Re: PMV31XN - dissipation?

Malcolm Reeves wrote:
On Fri, 27 Oct 2006 17:52:31 GMT, PeteS <peter.smith8380@xxxxxxxxxxxx>
wrote:


Malcolm Reeves wrote:

Hi,

I'm trying to work out the maximum dissipation of an nxp (formally
philips) part, the PMV31XN (otherwise known as beating your head
against a brick wall!). I used the nxp support form and selected
critical as this is holding me up, but I've yet to receive a reply 2
days later so its a good job is it only critical and not life and
death isn't it :-/.

I have the PMV31XN data ***, in that it says the part can dissipate
2W with Tsp at 25C and de-rates above that. Tsp stands for
Temperature at the solder point. I'd guess that is the drain lead BUT
the data*** doesn't specify. It is also a not very helpful figure.
Great if I attach and infinite heatsink to the drain lead I can
dissipate 2W but I'm all out of infinite heatsinks so how about some
practically useful figures, such as what can the part itself dissipate
WITHOUT an infinite heatsink.

I'd assume somewhere in the range 200-500mW but where? And how do I
calculate that if the data*** does see fit to tell me nor can I find
the data anywhere on the nxp web site. I'm considering replacing the
part with some other manufacturer all because of the lack of data!
When will manufacturers understand that good data is critical to
sales. As is good and prompt support!



Hi Malcolm

The thermal resistance of the part is 60C/W at DC to the pins. At 25C starting, then the junction will be at 145C for 2W dissipation, just within it's limits.

Of course, this will raise the temperature at the pins (by how much depends on the thermal resistance of the board. Assuming these are fairly small tracks (it's a SOT23 after all) I'll assume theta Track-ambient of 40C/W. That's a total of 100C/W junction to ambient - in other words we can convert the problem from junction to pin to junction to ambient.

I would not personally run the part higher than 0.5W - at that, I would expect a junction rise of about 50C from ambient and the pins will be at about 20C above ambient (derating the part anyway).

Interesting that they don't show theta j-a (via the case) at all. Knowing Philips (sorry, NXP) all that information will be in a separate data *** somewhere that they just expect you to know how to find.

If I felt less lazy I'd throw a spread*** together to show the limiting conditions for the device, but it's Friday and I've hda a full day of designing for a new board, so I going to sit with some wine :)


The assumption is the critical bit. What I looking for I guess
(besides venting my spleen about poor data *** and poorer support!)
is some philips data reference or someone else's data to work out pin
to ambient.

I'd worked out that theta j-p is 62.5C/W ((150C-25C)/2W). Why they
didn't just say that I don't know. It's not helpful to hide the
figures and the graph of power is just a waste of ink. It would be
more useful, as I seen on other parts, philips ones included, to quote
j-a and j-pin. That way if you need to get more out of the part you
can add copper and do more extensive calculations. If it works with
j-a then you don't need to do the calcs even if you add the copper).

It all comes down to needing a nxp figure for j-a for nxp sot23
package and/or some data for PCB copper area C/W to ambient.

Malcolm

I agree with you.

NXP (Philips in an even less useable guise as far their website is concerned, and *that's saying something*) has been notorious for not providing relevant details in the standard data sheets.

I'll have to dig back at work, but I seem to remember one Philips part where the package diagram, thermal response tables and even pinouts were in 3 separate datasheets apart from the core datasheet.

This is of interest in another way; I am probably going to move to an ARM7 device on a major device redesign, and Philips has parts, but if the docs are still going to be like this, I won't touch it. I don't care how damn good the part is, if it's not documented so I can design *easily* with it, I won't use it.

Cheers

PeteS
.

Quantcast