Arpi wrote:A thick probe works as a natural averaging filter. A naked thermocouple reacts very quickly and adds noise. If the thermocouple is placed in a draft of air, then it will add noise (up and downs). Noise is bad for pids and makes delta Temperatures difficult to read.

Rafael, do you have an opinion on grounded vs ungrounded TC? Should I apply the same reasoning and prefer the ungrounded probe (should be more stable, while a bit late)?

Hi

I am not very familiar but I think the ground is to protect against inductive signals (a braided shield used in wires for noisy industrial environments with motors,etc) going around the long metal wires of the thermocouple (they are like antennas that pick noise). If you get a probe with a braided metal jacket (like I have) then I think you won't need a grounded probe. The probe would ground itself if you use a metal adapter attached to the roaster.

Cheers

Arpi wrote:I am not very familiar but I think the ground is to protect against inductive signals (a braided shield used in wires for noisy industrial environments with motors,etc) going around the long metal wires of the thermocouple (they are like antennas that pick noise). If you get a probe with a braided metal jacket (like I have) then I think you won't need a grounded probe. The probe would ground itself if you use a metal adapter attached to the roaster.

From my understanding, induction is not a factor here and in this case, being grounded would bring electrical interference rather than reduce it.

Omega web site has good information on thermocouples:
http://www.omega.com/prodinfo/thermocouples.html
http://www.omega.com/techref/themointro.html
Their Probe configurator is also very useful: http://www.omega.com/config/probeconfig.html

The latter includes a good picture explaining the different types of thermocouple junctions:


From omega.com.

I'm not an electrical engineer but from what I understand, as the grounded thermocouple junction is soldered to the probe sleeve, it will react faster to temperature changes, but it could also be affected by grounding of the TC wires to ground through the sleeve itself (ground loop), hence my worries. As the Hottop frame is grounded, if I fix the probe to the frame (hole in the back of the roaster chamber), metal to metal, would I ground the TC junction, leading to incorrect temperature readings?

As for ungrounded junctions, Omega specifies the insulation resistance: >1 GΩ for a 1/8" sleeve (>500V). Therefore, an ungrounded probe would be electrically isolated from outside interference (provided that the meter is also isolated)... but would also be slower to react to temperature changes.

Remaining unsure about ground loop issues with grounded probes, I will probably avoid the risk and go with ungrounded probes.

PS. Omega has also a 13 pages PDF explaining in details how thermocouples work, for the coffeegeeks also interested in electronical geekery: http://www.omega.com/temperature/z/pdf/z021-032.pdf

I use a grounded probe for BT in my Hottop. In addition to slightly better response time, I have also convinced myself that stem effects should be reduced with a grounded probe. Since I currently use a 1/8" sleeved probe stuck through the bean chute cover (a la Randy G), I think stem effects might be pretty important.

I have not experienced any ground loop problems while using a hand-held meter (HH506RA or Fluke 54) or commercial PID controller (Watlow 93).

But I have experienced solvable ground loop problems using thermometry devices (TC4/arduino, e.g.) that are powered by a USB interface while a power adapter is plugged into the PC.

Ground loop problems appeared when using ungrounded (i.e. 2-prong plug) power adapters for the PC. The solution for me was to either use a 3-prong PC power adapter, or provide a separate earth ground for the USB's 5VDC (on the negative side, of course).

I think the problem arises from the fact that a 2-prong ungrounded PSU leaves the negative side of the DC supply voltage floating. So you end up with a potential difference between the negative thermocouple lead (which is at earth ground potential) and the GND circuit in your thermometry device. In my case, this resulted in a shift in temperature readings on the order of 10F, IIRC.

But a grounded PSU for the PC fixed the problem completely on the roaster.

In other instances, i.e. grounded probes on an espresso machine, running a ground wire from the GND circuit on the thermometry device to a ground prong on an outlet has also fixed the problem. Based on my experience I think that any reasonably low impedance connection from the machine frame to the GND of the thermometry device should resolve the problem caused by the floating supply voltage from a 2-prong PSU.

Jim

Louis.

In a general sense, going ungrounded will probably good for the Bean probe (BT) but for the Environment probe (ET) I would suggest at least a grounded or even exposed bead.
I do not know how quickly the air temperature can change in the Hottop and how relevant it is for you but in my roaster an ungrounded probe type thermocouple introduced a significant lag in the measurement of the ET. Sometime would show a temperature that was either much higher or much lower than the actual and making me overcontrolling and overshooting my targets significantly.
I have changed it to a thin grounded T/C and introduced ground loop issues. I have to now insulate the T/C attachement to the roaster to alleviate that problem. Still the faster response is well worth it
As Raphael stated, there is also the risk of overreactive T/C that do not fit well with controlling devices (PID), but I do not think that will be the case for your application.

JimG wrote:...running a ground wire from the GND circuit on the thermometry device to a ground prong on an outlet has also fixed the problem. Based on my experience I think that any reasonably low impedance connection from the machine frame to the GND of the thermometry device should resolve the problem...

I tried doing this on my roaster. Although it sometime seems to help, some other time it does not seem to make a differences. Not being an electrical anything, I am going at it with a trial and error method.

Jim, Sylvain, thank you, this is exactly what I was trying to understand.

JimG wrote:In other instances, i.e. grounded probes on an espresso machine, running a ground wire from the GND circuit on the thermometry device to a ground prong on an outlet has also fixed the problem. Based on my experience I think that any reasonably low impedance connection from the machine frame to the GND of the thermometry device should resolve the problem caused by the floating supply voltage from a 2-prong PSU.

So as the Hottop frame is grounded to the wall socket, having the grounded probe in contact with the frame might actually be a good thing, as long as the meter is also grounded to the same ground (probably a pleonasm) (through its own three prong power supply or with a separate ground cable)?

So the best would be an exposed probe for ET (I'll only need to ensure I don't accidently break it, especially if I choose to mount it on the chute cover, else I would go with a grounded probe) and a grounded probe for BT (unless I plan to use a PID to control the roaster: not in my plans), and ensure everything share a common ground to avoid interference from ground loop.

JimG wrote:I use a grounded probe for BT in my Hottop. In addition to slightly better response time, I have also convinced myself that stem effects should be reduced with a grounded probe. Since I currently use a 1/8" sleeved probe stuck through the bean chute cover (a la Randy G), I think stem effects might be pretty important.

Stem effect...

Always learning something! I found with Google this short white paper explaining the issue well: http://www.nanmac.com/documents/the-stem-effect.pdf.

I suppose stem effect would also be present for a BT probe fixed to the backwall of the roasting chamber (it could damp the temperature difference between the actual ET and BT, showing higher BT), while it would be less an issue with an ET probe through the same backwall (which should itself be closer to the actual ET).

How would one fix probes through the backwall with thermal insulation between the probe and the frame? Is there some COTS (commercial off the shelf) adapters to do so? High temperature resistant rubber grommit?...?

A grounded probe is not susceptible to ground loops when connected to a battery operated meter. There is no path for loop current flow. I've run grounded probes in the HotTop with out issue.

Any data connection to a PC should be optically isolated. USB 0v may be very different than Earth. It can vary between front and back connectors. Plugging in a USB drive will often RESET a PC due to improperly designed chassis / mobo / PSU

Test the system by connecting the meter to the probe with the DUT (Device Under Test) unplugged. Note the reading. Plug in the DUT. There should be no change in the reading. Unplug the DUT. Connect bench power supply. Should be no change. Disconnect bench supply. Connect PC. Should be no change. Connect bench supply. Should be no change. Connect DUT. Should be no change.

Most, if not all, meters use differential inputs which are noise canceling. Effectiveness depends on the circuit design, pcb layout and construction, parts tolerance and assembly. The parameter used to measure noise immunity is CMMR (Common Mode Rejection Ratio). Bigger is better.

Laboratories and places like recording studios which deal with µV level signals use a single point Technical Earth. Each piece of gear is connected to that and ideally there should be no earth connection between individual pieces of equipment.

GROUND is a term almost universally misapplied. An electronic component may have Signal, Chassis and Earth Grounds, all electrically isolated.

MaKoMo wrote:These drops are not an issue anymore with Artisan. So the Omega HH506RA works well in this setting. However, I updated to a 4-channel Voltkraft K204 (without memory for the cheaper price) just because I wanted to see more ET!

Now I am starting to use both in parallel, this gives me 6 curves in Artisan. The two extra once I spent for Ambient temperature and cooling air temperature.
M.

Can anyone point me to a retailer that offers Voltkraft K204 in the domestic U.S? I found some UK sites that sell it, but I was hoping to buy local. Also, is there a brand/model equivalent to the K204 that doesn't contain data logger memory as well, at the cheaper price?