Transistor equivalent

Status
Not open for further replies.
weeefff, what about 3x5 inches heatsink, now I not have enough drink cans and scrap aluminium, I will try to collect much more.
Any formulas to calculate heatsink size?, I not found yet
 
Last edited:
Well use 1 heatsink for 1 transistor, hmm, at least 6 pcs, need above 150 cans



Hi Nikolai,

Make the heatsinks from sheet aluminium, say 2.5mm thick.

Casting ally is not required.

Can you do a drawing of your intended case because, if a certain design, it can be used for a very good heatsink for the output power transistors. That would be best.

spec
 
Instead of guessing about a heatsink, go to www.digikey.com and search for Thermal Heatsinks where they have thousands with dimensions and thermal ratings. But they are not in a logical sequence and entering your dimensions did not find one. So go to the website of one of the many manufacturers listed.
 

Hi audio,

I would not be guessing, but a good suggestion all the same.

That site, I suspect, would not cover the case as heatsink approach, but haven't looked at it yet.

In the meantime could you do the dissipation calcs for each output transistor, both in the 40W and a 20W amp versions.? That would be help because it would save me doing it.

spec
 
Heatsinks and Thermal Budget

(1) Heatsinks

Overall heatsink objectives are:

(1.1) Conduct the heat away from the transistor mounting area by a thermal conducting material (ally or copper but stick with ally for explanation) with a low thermal resistance. This means thick for a sheet of ally.

(1.2) Once the heat has been conducted away from the transistor mounting area, present a large surface area to the air to efficiently transfer the heat to the air by convection. This is normally done by fins. The best fins are vertical because heated air rises, but there must be a free flow for the air through the fins (open top and bottom)

(1.3) The heatsink surface area must be in free flowing air as cold as possible. The transfer of heat from a heat sink to air is proportional to the difference between the heatsink surface area, the temperature of the heatsink, and the temperature of the air.

For example, if the heatsink surface is 60 deg C and the air in the area of the heatsink is also 60 deg C there will be no convection and no cooling of the heatsink. 60 deg C is not unusual inside equipments.

If the air was 70 deg C, again not uncommon in an equipment cabinet, the air would actually heat the heatsink up instead of cooling it!

There is stacks of data on the net and elsewhere about heatsinks and also about making heatsinks

Heatsink are specified by their thermal resistance in deg/CW, to free air This is a measurement of how much heat the heatsink can dissipate into the air A small heatsink would typically be 10 deg/CW and a large one 0.5 deg/C W

**broken link removed**


(2) Thermal Budget

Take an example:

A transistor has a maximum junction temperature of 170 deg C, as shown on the data sheet. That temperature must not be exceeded or the transistor will be destroyed.

The transistor is dissipating 50W. The air temperature in the vicinity of the heat sink is 50 deg C.

Thus, the temperature difference between the transistor maximum junction temperature and the air is:

170- 50 = 120 deg C

The transistor dissipation is 50W so the maximum thermal resistance of the heatsink is 120/50= 2.4 deg C W, a perfectly achievable figure.

It would not be wise to operate the transistor with a maximum junction temperature. You only do that if you absolutely have no option.

A reasonable max junction temp design aim would be 90% of data sheet Tjmax.

So now, Tjmax_design = 0.9 * Tjmax_data_sheet = 0.9 * 170 = 153 deg C. That is now the design maximum junction temperature.

Doing the heatsink calculations again:

Tjmax_design- Tamb = 153- 50 = 103 deg

Now the heatsink thermal resistance is, 103 deg C/50W = 2.06 deg C W, a lower figure you will notice.

For an actual design, the thermal budget calculations are exactly the same priciple, excpt you would also have to take into account the thermal resistances between the junction and case, and case to heatsink. When you include these thermal resistances the situation becomes much more critical.

A typical thermal resistances from junction to case for a high power TO-3 transistor, is around 0.8 deg C W and, using a mica washer between the case and heatsink, the thermal resiatance between the case and heatsink is 1.1 deg C W. Mica has the lowest thermal resistance out of the freely available and low cost group of insulating washers. It is also the most stable and has the best dialectric performance. Aluminium oxide has an even lower thermal resistance but is expensive, while Beryllium oxide is the best, but is very expensive and also toxic, so it is now banned.

Finally, the total thermal resistance between the junction and the air is, 0.8 + 1.1 + heatsink thermal resistance.

I will leave you to work out, if you like, the maximum heatsink thermal resistance that would do the job!




Links
(1) **broken link removed**

 
Last edited:
Most class AB amplifiers are about 60% efficient so a 40W amp would have a total draw from its power supply of 66.6W and heat with 66.6W - 40W= 26.6W, about 12W for each output transistor and about 1.3W for each driver transistor.
But music is not continuously at full blast so I think heatsinks calculated at 8W for each output transistor will be fine.
 
Thanks, you wrote too many information. But sorry I have learnt about heat transfer almost 1 year ago in my physics class and did alot of exercise, I felt tired of it.
After what audioguru said, I still want to make a heatsink calculated at 15W for 40W amp and 8W for 20W amps . Because I don't want use a cooler for heatsink because it noisy and make a lot of dust sticks on the heatsink after short time (good dust filter not cheap), dust also reduce heat transfer. Summer the north of Vietnam very hot so large heatsink will keep transistor not hot even without a cooler.
 
Thanks, you wrote too many information. But sorry I have learnt about heat transfer almost 1 year ago in my physics class and did alot of exercise, I felt tired of it.

not necessay to read- but you did say you were keen to learn! But how are you going to design, build tand test the heat sinking- the tail chasing approach?

After what audioguru said, I still want to make a heatsink calculated at 15W for 40W amp and 8W for 20W amps .

OK, but ag has not done tha calculations for the 20W and 40W amps.


You are confused; I said nothing about a cooler for class AB, quite the reverse. In fact I suggested using the case as a heat sink
 
Last edited:
not necessay to read- but you did say you were keen to learn!



OK



You are confused; I said nothing about a cooler for class AB, quite the reverse. In fact I suggested using the case as aheat sink
Case? It may be made from steel that has poor heat transfer, use seperated alum heatsink will be much better
 

ERRATA
(1) C1 & C9 should read 4m7F (4,7oouF) minimum
(2) 60V label should read 38V
 
Last edited:
Can you change schematics to PCB layouts?
I also want to know what kind of PCB should I use: glass or bakelite or other type? Now I only have bakelite
 
A typical substrate is FR4 glass epoxy, 1 oz copper. I used 2 oz copper for my amp.
1 oz, I think is the weight of a square foot of copper.
With ROHS (Reduction of Hazardous substances), the standard Pad coating is HASL or Hot air solder leveled.
Typical resist color is green. Resist, doesn't let solder stick there and generally covers all but the pads.
The silk screen is usually a component outline and designator to help stuffing the board.
There is also an outline/cutout layer,

I described basically a single-sided board.

Traces have to to have the cross-sectional area to carry the designed currents.

There is some info here: https://en.wikipedia.org/wiki/Printed_circuit_board

Making PCB's is an entirely different subject. Usually you send "Gerber filles, a drill file and some instructions" to the PCB manufacturer. Each manufacturer has a set of "design rules" that you must follow.

Some will accept Eagle. One such "cheap" fabricator is: https://www.itead.cc/open-pcb/pcb-prototyping.html
 
I can't no. I am new to Eagle and don't know how to do it at the moment- much as I would like to.



Fibreglass by a mile
Thank you anyway. I plan to print the schematic into paper, then I will use my "traditional method" to make PCB. draw copper track with permanent makers then etching.
 
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…