Temperature Stabilization for the
W1HUE QRP-PLUS Final Amplifier Mod

Larry East, W1HUE
Tucson, Arizona

(contact author)

Creative Commons License

Reprinted from the July 1999 ARCI QRP Quarterly
 

Beginning with the January 1997 issue of the QRP Quarterly, I published a series of three articles describing modifications to the Index Labs. QRP-PLUS Transceiver. In the second article of the series, which appeared in the April 1997 QRP Quarterly, I described how to replace the IRF510 MOSFET used in the final amplifier with an MTP3055E MOSFET for improved SSB linearity and output power. Unfortunately, my original circuit did not include any temperature compensation for the DC operating point of the MTP3055E, which exhibits an upward drift in its drain current under key-down conditions. With the zero output power drain idle current initially set at 200 mA, it will typically increase an additional 80 to 100mA over a 30-second period and then (more-or-less) stabilize. I did not consider the drift to be cause for concern at the time, but several folks who have performed the mod have expressed concern. In any event, the idle current is difficult to set when itís drifting all over the place!

The idle drain current of the original IRF510 also drifted upward, but not as much as the MTP3055. This is because the Ď510 is a much lower current device than the Ď3055, and the idle drain current is closer to its "zero temperature coefficient" drain current versus gate voltage point. This point occurs at a drain current of about 2A in the IRF510 and at about 7A in the MTP3055. Below this point, the drain current will have a positive temperature coefficient, and a negative coefficient above (thatís why "thermal runaway" is less of a problem with MOSFETs than with bipolar transistors).

I decided to see what could be done to reduce the drain current temperature drift. One possibility is to replace the MTP3055E with the newer MTP3055V; according to its data sheet, the "V" version has a lower temperature coefficient at drain currents below 1A. I have been told by others that the "V" works just fine in my modified amplifier circuit, but I have not tried one myself so I have no data on its drain current point drift. For temperature compensating an MTP3055E, the relatively simple circuit shown in Figure 1 provides excellent results. In this circuit, compensation is provided by an NPN transistor in thermal contact with the MTP3055E (how that is accomplished is described later). R1 is adjusted so that the change in the transistorís collector voltage with temperature reduces the MOSFET's gate bias to exactly cancel its drain current positive temperature coefficient. The voltage drop across the transistor is a multiple (determined by 1 + R1/R2) of its base-emitter voltage, V BE, which has a negative temperature coefficient. The change in the MOSFETís bias with temperature is therefore given by:

Equation

Note that this scheme only works if the magnitude of the MOSFETís drain current temperature coefficient is greater (and opposite to) than the base-emitter voltage temperature coefficient of a bipolar transistor. This happens to be the case for an MTP3055E in the drain current range of interest. However, this circuit will very likely overcompensate the drift in the original IRF510 amplifier. It will probably also overcompensate the drift in the newer MTP3055V. (If the drift in the idle current in your rig is less than about 30 mA or so, I wouldnít worry about it.)

Temperature compensation circuit
Figure 1.  Temperature compensation circuit.

I used a 2N3904 for the compensating transistor, but just about any small NPN transistor will work (do not use a Darlington type, however). R2 should be in the range 1K to 10K, and the maximum resistance of R1 should be 20 to 30% larger than R2. For example, if you use a 5K pot for R1, then 3.9K would be an appropriate choice for R2. The ferrite bead is probably an overkill, but I wanted to make sure that no RF reached the 78L05. The transistor is epoxied to a 0.25-in. wide, 0.064-in. thick piece of brass (obtainable from most hobby shops) fabricated as shown in Figure 2 (copper or aluminum will also work Ė maybe even better than brass). The transistor should be in "intimate contact" with the metal; donít depend on epoxy for good heat transfer! The metal strip holding the transistor is mounted over the MTP3055E mounting flange; use heat sink compound between the two to ensure good heat transfer. R1, R2 and the 0.1 uF bypass cap can be mounted on a small piece of perf-board suspended above the transistor by its leads; a little epoxy can be used to hold it in place. Make sure that R1 can be accessed for adjustment, and that none of the components touch the cabinet or the PC board above the RF amplifier board.

Transistor mounting bracket
Figure 2.  Compensating transistor mounting bracket.

This mod requires some minor surgery to the RF amplifier board:

  1. You must remove the existing bias adjusting pot from the board. Bend the ground leg of the pot 90-degrees so that it sticks out behind the pot, then remount the pot to the board using the other two legs.
     
  2. Remove the 78L05 regulator and bend the common lead (thatís the one in the center) 90-degrees so that when the 78L05 is remounted, it points toward the center of the board.
     
  3. Drill a small hole through the board so that the 0.1 uF bypass cap shown in Figure 1 between the common lead of the 78L05 and ground can be mounted on the top of the board and soldered (through the hole) to the ground plane under the board.

The 0.1 uF cap that bypasses the output of the 78L05 (not part of the original circuit) can be mounted under the board.

After installation, set R1 to its minimum resistance and the bias pot set at its full clockwise position. Slowly adjust the bias pot for an idle current of about 200 mA (thatís in addition to the 400 to 450mA the rig draws in transmit with the MOSFET biased completely off). Remember that the RF drive must be reduced to minimum when making this adjustment. You should note an upward drift in the idle current. Unkey the rig and increase the resistance of R1 slightly (about 1/8 turn). Wait a minute or so for the MOSFET to cool, then key the rig and readjust the idle current using the bias control. The drift should now be less, maybe even downward. Continue this process until the idle current remains steady. If you "overshoot" and need to reduce the resistance of R1, reduce the bias slightly (rotate the bias pot clockwise) before re-keying the rig to prevent excessive MOSFET drain current. When R1 is properly adjusted, the idle current should stabilize within a few seconds and then remain constant within a few mA for as long as the rig is keyed.
 

———— Copyright © 2005, 2012 by Larry East, W1HUE ————
Creative Commons License

Page last updated on June 18, 2012