| |
Main
Date: 28 Jul 2006 02:17:44
From: Sammus
Subject: Silvia thermowell
|
Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
silvia) and was wondering if someone could offer some information on
drilling and tapping the boiler to accomodate a thermowell, in
particular, Can I get a themowell with a built in thermocouple thats
suitable for this? ie v.short response time, threaded and the right
length?
Have many people done the mod like this? and if so, how much of a
difference will it make? would I need to get one custom made or is
there a particular model/supplier I should ask, thanks for any help
people! PS anything available in Australia would help...
|
|
| |
Date: 28 Jul 2006 08:45:16
From: Randy G.
Subject: Re: Silvia thermowell
|
"Sammus" <Sammus187@gmail.com > wrote:
>Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
>silvia) and was wondering if someone could offer some information on
>drilling and tapping the boiler to accomodate a thermowell, in
>particular, Can I get a themowell with a built in thermocouple thats
>suitable for this? ie v.short response time, threaded and the right
>length?
>Have many people done the mod like this? and if so, how much of a
>difference will it make? would I need to get one custom made or is
>there a particular model/supplier I should ask, thanks for any help
>people! PS anything available in Australia would help...
>
It might be a fun experiment, but as others have stated, the
performance in the end will not be improved in a noticeable way.
The benefit of having a PID equipped machine is that you have the
ability to choose a stable temperature before you begin the pull. Once
you hit the brew button the design of the machine takes over. Since
the heating element's function is binary (it is either on OR off),
after starting the pull the element is turned on by the PID because of
the drop in temperature and will stay on during the pull. As Greg
mentioned, the stored thermal energy in the heating element will cause
a bit of overshoot during the recovery regardless of how the
temperature is sensed. A low-mass thermocouple in the water may help
decrease that a bit, but all the work involved in adding that style of
thermocouple will not change the way the shot temperature is
controlled.
Randy "be happy - make coffee" G.
http://www.EspressoMyEspresso.com
|
| |
Date: 28 Jul 2006 07:09:00
From: daveb
Subject: Re: Silvia thermowell
|
seldom that I agree, but mr. smith is right on point.
Dave
93pids
Ian Smith wrote:
> On 28 Jul 2006 02:17:44 -0700, Sammus <Sammus187@gmail.com> wrote:
>
> > Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
> > silvia) and was wondering if someone could offer some information on
> > drilling and tapping the boiler to accomodate a thermowell, in
> > particular, Can I get a themowell with a built in thermocouple thats
> > suitable for this? ie v.short response time, threaded and the right
> > length?
>
> Since you're driving to a steady state, why do you want a thermowell
> (and indeed one with a v short response time)? Before the shot,
> you're at a uniform temperature (that's the point of putting a PID in)
> so there's no need to detect rapid changes - there aren't any. Once
> the shot starts, you're not doing any useful controlling, so the
> effort of monitoring temperature fluctuation is wasted.
>
> So why go to the bother? What are you trying to achieve?
>
> > how much of a difference will it make?
>
> Roughly - absulutely none whatsoever, I would think. I suppose it
> might help with bragging at your local coffee shop, and it's
> another point of failure, but that's about all.
>
> regards, Ian SMith
>
> --
>
|
| |
Date: 28 Jul 2006 07:07:36
From: gscace
Subject: Re: Silvia thermowell
|
I did it and there's really not too much difference. The problem is
that the Silvia heating element stores a lot of heat inside it if the
element is active for any length of time, such as when recovering from
brewing. So the water temperature overshoots on recovery and it's
tough for the PID controller to handle it well, regardless of the
temperature sensor position. There's not too much that you can do to
fix the problem except for preheating the feedwater to the boiler,
which will mean that the brew boiler heating element will have to work
less hard when recovering.
-Greg
Sammus wrote:
> Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
> silvia) and was wondering if someone could offer some information on
> drilling and tapping the boiler to accomodate a thermowell, in
> particular, Can I get a themowell with a built in thermocouple thats
> suitable for this? ie v.short response time, threaded and the right
> length?
> Have many people done the mod like this? and if so, how much of a
> difference will it make? would I need to get one custom made or is
> there a particular model/supplier I should ask, thanks for any help
> people! PS anything available in Australia would help...
|
| |
Date: 28 Jul 2006 06:02:45
From: Omniryx@gmail.com
Subject: Re: Silvia thermowell
|
I seriously contemplated this mod, as well, but came to the same
conclusion that Ian reached below. Why bother?
Will
Ian Smith wrote:
> Since you're driving to a steady state, why do you want a thermowell
> (and indeed one with a v short response time)? Before the shot,
> you're at a uniform temperature (that's the point of putting a PID in)
> so there's no need to detect rapid changes - there aren't any. Once
> the shot starts, you're not doing any useful controlling, so the
> effort of monitoring temperature fluctuation is wasted.
>
> So why go to the bother? What are you trying to achieve?
|
| |
Date: 28 Jul 2006 12:50:55
From: Ian Smith
Subject: Re: Silvia thermowell
|
On 28 Jul 2006 02:17:44 -0700, Sammus <Sammus187@gmail.com > wrote:
> Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
> silvia) and was wondering if someone could offer some information on
> drilling and tapping the boiler to accomodate a thermowell, in
> particular, Can I get a themowell with a built in thermocouple thats
> suitable for this? ie v.short response time, threaded and the right
> length?
Since you're driving to a steady state, why do you want a thermowell
(and indeed one with a v short response time)? Before the shot,
you're at a uniform temperature (that's the point of putting a PID in)
so there's no need to detect rapid changes - there aren't any. Once
the shot starts, you're not doing any useful controlling, so the
effort of monitoring temperature fluctuation is wasted.
So why go to the bother? What are you trying to achieve?
> how much of a difference will it make?
Roughly - absulutely none whatsoever, I would think. I suppose it
might help with bragging at your local coffee shop, and it's
another point of failure, but that's about all.
regards, Ian SMith
--
|
| |
Date: 28 Jul 2006 12:33:45
From: Coffee for Connoisseurs
Subject: Re: Silvia thermowell
|
Recent Silvia boilers come with a built in (and as yet unused) thermowell.
--
Alan
alanfrew@coffeeco.com.au
www.coffeeco.com.au
|
| | |
Date: 28 Jul 2006 15:05:10
From: Steve
Subject: Re: Silvia thermowell
|
On Fri, 28 Jul 2006 12:33:45 GMT, "Coffee for Connoisseurs"
<alanfrew@coffeeco.com.au > wrote:
>Recent Silvia boilers come with a built in (and as yet unused) thermowell.
How recent, approximately?
What would one look for to determine whether or not the machine has
one?
|
| |
Date: 28 Jul 2006 04:14:27
From: daveb
Subject: Re: Silvia thermowell
|
Sure you can. Drill and tap for a suitable fitting. Omega.com has 'em
and TCs that go with the fittings
But if you have ever been inside a silvia boiler you will find it is
mostly taken up with the heating element. You may or may not get a
better read on the temp this way.
???
Dave
92 pids
Sammus wrote:
> Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
> silvia) and was wondering if someone could offer some information on
> drilling and tapping the boiler to accomodate a thermowell, in
> particular, Can I get a themowell with a built in thermocouple thats
> suitable for this? ie v.short response time, threaded and the right
> length?
> Have many people done the mod like this? and if so, how much of a
> difference will it make? would I need to get one custom made or is
> there a particular model/supplier I should ask, thanks for any help
> people! PS anything available in Australia would help...
|
| |
Date: 28 Jul 2006 20:37:42
From: Sammus
Subject: Re: Silvia thermowell
|
Omniryx@gmail.com wrote:
> Greg, how much overshoot have you seen on the Silvia? I have never
> seen an (indicated) overshoot of more than 1 degree Celsius and it
> always drops to the target temp in less than sixty seconds...usually
> less than thirty. Am wondering if that is typical. I am using Jim
> Gallt's kit with a Watlow 93 controller and thermocouple on the outside
> of the boiler.
>
> Will
>
Wow thats so much better than what I get, I usually get 2-3F overshoot
and takes a minute or three to stabalize...maybe its my thermocouple
and positioning? see it at
http://webs.gallery.netspace.net.au/SilviaProject/DSC00620
I've been thinking I might get some thermal cement and stick it in the
little well the old Tstat sits in for better readings.
I wanted good response time to reduce this overshoot, I imagine that if
the heater is on and the TC is lagging behind a bit, this would
increase the overshoot. I was always a bit sus about my TC mounting, so
I might try adjusting this before I get to crazy with it :P
|
| | |
Date: 28 Jul 2006 22:18:45
From: Randy G.
Subject: Re: Silvia thermowell
|
"Sammus" <Sammus187@gmail.com > wrote:
>Wow thats so much better than what I get, I usually get 2-3F overshoot
>and takes a minute or three to stabalize...maybe its my thermocouple
>and positioning?
I find the overshoot is increased the longer Silvia sits idling.
Probably because the internals are so close in temp to the set point
that there is little to "absorb" the extra heat energy. I find that
employing the E-61 trick of pumping a little water through cools
things down in a hurry and I get better stability that way. The
recovery rate seems better between shots after that as well. Nothing
documented or quantified; simply observed behavior.
Randy "better than overshooting the runway!" G.
http://www.EspressoMyEspresso.com
|
| |
Date: 28 Jul 2006 11:23:58
From: Omniryx@gmail.com
Subject: Re: Silvia thermowell
|
Greg, how much overshoot have you seen on the Silvia? I have never
seen an (indicated) overshoot of more than 1 degree Celsius and it
always drops to the target temp in less than sixty seconds...usually
less than thirty. Am wondering if that is typical. I am using Jim
Gallt's kit with a Watlow 93 controller and thermocouple on the outside
of the boiler.
Will
gscace wrote:
> I did it and there's really not too much difference. The problem is
> that the Silvia heating element stores a lot of heat inside it if the
> element is active for any length of time, such as when recovering from
> brewing. So the water temperature overshoots on recovery and it's
> tough for the PID controller to handle it well, regardless of the
> temperature sensor position. There's not too much that you can do to
> fix the problem except for preheating the feedwater to the boiler,
> which will mean that the brew boiler heating element will have to work
> less hard when recovering.
>
|
| |
Date: 29 Jul 2006 09:09:12
From: jggall01
Subject: Re: Silvia thermowell
|
Sammus wrote:
> Omniryx@gmail.com wrote:
> > Greg, how much overshoot have you seen on the Silvia? I have never
> > seen an (indicated) overshoot of more than 1 degree Celsius and it
> > always drops to the target temp in less than sixty seconds...usually
> > less than thirty. Am wondering if that is typical. I am using Jim
> > Gallt's kit with a Watlow 93 controller and thermocouple on the outside
> > of the boiler.
> >
> > Will
> >
>
> Wow thats so much better than what I get, I usually get 2-3F overshoot
> and takes a minute or three to stabalize...maybe its my thermocouple
> and positioning? see it at
> http://webs.gallery.netspace.net.au/SilviaProject/DSC00620
> I've been thinking I might get some thermal cement and stick it in the
> little well the old Tstat sits in for better readings.
>
> I wanted good response time to reduce this overshoot, I imagine that if
> the heater is on and the TC is lagging behind a bit, this would
> increase the overshoot. I was always a bit sus about my TC mounting, so
> I might try adjusting this before I get to crazy with it :P
Hi, Sammus -
Have you mentioned which model PID you are using? Regarding the
overshoot you describe, I would look at the PID programming first, the
t/c second. Your t/c placement should work, IMO, as long as it is in
good contact with the brass (from photo, this looks to be the case).
I have set up machines with several different types of controllers and
have noticed a wide range in performance in regards to overshoot.
-- JGG
|
| |
Date: 29 Jul 2006 03:33:27
From: daveb
Subject: Re: Silvia thermowell
|
Sammus!
drop me a note. I have some suggestions.
DAve
93pid
|
| |
Date: 29 Jul 2006 22:02:00
From: Sammus
Subject: Re: Silvia thermowell
|
Howdy, am using a Fuji PXR3, I used the auto calibration function by
starting the autotune at the same time as hitting the brew button when
pulling a shot, I read somewhere that this would give the best
performance, that was a while ago now though, any other proven tuning
techniques?
>
> Hi, Sammus -
>
> Have you mentioned which model PID you are using? Regarding the
> overshoot you describe, I would look at the PID programming first, the
> t/c second. Your t/c placement should work, IMO, as long as it is in
> good contact with the brass (from photo, this looks to be the case).
>
> I have set up machines with several different types of controllers and
> have noticed a wide range in performance in regards to overshoot.
>
> -- JGG
|
| |
Date: 29 Jul 2006 12:45:20
From: daveb
Subject: Re: Silvia thermowell
|
how many of them?
jggall01 wrote:
> Sammus wrote:
> > Omniryx@gmail.com wrote:
> > > Greg, how much overshoot have you seen on the Silvia? I have never
> > > seen an (indicated) overshoot of more than 1 degree Celsius and it
> > > always drops to the target temp in less than sixty seconds...usually
> > > less than thirty. Am wondering if that is typical. I am using Jim
> > > Gallt's kit with a Watlow 93 controller and thermocouple on the outside
> > > of the boiler.
> > >
> > > Will
> > >
> >
> > Wow thats so much better than what I get, I usually get 2-3F overshoot
> > and takes a minute or three to stabalize...maybe its my thermocouple
> > and positioning? see it at
> > http://webs.gallery.netspace.net.au/SilviaProject/DSC00620
> > I've been thinking I might get some thermal cement and stick it in the
> > little well the old Tstat sits in for better readings.
> >
> > I wanted good response time to reduce this overshoot, I imagine that if
> > the heater is on and the TC is lagging behind a bit, this would
> > increase the overshoot. I was always a bit sus about my TC mounting, so
> > I might try adjusting this before I get to crazy with it :P
>
>
> Hi, Sammus -
>
> Have you mentioned which model PID you are using? Regarding the
> overshoot you describe, I would look at the PID programming first, the
> t/c second. Your t/c placement should work, IMO, as long as it is in
> good contact with the brass (from photo, this looks to be the case).
>
> I have set up machines with several different types of controllers and
> have noticed a wide range in performance in regards to overshoot.
>
> -- JGG
|
| |
Date: 30 Jul 2006 03:53:15
From: Phil Paintin
Subject: Re: Silvia thermowell
|
gscace wrote:
> I did it and there's really not too much difference. The problem is
> that the Silvia heating element stores a lot of heat inside it if the
> element is active for any length of time, such as when recovering from
> brewing. So the water temperature overshoots on recovery and it's
> tough for the PID controller to handle it well, regardless of the
> temperature sensor position. There's not too much that you can do to
> fix the problem except for preheating the feedwater to the boiler,
> which will mean that the brew boiler heating element will have to work
> less hard when recovering.
That suggests the time delay between the element being energised (and
de-energised) and that heat being detected at the t/c is too long for
the controller's time window -- or is that over-simplifying? If
different controllers vary in the maximum time window, it could make a
big difference I guess?
> Sammus wrote:
> > Hi, I'm not happy with my thermocouple on the outside of my boiler (PID
> > silvia) and was wondering if someone could offer some information on
> > drilling and tapping the boiler to accomodate a thermowell, in
> > particular, Can I get a themowell with a built in thermocouple thats
> > suitable for this? ie v.short response time, threaded and the right
> > length?
> > Have many people done the mod like this? and if so, how much of a
> > difference will it make? would I need to get one custom made or is
> > there a particular model/supplier I should ask, thanks for any help
> > people! PS anything available in Australia would help...
|
| |
Date: 30 Jul 2006 21:06:51
From: jggall01
Subject: Re: Silvia thermowell
|
Sammus wrote:
> Howdy, am using a Fuji PXR3, I used the auto calibration function by
> starting the autotune at the same time as hitting the brew button when
> pulling a shot, I read somewhere that this would give the best
> performance, that was a while ago now though, any other proven tuning
> techniques?
>
>
> >
> > Hi, Sammus -
> >
> > Have you mentioned which model PID you are using? Regarding the
> > overshoot you describe, I would look at the PID programming first, the
> > t/c second. Your t/c placement should work, IMO, as long as it is in
> > good contact with the brass (from photo, this looks to be the case).
> >
> > I have set up machines with several different types of controllers and
> > have noticed a wide range in performance in regards to overshoot.
> >
> > -- JGG
Sammus -
FWIW, my limited experience with that Fuji has been that 3F overshoot
is about what you will get. But the recovery time is acceptable, even
though you get some overshoot. And the stability at idle is very good
with the PXR3. But I have never been able to completely get rid of the
overshoot after pulling a shot.
If you can get 2 minutes to 2:30 recovery time after a shot, don't
worry about the overshoot. I doubt you will be able to significantly
improve the recovery time below 2 minutes, regardless of amount of
overshoot.
-- JGG
|
| | |
Date: 31 Jul 2006 09:52:19
From: Randy G.
Subject: Re: Silvia thermowell
|
"jggall01" <jggall01@yahoo.com > wrote:
>FWIW, my limited experience with that Fuji has been that 3F overshoot
>is about what you will get. But the recovery time is acceptable, even
>though you get some overshoot. And the stability at idle is very good
>with the PXR3. But I have never been able to completely get rid of the
>overshoot after pulling a shot.
>
>If you can get 2 minutes to 2:30 recovery time after a shot, don't
>worry about the overshoot. I doubt you will be able to significantly
>improve the recovery time below 2 minutes, regardless of amount of
>overshoot.
>
I have used two different PIDs and from the additional info gathered
here on AC, it seems safe to state you overshoot can be minimized, but
you do so at the price of a much longer recovery time between pulls.
It is sort of like driving a car- if you speed up to a point the odds
are that you will skid past it when you lock the brakes and have to
back up. If you drive towards the point slowly, you will be able to
stop right on the mark but it takes a lot longer to get there.
Randy "Andy- Greg? Where are you when we need you?" G.
http://www.EspressoMyEspresso.com
|
| | | |
Date: 31 Jul 2006 13:33:40
From: Johnny
Subject: Re: Silvia thermowell
|
"Randy G." <frcn@DESPAMMOcncnet.com > wrote in message
news:s0dsc21te7to1ombcu4umh20prsbip257q@4ax.com...
> "jggall01" <jggall01@yahoo.com> wrote:
>
> >FWIW, my limited experience with that Fuji has been that 3F overshoot
> >is about what you will get. But the recovery time is acceptable, even
> >though you get some overshoot. And the stability at idle is very good
> >with the PXR3. But I have never been able to completely get rid of the
> >overshoot after pulling a shot.
> >
> >If you can get 2 minutes to 2:30 recovery time after a shot, don't
> >worry about the overshoot. I doubt you will be able to significantly
> >improve the recovery time below 2 minutes, regardless of amount of
> >overshoot.
> >
>
>
> I have used two different PIDs and from the additional info gathered
> here on AC, it seems safe to state you overshoot can be minimized, but
> you do so at the price of a much longer recovery time between pulls.
> It is sort of like driving a car- if you speed up to a point the odds
> are that you will skid past it when you lock the brakes and have to
> back up. If you drive towards the point slowly, you will be able to
> stop right on the mark but it takes a lot longer to get there.
>
Quite a good analogy but my experience with controlling industrial
mechanical stages with PID's is that for PID control of a car that would not
be so, especially with modern anti-skid brakes and good tires. Most of that
problem is the driver using a foot instead of a PID ;-)
With stages, which tend to be high mass, moving and accelerating quite
quickly, you could aim for a spot and have it stop on it within a few
microns of the target in milliseconds.
The problem with Silvia from what I can glean from the rest of this thread
is that there is a large thermal hysterisis between the element and the
sensor, particluarly in the element itself. You don't have that on a motor
vehicle unless it is really really BIG, more like a battleship kind of size.
And then a battleship is a much beter analogy as brakes on almost all motor
vehicles tend to be up to the task, but on a sea vessel all you can do to
slow down quickly is reverse thrust, a not very efficient process even when
going forward.
So Silvia has a lot of intertia, hard to get going and then once going, hard
to stop. Greg says the element stores heat, so you have this lag when the
element is switched on, the heat from the element has to heat up whatever is
storing the heat to get to the water. Then when the heat is switched off,
that storage keeps supplying residual heat to the water longer than you want
and overshoot occurs. At least that's the theory.
I like to use the analogiy for the 'I' of PID to a rubber-band (or clock
spring). So it's like pulling a heavy weight with a rubber band: you keep
moving, stretching the band and nothing happens until you finally wind-up
enough over time to overcome the interia and it starts to move. But then it
has less resistance so you have to act quickly to back off the tension to
avoid it turning into a catapult. That's the 'D' part which acts based on
how quick things are changing. Trouble is if you back off too quick you
lose it, so you have to do it 'just so': tuned. As you can see from the
analogy, this is worst right around the target, where that big old inertia
weight wants to keep moving as you try to stop on the dime, overshoots, so
you go back the other way and it overshoots again. Oi! The 'P' part is just
the direct force you put to it based on the error(difference between where
you are and where you want to be) so it becomes insignificant as you get
real close to the target.
What you need with Silvia and the lag is a predictive PID, preferably
psychic ;-)
Anyway, somehow you need to take account of the thermal intertia in the
feedback loop between the PID sensor and the heating element. This also
includes the water in between the element and the sensor just to confound
things.
What one might do to overcome thermal lag is to increase the D term but, as
anyone who has tried to tune manually knows, there's a point where the D get
so big that it causes a feedback loop and you get oscillation in the system
and it doesn't settle to a steady point, keeps umming and erring about where
it wants to be.
Of course the terms all interact with one another so you have to reduce the
other terms to prevent instability. Evn then there is a point at which you
cannot add any more D without causing instability even with reducing the
other terms. It's a balancing act and there will be a best case scenario
for a particular system for a particlur required response. There's a lot of
info out there about this if you search. Have fun and caveat emptor.
And, as someone else pointed out, the other way to bypass all this is to
have the supply water already at the temp you need :-) An approach that Andy
proposed here:
http://groups.google.co.in/group/alt.coffee/msg/48f5560fc4d1bf39 and has
also posted about on HB
|
|
