[time-nuts] Adding adjustment pot to 5680

Robert Benward rbenward at verizon.net
Wed Jan 11 14:12:02 UTC 2012


Arthur,
800mW will give you a 52C rise in junction temperature over ambient when not 
mounted to anything.  The 5680s get pretty hot, probably 50C internally (has 
anyone measured it?).  That will put you at 102C.  Make sure the flavor of 
7805 you selected is rated for that (most are rated for 125C).  Not all are, 
and not all have thermal shutdown.  As for reliability, without any 
derating, this won't give you much margin.

Don't forget the cap on the output.

Bob



----- Original Message ----- 
From: "Arthur Dent" <golgarfrincham at yahoo.com>
To: <time-nuts at febo.com>
Sent: Tuesday, January 10, 2012 10:50 PM
Subject: [time-nuts] Adding adjustment pot to 5680


>You are driving the 7805 near its limit. As a rule of thumb,
>a TO-220 case can disipate about 1W of power, if it's
>free-standing. You have max 100mA with a voltage difference of
>10V, resulting in 1W max. Ie the 7805 will be running at a
>considerable high temperature. As you have there a good heat sink
>already (the case), i'd mount the 7805 against the case and use
>wires to connect it.
> Attila Kinali
++++++++++++++++++++++++++
I had considered mounting the 7805 on the vertical heatsink you see
to the left of the 7805 in my photo, using a short screws in either side
of the the tapped hole that is already there and use leads to connect
the 7805 to the connector pins, but felt it wasn't necessary. Directly
connecting the regulator to the DB-9 pins was much simpler.

FYI, the load is not 100Ma as you assumed, that is the max rating of
the 78L05 that I didn't use. The actual load the 7805 sees in the 5680A
is about 80Ma so even at the higher temperatures inside the case I
don't see a problem. You are correct that it might be a better practice
to mount the regulator on the heat sink, especially if this were a
production item, but in this case I felt it was overkill.
http://farm8.staticflickr.com/7013/6668466093_90782cf7e9_b.jpg


-Arthur





More information about the time-nuts mailing list