38S: IRF510 5W MODS/Output Filter


From: Paul Harden (pharden@aoc.nrao.edu)
Date: Sun Jan 26 1997 - 16:51:04 EST

NorCal 38-Special - 5W IRF510 Version

The IRF510 5W mod was added to my prototype version and bench tested
carefully to address some of the reports of smoking IRF510's, chirpy
output, instability and spurs every few KHz around 10MHz. This is the
result of those lab tests and is PRELIMINARY at this point. I need to
get my 38S packed up, taken back home, and hope to get some good QSO's
on it this sunday afternoon, to verifiy the results in the REAL WORLD.

For clarification, below is the output filter schematic. I am designating
the optional capacitor for the filter input as C501, as that is how it
is labeled on the 38S PCB (right next to L3). It is shown as C505 on
the IRF510 mod, 38S instruction book p. 11, but C501 on the PCB.
This actually consists of TWO different mods, where the second is just
an extention of the first.


1. Make sure you cut the trace as explained in instruction #1 on page 8
   in the 38S manual. Failure to do this will put +12v on the HC240
   outputs and surely cause destruction of the I.C.
2. Wind L101 per the instructions ... 27T on the T37-2 core.
   (KD7S and NA5N believe in the "wind-em-tight" method. My L101 with
   27T was measured at 3.2uH with 12pF stray capacitance on a H-P 4332A
   L-C-R meter).
3. This IRF510 5W amplifier is based on DIRECT coupling to the HC240
   DO NOT INSTALL .1uF, C101; place a wire link in place of C101 for
      direct input coupling.
4. Install remainder of IRF510 components per instructions, page 8.
5. The value of C26 should be increased from 220pF to around 330pF for
   Xc=50 ohms at 10MHz. This is easiest done by soldering a 100pF cap
   across C26 on the circuit side of the board.
6. INSTALL OPTIONAL CAPACITOR C501; the value should be from 330pF up
   to 560pF, the mirror value of output capacitor C29. (Output filter
   caps C28 and C29 are incorrectly labeled C25, C26 on the IRF510
7. If possible, monitor input current to the 38S during keydown to
   ensure the IRF510 is not drawing >1A. If so, there's a problem.
   (Check the dc voltage on the gate; should be 0v on key up).
   Normal operation, the current draw will be 400-600mA, depending upon
   the HC240 drive and final output power.
8. Readjust TC2 for a compromise between maximum output power and good
   sounding tone in the earphones. It should not sound raspy. Or,
   monitor output purity on another receiver for a nice pure tone.
   Adding C501 also makes finding the proper TC2 tuning easier and
   more distinct.

For clarification of the mods and component designations, the output
filter is drawn below:

                   | -----> to receiver
                  L101 |
                   | |
 From Q2 | C27 | L3 L4
 HC240 --- IRF510 -*-- .1 ---*--- T37-2 ---*--- T37-2 ---*----> ANTENNA
 PA | | |
                             | | |
                           C501 C28 C29
                         330-560pF 820pF 560pF
                             | | |
                             | | |
                            gnd gnd gnd

This mod direct couples the IRF510 to the HC240 output for class C
operation; no current will flow until the HC240 output goes HI (>0v).
Reflected power in the output filter seems to easily transfer to the
HC240 input through the CMOS structure. Increasing C26 to 300pF
terminates the HC240 input to 50 ohms and attenuates many of the low
frequency spurs. Without C501, the 2nd harmonic was -24dBc; C501 thus
restores the 2nd harmonic to -40dBc in addition to removing the
spurious emissions around 10MHz and gives the amplifier stability.
I see no "chirping" or phase noise with this scheme in the lab, but
will verify on the air shortly.

The above MOD is MANDATORY in terms of removing R101 and C101. With
capacitor coupling, R101 puts about +4v on the gate, which on a
marginal IFR510, would put it perilously close to exceeding the maximum
drain current when RF is applied. C501 is probably mandatory to ensure
FCC compliance for the 2nd harmonic power. C501 and C26 mods together
ensure a cleaner output, better filter impedance and good stability.

The above output filter was analyzed using wide band noise source and
a RF Vector Impedance Analyzer (see 38S stock mod for discussion). The
above output filter yields:
   Zin = 36 ohms -j50 (looking into C27 towards antenna)
   Zout= 48 ohms -j85 (at antenna terminal looking in)

MOD #2

This mod is a continuation of the above and is preliminary at this
point until I verify its effects on the air. On the test bench, it
cleans up the output even further, gives the output filter some
additional voltage gain for higher output, and moves where the antenna
power to the receiver is pulled off to prevent overloading the NE602
receive mixer with the ten times increase in output power the IRF510

1. ADD an additional 200-270pF across the 820pF, C28 (for a net
   capacitance of C28 now 1100-1200pF, or about 1/2 the 560pF input
   and output capacitance of C501 and C29).
2. REMOVE C1 from the junction of C27, C501 and L3 and relocate to
   the junction of L3, L4 and C25. C1 is the capacitor that couples
   the RF to the receiver input. I physically removed C1 from the
   board and soldered a new .1uF on the solder side from the original
   C1-D1 hole to the L3-C28 junction.
3. Readjust TC2 for best tone and output, blah, blah, blah.
   (REPLACE TC2 if you've adjusted it as much as I have in the
    past 2 days -hi).

With C501 and the extra capacitance at C28, the junction of L3-L4 is
now at a lower voltage node than previously, which supplies less
p-p RF voltage to receiver for a nicer sounding sidetone. It drops
the MDS by about 2-3dB, but the resultant MDS on mine still measures
in at -125dBm, or still plenty of sensitivity to the receiver. BUT,
a little on the air will confirm this.

With MOD #2 the Zout is scarcely changed (46 ohms at -j80) and the
Zin of the filter is lowered even further to 16 ohms at -j80. The
16 ohms is closer to the ON resistance of the IRF510 than previous
filter Zin impedances, and the -80 degrees of (capacitive) phase shift
in the filter is the same on both ends, meaning it should be stable
and won't chirp. (One cause of chirp is when the input and output
impedances of a filter are at different phase shifts, or capacitive
values, which causes the voltages to try and normalize when RF is
applied, causing the load on the PA amplifier to change and hence the
output filter to cause a momentary shift in frequency - or a "chirp").

The above mod is NOT MANDATORY, and again, preliminary at this point.
If you do try it, I would appreciate hearing from you. It is fairly
simply to perform, and DE-perform if you have to -hi.

I was doing these lab tests late saturday afternoon and evening, and
watching my QRP-L mail ring with all the traffic about working AK.
And there I was, with MY 38S scattered in pieces all over the work
bench (and of course no antenna here). But I was having fun.

Hope this helps some of you experiencing various forms of instability
with the IRF510 mod, and always interested in the observations and
findings of others.

72, Paul NA5N

Search QRP-L Archives

[ QRP-L Archive | ]
[ 1993 | 1994 | 1995 | 1996 | 1997 | 1998 | 1999 | 2000 ]


This archive was generated by hypermail 2b29 on Fri Jun 02 2000 - 11:32:50 EDT