Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Heatsink and MOSFET basics and design Help

Status
Not open for further replies.

gmcjetpilot

New Member
I want to switch 60 watts (inductive) both in on-off cycles (flash p=0.666 sec, 50% duty ~
1.5 Hz) and steady. Thanks. I don't want forced air cooling... but want to see what we
can do.


  1. 12-14.1 volt DC system
  2. ASSUME max heatsink Temp 60C.
  3. Case is a small 3.5" x 2.5" x 1.7" (can make case bigger)
  4. Assume ambient T=25C, and Tjunc = (as low as possible <= 120C)
  5. MOSFET (P-channel, because it has to be, please no why N channel)
  6. 60 watt On-Off (flash) and On steady (would like 100 watt but lets start w/ 60W)
  7. The MOSFET P's considered for reference (SEE below)

Heatsink calculation:

Tr = Th - Ta = 60 - 25 = 35°C
Ph = Vh * Ih = 14 * 4.3 = 60W
Rth = Tr / Ph = 35 / 60 = 0.58°C/W (that seems low, forced air needed?)

Going from junction and adding all the R's

Rth(j-a) = 0.5 + 1.1** + 0.58 = 2.2 C/W
T(j-a) = Pd x Rth(j-a) = 60W x 2.2 C/W = 132C temp = that is a little hot but OK.
** metal metal paste or mica+paste, special care taken.

Then WHAT KIND OF HEAT SINK WOULD YOU USE?

I was going to do a DIY copper or alumunum (painted black, except at transistor where
it is polished with paste). Rth should be less than 1.0..... I wanted to keep the heatsink
in the case with some holes. I am seeing to get 0.60 CW heatsink I need an external
heatsink, honking big heat sink. Comments? Ideas.

PROBLEM! I need something small and light... Are my calculations way out to lunch?
Is heatsink'ing a MOSFET with 4.3 amps at 14 volt souce, some big deal? I need a
heatsink, but how big.... thanks.

* Ref transistors (not married to them but T0-220 or through hole is perferred).

SPP80P06P
Continuous Current (Tc=110deg) 64 amp
Power dissipation (TC = 25 °C) 340 W
Thermal resistance, junction - case RthJC = 0.4
(max Junction 175 deg)

STB80PF55/STP80PF55
ID Drain current (continuous) at (Tc = 100°C) 57 amp
PTOT Total dissipation at (TC = 25°C) 300 W
Rthj-case Thermal resistance junction-case max 0.5 °C/W
(max Junction 175C)
 
How long is this unit working for in a 24 Hour period?

To get the lowest heat generation you would need a Mosfet with a VERY low Drain Source on resistance.

Thus less cooling and Aluminium parts necessary. I am out of the warp now....but I know back in the early Nineties, N Channel Mosfets
with very low Drain Source on resistance were available @ a reasonable price. BUZ11 comes to mind. I don't know about P Channel now though.
 
Last edited:
You have a 60 Watt panel. You want to switch on/off 4.3 amps into a 14 (12) volt battery. Using a SPP80P06 MOSFET. RDSON=0.023 ohms.
If watts= I^2xR then the power loss is under 1/2 watt. Why the big heat sink? What do I not understand?
 
How long is this unit working for in a 24 Hour period?

To get the lowest heat generation you would need a Mosfet with a VERY low Source Drain on resistance.

Thus less cooling and Aluminium parts necessary.
Yes "ON" means 24/7. Continuous is continuous, 4 hours or 40 hours, no difference..... Also low ON resistance is good.... I picked the lowest of the low.....

I'm assuming (in my calculations) the transistor has to dissipate the full POWER it is switching, in this example 60 watts. However the resistance is low 0.023 ohm :: take [(4.3 amps)^2] x 0.023 ohm = 0.43 watts; that is easy, verses 60 watts. Doing some calculations based on above I need about ~170 CFM to cool this thing on a 4x4x1.25 inch sink (to get down to 0.6 C/W) ..... That is not unrealistic for my needs.... I think my calculations are off..... Are you familiar with sizing heatsinks or just kind of tossing ideas around?
 
Last edited:
You have a 60 Watt panel. You want to switch on/off 4.3 amps into a 14 (12) volt battery. Using a SPP80P06 MOSFET. RDSON=0.023 ohms.
If watts= I^2xR then the power loss is under 1/2 watt. Why the big heat sink? What do I not understand?

THANK YOU!!!!!! That is what I was wondering.... Is that the proper calculation. As I just wrote tvtech:

....example 60 watts. However the resistance is low 0.023 ohm :: take [(4.3 amps)^2] x 0.023 ohm = 0.43 watts; that is easy, verses 60 watts.

You are saying I only need to dissipate 1/2 watt? That is easy! I don't have to dissipate the full 60 watts.... That makes sense but was not sure. If you are sure that is GREAT! However I have to ask are you sure, because every calculation example I see is using POWER based on the DRAIN load......

Ever see FET's and MOSFET's on a big audio amp of 60 watts..... the heat sinks are large.... (and that is audio not continuous).

That means I need a Rth total of only 40 C/W at worst case for 100 watts light.... (about 2.3 watt dissipated). That will be easy.
 
Last edited:
Its not that easy to switch a Mosfet properly though if in a Linear application. Like what you are trying to do.

I have forgotten what I did....suffice to say you have to raise the Gate Voltage.

And all worked well with minimal losses. It's a long time ago, but the experts here will explain better than I ever can.

Spot of Alzheimer's setting in here. Like my Dad.

Thanks and Cheers

Be good folks
 
Last edited:
....example 60 watts. However the resistance is low 0.023 ohm :: take [(4.3 amps)^2] x 0.023 ohm = 0.43 watts; that is easy, verses 60 watts.

You are saying I only need to dissipate 1/2 watt? That is easy! I don't have to dissipate the full 60 watts.... That makes sense but was not sure. If you are sure that is GREAT! However I have to ask are you sure, because every calculation example I see is using POWER based on the DRAIN load......

Ever see FET's and MOSFET's on a big audio amp of 60 watts..... the heat sinks are large.... (and that is audio not continuous).
You are just switching the power on and off (if I understand correctly), so it's similar to a mechanical switch. There's no power dissipated when it's off and the power dissipated when it is on is just the "on" resistance of the switch times the load current (neglecting the small loss as it's switching).

Audio amps require large heat sinks because they are operating in the linear mode where they act somewhat like variable resistors to control the load (speaker) current. That, of course, dissipates much more power.
 
You are just switching the power on and off (if I understand correctly), so it's similar to a mechanical switch. There's no power dissipated when it's off and the power dissipated when it is on is just the "on" resistance of the switch times the load current (neglecting the small loss as it's switching).

Audio amps require large heat sinks because they are operating in the linear mode where they act somewhat like variable resistors to control the load (speaker) current. That, of course, dissipates much more power.

You lost me.... ON or OFF..... continuous.... Obviously OFF is not the issue. That is a no brainer.

ON it has to pass say 4.2 amps (60 watt drain) continuiys. As I understand it, only a small bit of that is dissipated because of the low transistor resistance. OK

As far as music, if it's peak current max is 4.2 amps OR LESS (based on the music), I don't see how that is worse than continuous at 4.2 amps. It seems music amplification would be less demanding, as music varies in dynamics, thus the power required varies. You say "linear mode" is worse. OK

I am also SWITCHING the light on and off with a pattern, square wave, at low Freq, 50% duty, it is again just ON and OFF..... I would think this is less critical than ON continious.... due to the OFF time.

Regardless the calculations and heatsink sizing is much more realistic and do-able now. In fact it's going in an aluminum case and I think I'll isolate the transistor (mica + paste) and use an aluminum case as a heatsink.
 
Last edited:
The audio is constantly varying a level, seldom full on or full off. This type of operation produces heat in a mosfet. Mosfets make little heat when full on or full off. The faster you can switch between the on - off state the cooler it will run. A square wave is like that, a triangle wave keeps the on - off slower so more heat.

Here's a good link about mosfets and heat sinks and how to pick them; http://mcmanis.com/chuck/robotics/projects/esc2/FET-power.html
 
The audio is constantly varying a level, seldom full on or full off. This type of operation produces heat in a mosfet. Mosfets make little heat when full on or full off. The faster you can switch between the on - off state the cooler it will run. A square wave is like that, a triangle wave keeps the on - off slower so more heat.

Here's a good link about mosfets and heat sinks and how to pick them; http://mcmanis.com/chuck/robotics/projects/esc2/FET-power.html

That is a great link. I read it all, understand. For historical purpose if any one else has the same question I include two more links.... (In no particular order).

**broken link removed**
https://www.electro-tech-online.com/custompdfs/2011/07/heatsink.pdf
http://robots.freehostia.com/SpeedControl/Mosfets.html
 
To use I^2 * R for power calculations you need to use the RMS value of I

You lost me.... This MOSFET is switching DC or Square wave (DC) at very low Freq .... RMS is peak.

Peak would be the way to go for a sine wave? OK, good to know.... Are you talking about something else?

FOLLOW UP TO YOU OR OTHERS.... Related Design Questions.
I plan on bolting the MOSFET P (two of them) right to an aluminum project box (approx 4"x3"x2") with mica and paste. I might add a typical TO-220 bent up heat sink inside. I figure that should be more than enough to handle the 2.5 watts dissipation. We shall see.

Also should I solder the MOSFET direct to board or use a socket?

To ease PCB layout I was thinking of mounting the MOSFET OFF the PCB, just on the metal case. The MOSFET would be connected with leads or some kind of header, to the PCB? (Do they make these jumpers?) This would allow me to locate the MOSFET anyway I want. Laying down relative to case height. I could get away with shallow box that cost less. The PROJECT boxes in aluminum get expensive quickly in larger sizes. I would like to keep it small. The min internal dimensions right now are 2.5" x 2" x ~1.5". The bent up alum boxes are not much cheaper in the small sizes. Bottom line cost is a factor and want to make it nice, semi pro but not gold plated.
 
It's DC. No RMS involved. Ever.

A design I did way back in the early Nineties....

A done and dusted BUZ11 design, running from DC. 10 Amps from a 10.6V to 14.4V supply for ever. Heatsink from Maplin. Small little thing.

Plastic case is about 60MM by 60MM by 40MM.

Switching frequency around 7 to 10 Hertz. CMOS 555 doing the timing job. And a simple (cannot remember the one)...digital IC used as a Voltage Doubler to raise the Gate Voltage of the BUZ11 above the Source Voltage.

High side driver I believe.
 
Last edited:
It is a bit unclear what you are trying to do. You mention switching an inductive load first then later mention audio.

If you are switching the power supply to an audio amp on and off at 0.666 secs there will likely be issues with bias setting on the amp that will create loud pops into output speaker and possibly damaging speakers. If you are switching a motor at 0.666 secs to vary speed it is too low a PWM rate.

When switching an inductive load you have to also be concerned about inductor discharge kick back which can create high voltages that exceeds the voltage breakdown voltage on the MOSFET's.

If you mention the audio amp just as a comparison there is a big difference if the MOSFET's are operating in linear mode versus switching mode. In switching mode they are either totally ON or totally OFF where there is low resistance (0.023 ohms in the part you specified) resistance across the MOSFET.
 
Last edited:
It's DC. No RMS involved. Ever..

NO kidding! Ha ha. You brought it up not me. I stated DC switching very first post. Thanks for your input....


It is a bit unclear what you are trying to do. You mention switching an inductive load first then later mention audio.
It is very clear to me and others. It is OK. There is NO AUDIO... I did not mention it... this was brought up by another member, as an example to explain heat dissipation for switching vs constant (switching) current. We are good here.... The thread is done, question answered. Thanks.

If you mention the audio amp just as a comparison there is a big difference if the MOSFET's are operating in linear mode versus switching mode. In switching mode they are either totally ON or totally OFF where there is low resistance (0.023 ohms in the part you specified) resistance across the MOSFET.

In the famous words of Nick Cage in Moon Struck, "I know that now". Yep we already covered it, but thanks for repeating and confirming. Yes we are switching not in linear mode. Good stuff.

STOP THREAD DRIFT //// A THREAD IS A TERRIBLE THING TO WASTE!
 
Last edited:
You lost me.... This MOSFET is switching DC or Square wave (DC) at very low Freq .... RMS is peak.

No, for a unipolar swing of current, RMS is not peak it's Ipk*sqrt(duty cycle), the Iac(rms) current is Ipk* sqrt(duty cycle*(1-duty cycle)

Don’t confuse a alternating square wave with a square wave, they are not the same thing.


It's DC. No RMS involved. Ever.
Yes, there is.
 
Last edited:
No, for a unipolar swing of current, RMS is not peak it's Ipk*sqrt(duty cycle), the Iac(rms) current is Ipk* sqrt(duty cycle*(1-duty cycle)

Don’t confuse a alternating square wave with a square wave, they are not the same thing.
Nope I am not confused and I am very aware alternating square wave is not the same thing as audio or sine waves. Read what I wrote, but thanks for the caution.

It's DC. No RMS involved. Ever.
Yes, there is.
Please stop.... for constant DC current, Vavg = Vpk = Vrms, don't be petulant.
Google: en.wikipedia.org/wiki/Root_mean_square

Now for switching DC, sine, sawtooth, square there is a RMS. However I am not working with it in a SIGNAL mode. For a square wave:

**broken link removed**

**broken link removed**

**broken link removed**
So, Vrms = Vpk

**broken link removed**
So, Vavg = Vpk

If I want thermal info for switching of the transistor it is in the manufactures specs.
Transient Thermal Response Curve for a Square Wave Pulse.
That data is provided and I looked at it.

Where Zθ JC(t), Thermal Response, and:
Tjm - Tc = PDM * Zθ jc( t )
Zθ jc(t) = (*X*) C°/W Max. (for faster square wave switching Zθ jc(t) goes down)

*X* - For the three transistors I am considering, depending on which one I use, with my slow switching, at 50% duty, square wave, Zθ jc(t) ~= 1. It's basically ON or OFF, not really switching a signal, so there is no additional Power Dispassion needed. In fact it's a benefit, the transistor will run cooler or about the same as constant current.

I have degree's in engineering and math, LSU undergrad, UW grad school. Not sure why you are arguing with me. My question was answered 5 posts ago. I was not sure about power dissipation of the MOSFET and heatsink calculations. I am clear on that now or as much as I need to be. This is a simple deal. No need to make it more complicated. I am sure you know what you are talking about most of the time. I do thank you for your interest in helping, but we are past that. Please don't be pedantic or state more non sequiturs.
 
Last edited:
Nope I am not confused and I am very aware alternating square weave is not the same thing as audio or sine waves. Read what I wrote, but thanks for the caution.

I'm afraid you are confused...

First I never said anything about audio or sine waves in my post. Then there is the matter what your waveforms really are. While that's a pretty picture you posted, you only have the positive portion of that waveform. You have "chopped" DC, your current is unipolar... in never swings negative so the associated foumulas in your picture don't work with what you described in your first post. As soon as the DC was "switched", it stopped being "just DC".

The closest references at hand for the correct formula are:

Fundamentals of Power Electronics (second edition) Erickson & Maksimovic, page 806 fig A4
Switchmode Power Supply Handbook, Kieth Billings page3.85 fig 3.4.10

And my degree is from NU and I've been designing SMPS (DC-DC's) for the last 20 years and guess what the front end of those puppy's are... CHOPPED DC with an inductive load...
 
Last edited:
Wait a sec... I’m wrong…the input voltage is chopped DC and has nothing to do with it. The current with an inductive load is a ramp… that’s what you need to calculate the RMS value of to figure the Irms^2*R.

Sorry for going off on a tangent...
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top