<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Thanks again for all the ideas and references! The probes are $10
each from an inexpensive source (Digikey wants roughly 2-3X that
much). Either the super glue or Colin Ladd's method with "cast
alumina" sounds like the way to go for a couple of the probes.<br>
<br>
But I got the probes to stay in place for a change and an actual
soldering run exactly matched a dry run with no paste: except for
being about 30 seconds out of phase the board temperature in four
places nearly exactly matched the Kester profile I've been using
(but the paste is actually made by Nordson EFD). Just 72 hours ago
I would have insisted this oven couldn't begin to accomplish this,
but it did. No doubt different board sizes, variety of components,
places in the drawer, etc, still bring about a lot of variability,
but I may have finally gotten the upper hand.<br>
<br>
The three key fixes are:<br>
<br>
1) Plugging up the air leaks<br>
2) Arranging for the PCB to be held by its edges so there is air
under it as well as above it<br>
3) Giving the oven intervals that are 2/3 as long as it thinks
they are (i.e. each of it's 10 second temp periods actually lasts 15
seconds)<br>
<br>
Here's what the latest board looks like:<br>
<br>
<a
href="http://triembed.org/blog/wp-content/uploads/2014/10/ocxo-div-0100.jpg">http://triembed.org/blog/wp-content/uploads/2014/10/ocxo-div-0100.jpg
</a><br>
<br>
And here's the thermocouple data (the four probes agreed so closely
they rendered as a single line. The graph to the left is the desired
profile).<br>
<br>
<a
href="http://triembed.org/blog/wp-content/uploads/2014/10/ocxo-graph.pdf">http://triembed.org/blog/wp-content/uploads/2014/10/ocxo-graph.pdf</a><br>
<br>
<br>
-Pete<br>
<br>
</body>
</html>