Transistor oven.

[Herb]

Hi.
I wish to use a monolithic transistor array in a circuit where the die
temperature is temperature stabilized. I am hoping to find a circuit
design for a 'transistor oven' whereby one transistor is used as a
temperature sensor eg by monitoring the Vbe, and one of the transistors
(possibly the same one) as a heater to raise the temperature to ~50C. I
would appreciate if someone would suggest a circuit to achieve this. I
am looking for a circuit with the minimum of external components.
Temperature accuracy of +/-5% sould be sufficient.
TIA
HMV

 

[Ian Stirling]

Hi.
I wish to use a monolithic transistor array in a circuit where the die
temperature is temperature stabilized. I am hoping to find a circuit
design for a 'transistor oven' whereby one transistor is used as a
temperature sensor eg by monitoring the Vbe, and one of the transistors
(possibly the same one) as a heater to raise the temperature to ~50C. I
would appreciate if someone would suggest a circuit to achieve this. I

Consider the problems of differential heating - if you heat it
using internal components, then you're going to have fairly significant
thermal differentials across the die.

am looking for a circuit with the minimum of external components.
Temperature accuracy of +/-5% sould be sufficient.

+-5% means nothing.
+-5% in F, C, K, R, ...

 

[Rene Tschaggelar]

+-5% means nothing.
+-5% in F, C, K, R, ...

There is only one physical temperature unit, and
this is Kelvin. All the others are history biased
convenience units.

Rene

 

[Glenn Gundlach]

Hi.
I wish to use a monolithic transistor array in a circuit where the die
temperature is temperature stabilized. I am hoping to find a circuit
design for a 'transistor oven' whereby one transistor is used as a
temperature sensor eg by monitoring the Vbe, and one of the transistors
(possibly the same one) as a heater to raise the temperature to ~50C. I
would appreciate if someone would suggest a circuit to achieve this. I
am looking for a circuit with the minimum of external components.
Temperature accuracy of +/-5% sould be sufficient.
TIA
HMV

I did exactly that for a log amp about 10 years ago. Used a MAT-04 with
one transistor as the temp sensor, 2 transistors as the heaters and the
4th as the log element. The circuit was actually from Jim Williams. I
used 1/2 AD708 as the heater driver and the other half as a lamp
driver. The LED is red when under temp and turns green at temperature.
Using multiple meters on the unit as temp readout and drive level,
placing a finger on the package causes the drive to increase with no
measurable change in the temp monitoring point. IIRC 469 mV corresponds
to 69 C. From room temp to operate takes 10-15 seconds. The heaters run
on 12 volts with the drive limited to 30mA to not damage the
transistors during initial warm up. Actual maintenance current is only
a few mA when warmed up.
GG

 

[Jim Thompson]

Consider the problems of differential heating - if you heat it
using internal components, then you're going to have fairly significant
thermal differentials across the die.

+-5% means nothing.
+-5% in F, C, K, R, ...

Die mount over an insulator. Metal can package, evacuated.

Resistor (heater) pattern around edge of die.

It's been nearly 40 years now, but I believe we were doing about
0.05°C stability.

...Jim Thompson

 

[Gerhard]

Hi.
I wish to use a monolithic transistor array in a circuit where the die
temperature is temperature stabilized. I am hoping to find a circuit
design for a 'transistor oven' whereby one transistor is used as a
temperature sensor eg by monitoring the Vbe, and one of the transistors
(possibly the same one) as a heater to raise the temperature to ~50C. I
would appreciate if someone would suggest a circuit to achieve this. I am
looking for a circuit with the minimum of external components. Temperature
accuracy of +/-5% sould be sufficient.
TIA
HMV

See "Self-Heated Transistor Thermostats Individual Components"
Electronic Design, Feb 9, 1998. It employs a TIP41 as the heater-sensor.

Gerhard van den Berg

 

[Ian Stirling]

Die mount over an insulator. Metal can package, evacuated.

Resistor (heater) pattern around edge of die.

It's been nearly 40 years now, but I believe we were doing about
0.05?C stability.

Right.
I'm not arguing that this is impossible - but that it does sound a tad
unlikely that this is the sort of scheme proposed.

 

[Jim Thompson]

Right.
I'm not arguing that this is impossible - but that it does sound a tad
unlikely that this is the sort of scheme proposed.

Well-insulated is most of the trick.

And proportional control is a necessity. Heaters banging on and off
wreak havoc in analog circuits.

...Jim Thompson

 

[josh lawton]

I think the temperature must be higher unless you really want to limit
the application. I would take transistor and short across vdd.
Regulate the current to obtain the desired temperature.

 

[josh lawton]

Jim Thompson wrote:

Well-insulated is most of the trick.

And proportional control is a necessity. Heaters banging on and off
wreak havoc in analog circuits.

...Jim Thompson

I would love to see how you can reach this without some calibration.

 

[Spehro Pefhany]

Jim Thompson wrote:


I would love to see how you can reach this without some calibration.

Why would *stability* require calibration?


Best regards,
Spehro Pefhany

 

[josh lawton]

I think you must interpet things differntly. I assume .05?C stability
is the tolerance. I would assume the circuit is stable, otherwise why
build it?

 

[Jim Thompson]

I think you must interpet things differntly. I assume .05?C stability
is the tolerance. I would assume the circuit is stable, otherwise why
build it?

0.05°C was the measured _variation_

As Spehro points out, it is only the _absolute_value_ that needs
calibration.

...Jim Thompson