On the old Vibe:
On the Rotary:
Granted, the two machines are different, but they share identically sized boilers, identical groups, and identical heat exchangers. Both were PID'd in essentially the same manner. Why the difference? the only obvious difference that should matter for shot temperature stability is boiler AUTOFILL, present in the plumbed-in rotary but absent in the vibe machine, whose boiler must be manually refilled with the aid of a sightglass. Autofill tends to be sensitive and anyone who owns a machine with autofill knows that the autofill kicks in at unpredictable times. Anything that diminishes boiler fill (like frothing, or using boiler water for drinks) will increase autofill activity.
I discussed these findings initially with Jim Schulman, who agreed that the autofill was the likely culprit in making the plumbed in rotary machine less temperature stable on a complex shot series that included frothing. We discussed possible ways to defeat this such as having a switch one would use to defeat the autofill normally but allow the autofill to work once daily, or however often one desired or when one chose (i.e. not when making espresso drinks). The risk of that of course is that you forget to turn the autofill on again so the element eventually burns out when you empty out the boiler. Of course there are other options such as putting in on a timer, something Jim also brought up.
I had a conversation with Michael Teahan in Los Angeles today, who as always is a fount of espresso knowledge. Rather than using an on-off switch, Michael suggested using a relay to make sure the autofill doesn't actuate DURING A SHOT, but functions otherwise. He mentioned additional options for making this sort of thing work, such as slowing down the inflow into the boiler from the autofill, but the relay seemed the easiest to impliment.
Before actually trying to impliment this modification, however, I thought it would be worthwhile to repeat the shot series with the autofill bypassed. Here is a picture of the autofill circuit in my machine, on the left side of the boiler:
The autofill probe (as can be seen) is insulated in a sleeve and sticks into the boiler where it contacts boiler water; when it does, it completes a circuit with ground, and turns off the autofill as it senses the correct level. Here is how I defeated it, temporarily for purposes of conducting this experiment:
The temporary wire bridges the circuit and shuts off the autofill. This was confirmed by draining 8 oz of boiler water out and seeing that the boiler autofill did not actuate. I then reversed the change (removed the wire), obviously with the machine unplugged, then replugged in the machine and the autofill immediately acutated for about 15 or 20 seconds to refill the boiler.
With the autofill disabled, I repeated the shot series exactly as before. Here are the new shot curves:
As you can see, the shot temperature curves are now more tightly grouped, with a reduction in both the maximum and minimum shot temperatures. I haven't done any mathematical evaluation of these data, but visually it appears to me that the variability in shot temperature has been reduced by at least a third.
What remains to be determined is what is the best way to get this level of improvement in actual shot temperatures without risking burning out the element or causing a housefire
My impression is that a simple relay, powered by the input solenoid wires (as suggested by Michael Teahan) is the right approach, however I'll have to install one and test again to confirm that this very simple modification in fact has the desired result.
ken
p.s.: This is crossposted on alt.coffee; please do not respond in both threads
