On Thu, 01 Dec 2005 21:53:54 -0600, jim schulman
After playing with it most of the night, I now have reservations on
how useful this is as a **final** temperature control on single or
double boiler espresso machines. I doubt what I say will affect it's
usefulness as a pstat replacement, although Ken's experience, see
below, may indicate that even here a PID may be better.
The problems is not the deadband and the consequent loss in accuracy,
but that it turns on and off for longer periods of time (i.e cycle
times around 1 minute, rather than the 1 second of an ssr or 1/60th
second of an SCR).
Here's what I found when I rigged the PF (this is neither Schomer nor
Scace, just a TC in the PF with some waterproof packing -- I use the
spring lever to blow the entire shot volume through in less than a
second and get the average shot temperture by taking the high
reading).
Using a PID or dimmer, the PF reading was always a fixed offset away
from the boiler reading -- so the boiler reading allowed one to know
average shot temperature within 1F at worst. Not so with the
temperature switch. The difference between its reading and the shot
temperature dependended on whether the heat was on or off. With the
sensor further away from the heater than the intake of the piston, the
'heater on' readings were too low, and 'heater off' readings too high
-- the shot temperatures, being closer to the heat source, swung more
dramatically. If the sensor was placed closer to the heat, the
opposite happened. Moreover, the closer to the heat the sensor is
placed, the more overshoot one reads.
When I finally found a spot for the TC that remained perfectly
correlated with the shot temperatures (within 1F), the performance of
the switch had degraded considerably-- set at 200F the actual
operating range ran from 198F to 204F. Hugely better than button
stats, and a little better than bulb stats and pstats, but not in the
same league as a well placed PID.
I ended up punting -- I put the dimmer back in series with the unit,
and set it down after warmup, so the heat remains on for long periods
as it traverses the 199 to 200 interval. This reduces the subsequent
overshoot back to 201. However, it's a concession of defeat for the
concept of using this for very tight control, since with the added
dimmer cost, one may as well get a proportional controller.
Adding a PID to Ken's Junior boiler improved the inter-shot stability
of the Cimbali considerably when starting after an idle. I always
thought with all the buffering between boiler and brewhead, the effect
of an on/off control is minimized. However, a pstat looks a lot better
than it may be -- by recording pressure, it's reading an average
temperature for the enitre boiler -- a very stable figure. The HX
itself may be much more influenced by the heater or other local
conditions in the boiler. In other words, the 2- 3C swing read by the
pstat may be concealing much larger ones at the HX itself; just as the
nice readings I had last night concealed much larger swings at the
intake of the Pepina's piston/cylinder.
This opens up one of the problems of espresso machine control in
general. The sensors for control do not read shot temperature, but
something upstream. The obvious solution is to find the spot that
correlates best with shot temperature over the entire heating cycle
for on-off control, or the full wave-length of a PID controller (the %
output values on these mostly graph out as decaying amplitude
sinewaves with fixed wavelengths).
The peppina makes this easy since there's no refill, and the cylinder
isolates one shot's worth for measurement. I think the same can be
done with pump machines by putting a sensor in a blind filter. Run
this to get an idea of what's going on at the head for different
sensor placements.
--
jim schulman
<jim_schul...@ameritech.net>
