Re: RF power chain mods and improvements..

K9HZ <bill@...>

I guess i don't understand where you are going with the software... but wouldn't it be easier to convert your display work from the I2C buss?  Then you have a whole host of pins to work with... freed up from the parallel data lines of the ex-display?


Dr. William J. Schmidt - K9HZ

On Jun 6, 2018, at 11:42 PM, ajparent1/KB1GMX <kb1gmx@...> wrote:

One of the factors not reported was the consistency of the gain vs spur. Simple put
the gain and the spur interacted even when I held the level to the mixer constant.
I'd lower the gain and the spur would get worse as an example.   It varied 
greatly with supply voltage as well.

Reason sluggish 3904s (low FT, Maybe Low  HFE).  Remember we are using a
300mhz device and that is a typical value and can be as low as 250 (batch and
vendor) or much higher.  I changed them out for BFR106 (Q20, Q21, Q22), Seems
the 3904s were not up to the task of keeping the amp load independent. 

The net result was what others had reported as lower spur production, in mine it
dropped the 10M spur nearly 12db.  So it was either distortion in the  45Mhz amp
or its ability to be a low output impedance which should be with emitter follower
with the 47ohm making it look near 50 ohms.  I didn't  try 2n2963A as I didn't
have any left in SMT but I suspect they would have been better.

For now I'm taking a break on output power and addressing other things.  The
power thing stands at 10W at 80m, 7W at 10M with the spur 46db down.  

I'll list all the changes as my unit stands while I do pop/click and mod then agc.

For now mods I have in place, I'm not 100% happy but that is what
us experimenters enjoy.  Likely others to come. ;)
Replace and reset bias OF Q90 with 2n2369A or better (BFR106),
Set bias for 18ma (r86 2.2K).  Make C81 470pf, and r83 8.2ohms.

Change C82 to 470pf (peaking for 10M).  

T9 is now 9 turns #28 bifilar (one winding was never used).

All the 2n3904s in the predriver and driver replaced with 2n2222A or better.
Hint I mounted them on the underside with very short leads as it makes
the top less crowded.  

Driver amp has 2.7uh SMT choke in series with R86 and that is now 1.6Kohms
R87, r88 Both have a 270PF cap in parallel with them.

R911,r96, R941, R942 emitter resistors for driver reduced to 11ohms (add parallel 22ohm).
R92, R95 replaced with 1.5K ohm in series with 2.7uh choke.  (resets bias and adds peaking.)

T10 has one less turn.
T11 reduced by 1 turn.

45mhz IF replace (Q20, Q21, Q22) with BFR106.  Change R23 to 18ohms.  Change R27 to 39 ohms.
!2mhz IF amp change R43 to 18 ohma and put a .1uf in series with 680 ohms across R41.

Replaced Q10 with BFR106 for better receive gain at 20mhz and up.  (I may change out the two 3409s).

Add LM7806 regulator in handy location to limit voltage to ta2822 to 6V on pin 2.
Add 1K resistor from negative side of C77 to ground (prevents pop when pluging in
headphone or speaker)

Add 16K resistor from base o ground of Q70 for better overload performance.

The one change that really helped the transmitter was going to the 2n2222A (TO18) part.
Before that it was 13W on 80 and barely 7 on 40 and less on 20.  After that change alone
40 was near 10W and 20 was about 8.5W.  This has been tried on another unit that
was weak on 40M and it really helped with no other tricks.

Enough for now.  I plan to spend a few days re-rolling the software off the V4.3 release
for my user interface which will not include CAT, CW or a rotary encoder.  Without CW
mode the sidetone CW-Tone and CW-KEY pins are usable for general IO by removing
R256 and R105.


Join to automatically receive all group messages.