Krell KSA 100mkII Clone

[Fix]

.... That's why I need some help understand the data sheet...
http://www.onsemi.com/pub_link/Collateral/MJ15003-D.PDF


It says:
High Safe Operating Area (100% Tested) − 5.0 A @ 50 V

What is Tc 25c*
What is Tj 200 C*

If I understand the chart correct.

TC is @25c ( room temp ? )
And Tj, (TO-3 Case, or just junction.. ?) @ 200c
Then it's safe to use 5A @ 50v

But I would like to know how hot can the heatsink get without the MJ15003 will burn...

What is the magic formula...

[AndrewT]

http://www.diyaudio.com/forums/showt...195#post644195

Bensen gives a very good idea of the "magic formula"

I have a modified version that also does BJTs and Drivers as well as FETs.
Email me.

Note on page 2 of the 15003 pdf that the data is for Tc=25degC,
i.e. an infinite heatsink held at a temperature to ensure the case of the DUT is always @ 25degC

[BobEllis]

to go back to a basic level, OnSemi says the junction temperature (Tj) can be 200C. The thermal resistance between the junction and case (Rjc) multiplied by the power dissipated in the device determines the temperature rise above case temperature.

In Andrew's infinite heat sink example, we maintain the case at 25C. Pump 250 watts into it (50V, 5 A) multiplied by Tjc of .7 and you get junction temperature 175C above the case, or 200C.

In real life, you are talking about nearly 60C on the sinks. There is a thermal resistance of roughly .7 between the sink and the case due to the isolator. For improved reliability you'll want to limit junction temperature to at least 50C below rated. Some say use 100C junction temperature, but most people are using 150C rated devices.

That means that you can only use enough power to raise the junction temperature 90 degrees. The thermal resistances junction to case and case to sink add, so expect a rise of 1.4 degrees per watt. Crunching the numbers 90C / 1.4C/W = 64W safe dissipation per device.

These calculations give results similar to the de-rating graphs in the data sheet.

This will give you a rough feel for how hard you can push your devices. SOA calculations are better.

[Fix]

I've found a Swedish page where the calculations are shown.

Tc 60 Case Temp
Tj 200 Junction Temp
Tjc 0,7 C/W TO3
P 200 Watt

Tc = Tj - Tjc * P

If I get it, Case temp @ 200W must not be over 60c
This is for 1x TO-3 MJ15003, and we're using more in this amp, so each transistor is @ 25w ?

And then 25c @ 250W / 50V/5A is True... needed to keep under 25c...

To make it easy, keep heatsink under 60c then there will be no problem at all.

Am I right ?

[AndrewT]

Originally posted by Fix
Am I right ?

no.

[BobEllis]

Originally posted by Fix

Tc = Tj - Tjc * P

That means the device case is at 60C - there is another P * Tcs (case to sink thermal resistance) to contend with. At 200W/device this is roughly another 140C, so you can be safe operating at 200W on a sink at -80C. Easy to do with a bit of liquid nitrogen.

With each device at 25W (.5A @50V), you'll see ~35C rise from sink to junction. From a steady state dissipation perspective you are safe on a 60C sink.

However, we aren't using amps at steady state DC conditions, and you need to compare your load line to the device's SOA to really know if you are safe.

[Fix]

Tc = Tj - Tjc * P

60 = 200 - 0,7 * 200

I've read the text again, and it says that Device Case shall not exced over 60C to be safe.

If you take same formula
25 = 200 - 0,7 * 250
Same as rated in data sheet TC = 25C @ 250W

So if you want to use 250W, TC shall not get over 25C with cooling ?

Here is the link
http://web.telia.com/~u93007940/

It also describes how to calculated size of heatsink etc.

[BobEllis]

I cannot read Swedish, so I cannot directly comment. However, it seems that the author you cite has made two assumptions that are not valid in the real world (or you have misread him).

First, he assumes that you want to run your devices right at their maximum ratings. This is a bad idea. For the sake of reliability you want to keep your devices significantly cooler than their maximum rated temperature. The cooler they are the less likely they are to fail. Not alll devices will operate reliably at their rated maximums.

DIYers (and smart people like Nelson Pass) often use 50% of maximum rated as an operating point. That's why I recommended a target maximum junction temperature of 150C for your 200C rated devices.

Second, your reference (or the way you are reading it) ignores the fact that there will be a temperature rise between the heat sink and the device case. Not likely to happen in the real world. Even if you don't use an isolator. you will get some thermal resistance between the sinks and the case. Some isolators do better than the .7C/W that I used as an example, some worse. You need to take that rise into account.

Edit: (can any swedish speakers help here?)

Looking at Mathias' page I think you missed his reference to the thermal resistance from the case to sink - there are three factors getting from junction to air in this equation:



http://web.telia.com/~u93007940/

Först effekten:

formel6.gif (1,1 Kb)

TJ är kristalltemperaturen (max 175 °C)
TO är rumstemperaturen (25-50 °C) TJC är JUNCTION TO CASE värmeledningen (°C/W för TO-3=1,2 och för TO-220=2)
TGK är värmelednigen från transistorhölje genom luft, glimmerskiva och kiselfett. Luft beräknas vid normal sammansättning till 0,3 och glimmerskiva till 0,7. Påläggs kiselfettet på båda sidor om glimmerskivan förbättras egenskaperna 0,2-0,3.
TK är kylflänsens "till luftström" värmeledning. Dess storhet är helt beroende av profilutseende och storlek.

[AndrewT]

Originally posted by BobEllis

Hi,
Tjc = Rth j-c (junction to case)
Tgk = Rth c-s (case to sink)
Tk = Rth s-a (sink to ambient)

You need to find the values of these three resistances and then find what power you want/need to dissipate

[NoProzac]

Need some help. What type is this?