[clug] Tridge's coffee contraption - photos

Alastair D'Silva alastair at newmillennium.net.au
Tue Dec 1 02:51:43 MST 2009


> -----Original Message-----
> From: linux-bounces at lists.samba.org [mailto:linux-
> bounces at lists.samba.org] On Behalf Of David Cottrill
> Sent: Tuesday, 1 December 2009 6:33 PM
> To: Alastair D'Silva
> Cc: linux at lists.samba.org; Felix Karpfen; Paul Mackerras;
> hugh at blemings.org
> Subject: Re: [clug] Tridge's coffee contraption - photos
> 
> Now that we are truly off topic I think I see Pauls point.
> However... by calculating the inherent delay in the RC circuit formed
> by the series resistor and the input capactance we can adjust the PID
> output timer on the triac to account for this known error and it will
> likely be small enough that it will not be a problem.
> 
> If I've made any dud assumptions then let me know.
> 

You are right, there will be some lag. However, this lag is also in place in
the transformer situation, all other things being equal. From the source's
point of view, the impedance of the load will be similar (the variance
coming from there not being a big inductor in series with the voltage
source) - the input capacitance of the microcontroller and the (effective)
series resistance from the source will be identical.

(Disclaimer: I haven't touched maths to any significant degree since uni, so
please double check my work :) )

You are also right, in that delays can be accounted for in software. In
fact, since there is a feedback loop in place, this is already done for you
:) The end effect of any delay that I can see would be to clamp the maximum
duty cycle to something less than 100%. Assuming a 1ms delay per half cycle
(which would be huge), you would only get 90% time on, lost from the part of
the wave near 0V. Since power is proportional to the square of the voltage
(in a purely resistive load, which we'll say the hot air gun is for
simplicity sake), we get power loss = sin(0.1 * pi)^2 = 0.095, or 10%.

It would be interesting to hook up the original circuit to a CRO to see what
the effective duty cycle is at 100% power.



More information about the linux mailing list