A BT speedometer. Generally, I watch the BT rise and count for 10-20secs. to guesstimate how fast I'm going with the power being applied by the variac or have an general sense from the graph but don't always hook up the computer. I use a count-up timer also. Is there a meter that could display the rate of BT rise? A display with 2 digits. Something that could split off the existing BT TC or pair 2 up if necessary.

When I was roasting with a variac, I prepared a 30 second by 30 second profile chart on a piece of paper. This string of numbers were my setpoints. I twiddled the variac to make sure the actual temperature stayed at or close to them throughout the roast. I also recorded the actual temperature at each 30 second checkpoint to see if I was staying close, getting closer, or drifting away

It took a while to learn how fast the temperature responded to the variac twiddles, but once I had that pat, I got pretty good at this style of "manual ramp/spoak control"

When Fuji came in with ramp soak controllers at 1/10th the price they had ever been before, I decided to let the black box do all the work

My point is you don't really need a rate of rise meter, if you are controlling manually, pencil and paper is no extra work; if you have auto control, the meter is inside the black box.

The advantage of manual control is that I always tweaked the variac up, never down, so I was guaranteed never to have falling environmental temperatures. if I ran ahead of the profile, I'd just keep he hea steady, until the desired profile caught up again with the actual temperature.

That is a good question - without using a computer it seems to be difficult. Any "split" off of the BT probe would be required to have some sort of memory to hold a BT temperature so you could compare the temp change over time. Of course, as jim suggests, you can use your own memory with a pen and pencil. If you have a laptop, but can't hook up your meter, you can still make a simple spreadsheet to manually enter the temps at 15 or 30 second intervals and get a rate of change. If you can hook your meter up to a computer, I've generated a spreadsheet that performs a data call on the file as it's being written to, for realtime graphing and derivative generation. Every once in a while, the spreadsheet tries to access the file at the precise moment it's being written to and I have to clear the error (takes about 2 seconds - it's one button press). I also am running the spreadsheet on a slow laptop - intel atom processor - so with a core2 duo it should have much less trouble. If you're interested in the spreadsheet i can email it. I manually control BT via ET for my roasts based off this spreadsheet and I can hit +-15 seconds or +-3 degrees pretty easily.

another_jim wrote:The advantage of manual control is that I always tweaked the variac up, never down, so I was guaranteed never to have falling environmental temperatures. if I ran ahead of the profile, I'd just keep the heat steady, until the desired profile caught up again with the actual temperature.

I do have to dial down the variac around first crack and beyond. But I do maintain a positive BT rise and a level or positive roast chamber ET at all times.
Spencer
My datalogger can connect to the computer and figured the calculation could probably done there but often I don't hook up to the computer unless I'm really interested in analyzing the data. I get nervous having my main laptop around the roasting environment. Maybe someday I'll add a small screen used laptop dedicated to my roasting setup.
As Jim suggested, it's not that hard to do the basic calculation on the fly. I generally will just watch the BT degrees rise while counting say for 15secs. and multiply times 4. My BT TC is a fairly fine bead tip that reacts quite quickly so I have a pretty good idea how fast it's progressing from experience.
The thought of just looking at a display and seeing "18" as say the present BT rise speed during ramp seemed like a sweet tool.
I'm definitely a variac vs PID type person, and like the manual adjust of the dial along with the "kill a watt" meter for voltage level numbers.

You might try a simple analog RC differentiator circuit to display a rate of change connected between the thermocouple adapter and a voltmeter.

I don't know how you'd calibrate it, but with time you'd get a feel for the numbers your seeing versus your graphed profile.

-Chris

yakster wrote:You might try a simple analog RC differentiator circuit to display a rate of change connected between the thermocouple adapter and a voltmeter.

I don't know how you'd calibrate it, but with time you'd get a feel for the numbers your seeing versus your graphed profile.

I was going to suggest the same thing, along with a Fluke 80TK thermocouple module to amplify the t/c output.

But, after further consideration I decided that the fluctuations in the input signal might render realtime differentiation meaningless. I have plotted dBT / dt for a few of my roasts and found that an awful lot of smoothing is required to make the results useful.

If there is a way to incorporate smoothing into the analog circuit, then this would be a nice solution, IMO.

Jim

Seems as though this unit does it? Part way down the list it states " Math Expressions: Display the Rate Change, Max/Min. or Timed Average of Measurements" meter
I would just need to find a used one on ebay for a fraction of the price

+1 on using a PID instead.

But if you're set on doing it manually with a meter, the Fluke 189 might be suitable- I couldn't make sense of the specs: http://us.fluke.com/usen/Products/Fluke+187+189.htm?catalog_name=FlukeUnitedStates

JimG wrote:I was going to suggest the same thing... but, after further consideration I decided that the fluctuations in the input signal might render realtime differentiation meaningless. I have plotted dBT / dt for a few of my roasts and found that an awful lot of smoothing is required to make the results useful.

If there is a way to incorporate smoothing into the analog circuit, then this would be a nice solution, IMO.

This is basically how an RC network works, by smoothing out rapid changes. There is a fixed time constant determined by the values of the resistor and capacitor in the RC network that determine how long it takes to charge up the capacitor to a given voltage, the differentiation is just an end result of this smoothing because the voltage doesn't immediately jump up.

I think this method is worth a try if the PID is out of reach.

I will qualify this with the fact that it has been quite a while, though, since electronics classes, and I'd have to do more reading before I'd do this myself.

-Chris

I'm able to report success with a breadboarded RC circuit that measures and displays BT rise per minute. While my simple prototype has some minor issues, and is not terrifically convenient to lug out to the roaster in its present state, I am satisfied that the basic design is good and can be effectively applied.

For convenience, I used a Fluke 80TK adapter to amplify the raw thermocouple signal. This could just as easily be done with one of the several IC's available for this purpose.

The 80TK outputs a signal, in millivolts, equal to the temperature sensed by a type K thermocouple. I took this millivolt signal and ran it through a very simple differentiator circuit comprised only of an op amp, capacitor, and resistor.

For the prototype, I used C = 1000uF, and R = 60k ohms. This gave me a direct readout in degrees per minute on the millivolt scale of my DMM. I had expected that the displayed rise / min values would have too much noise to be of use, but I was wrong. Even without doing any filtering, the blue line in the graph below shows that the rise / min values are fairly smooth.



In the above graph, the blue line represents the values of rise / min that were derived from the differentiator circuit and displayed on the Fluke 189 DMM in real time. The red line represents rise / min values that were computed, post-roast, using the temperature profile that was captured during the roast.



The above graph is just a simple record of the roast temperature profile. The blue line is the environmental temperature, and the red line is the bean temperature. These data were recorded on a Fluke 54 II.

Jim