OpenBCM V1.08-5-g2f4a (Linux)

Packet Radio Mailbox

IZ3LSV

[San Dona' di P. JN]

 Login: GUEST





  
G8MNY  > TECH     26.12.23 08:50l 111 Lines 5128 Bytes #999 (0) @ WW
BID : 63341_GB7CIP
Read: GUEST
Subj: AF Waveform Generator (RS)
Path: IZ3LSV<I0OJJ<EA2RCF<LU9DCE<GB7YEW<VE3CGR<VE7ASS<VE2PKT<VK4OT<VK6ZRT<
      GB7CIP
Sent: 231226/0744Z @:GB7CIP.#32.GBR.EURO #:63341 [Caterham Surrey GBR]
From: G8MNY@GB7CIP.#32.GBR.EURO
To  : TECH@WW

By G8MNY                                      (Correction Dec 20)
(8 Bit ASCII graphics use code page 437 or 850, Terminal Font)
These devices are usually based around the ICL 8038CC or ICL 8038BC waveform
generator ICs. If a frequency ratio of 1000:1 for 20Hz to 20kHz on 1 scale are
used (max possible!), there can be problems with set square 1:1 ratio at the LF
end. I found a simple bodge was very effective at solving this.

         1N4002    +12V
+12.7VÄÄÂÄÄ´>ÃÄÄÂÄÄÂÄÄÄÂÄÄÄÂÄÄÄÄ100KÄÂÄ -12V
     Set³       ³  ³   ³   ³     /³\ ³ Sine
    20HzÀÄ1KPotÄÙ  ³   ³u1 ÃÄ100KÄ)ÄÄÙ Clippers
          /³\      ³  ===  ³+ /³\ ³
     ÚÄÄÂÄÄÁÂÄÄÄ¿  ³   ³ ÚÄÁÄÄÄÁÄÄÁÄÄÄÄ¿ Sine
     ³  ³ Freq 2K7 ³   ³ ³ 6   1  12  2ÃÄÄÄÄÄÄÄÄo<ÄÄÄÄ¿      +12.7V
     ³ 2K7 Pot<ÄÁÄÄ)ÄÄÂÁÄ´8        Tri3ÃÄÄÄ15KÄÄo     ³         ³
   Freq ³  25K     ³ 56K ³            9ÃÄÄÂ150KÂo Level      3³\³7 TL081
    Pot<ÁÄÄÄÙ      ³  ³  ³ 4  5  10 11 ³Sq³    ³    25K<ÄÄÄÄÄÄ´+ \
    25K            ³100K ÀÄÂÄÄÂÄÄÂÄÄÄÂÄÙ  ³    ³    Log       ³   6>ÄÂÄ220Äo)
     ³             ³PresetÄ´  ³ ===C ³-  15K   ³      ÃÄ15KÄÂÄ´-1/   ³   Output
     1K            ³      4K7 ³  ³4n7³    ³    Ão<ÄÄÄÄ´     ³2³/³4   ³
   Preset          ³  Set  ³ 4K7 ³   ³  +12V   Ào    _³_    ³  -12V  ³
     ³set 20kHz    ÀÄÄÄÄÄ>1K  ³  ³   ³          o           ÃÄ39KÄÄÄÄ´
    _³_               1:1 pot ³  ³   ³                      ÀÄÄÄÄÄ´ÃÄÙ
                           ÀÄÄÙ  ³   ³                           3p3
                       -12VÄÄÄÄÄÄÁÄÄÄÙ

I put in 56K & 100K preset for fine adjustment of LF correction from the
frequency pot slider to pin 4 "set square". This keeps the square wave ratio at
1:1 over the 1000:1 frequency range, which is needed to keep the sine wave even
harmonic distortion low. My circuit was on a modifed R.S. PCB.

INSIDE THE IC
The ramp currents into the cap on pin 10 are controlled by the voltage on pin 8
with up & down currents balanced by the set square current inputs on pin 4 & 5.
The ramp voltage on the capacitor is buffered to become the triangle output.
The charge up down level threshold toggle becomes the square wave output. The
sine wave output uses triangle wave & 2 clippers with thresholds set on pin 1 &
12 to produce best shape.

FREQUENCY SCALE
I changed the linear frequency scale (that was a decade scale) to a log scale
over the 3 decades. This is done by using a ganged 25K linear pot, instead of
the single 10K lin, a good near square law scale results & with 2K7 loads on
each, makes a pseudo log scale resulting 700 Hz in the middle & a usable LF
end. This is better than using a log pot as the are not very log actually!

ÚÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄ¿
³        500 1k                          ³
³     300 /~~\ 5k      _       _         ³
³ !  100 (Freq) 10k   (_)     (_)    (ú) ³
³on    20 \__/ 20k  Si Tr Sq  Level  Out ³
ÀÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÄÙ

N.B. the 20Hz LF frequency setting preset is very critical & temperature
sensitive as the offset of few 100mV above the +ve rail used by the IC with the
+ve rail diode is used.

PIMPLES
Another problem I fixed was the pulse on top of the Sine & Triangle waveforms
due to crosstalk from the large square wave edge. I used an earthing contact on
a spare mode switch to short out this signal (after 150K) then the problem goes
away. In the circuit all the waveforms are about the same peak level with the
level pot at max.

     _n_      Crosstalk   n
    /   \     Pimples    / \
   Þ     Ý     Þ        /   \     /
         Þ     Ý             \   /
          \_ _/               \ /
            u                  u

SLEW RATES
Here the buffer OpAmp is a fast slew rate one, not the original 741. The 3.3pF
on the NFB arm is another select on test, for perfect non ringing 20kHz square
wave at 7V RMS (20V P-P).

  741 Internal           Untamed Ringing         400kHz Bandwidth
 Compensation &            Fast OpAmp             Limited OpAmp
  Low Slew Rate           Good Slew Rate          Good Slew Rate
   ___                     !Ü____                   _____ 
  /   \ 20kHz              ³ß~   ³ 20kHz           ³     ³ 20kHz
 /     \     /             ³     ³     ³           ³     ³     ³
        \___/                    ³Ü____³                 ³_____³
  Square Wave!                   !ß~
                         >400kHz ringing

SINE DISTORTION
These Waveform ICs are not the lowest of distortion, but figures as low as 0.5%
(-46dB) can be achieved at 1kHz & 7V RMS, sometimes with the more stable BC IC,
but for the whole 20Hz - 20kHz only about 2% (-34dB) is possable. But with some
care is needed to achieve this with the 2 sine clipper preset pots & the
frequency sensitive 1:1 set square pot.

The main advantage of these oscillators is their level flatness, typically
ñ0.01dB (1%) & no bounce of the Wien Bridge types. Also as they are DC
controlled & can be used for sweeping tests locked to a scope/chart recorder
etc.


Also see my bul on "Simple 1kHz AF test osc" for details on waveform structure.


Why don't U send an interesting bul?

73 de John G8MNY @ GB7CIP




Read previous mail | Read next mail


 04.11.2024 23:02:48lGo back Go up