Click here to go to our main page on heat and temperature effects
Click here to go to our main page on attenuators
Click here to view an amplifier temperature compensation example (new for March 2022)
New for May 2020. Temperature-variable attenuators are often used to compensate for amplifier gain changes over temperature.
History of temperature variable attenuators
The original patent US5332981A (July 26, 1994) was issued to EMC corporation, and they trademarked the name "Thermopad®" which seems to have gone the way of "Xerox" did for copiers. Here is the abstract from the patent:
An absorptive temperature variable microwave attenuator is produced utilizing at least two different thick film resistors. The temperature coefficients of the resistors are different and are selected so that the attenuator changes at a controlled rate with changes in temperature while the impedance of the attenuator remains substantially constant. Substantially any temperature coefficient of resistance can be created for each resistor by properly selecting and mixing different inks when forming the thick film resistors. Furthermore, attenuators can be created having either a negative temperature coefficient of attenuation or a positive temperature coefficient of attenuation.
The two inventors are Joseph B. Mazzochette and John R. Steponick. If you happen to see them, tell them we said thanks!
Over the years, EMC merged with Florida Labs and then was acquired by Smiths Interconnect. The patent expired, but there are other patents that followed it. PSemi (now Murata) used the technique within a digital step attenuator to compensate states over temperature (patent application US20180102763A1).
Using temperature-variable attenuators
As you should know, amplifier gain always has a temperature coefficient, and for solid-state amplifiers such as MMICs, that coefficient is always negative. So 25 dB gain at 25C might be 23 dB at +85C, and 27 dB at -55C. The temperature coefficient of such an amplifier is:
GainTempCo=(GainHot-GainCold)/(TempHot-TempCold)
Note: always use the end points of the data, in this case, -55 and 85C, to compute the tempco. Temperature relationships are never perfectly linear, this will spread out the error. In this example the temperature coefficient works out to -0.023 dB/C.
Now go and pick out an attenuator that can provide the opposite effect. Temperature-variable attenuators are specified in dB/dB/C, you need to pick a dB value that will provide the require range. Example: pick 6 dB nominal, with 0.004dB/dB/C. Multiply together and you get 0.024dB/C, which is close enough for government work.
The full value of this attenuator will have to be absorbed in the design. Captain Obvious provides these helpful hints: in a receiver, put the attenuator somewhere after the LNA, even in the IF path. In a transmitter, locate it back around the pre-driver stage where it will see very low power.
Risks and limitations of temperature-variable attenuator gain correction schemes include:
- The attenuators might not be at the same temperature as the amplifier in your module, if they are physically far apart.
- It is best to consider this a steady-state solution, during warm-up you may see that the attenuator lags behind the amplifier.
- In transmit applications you should consider that dissipating RF power in the attenuator may cause it to self-heat.
- Like other thick-film components, they don't seem to work above 18 GHz.
- No one seems to possess measured S-parameters of temperature-variable attenuartors, or an equivalent circuit model of them that you can insert into designs.
Amplifier temperature compensation example
This is on a separate page,new for March 2022.