G8MNY  > TECH     30.03.25 08:15z 128 Lines 5505 Bytes #97 (0) @ WW
BID : 29141_GB7CIP
Subj: PIC Freq Counter Mods
Path: ED1ZAC<ED1ZAC<GB7CIP
Sent: 250330/0804Z @:GB7CIP.#32.GBR.EURO #:29141 [Caterham Surrey GBR]
From: G8MNY@GB7CIP.#32.GBR.EURO
To  : TECH@WW

By G8MNY                                 (Updated Feb 09)
(8 Bit ASCII graphics use code page 437 or 850, Terminal Font)

After making a PIC frequency counter from EI9GQ's design in October 2006
Radcom, as an "on the night club construction contest", I found some faults
with it/the club kit provided.

USELESS PREAMP
The original circuit has very poor gain up to 1V RMS input was needed to get
the 4V p-p logic level needed to drive the PIC properly. I could not discover
exactly why, it was not just the biasing, as the HF frequency response was
also very poor!

So I replaced the supposedly 550MHz BF199, for a 500MHz 2N2369A & rebiased it
to give the best input sensitivity (around 18k base to collector).

I also protected it from overload (when the display goes funny & needs
repowering to reboot the PIC) with 2 back to back signal diodes on the input
BNC socket. And I increased the LF range by changing the 470pF input cap to a
10nF so that frequencies down to 1kHz could be measured (software disabled
below that!)
                               ÚÄÄÄ+5V
                              560R             Frequency      Sensitivity
        Protection        ÚÄÄÄÄ´                1kHz - 10kHz     100mV
        Diode            18K   ÃÄ270RÄ>PIC     10kHz - 10MHz      10mV
BNC     Clipping     10n  ³  ³/                10MHz - 30MHz     100mV
I/P  o)ÄÄÂÄÄÄÂÄÄÄÄÄÄÄÄ´ÃÄÄÁÄÄ´2N2369A          30MHz - 45MHz     600mV
Socket³ _³_ _³_  open        ³\e
      ³ \_/ /_\  wires         ³
      ÀÄÄÁÄÄÄÁÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÁÄÄÄ
       2x 1N4148          PIC PCB

FREQUENCY ACCURACY
On the construction day, all the kits read a 10MHz reference about 1.5kHz high.
This meant the 10.240MHz timebase crystal has to be slowed down. EI9GQ'a write
up mentions the 2 fixed 22pF caps need to be selected for your crystal. I
changed my original crystal for a recovered surplus one (different batch) &
again had the same frequency. So I swapped the 22pF caps for slighty larger
values, but I found the best result was to add a small ceramic 22pF across the
crystal. This gave a nice +/- 200Hz range @ 10MHz on the fine Cal trimmer.

       PIC
    ³ 10.240 ³
    ÃÄÄ´[]ÃÄÄ´
    ³        ÃÄÄÄ¿
    ÃÄÄÄ´ÃÄÄÄ´   ³\
    ³  22pF  ³   ³/\
   ===      === === Cal
22pF³    22pF³  /³ Trimmer
  ÄÄÁÄÄÄÄÄÄÄÄÁÄÄÄÁÄÄ

As there is no temperature compensation of the reference crystal, I also tried
some negative temperature 22pF caps, but these would not tune on frequency for
me! :-(

SOFTWARE PRECALLER
I did find that there is a software problem affecting the accuracy when the
counter uses its software selected PIC prescaller. Eg. if you calibrate the
counter at say 4MHz & retest it at 40MHz, there is a small under read error of
greater than 10x of the count. This may be due to the software loosing some
counts cycles when it does its cleaver stuff counting buffer over loads etc.
But normal counters also connected, behave quite normally on this test.

BOXING UP
I found a second hand plastic box that was just big enough (120 x 65 x 40mm),
as the smaller plastic & diecast ones I had were all too small.

I mounted the display with 6BA stand off pillars that were exactly the same
depth as the LCD display, but I had to drill out the metric PCB holes to 6BA
clearance.
                 display cut out
      ====================================
     ³³    [] |~~~~~~~~LCD~~~~~~~~| []   ³³     []=threaded stand off pillars
     ³³heat===========================   ³³
     ³³sink--screen---                   ³³ Input
POWER³³ reg===========     room for      ³³ BNC
JACK o)         PIC         nicads       ((o)
      =====================================
                   Box Bottom

I then stacked the PIC PCB with long 6BA bolts on that. But that was actually a
mistake & I needed a double insulated screening plate between the 2 PCBs with
6BA nut separators, to stop the display crosstalk noise from upsetting the
counter reading.

POWERING
I added a series idiot diode on the 12V feed & initially a heatsinked the
regulator as I was powering bright backlight LED from 5V.

But later I used a small nicad battery option with no LED, but charging via the
LED. PCB track cuts were needed for this circuit arrangement.

        2x  1N4001            ____ +5.0V               ON/OFF
+12V o)ÄÂÄ´>ÃÄÄÄÄÄÄÄÄÄÄÄÄÄÂÄÄ´7805ÃÄÄÄÄ¿ N.O. relay   SWITCH
      ³ ³                ===  ~~³~      ÄÄÄÄÄÄoÄÄÄÄÄÄÄo\______+5V to
      ³ À´>ÃÄÂÄÄÄÄÄÄ¿     ÀÄÄÄÄÄ´      ³ N.C.              ³   PCB
      ³   ÚÄÄÁÄÄ¿   ³           Á      ³                  ===
     33R  ³Small³  _³_ Charge          ³                   Á
      ³   ³ 5V  ³  \_/=Backlight       ³
      ³   ³Relay³   ³  LED             ³
      ³   ÀÄÄÂÄÄÙ   ÃÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÙ+4.8 to 5.5V
      ³      ÀÄÄÄÄÄÄ´
      ³           +_³_ NICADS
      ³             Ü    4x
      ³             ³  220mA/H
      ÀÄÄÄÄÄÄÄÄÄÄÄÄÄÙ

When the relay is released with no 12V, power comes from the wired in nicads.

Charging is through the Relay & the bright LCD display LED, & is enough current
to recharge in 24 hrs. Also running the LED only from 12V via a 1W 33R to the
nicad battery instead of the original 10R from 5V will remove the need for a
regulator heatsink.


See other tech buls "Watson 'Hunter' Counter Mods", "Old Venner Counter Type
TSA3334", "Marconi Counter Type 2434A", "198kHz Off Air Standard", "Off Air
Lock for Ref Osc" & "Calibrating Frequency".



Why don't U send an interesting bul?

73 de John G8MNY @ GB7CIP


Read previous mail | Read next mail