of your calcs
or just build a "lab amp"
I would like to see your 4 ohm Hi-fi speaker
Last edited:
By just assuming things, there's no point in all your calculations.
For a nominal 0.35C/W and 25C ambient, Ts will be closer to 60C. (see graph example)
(there are various tools on the web that will estimate the thermal resistance of the heatsink for you)
I gathered you have JMlab Spectral loudspeakers.
Those have a sensitivity of ~93dB (at 2.83V)
As they're 4 ohm loudspeakers, 90dB for 1W.
105dB sound pressure would require a little over 30W continuous power in 4 ohm.
10W class A power in 4 ohm is sufficient for 100dB sound pressure at 1m distance.
Examples of serial production class A power amps that look similar to your intentions are the Forté Audio models 1 (40W/8, bjt input stage, 4 pairs of output devices) and 1a (nominal 50W/8, JFET input, 5 pairs of A1302/C3281)
Those are biased 50% in class A, the 1.25A bias level of the 1a would make your loudspeakers reach 101dB in class A.
(Your max output power calculations for 4 ohm are somewhat meaningless. Best to expect of an amplifier that does 25W/8 continuous, is ~40W in 4 ohm for 1% THD)
Attachments
Well, if data is not available, I have to assume and derate a little. I am using 0.4deg/W, not 0.35.
How do you read the graph? My calculation is 55deg, somewhat close to 60deg.
Sounds like the Forte Audio model 1a is close to what I am doing. 5 pairs, 1A. I use MJL3281/1302, maybe is the same as 2SC3281/2sC1302 with different casing.
I am still considering lower the transformer voltage to 18V-18V and bias to 300mA per stage. So I'll be running 1.5A idle and get 18W Class A output.
Yes, the JMLab is quite efficient.
0.35C/W times 1.34 (30C), divided by 1.05 (70C).
Makes ~0.45 C/W
For 75W dissipation : 58.5C at 25C ambient.
(you don't want 58.5C, or 60)
You lost me on the graph. What is the graph for?
58deg too high? why? I calculated the case temperature is 62deg.
What is the target max temperature I should shoot for? that will tell me exactly what transformer and what idle current I should use.
Thanks
To derate the thermal resistance of the heatsink for operating temperatures other than the one used by the manufacturer to determine Rth sink-ambient.
In this case, Conrad heatsinks in Australia, that measures heatsinks at 80C above ambient.
(which can be read in the graph, derating factor of 1 for a dT of 80C)
Personally, I try to never go above 50C at 25C ambient.
(with loudspeakers which are low impedance, and require 10 times more class A power than yours)
Last edited:
post67 shows the derating curve which is applied to Rth s-a when deltaT is not equal to the manufacturer's specified value.
i.e. 0.35C/W @ 80C becomes >0.525C/W for 25C deltaT
Now apply the power to predict the sink temperature.
Until you become proficient at this and gather some experience in how it all adds up, it will be easier to revert to an earlier proposal you made
your earlier proposal was for a ClassAB amplifier. You have moved to a full ClassA amplifier and now due to heat concerns you are moving back to a ClassAB amplifier.
Build the ClassAB amplifier. Adjust the bias and measure some temperatures.
That will be the start of you gathering some real data on these unknown heatsinks.
this is where you have gone wrong
looking at the post67graph (other manufacturers could be different and different SHAPES of sink will be different) the de-rating factor for 25C deltaT is >1.4
i.e. 0.35C/W @ 80C becomes >0.525C/W for 25C deltaT
Now apply the power to predict the sink temperature.
Until you become proficient at this and gather some experience in how it all adds up, it will be easier to revert to an earlier proposal you made
your earlier proposal was for a ClassAB amplifier. You have moved to a full ClassA amplifier and now due to heat concerns you are moving back to a ClassAB amplifier.
Build the ClassAB amplifier. Adjust the bias and measure some temperatures.
That will be the start of you gathering some real data on these unknown heatsinks.
Last edited:
Alan,
I do very similar and assume my worst case summer Ta is <=28°C
This gives me a starter value for external heatsinks and I assume an internal Ta of 35°C to 50°C, depending on other internal dissipations.
If the sink is @50°C when Ta=25°C, then when Ta rises to 35°C, Ts becomes ~60°C
Tc will be slightly above, say 65°C, to allow for Rth c-s.
Now add on a little for tolerances. I allow 5C for total tolerances giving in this example worst case Tc of 70°C
de-rate the transistor using the standard manufacturers' formula:
DF = {150C-Tc} / {150C-25C} which for 70°C gives DF ~ 0.64
A 150W device becomes an effective 96W device during summer use.
Euvl ? did a thorough analysis in Linear Audio.
There are many, not as thorough, in this Forum.
Many heatsink datasheets show a similar graph to post67. The axes may be different, but they allow one to predict the actual temperature when sink and power are the two knowns.
There are many, not as thorough, in this Forum.
Many heatsink datasheets show a similar graph to post67. The axes may be different, but they allow one to predict the actual temperature when sink and power are the two knowns.
To actually calculate the thermal resistance correction factor for varying temperature, you'd have to take an advanced thermodynamics class.
During my years at university, MSc EE students were not even required to do Thermo 101.
But bottomline it's just a correction factor, all one has to do is pick a delta T and multiply the nominal Rth of the heatsink by the Cf.
Various manufacturers have a Cf table in their technical paper, but heatsink manufacturers are free to pick a temperature above ambient for Rth measurements.
Some choose 60C above ambient, while Aussies pick 80C.
Aavid Thermalloy uses 75C : - heatsink design tools
(you can find a graph for this table at Infineon)
Step 1 in your case is to figure out the nominal number for the heatsinks of your amp case.
Rod Elliott e.g. has an Excel file on his homepage to get an approximation.
Or you can determine an accurate Rth graph of your heatsinks by clamping a (metal case) power resistor to them, and running a current through it with help of a lab power supply.
You guys could have told me so I don't have to go through all the calculation.
Well, in that case, I have pictures in post #45. Please tell me what is the max power dissipation I can have at idle and I just work back to get the transformer and bias current. 65W?
Well, in that case, I have pictures in post #45. Please tell me what is the max power dissipation I can have at idle and I just work back to get the transformer and bias current. 65W?
Can you explain why you X1.34 and then divided by 1.05. I have no idea why you do that.
Why times 1.34 for 30deg C? Are you assuming ambient temp?
You assumed a temperature rise of 70C for your heatsink in your posts (200W dissipation)
At 80C above ambient, your heatsink will have a lower Rth number than for 70C.(the heatsink becomes more efficient)
According to the graph, it differs by 5%.
If your heatsink has an Rth of 0.35C/W at 70C above ambient, it will be ~0.333 C/W (0.35/1.05) at 80C above ambient.
For 30C above ambient, the correction factor is 1.34
That number is in comparison to the heatsink measurement at a delta T of 80C.
So : 1.34 times the 0.333 C/W
At 80C above ambient, your heatsink will have a lower Rth number than for 70C.(the heatsink becomes more efficient)
According to the graph, it differs by 5%.
If your heatsink has an Rth of 0.35C/W at 70C above ambient, it will be ~0.333 C/W (0.35/1.05) at 80C above ambient.
For 30C above ambient, the correction factor is 1.34
That number is in comparison to the heatsink measurement at a delta T of 80C.
So : 1.34 times the 0.333 C/W
- Status
- Not open for further replies.