Morello said:
It would depend on the application.
It would depend on the max Tj of the device.
For hifi audio, with TO3 or TO247, 150 C is Bad.
For non-audio and a device with max Tj 200 C, it would be OK.
Myself try to hold a 200 max device below 130 C for AUDIO
this is calculated with safety-margin included (worst case).
I use nothing lower than 40 C as room temperature.
As I, different from Nelson, not is in ULTRA-FI,
but only in some sort of HIGH-FI, this will do.
"It's all about the money"/ Meja
Every 10 C you lower the Tj, will cost you money, more heatsinks/output devices.
Temperature of chip= Tj
Wallet thickness = Wth
The coefficient (or is it a function ???) can then be described as Tj/Wth
and
the lower Tj - the higher Wth
😀 😀 😀 😕
Halo halojoy
What is the best temperature if I am poor?
What is the best temperature if I am rich?
Your simple and easy advice would much appreciate.
JH
What is the best temperature if I am poor?
What is the best temperature if I am rich?
Your simple and easy advice would much appreciate.
JH
Despite the heatsink topic being 'beat to death' I see no one putting it in simple layperson terms. Here's my novice-level misperceptions on what I need for my Aleph 5. The Aleph 5 dissipates 300w total. So, I need dissipation of 150W per side. Starting at 20C and according to NP I'm not exceeding 50C. That leaves me with a difference of 30C. So, 30 divided by 150 leaves a Rth of .2. Now Aavid Thermalloy has a great little utility that allows you to adjust the length of any of their extrusions and it recalcutates the Rth. As a very simple example of what I think I could use on each channel of my Aleph 5 you can look at the following:
Rth of a 14 inch section of 68290
The only thing I'm left in doubt about now is how to add the Rth's of multiple smaller sections.
Comments?
Suggestions?
Corrections?
Regards,
Dan
Rth of a 14 inch section of 68290
The only thing I'm left in doubt about now is how to add the Rth's of multiple smaller sections.
Comments?
Suggestions?
Corrections?
Regards,
Dan
say you have 1.0 C/W sections at home.
and you want a combined section that has 0.2
If all sections are the same
the combined section's Rth is 1.0/number of sections
so in this case you need 5 sections 1.0/5= 0.2
If you have two 0.6 sections
the combined will be half. That is 0.3, as 0.6/2= 0.3
2 sectors are twice as effective coolers
3 sectors are tree times as effective
so they have two and three times lower Rth, than one.
It is like parallelling R, resistors,
Two 10 Kohm, will lead twice as much current,
so the combined resistens is 5 kOhm, half of one resistor
and you want a combined section that has 0.2
If all sections are the same
the combined section's Rth is 1.0/number of sections
so in this case you need 5 sections 1.0/5= 0.2
If you have two 0.6 sections
the combined will be half. That is 0.3, as 0.6/2= 0.3
2 sectors are twice as effective coolers
3 sectors are tree times as effective
so they have two and three times lower Rth, than one.
It is like parallelling R, resistors,
Two 10 Kohm, will lead twice as much current,
so the combined resistens is 5 kOhm, half of one resistor
One thing people are ignoring in this thread is the placement of devices on the heatsink. I've seen pictures on this very website where four or six TO-247 packages are mounted right in the center of a huge heatsink, with lots of metal jutting out on every side. Unfortunately these people are not getting the heat handling abilities they paid for, because the center of that sink is going to be pretty toasty, and the fringes rather less so. It is important to spread the heat load around on the sink, paying attention to the effect of edges and so forth.
that is very correct jwb!
Also when you use several sections connected together
some of the heat dissapating areas are blocked
by the section nearby.
So my the combined section's Rth is somewhat higher
than plainly Rth/number of sections.
This is why the datasheets of a heatsinsk
doesn't show a straight linear falling line
in the diagram. Those C/W - lenght in millimeter sheets.
One could also say that the [surface area]/volume gets lower
for long profiles. mm^2/mm^3
Worst use of aluminium would be a bowl (ball) of the material.
It is the shape with the smallest area/volume relationship.
On the other hand a bowl is the most effective shape of an animal
to keep from losing body-temperatur.
So in cold environment an animal like a ball would need less food to survive. Because of least energy losses.
Ideal heat dissapator would be a material with very low heat-transporting coefficient. With a very large area placed very near the spot where the heat comes in.
This is what jwb pointed out.
Also if we could place the heatsink outside the cabinet,
say with 10-20mm room between box and heatsink, this would
allow that side of heatsink to be cooled by circulating airflow.
This have a negative aspect.
An open space could eventually allow something to short circuit.
Like a screw driver coming into contact with leads of transistor.
Also when you use several sections connected together
some of the heat dissapating areas are blocked
by the section nearby.
So my the combined section's Rth is somewhat higher
than plainly Rth/number of sections.
This is why the datasheets of a heatsinsk
doesn't show a straight linear falling line
in the diagram. Those C/W - lenght in millimeter sheets.
One could also say that the [surface area]/volume gets lower
for long profiles. mm^2/mm^3
Worst use of aluminium would be a bowl (ball) of the material.
It is the shape with the smallest area/volume relationship.
On the other hand a bowl is the most effective shape of an animal
to keep from losing body-temperatur.
So in cold environment an animal like a ball would need less food to survive. Because of least energy losses.
Ideal heat dissapator would be a material with very low heat-transporting coefficient. With a very large area placed very near the spot where the heat comes in.
This is what jwb pointed out.
Also if we could place the heatsink outside the cabinet,
say with 10-20mm room between box and heatsink, this would
allow that side of heatsink to be cooled by circulating airflow.
This have a negative aspect.
An open space could eventually allow something to short circuit.
Like a screw driver coming into contact with leads of transistor.
jwb,
at www.fischerelektronik.de (heatsinks) they explain how the Rth is measured. They use one (1) device placed in the center of the heat sink to bring the heat into the sink.
This means that you can actually improve the given Rth by spreading more devices over the heatsink.
Looking at the figures for Rth from different manufacturers of (almost) the same heatsinks I think everybody measures this in a different way.
william
at www.fischerelektronik.de (heatsinks) they explain how the Rth is measured. They use one (1) device placed in the center of the heat sink to bring the heat into the sink.
This means that you can actually improve the given Rth by spreading more devices over the heatsink.
Looking at the figures for Rth from different manufacturers of (almost) the same heatsinks I think everybody measures this in a different way.
william
I have a question.
We use sometimes L-shaped long bars to mount the transistors
onto the heatsink. That bar can be as long as the heatsink.
What will happen, if we put the insulting material
between the bar and the heatsink. That is a very large insulator area.
It is like making the body of the transistor very large.
Is there any advantages or negatives?
We use sometimes L-shaped long bars to mount the transistors
onto the heatsink. That bar can be as long as the heatsink.
What will happen, if we put the insulting material
between the bar and the heatsink. That is a very large insulator area.
It is like making the body of the transistor very large.
Is there any advantages or negatives?
halojoy said:
I beg to differ with the above theory because the Aavid Thermalloy website forces you to quadruple the length of any given extrusion to halve the Rth.
Regards,
Dan
The reason for the calculations given by the Aavid site is that they assume a single point source heat radiator in the middle of the heatsink. The longer you make the sink, the less effective the periphery of it becomes at radiating heat due to the thermal resistance and lack of infinite conductivity of the aluminum.
If you mount a bank of 8 semiconductors spaced equidistant from each other on an 8" sink, the effective Rth of the heatsink will be 1/8 that of a 1" piece. (for all practical purposes, each semiconductor can be thought of having its own 1" heatsink.
Consider a hypothetical single heatsink 1" long with a point source radiating 10W into the middle of the sink. Let's say it has a hypothetical temperature rise of 10 deg C for a thermal resistance of 1 deg C per watt. if you take 8 of these and bind them together with some threaded rod to make one large aggregate heatsink, it will still only rise 10 degrees C, but will be dissipating 80W, for an effective thermal resistance of 0.125 deg c/w. (so 1/8 that of a single one)
The lesson learned from all this of course is to always space out your power semiconductors as much as you can across the heatsink to minimize the temperature gradient between the heat sources and the edges. This becomes more important with higher power dissipations as well (like with class A amps.)
If you mount a bank of 8 semiconductors spaced equidistant from each other on an 8" sink, the effective Rth of the heatsink will be 1/8 that of a 1" piece. (for all practical purposes, each semiconductor can be thought of having its own 1" heatsink.
Consider a hypothetical single heatsink 1" long with a point source radiating 10W into the middle of the sink. Let's say it has a hypothetical temperature rise of 10 deg C for a thermal resistance of 1 deg C per watt. if you take 8 of these and bind them together with some threaded rod to make one large aggregate heatsink, it will still only rise 10 degrees C, but will be dissipating 80W, for an effective thermal resistance of 0.125 deg c/w. (so 1/8 that of a single one)
The lesson learned from all this of course is to always space out your power semiconductors as much as you can across the heatsink to minimize the temperature gradient between the heat sources and the edges. This becomes more important with higher power dissipations as well (like with class A amps.)
halojoy said:
Well, if the mounting tabs of the transistors are all at the same potential and there's no need for them to be isolated from each other, this is a very effective way of reducing the thermal resistance between the transistors and the heatsink.
Becuase of the larger surface area, you'd be much better off with a large insulator between the L-shaped extrusion and the heatsink. As said, this only serves to isolate the heatsink from the transistors, not the transistors from each other. If there isn't too much of a DC potential on the transistor mounting tabs, you'd be better off using nylon spacers/hardware to isolate the heatsink from the chassis and/or other heatsinks at different potential (say the two rails of a power amplifier) and ignore the insulator altogether, as long as the voltage isn't high enough to harm a stray finger. (although if your dog decides to lick the heatsink, he probably won't be too terribly pleased with the result. 🙂
After reading all of above, I just want to know the following, for my own peace of mind: (This is on Aleph30)
1.) Temperature on FET metal tab - IRFP250 - is 90°C, temp. next to FET on heatsink is 70°C, can I get away with this or is it in danger zone?
2.)What is exceptable temp. on rectifier bridge, or just next to it on heatsink?
3.)What is excaptable temp. on body of Q1, Q2, Q3?
4.) I fitted a thermistor on primary line after fuse, body temp.= 57°C, is this O.K.?
Thanx for help.
1.) Temperature on FET metal tab - IRFP250 - is 90°C, temp. next to FET on heatsink is 70°C, can I get away with this or is it in danger zone?
2.)What is exceptable temp. on rectifier bridge, or just next to it on heatsink?
3.)What is excaptable temp. on body of Q1, Q2, Q3?
4.) I fitted a thermistor on primary line after fuse, body temp.= 57°C, is this O.K.?
Thanx for help.
Some practical thinkings.
Yes, you are right.
If they are joined together.
And that is what they mostly are.
But 4 sections, separated from each other
will have a combined Rth of Rth/4,
as long as the heat dissapated from one section
does not come in contact with another section.
This is theory. In real practice we do not build amplifers
where heatsinks are spread out over a big area.
And in real practice, the sides of the heatsinks
that are on the inside of the amplifier
does not contribute much to the cooling.
dantwomey said:
Yes, you are right.
If they are joined together.
And that is what they mostly are.
But 4 sections, separated from each other
will have a combined Rth of Rth/4,
as long as the heat dissapated from one section
does not come in contact with another section.
This is theory. In real practice we do not build amplifers
where heatsinks are spread out over a big area.
And in real practice, the sides of the heatsinks
that are on the inside of the amplifier
does not contribute much to the cooling.
Opinions please. .
Any opinions on these questions - any one? Please..
Thanx for help.
JDeV said:
Any opinions on these questions - any one? Please..
Thanx for help.
They are OK but maybe just a little on the high side. If the temp on the heat sink a little distance away from the FET is 60 degrees or less you are hot but within limits IMO. If you are at 50 degrees in that spot then you are just fine.
/UrSv
/UrSv
Peace of mind is not bestowable
It's different with each person.
If you want peace of mind, I suggest that you buy spare FETs.
If you still have trouble sleeping, get bigger heatsinks (you did say you read that stuff, right?).
As for what temperature is right for Q1, Q2, Q3 in an Aleph 30, that's like asking "what is the price of peas in Paris?" (maybe Nar knows the answer😉 ) Have you read the manufacturer's data sheets and tried to evaluate it?
Ok, I'm being a little tough on you, but this is engineering, where not everything has a definite answer. Typically, in engineering, if we can't evaluate something, or don't feel like making the effort, we...
1) Try something and see if we're satisfied with the results.
OR
2) Make a conscious effort to over-design, so we can sleep at night.
OR
3) Look at what others have done successfully (go back and read that stuff again; do a search for all posts by Nelson Pass with the word "temperature" in them). Then emulate them.
So, try one of those approaches. Then ask us some easier questions (that we might know the answers to).
Good Luck / sleep well
It's different with each person.
If you want peace of mind, I suggest that you buy spare FETs.
If you still have trouble sleeping, get bigger heatsinks (you did say you read that stuff, right?).
As for what temperature is right for Q1, Q2, Q3 in an Aleph 30, that's like asking "what is the price of peas in Paris?" (maybe Nar knows the answer😉 ) Have you read the manufacturer's data sheets and tried to evaluate it?
Ok, I'm being a little tough on you, but this is engineering, where not everything has a definite answer. Typically, in engineering, if we can't evaluate something, or don't feel like making the effort, we...
1) Try something and see if we're satisfied with the results.
OR
2) Make a conscious effort to over-design, so we can sleep at night.
OR
3) Look at what others have done successfully (go back and read that stuff again; do a search for all posts by Nelson Pass with the word "temperature" in them). Then emulate them.
So, try one of those approaches. Then ask us some easier questions (that we might know the answers to).
Good Luck / sleep well
Thanx Vince
Now that will give me peace of mind or was that a "piece of your mind." 😉 Big temperatures could be normal in alot of cases but could also indicate big problems.
I hoped someone took these measurements and just wanted to compare my values - with the idea to save on those experimenting costs.
But I suppose by experimenting 1 will learn 1st hand - at a cost most propably.
Easy question for Vince: What is the price of peas in Philadelphia? 😀 (Don't attack - only kidding.)
Now that will give me peace of mind or was that a "piece of your mind." 😉 Big temperatures could be normal in alot of cases but could also indicate big problems.
I hoped someone took these measurements and just wanted to compare my values - with the idea to save on those experimenting costs.
But I suppose by experimenting 1 will learn 1st hand - at a cost most propably. Easy question for Vince: What is the price of peas in Philadelphia? 😀 (Don't attack - only kidding.)
You're pushing, buddy!
You're a good sport, JDeV!
So, without being an expert (I haven't built this amp) I'll give you some more opinions.
I think your FETs are hot. I looked at your amp in the other thread, and it looks nice. Heatsinks look like they might be big enough to me. If you haven't done so, use heatsink grease between any pieces of metal where they are joined. I can't quite make out how your devices are mounted, but I think you have several pieces of metal there.
They might last a long time, I don't really know. I am basically going on NP's say-so that he likes to run them with the sinks at 50C for commercial products, and he seems comfortable with 60 at home. But if it's too hot to touch, I'd be wary. There are others here who have lots of experience and may think it's ok; I just use NP as my reference on these issues because I know his credentials and that he's a no-nonsense guy.
Do you know what is the surface area of your heatsinks? From my experience with my Zen 4, I think that something about 10in^2/watt is a good number, assuming that the fins are spaced reasonably and the sinks aren't too tall. (I'd say yours fit those parameters) With that sort of heatsink I get a temp rise of about 30 on my amp. It runs at about 50C with an ambient of 20. BTW, I wouldn't include the back-side when figuring their area, and that's 63cm^2/watt.
For the bridge, If you haven't done so, bolt it to the bottom plate of the amp, and it shouldn't get more than "pretty warm" and it will be fine. I wouldn't put it on the heatsink with the FETs. It might actually get hotter there than it would be by itself.
57 on the Thermistor is ok, I would think. It isn't an active device, and that's hot but not crazy. I think mine is in that vicinity. Hot, but not godawful hot. Power resistors can run hotter than that, and it's basically a resistor.
As for acceptable temperature on Q1,2,3: That's the input differential pair, and it's current source. The current through Q1 should be 25ma according to the manual, and the voltage across it should be about 15V. That's less than half a watt dissipated. I don't think it should even need a heatsink, or maybe just a very small one on the board. Similar for the diff pair, but maybe the voltage across them is bigger. You haven't put them on the heatsinks with the output devices, have you? They should have their own little heatsink on the ckt board, and the temps should be low as a result. Look at what Nelson uses here on this page: http://www.diyaudio.com/forums/show...748&perpage=15&highlight=camera&pagenumber=39
IRF's website has the data sheets for everything, and you can see the max junction temperatures there. Probably should shoot to be nicely below them. eg. If they say max 150, I'd like to not go above 100 (or so). They could be figured knowing how much power you are dissipating and the temperature of the package. Tj=Tcase + (Rjc x Pd). Pd is how much power the device is dissipating.
I've just now gone back and looked yet again at your heatsinks. One channel's worth may be about the same size as the sinks I'm using for my Zen 4 (~800 in^2/channel), but they are dissipating somewhere about 50% more heat. So, instead of a 30 degree rise like I get, you might get 45. Add that to 27, your ambient, and we get 72. You say that you get 70 on the sink next to the device. Makes sense.
The price of peas in Philly?
Who knows?
I did DIY peas last summer.
I'm sure that I didn't save any money, but they were tasty.
Oh, yeah. One more thing- How about trying to take those covers off the heatsinks? You might get a few degrees to the good. Make you sleep better
You're a good sport, JDeV!
So, without being an expert (I haven't built this amp) I'll give you some more opinions.
I think your FETs are hot. I looked at your amp in the other thread, and it looks nice. Heatsinks look like they might be big enough to me. If you haven't done so, use heatsink grease between any pieces of metal where they are joined. I can't quite make out how your devices are mounted, but I think you have several pieces of metal there.
They might last a long time, I don't really know. I am basically going on NP's say-so that he likes to run them with the sinks at 50C for commercial products, and he seems comfortable with 60 at home. But if it's too hot to touch, I'd be wary. There are others here who have lots of experience and may think it's ok; I just use NP as my reference on these issues because I know his credentials and that he's a no-nonsense guy.
Do you know what is the surface area of your heatsinks? From my experience with my Zen 4, I think that something about 10in^2/watt is a good number, assuming that the fins are spaced reasonably and the sinks aren't too tall. (I'd say yours fit those parameters) With that sort of heatsink I get a temp rise of about 30 on my amp. It runs at about 50C with an ambient of 20. BTW, I wouldn't include the back-side when figuring their area, and that's 63cm^2/watt.
For the bridge, If you haven't done so, bolt it to the bottom plate of the amp, and it shouldn't get more than "pretty warm" and it will be fine. I wouldn't put it on the heatsink with the FETs. It might actually get hotter there than it would be by itself.
57 on the Thermistor is ok, I would think. It isn't an active device, and that's hot but not crazy. I think mine is in that vicinity. Hot, but not godawful hot. Power resistors can run hotter than that, and it's basically a resistor.
As for acceptable temperature on Q1,2,3: That's the input differential pair, and it's current source. The current through Q1 should be 25ma according to the manual, and the voltage across it should be about 15V. That's less than half a watt dissipated. I don't think it should even need a heatsink, or maybe just a very small one on the board. Similar for the diff pair, but maybe the voltage across them is bigger. You haven't put them on the heatsinks with the output devices, have you? They should have their own little heatsink on the ckt board, and the temps should be low as a result. Look at what Nelson uses here on this page: http://www.diyaudio.com/forums/show...748&perpage=15&highlight=camera&pagenumber=39
IRF's website has the data sheets for everything, and you can see the max junction temperatures there. Probably should shoot to be nicely below them. eg. If they say max 150, I'd like to not go above 100 (or so). They could be figured knowing how much power you are dissipating and the temperature of the package. Tj=Tcase + (Rjc x Pd). Pd is how much power the device is dissipating.
I've just now gone back and looked yet again at your heatsinks. One channel's worth may be about the same size as the sinks I'm using for my Zen 4 (~800 in^2/channel), but they are dissipating somewhere about 50% more heat. So, instead of a 30 degree rise like I get, you might get 45. Add that to 27, your ambient, and we get 72. You say that you get 70 on the sink next to the device. Makes sense.
The price of peas in Philly?
Who knows?
I did DIY peas last summer.
I'm sure that I didn't save any money, but they were tasty.Oh, yeah. One more thing- How about trying to take those covers off the heatsinks? You might get a few degrees to the good. Make you sleep better
Am sleeping well 😉
Hi Vince
I knew you just pretended to be a tough nasty guy. 🙂
Actually I thought everything over again, did some calculations and decided (after reconcidering your comments) to reconnect the fans, but replacing the existing "out of old PC powersupplie" 1's with the highly aclaimed (and highly )
Papst 8412NGL fans. All temperatures are now down by about 20°C and I really slept well. 😉 BTW - these new fans are not really much an improvement on the noise issue, IMO 🙁 Expected much more for $30 worth of fan.
Thanx again for your time and comments.
Oh yes... My FETs are mounted directly onto heatsink with Silica pads, they are clamped with a Aluminium angle bar - which should help to absorb more heat as well. I also plan to reduce the bias to reduce the heat.
Hi Vince
I knew you just pretended to be a tough nasty guy. 🙂
Actually I thought everything over again, did some calculations and decided (after reconcidering your comments) to reconnect the fans, but replacing the existing "out of old PC powersupplie" 1's with the highly aclaimed (and highly
) Papst 8412NGL fans. All temperatures are now down by about 20°C and I really slept well. 😉 BTW - these new fans are not really much an improvement on the noise issue, IMO 🙁 Expected much more for $30 worth of fan.
Thanx again for your time and comments.
Oh yes... My FETs are mounted directly onto heatsink with Silica pads, they are clamped with a Aluminium angle bar - which should help to absorb more heat as well. I also plan to reduce the bias to reduce the heat.
- Status
- Not open for further replies.