The new analog PCB is working well for me (haven't gotten feedback yet from Randy and Ed). There are still a couple of unpopulated boards on my shelf if anyone is interested.

Jim

EDITED 4/4/2010 to make sure its clear I am speaking of the analog version of the circuit. Andrew's digital version is showing great promise, too.

Progress report on the digital rate of change of temperature meter

It is going well.

I have my prototype reading the thermocouple 4 times a second and displaying it on the LCD along with the ambient temperature. Depending on the thermocouple type and construction, the temperature range is -270 degrees C to 1200 degrees C.

It's not yet doing the rate of change for the thermocouple, but that's really just a minor bit of coding. I also have yet to code the RS232C serial output to a PC/bluetooth modem.

The circuit uses no surface mount components, no user calibration is required, and there are no trimpots to adjust, not even for the contrast of the LCD. Only a single precision component is required, an LT1634CCZ-5 precision voltage reference, cost USD2.50 (plus shipping) from Linear Technology. Resistors are all 5% tolerance. I have even eliminated the 32.768kHz watch crystal that I was using earlier, and I'm glad of it because they are fiddly small things as bad as any surface mount component. The other ics are a CMOS 4007, and an LM358 dual op amp. There is also a single NPN transistor.

The current consumption is 10.5mA - sorry it's so high, but I have to run the PIC at 32MHz which chews up the power more than I'd like. A normal 9V battery should run it for around 15 hours, or you can power it from an AC or DC wall wart, 9 - 15V. If you want a backlight on the LCD you'll have to run the thing on a wall wart, since backlights consume 100mA and up. I'll probably add a battery test facility after everything else is working - the PIC has an onboard 10 bit ADC that we aren't currently using.

The microcontroller is a PIC 16F1827, and they cost USD1.52 each (plus shipping) from Microchip Direct. It can be programmed using Microchip's own PICkit 2 or PICkit 3, or a cheap PICkit 2 clone programmer (the PICkit 2 design is open source), or I will be able to send you a programmed chip for USD10.00 (the extra covers my handling).

The unit has a single pushbutton that allows you to change the settings: temperature units (degree C or degree F, and I'll probably add kelvin. Anyone want Rankin?), thermocouple type (K, J, T currently supported, requests for any others will be taken kindly), and display contrast. Settings are stored in non-volatile memory, so you won't have to fiddle with them often.

Cheers,

Andrew

Very cool. I can't wait to try one.

Jim

Okay, so what's the process required to acquire this wonderful tool?
I'm wetting myself in excitement.

I will be interested in this utility too.

The thread above is very exciting. A few months ago I have started with two roasters: old little Probat testing device built in 1970, and 2-year old manual Probat L5. And just at the beginnig of our work, I discovered that such a speedometer will be great help.

Andrew,

is it looking like all of the required components could be found from just one or two suppliers? I haven't dabbled in electronics much in the past twenty years (besides making this clock from a kit: http://www.ladyada.net/make/monochron/), but I do recall the pain of having to seek out the various components from many different suppliers.

AndrewPartridge wrote:... or I will be able to send you a programmed chip for USD10.00...

Or since it sounds like you're game for providing components yourself, any chance you'd also be willing to sell all the required components together as a 'kit'? (with any handling fees you feel is fair)

Rama

That is one cool clock, Rama!

Thanks for your interest in the digital temperature rate of change meter everyone.

There are still a few steps before you can lay your hands on one of these - I have to finalise the firmware and hardware (in particular, I'm waiting on an LT1634-5 sample that I ordered only a couple of days ago when the design changed and made the LT1634-4.096 I had unsuitable), then I want someone - Jim? - to lash one up on a solderless breadboard and actually use it to make sure it will do the job. I am not a coffee roaster, so it's conceivable I might have missed something critical in the design.

And lastly we need to design a PC board for it, and that will force us to decide on a specific enclosure and also the specific model of display and pushbutton (I'm assuming we want to mount all the components direct to the PCB).

Apart from the PCB, all parts are available from Digikey in the US, with the possible exception of the LT1634-5 voltage reference in the TO-92 package, which is a non-stock item. I think we can design the PCB to accommodate the S8 surface mount version as well (it only has 2 pins connected), which is stocked by Digikey. Or if you don't want to brave surface mount, you can buy the TO-92 package version direct from Linear Technology.

I'm personally relying on Linear Technology giving me a sample of the TO-92 LT1634-5, because their international shipping charges are over the top for me (USD35 anyone?). By the way, those high shipping charges make it uneconomical for me to put together kits of parts myself. I hope Linear Technology doesn't charge as much for delivery to US addresses.

Cheers, Andrew

AndrewPartridge wrote:.... then I want someone - Jim? - to lash one up on a solderless breadboard and actually use it to make sure it will do the job. I am not a coffee roaster, so it's conceivable I might have missed something critical in the design.

You can count on it.

Jim

I think I'm past the point of return. I'm thinking rates . I can easily calculate and predict when I will be arriving at certain BT points. I can now make adjustments much sooner if I'm off track. Starting to be able to notice when I'm pushing the beans too hard and am just over heating the surface. JimG do you have a smoothed out rate graph of a full roast?

Ed -

Attached below are a couple of graphs from a February roasting session for the Brazil beans in the HB roasting competition. The first graph shows the rate of rise. The second shows the BT and ET during the roast.

This roast came out flat. I believe the reasons for it being so flat are the low-ish ET after 1st crack, and the fact that I allowed the rate-of-rise to drop below 10F per minute at the end of the roast.

Based largely on the advice you provided in our recent exchange of PM's, I've altered my roast profiles a little. Now, I maintain a higher ET after 1st crack (usually 440F to 450F), and I work to avoid having the rate-of-rise drop below 10F per min. Am getting better results, but still working to improve!

One of the consequences of this change in profile, however, is a shortening of the time between 1st and 2nd. I am finding it impossible to both avoid stalling/flattening the roast and stretch the gap to 4 minutes.

Jim