General Goal:

Physical characteristics: (Drawing #1 - 14KB jpeg file)
The unit will consist of a thermal grid (approximately 8x10 cm), made of narrow strips (approximately 3 per cm) that can each be set at different temperatures.
Thermal Patterns: (Drawing #2 - 65KB jpeg file)
Initially, only two temperatures will be available, but it will be possible to assign either of the two temperatures to each element in order to produce any of the thermal patterns (Drawing #2).

First Prototype (feasibility study): (Two small Peltier per element).

Construction: (Drawing #3 - 32KB jpeg file)
The unit would be made of 24 or 27 pieces of brass stock, 3/32" square, 10cm long. The temperature of each element would be controlled by two small thermoelectric (Peltier) devices set in a pattern as seen in the drawing #3.
One element with 2 small Peltier devices (Photo #1 - 12KB jpeg file) was assembled and tested.  The results were very disappointing: the temperature gradient along the element was too great  and this design was immediately abandoned.

Second Prototype (feasibility study): (Each element made of a 5cm long 3/32" square brass stock).

Construction:
This unit is based on a 5cm x 5cm Peltier device with a flat 5cm x 5cm brass water tank as a heat sink on one side, and 15 elements made of 3/32" square brass stock on the other. Each element (Photo #2 - 179KB jpeg file) has a thermocouple soldered through a small hole at its middle and a 10 cm long Nichrome heater wire is threaded through its whole length in both directions.  The tube is filled with thermal epoxy (Delta Bound 152-K-A by Wakefield Engineering).
Thermal Pattern:
Initially, only two temperatures will be available: a lower temperature that will be determined by the Peltier device and a higher temperature delivered by those elements that have their heaters turned on.
Feasibility Test: (Drawing #4 - 16KB BMP file)
Four elements as seen in photo #2 were epoxied (using thermal epoxy) to the Peltier device leaving a 1/32" air gap between the elements.
The Peltier temperature was controlled by means of a prototype unit designed in 2003 for the "Oral Hydration" project.  This device connects to a PC through a USB port.
A preliminary Visual Basic program was written to control the Peltier (Software Panel #1 - 42KB jpeg file) One of the element thermocouples and a special thermocouple, epoxied to the heat sink side of the Peltier, were used for Peltier temperature feedback.
The heaters in the two elements adjacent to the Peltier control element were connected to a variable power supply and the thermocouple of one of them was connected to a thermometer for manual heat control.

May 2, 2007 - With this setup we could only reach a temperature differential of 10°C, at which point the current through the nichrome wires was 1.2Amp and the insulation on the wires started to melt.  However the design should work after some modifications: using 3/32"channels rather than square tubing would make it easier to build the elements and may permit doubling the length of the heater wires and more evenly packing the assembly with epoxy. The heater wires should also be welded to conductor wires before they exit the epoxy thus keeping them at a lower temperature.

Element Pattern: (Drawing #4 - 16KB BMP file)
#1 - No Heat
#2 - Heat - Secondary Temperature
#3 - No Heat - Peltier Temperature = Primary Temperature
#4 - Heat

May 10, 2007 - The assembly was reworked using the same elements but after the Nichrome wires had been shortened, soldered to conductor wires and epoxied to the side of the assembly.
The test was repeated using the same "element pattern" as above.  Although a much higher current could be used without any problem (2.2  vs 1.2 Amp) only a differential of close to 14°C could be reached as seen on the (improved) software window (Software Panel #2 - 40KB jpeg file). The current through the Peltier was then 2.8Amp, the maximum attainable with the current Peltier controller.

The test was repeated using a different element pattern: (Drawing #5 - 16KB BMP file)
#1 - Heat - Secondary Temperature
#2 - Heat
#3 - No Heat
#4 - No Heat - Peltier Temperature = Primary Temperature
A temperature differential of 30°C could then be reached without any problem.  The heater current was just below 2.0Amp and the Peltier current was only 1.1Amp as seen on the software window (Software Panel #3 - 40KB jpeg file).  Notice also that, in this case, the Peltier temperature was equal to the command temperature.

Remarks:
1 - The water bath temperature was around 8°C.
2 - The "Peltier Temperature" on the "Software Panel" is actually the temperature read by a thermocouple connected to an element with a disconnected heater: not the actual temperature of the Peltier.

Conclusion:
1 - A larger gap and insulating material between elements will most likely be necessary, as suggested by Dr. Mark Hollins. But there's hope. 
2 - Keeping the heating elements farther from the Peltier device would also most probably minimize the influence of each element on its neighbor.

Third Prototype (feasibility study): (Each element made of a 5cm long 3/32" brass channel on top of a 3/32" brass square stock).

Construction: (46KB JPEG file)

The difference between this unit and the previous (second) one is in the construction of the elements.
1 -  Instead of  feeding the thermocouple and heater wires through a square brass stock, a brass channel was used ( HTML file).  This gave better control over the positioning of the heater wire inside the conduit by holding the wires in the epoxy, against the channel ( Construction Photo).
2 - The assembled channel was then epoxied on top of a piece of square brass stock, thus doubling its height, to keep the heater farther from the Peltier.  The thermocouple was threaded through a hole in the piece of square stock.
3 - The joints between the heater and conductor wires were encapsulated within a short piece of 1/8" channel to minimize heat exchange between elements. 
4 - Ceramic tape by Cotronics Corp. was used as an insulation between elements.

Test:

June 28, 2007 - The same test pattern shown in drawing #5 used May 10, was repeated with the new elements (Drawing #6 ). There were 4 elements when the tests were started, but when increasing the heater current to some value between 1.5 & 2.0 Amp., a short developed in both heated elements. One element was removed.
1 - The difference between the two temperatures (~10°C) changes little when changing the Peltier temperature without changing the heater current (set @ 1.0Amp): Panel 1 and Panel 2.
2 - Increasing the heater current to 1.5Amp.increased the difference in temperature to 25°C: Panel 3
3 - Increasing the  heater current some more increased the difference in temperature but caused a heater element fault before it could be recorded.

One faulty element was removed for inspection (see Meltdown Analysis below), and some more testing was performed using only 3 elements. The center one was at low temperature (Peltier temperature) the other two on either side were heated (Element Temperature).

4 - Heater current set at 1.6 Amp.gave a 25°C difference in temperature  Panel 4
5 - Heater current set at 1.8 Amp.gave nearly 32°C difference in temperature  Panel 5
Notice that the Peltier temperature was lowered in order to avoid any more meltdown.

Meltdown Analysis:

Photos of heater element after meltdown:
1 - Heater element in its entirety. (Meltdown #1)
2 - Connection with conductor wire that had signs of excessive heat damage (after removal of melted insulation ).( Meltdown #2)
3 - The other connection also with obvious heat damage.(Meltdown #3)

Photos of heat tests of various insulations :
1 - Tefzel melts at 1.3 Amp.  The green jacket (made of Tefzel a type of Teflon) that, so far, has been used throughout the experiment for heater wire insulation (from Vector 28 gage wire).
2 - Omega black Teflon jacket possible replacement for the Tefzel melts at 1.5 Amp.(Omega PTFE protective tubing TF-BK-30).
3 - A blue TFE Teflon wire insulation, also a possible replacement for the Tefzel did quite well at 2.0Amp. This jacket was stripped from a piece of Belden 83001 hookup wire.
4 - A piece of black shrink tubing next to a piece of transparent shrink tubing (that turned dark from the heat) they both also did well at 2.0 Amp. (Black shrink tubing by 3M).

Conclusion:

Element rework:
1 - Blue Teflon insulation
2 - Black shrink tubing over the solder joints
3 - Blue Teflon conductor wires (instead of the white ones).
4 - Double the length of the heater wire if there is enough room inside the brass channel.

Fourth Prototype (feasibility study): (Heating elements made of coiled Chromium wire from K type thermocouple).

Construction:

The K type thermocouple wires that have been used throughout this investigation have one Alumel conductor and the other is Chromel an alloy of Chromium and Nickel just like the Nichrome (Nickel Chromium) heater wire we use. Since the thermocouple conductors come with a Teflon insulation and their overall outside diameter is significantly smaller than any we could make, we decided to give it a try and use it as heater wire.  Because of its small OD (Outside Diameter) we were able to wind it onto a 1/64" piano cord thus forming a tiny coil that could fit inside our brass channel stock. The effective length of the wire was 4 times the length of the grill element thus twice the length of the heater wires we previously used.  Another advantage of using a coil is that the heater wire does not crisscross itself thus avoiding hot spots that are potential cause for over heating and shorts. This truly sounded promising.

Test:

July 23, 2007 Unfortunately, the results of the test were the most disappointing. With a 2 Ampere heater current we could only reach a 20°C differential (Test#1Test#2 - Test#3 - Test#4 ). 

Conclusion:

Apparently the coil was not a good idea. We assume that heating the sides and bottom of the brass channel instead of concentrating the heat at the top caused too much wasted heat for which the Peltier had to compensate. We obviously have to give up the coil idea and go back to our previous design. 
1 - We will first see if the Chromel wire, that is much easier to solder than the Nichrome wire, can satisfactorily work when not coiled. 
2 - Then, if the Chromel wire is not satisfactory, we will look at the commercially available pre-insulated heater wires, such as the "Formvar-Insulated Nichrome Wire" available from "A-M SYSTEMS,INC.".
      Or MINCO custom Thermo foil heaters with optional built-in temperature sensors. Although this does not look so good (Designer Kit PDF file).

Fifth Prototype (feasibility study): (2 heating elements made of straight Chromel wire from K type thermocouple, 2 made of Tefzel insulated Nichrome wire).

Construction:

Element Construction This prototype is similar in construction to the Third Prototype but only 2 of the heater wires are made of Tefzel insulated Nichrome wire, the other 2 are made of Chromel from Omega K type thermocouple wire. We want to compare the Chromel and Nichrome as heating elements. We expect that the Nichrome would require less current than the Chromel but the difference in insulation may negate the difference in wire.  The reason for going back to Nichrome and third prototype construction is what we uncovered when taking the previous heating elements apart. To take the heating elements apart, we apply a current of 2 Amperes to the heater wire for a few seconds, until the high heat softens the Delta Bond epoxy.  Despite the high heat, the Tefzel insulation was in amazingly good condition, and we assume that, as long as the connections are kept  epoxied inside the brass channel, there should be no melted insulation problem. 
    To make sure the heater wires were as close to the surface of the brass channel as possible, pieces of common hookup wire were used as stuffing material and the elements were squeezed by means of Binder Clips while the epoxy cured.

Test #1: August 20, 2007. Back to Nichrome wire heating elements. Drawing #6

Fifth Prototype Nichrome heater wire results
View the entire test panel:  1.7 Amp. test - 1.8 Amp. test - 1.9 Amp. test
As shown in drawing #6, the "Element temperatures" indicated on the test panels above were measured at the surface of the heated element sandwiched between 2 colder (not heated) elements.  The temperature of the other element, the one that had one side at room temperature (Drawing #6b) was at  68 °C when a heater current was 1.8 Amp.as shown on this panel.

Test #2: Chromel wire heating elements (straight: not coiled). Drawing #7

Fifth prototype with Chromel heater wires
View the entire panel: 1.7 Amp test - 1.8 Amp test - 1.9 Amp test - 2.0 Amp test -

Conclusion:

The Chromel is obviously not the way to go. 
The new technique for attaching the Nichrome and conductor wires together and making sure the heater wire is as close to the brass as possible seem to make quite a difference.

To Do:

We will now make a dozen more elements and see if the Peltier can handle them. 
Then, later, we will need an electronic monster to drive all this.

First 5cmX5cm Model: (22 January 2008).

Finally resuming this project building 12 heating elements following the instructions for the Grill Element #2.
As was mentioned in the instructions for the Grill Element #2: "This single procedure needs more accuracy in weld timing and wire flattening and positioning, but, with practice, seems to be the answer." We had little problem welding 5 heater wires on the first day, but on  the second day the "Weld Energy" of 3 was too high and caused the wire to actually be projected across the floor. "Weld Energy" of 2 was still insufficient to melt the copper wire.  Adding a variable transformer on the UNITEK power line and fine tuning the voltage to somewhere between "Weld Energy" of 2 and 3 seems to have help: we were able to resume welding the heater wires.
It appears that 90 VAC is working quite well leaving the power on continuously. We will try this again Monday hoping we do not get any more bad surprise.

Well, we did encounter more difficulties with the welding, although we manage to make the 13 heaters needed + 3 as backups. In the process we lost  some Nickel-Chromium wire and Tefzel insulation both becoming too short to be used after several welding attempts. 

A solution to this problem would be to weld the 1" piece of copper wire to a very short piece of Nickel-Chromium, resulting in little waste when the weld fails. Then, when we have a good weld, weld the heater wire to the short piece of Nickel-Chromium, and cover the whole thing up with the Belden jacket (see illustration).  However, this procedure will present a problem if the extra piece of Nickel-Chromium wire is not properly  physicaly connected to the heat sink.

Grill Thermode 5cm x 5cm Model (April 7, 2008).

Started in May 2007
Last revision: 2008 April 7