| |
Main
Date: 05 Aug 2007 11:57:08
From: Robert Harmon
Subject: Q: programming a PID for espresso machines.
|
I've put PID's on various machines, using the self-programming features of
Cal 3200 & Watlow 935 controllers. If I knew how to fine tune the
controllers I'm sure I could squeeze a bit more performance from them. But
is it really worth it to go to this additional trouble?
If it's worth it where does one learn how to make the adjustments in the
different controllers, since I'm assuming that each manufacturer has their
own proprietary labels for the different functions & presents them in
varying sequences for programming. Has anyone set out a step-by-step guide
for programming the different models, i.e., Watlow 935 (or 93 or SD or ??),
Love/Dwyer, Cal, Fuji, etc.
You know what I mean:
1) push button "A" once = > *msg appears*,
2) push button "A" twice = > *msg appears*,
3) push button "B" once = > to end process.
Or some such sequence of instructions. Randy has some Fuji instructions & if
anyone else has them posted I'd appreciate knowing where.
--
Robert Harmon
--
http://www.tinyurl.com/mb4uj - My coffee pages.
http://www.tinyurl.com/2tnv87 - My 'Guidelines For Newbies' page.
http://www.tinyurl.com/2cr3e2 - I have things for sale here.
|
|
| |
Date: 07 Aug 2007 21:53:59
From: jggall01
Subject: Re: Q: programming a PID for espresso machines.
|
jggall01 wrote:
> The parameters in that document are for a Fuji PXR3, a controller
> which accepts P as a unitless fraction of the measured range (not as a
> temperature value as do most other controllers).
I need to make a correction to the above. The P value on the Fuji is
given in per cent. So a value of 0.5 would result in P = 1F, which is
perhaps what Greg intended after all.
Sorry for the mistake.
I think I was still reeling from something I read here today about
Dave agreeing with something I wrote. Don't see that every day.
Jim
|
| |
Date: 07 Aug 2007 19:29:55
From: jggall01
Subject: Re: Q: programming a PID for espresso machines.
|
On Aug 7, 3:01 pm, "I- >Ian" <some...@nowhere.com> wrote:
> Next a set of manual parameters similar to the ones posted HERE http://tinyurl.com/2aaax6
> P = 0.5 / I = 225 / D = 4 / IoF = 8.5%
The parameters in that document are for a Fuji PXR3, a controller
which accepts P as a unitless fraction of the measured range (not as a
temperature value as do most other controllers). I believe the
document may have a typo (maybe Greg S. can either confirm this or
correct me) and that the intended value of P was probably 0.05, or 5%
of the measured range.
Combined with the other parameters in that document, the operative
value of P would have been 10F, or 5% of the measured range (50F to
250F). That (10F) is well within the range I think many of us use for
P.
Not saying there's any reason you should abandon what's working well
for you, but I think it might have been a happy coincidence.
Jim
|
| | |
Date: 08 Aug 2007 16:56:03
From: I->Ian
Subject: Re: Q: programming a PID for espresso machines.
|
On Tue, 07 Aug 2007 19:29:55 -0700, jggall01 <jggall01@yahoo.com >
wrote:
>On Aug 7, 3:01 pm, "I->Ian" <some...@nowhere.com> wrote:
>
>> Next a set of manual parameters similar to the ones posted HERE http://tinyurl.com/2aaax6
>> P = 0.5 / I = 225 / D = 4 / IoF = 8.5%
>
>The parameters in that document are for a Fuji PXR3, a controller
>which accepts P as a unitless fraction of the measured range (not as a
>temperature value as do most other controllers). I believe the
>document may have a typo (maybe Greg S. can either confirm this or
>correct me) and that the intended value of P was probably 0.05, or 5%
>of the measured range.
>
>Combined with the other parameters in that document, the operative
>value of P would have been 10F, or 5% of the measured range (50F to
>250F). That (10F) is well within the range I think many of us use for
>P.
>
>Not saying there's any reason you should abandon what's working well
>for you, but I think it might have been a happy coincidence.
>
I'd calculated the 1°F P from the 0.5% and the 50 to 250°F range in
the document and used it initially.
Since the 0.5°F is the lower end of P for the 32B, I figured what the
heck, give it a go.
The only confirmed benefit of the narrow band is the element indicator
blinks less ;-)
|
| |
Date: 07 Aug 2007 12:01:14
From: I->Ian
Subject: Re: Q: programming a PID for espresso machines.
|
On Sun, 5 Aug 2007 11:57:08 -0500, "Robert Harmon"
<Texas_Coffee@earthlink.net > wrote:
>I've put PID's on various machines, using the self-programming features of
>Cal 3200 & Watlow 935 controllers. If I knew how to fine tune the
>controllers I'm sure I could squeeze a bit more performance from them. But
>is it really worth it to go to this additional trouble?
>
>If it's worth it where does one learn how to make the adjustments in the
>different controllers, since I'm assuming that each manufacturer has their
>own proprietary labels for the different functions & presents them in
>varying sequences for programming. Has anyone set out a step-by-step guide
>for programming the different models, i.e., Watlow 935 (or 93 or SD or ??),
>Love/Dwyer, Cal, Fuji, etc.
>
>You know what I mean:
> 1) push button "A" once => *msg appears*,
> 2) push button "A" twice => *msg appears*,
> 3) push button "B" once => to end process.
>
>Or some such sequence of instructions. Randy has some Fuji instructions & if
>anyone else has them posted I'd appreciate knowing where.
>
>--
>Robert Harmon
Over the past year, mostly for my own education, I've experimented
greatly with Love 32B PID parameters for the Vibiemme Domobar Super
HX. [Grounded T thermocouple immersed in the boiler]
First iteration was a week long tuning to get the most stable P
parameter, then adust I and D for maximum stablility. In the end, the
boiler idle temperature held within a few tenths of a degree F.
Recovery is slowest with this setting.
P = 7.6 / I = 90 / D =15 / IoF = 8.5% [ IoF parameter bypass the PID
when at set temperature to maintain set point ]
Next I autotuned, starting at about 210°F for a set point of ~233°F
Boiler temperature was not as well regulated, but recovery to setpoint
was faster. IIRC, P = 8.4 / I = 77 / D = 19 / IoF = ~11%
Next a set of manual parameters similar to the ones posted HERE
http://tinyurl.com/2aaax6
P = 0.5 / I = 225 / D = 4 / IoF = 8.5%
With these parameters, the boiler temperature recovers most rapidly,
but with much less control, overshooting as much as 3°F
Bottom line, for home use it makes almost no difference to the shot
temperature as measured with Eric's Thermometer Adapter.
http://users.rcn.com/erics/
and an in-basket thermocouple
http://www.ielogical.com/coffee/BasketTC.JPG
The thermal inertia of the massive Vibiemme e61 swamps the minor
fluctuation of boiler temperature.
NOTE : The Vibiemme takes a about 52 minutes for the group temperature
to stablilize regardless of PID setup.
In the end, I've chosen to stay with the P = 0.5 / I = 225 / D = 4 as
it 'seems' to keep the group temperature slightly more even. My guess
is the thermosyphon 'likes' a bit of instability.
YMMV
|
| |
Date: 06 Aug 2007 03:07:39
From: jim schulman
Subject: Re: Q: programming a PID for espresso machines.
|
On Sun, 5 Aug 2007 11:57:08 -0500, "Robert Harmon"
<Texas_Coffee@earthlink.net > wrote:
>If it's worth it where does one learn how to make the adjustments in the
>different controllers, since I'm assuming that each manufacturer has their
>own proprietary labels for the different functions & presents them in
>varying sequences for programming
I've experimented with various tunings on a pair of Silvias I use for
espresso testing. The autotune algorithms (which are close enough to
the TLC manual tuning algorithm to be identical within the margin of
error) are very good. Going to more aggressive manual tuning
algorithms like ZN generally increases the overshoot a lot for little
gain in more rapid stabilization.
If you ever program a roaster, where minimizing the up and down on the
environmental temperature is as important as holding bean temperature
to the profile, manual tuning does pay dividends -- use an overdamped
setting by cutting the D & I amplitudes in half (this may be a
doubling of the I if the controller gives it in minutes or seconds, as
opposed to repeats). Don't change the P.
Basically, if you want the controller to settle to one temperature,
there isn;t much gold in the manual tuning hills. The story changes if
you are prepared to have the controller permantly hunt within a very
narrow band. Greg Scace posted on this a few years back, you can find
it via google. On the Silvia, for instance, setting the controller as
an on/off tstat with no deadband will create a roughly 12C
oscillation. You can get it down to 4C, at roughly 1/4 of the time
period, by setting the P to around 8 to 10C (1/3 of the autotune
range, or 3 times the amplification), setting the I to zero, and
setting the D very high (4 to 8 minute), so it reverses the heat as
soon as the temperature starts to change. (these figures are iirc and
rough)
I never explored this thoroughly, but if you want the temperature on
an unmodded Silvia to recover to within this range immediately after a
shot is pulled, rather than overshooting, and then floating for a long
period above the setpoint, so you can bang shots out without pause,
this may be the only way to do it.
|
| |
Date: 06 Aug 2007 04:57:00
From:
Subject: Re: Q: programming a PID for espresso machines.
|
Andy,
glad to see you see you say that. I've played around with the settings
on Fuji px series, but have found(at least for my projects)leaving it
set as is works best
John
|
| |
Date: 05 Aug 2007 21:37:39
From: Andy Schecter
Subject: Re: Q: programming a PID for espresso machines.
|
Robert Harmon wrote:
> I've put PID's on various machines, using the self-programming features of
> Cal 3200 & Watlow 935 controllers. If I knew how to fine tune the
> controllers I'm sure I could squeeze a bit more performance from them. But
> is it really worth it to go to this additional trouble?
Personally, I wouldn't bother.
Some folks program in extra derivative control (the D in PID). This makes the
controller more responsive to rapid changes, like when you're pulling a shot.
But the downside is that the control becomes very nervous (ie, less accurate)
around setpoint.
PID control is 50+ year old technology, and it can only do so much. The fact
that you can buy a new Japanese control for $130, a Chinese one for $50, or an
eBay find for $30, is remarkable.
But still, PID is feedBACK control: it waits to see an error, then tries to
correct for it. FeedFORWARD control would be a helpful addition: "you're
pumping 50ml of 80F water into the boiler to pull a shot, so we anticipate
we'll have to add x BTU's of heat to compensate." But feedforward control
doesn't come so simply.
--
-Andy S.
http://www.flickr.com/photos/andy_s/sets/
|
|
