Roasting ceramic beans
At various stages, I have used a dummy load for testing the performance of the
controller and the roaster. Generally,
I used a 200g batch of ceramic beans - the sort used for
blind baking flan cases, if you're into that sort of thing (I'm not, but I
found them in the kitchen, and my wife didn't seem to notice their
absence). I'm not sure it makes much difference, to be honest - I suspect
simply running the roaster empty would give much the same performance, because
I think the heat capacity of the beans is insignificant compared to that of
the roaster itself and the rate of loss through the walls.
Early development
As noted on other pages, the first run of the roaster (before it had any
proper profile control) severely over-did some beans. Therefore, when
the controller
was 90% done and had temperature control for the roast, but
no eject or cooling, I decided
to do some testing without destroying perfectly good coffee beans.
This is the very first controlled run.
The red line is the target profile, the blue line the achieved.
There's a big lag at first
(but I'd
limited maximum heat input during that stage to 80%, which may affect it).
When
the ramp rate slows down the temperatures catch up within about a minute.
There's some oscillation then, which dies out, but then when it gets to the
third ramp segment the oscillation starts building up again. I was using
a progressively tighter proportional band for each step in turn (just for
fun - no good
reason), so maybe I need to tune that.
This is another dummy load (ceramic
beans). Compared to the above, I've taken off the 80% power limit in the
first segment, and put the same (15 degree) proportional band on all steps.
There's less lag in the temperatures at first (as expected if it has more
power available). It catches up slightly quicker, and settles at about the
same rate. I don't know what the shorter period oscillation is on this.
The later segments are much better - presumably the proportional band I used
first time was too tight.
The cooling phase at this time comprises me opening the door and
using a hand-held blower to blast ambient air into the oven. It's not
particularly efficient, as you can see.
Investigating proportional band
|
So, I speculated above about the effects of proportional band. This is three
runs of a very artificial profile, identical except for proportional band. In the
upper graph,
it's set to 20 degrees. In the middle, it is 5 degrees. At the bottom is a
1 degree proportional band, which is as close as the controller will go to
on/off control (I didn't program it to work with a zero band).
I note that actually, the control achieved is not so different - the
maximum error is about the same. Certainly, there looks to be less
difference between these three than between the two 'early development'
graphs above. Basically, I'm not sure there's much conclusion to be drawn
from this test - except possibly that there's little point worrying about
the proportional band.
|
back to the main oven roaster page
back to the top roaster page
back to Ian's contents page
back to www.astounding.org.uk top contents
page
To comment on anything (please do) email
ian@astounding.org.uk