Norcal 40A

All the following is used with permission of the author. Thanks guys.

Increasing RF Output
Gary Surrency AB7MY:

I have spent some considerable time increasing the RF output of the NC40A. The following modifications have worked well for more than two years, and do not require any surgery to the PCB or anything drastic like that.
You will need to get some MRF237′s or 2N3924′s for the PA stage. Either one will achieve 5 watts output, with some additional mods to be described. My sources for these have recently dried up. The 2N3924 is a direct drop-in, whereas the MRF237′s C and E leads are reversed from the PCB layout. There are two ways to accomodate that. You can put the ’237 on the bottom of the PCB, and press its case against a small puddle of thermal grease on the inside of the bottom cover as you solder the leads from the top. This adequately heatsinks the PA at 5 watts.The ’237 case is grounded, so there’s no need for insulation.

Or, you can bend the Base lead between the E and C leads, and rotate the ’237 180 degrees so that the leads now fit the PCB pinout and install it from the top. That allows the stock heat sink to be used, but avoid long keydown operation or the PA will overheat at 5 watts quickly. Ordinary intermittent CW duty cycle does not cause the PA to overheat, but a larger heat sink is preferable anyway. I’ve been using mine with the stock heatsink on a 2N3924 during several QRP contests, and it has not failed.

Both devices are designed for 12-13 volt operation, instead of 28 volts like the 2N3553, etc. Better efficiency at 12-13 volts is the result, but they are rather fragile electrically and subject to SWR damage if your antenna is not well matched. This is due to low breakdown voltage ratings resulting from the original application these devices were designed for. But they work very well if you remember to be careful.

You also have to modify the low-pass filter (LPF), so that it presents the proper, lower impedance load to the PA for 5 watt operation. I found that the following changes were necessary:
Component Old Values New Values
C45 330pf 470pf
L7 18t on T37-2 16t on T37-2
C46 820pf 820pf (no change)
L8 18t on T37-2 18 or 19t on T37-2
C47 330pf 470pf
Good ceramic discs are OK, as that is what I used. But you could use silver mica or polystyrene caps if so inclined, as they have better Q and less loss.

By using an antenna analyzer, such as the MFJ-259B, you can look into the antenna jack and determine if the PA collector load is correct. Use a test resistor of 15-19 ohms tacked across the PA transistor collector to ground. A single 15 ohm, 18 ohm, or pair of 33 ohm or 39 ohm 1/4 watt resistors in parallel is fine.Leave the power off, and the PA device installed. Tune the analyzer to 7.040 MHz, and read the SWR and reactance.

With the LPF values from above, you should be able to make slight adjustments to the toroid windings to get the analyzer to “see” 50 ohms resistive, with no X value or reactance. Now the PA is set up to operate at 5 watts under ~13.8 v so it can work “hard” enough to develop 5 watts. Using the analyzer is a “real-world” way to verify the theory, and the although the measurements seem to be pretty “touchy” as you adjust the LPF components, the actual PA operation will not be so sensitive to small errors. If you get close – it is fine.

Also, the reduced collector impedance allows the stock 36v zener at D12 to be retained, since the peak collector voltage is held down by the lower collector Z. That demonstrates the LPF is matching the 50 ohm antenna load correctly to the 15-19 ohm collector load, as we just measured with the analyzer. If these steps aren’t taken, the stock zener will overheat at higher operating voltages. Also, the complexity of a tapped collector choke is avoided, and the associated wiring nightmare that would be required with the original PCB layout.

I have used the antenna analyzer / PA collector test resistor method with every rig I own, to verify and if necessary, make the required LPF changes to correct any PA to antenna load matching problems. I posted this procedure on the QRP-L some time back, with the original idea coming from “Solid State Design” by Wes Hayward and Doug DeMaw. I just use a different approach with the availability of modern test gear.

Don’t forget to remove the 15-18 ohm collector-to-ground test resistor before connecting the DC power again, and *don’t* transmit into your antenna analyzer by forgetting to remove it and re-attaching your antenna or dummy load and wattmeter.
You have been warned!

A little more drive is also required, so the following steps are needed:

I changed Q5 from a J309A to a J310. Select the best J310 if you have several to try.

The 100 ohm resistor at R14 is not changed, due to the low impedance input from T1 that is OK, and we don’t want to reduce the already marginal drive that we have for a larger PA stage.

A MPS2222a from Radio Shack, etc., was used for Q6 instead of the original PN2222a. Again, hand selecting a “hot” device is recommended while observing the wattmeter indication.

If 5 watts is still not possible (and it should be now) you can do this:

Some VFO’s or NE602′s used at U4 may be a little down on output. If this seems to be the case, you can replace the 5 pf cap used at C31 with another one, or use a slightly larger cap, of no more than 8-10 pf. Some of the 5 pf caps I have tested were low in value, preventing sufficient VFO signal into U4. But an NE602 is very easily overdriven into spurious output, especially with the simple one-pole bandpass filter following U4 in the NC40. I found that 5.6 pf to 6 pf at C31 would almost always help a low VFO signal into U4, and still prevent spurious mixer output. Radio Shack carries a package of small pf value caps if you need some.

If you have a small pot of 200 ohms, it is a better choice for the drive trimmer, R13.

If you have leading edge audio clicks in the headphones with a KC-1 installed, a 100 ohm resistor in series with the Mute lead from the KC-1 to the anode of D2 on the NC40A PCB will cure it.

After these mods, reduce the drive for about 1 watt of output and then adjust the TX bandpass filter trimmer C39 for maximum indication on the wattmeter. Keep the output at 1 watt or less to avoid overheating the PA as you tweak the trimmer for maximum drive. Allow the PA to cool a bit, then quickly set the drive pot to any output level up to 5 watts. Make sure your antenna is a good 50 ohm resistive load before operating, since the higher power capability will quickly fry the PA, even with protective zener D12 in place. That is the only caution when using these TO-39 PA devices, since they aren’t very tough. I always use my MFJ-259B to pre-match the antenna before transmitting.

You could install a larger PA transistor in place of the MRF237 or 2N3924, as they are getting hard to find. But suitable devices like the 2SC2078, 2SC1969, or MRF476 are all in TO-220 power tab packages, and require the leads to be bent oddly to fit the PCB layout. The extra drive and LPF changes described above are adequate to achieve 5 watts or more with these larger RF transistors (I know – I tried it), but they don’t allow the same simplicity of the original NC40A design. The original 2N3553 will almost achieve 5 watts with all of the other mods, so you might try it that way with your stock transistor. If yours has the 2SC799, it simply won’t handle the extra power.

On my NC40A with all of these mods, I actually have to reduce the drive pot to avoid exceeding 5 watts of output at 13.8 v to 14.0 v. Even with a gel cell during portable operation – it is still possible to get 5 watts output by increasing the drive pot setting.

With the stock collector choke in the rig, 5 watts is getting pretty close to its current rating. But it only gets slightly warm during normal operation at or below 5 watts. You could probably gain a bit of output by replacing the small, stock 18uH choke with one made from a ferrite core, such as a T37-43 or T37-61 wound with 8-10 turns of #24 wire. That was the case with the SW-40+. The series resistance and losses would be less, allowing more current for the PA and more RF into the antenna. Mine still has the itty bitty stock choke, since I haven’t gotten ’round to it, yet. ;-)

Oh yes, I’ve also installed a 10-turn tuning pot and it is a very nice improvement over the stock pot. You’ll need the KC1 installed as I have, or some other method to determine the frequency. Works very FB.

Multi-turn Tuning Potentiometer10tpot
This modification assumes you have the KC1 Frequency Annunciator installed in your Norcal 40A. There are several Bournes multi-turn potentiometers available from Mouser that will provide greater tuning resolution. Pick your pot based on the resolution you want, bearing in mind that the 10T version requires a lot of spinning to go from one end of the Norcal’s band to the other. If you mainly operate on 7040 kHz, this may be the best choice for you. I like to hang out on the low end of the band with frequent excursions to the QRP watering hole, so I chose the 5T version and believe it to have been the right choice for the way I operate. Here are the part numbers and prices of the 5T and 3T potentiometers:

3 turn Bourns 10K ohm – 652-3543S-1-103 at $22.62
5 turn Bourns 10K ohm – 652-3545S-1-103 at $23.30

Either of these two parts will install into the Norcal with minimal fuss. For physical installation, the hole in the front panel must be enlarged by 1/16th of an inch. After doing this, secure the pot to the front panel with the pot’s leads facing up. Due to the increased length of the multi-turn pot, it is not a drop-in replacement. In other words, the leads of the new pot won’t fit into the circuit board like the original one does. I simply used a 3 conductor piece of ribbon cable 3” long to connect the pot terminals to their respective thru holes on the board (see photo).

Depending on which end of the band I’m on, I get a tuning resolution of 12 kHz/revolution (bottom of band) to 7 kHz/revolution (top end). This is a vast improvement over the stock potentiometer’s 70 kHz/revolution and greatly increases the operability of a fine rig.