Log Sweeper (I) Prototype

April/26 2016

JUMP TO CONCLUSION (Oct 2016): DBM Frequency mix generator, and don't beveling the word said here, it's obsoleted)

I built Power Meter: AD8307 and the Sawtooth: Ramp signal source long time ago, but each of them have own standalone case.

 The ramp signal generator had been completed  for a long time . This case is not very fit for my work place,  and hard to install module into it.

recently, i got several suitable cases, JXD-11 old signal generator for students. That case is the one i like most, good quality, enough spaces, module structure. 

Power Supply

AD8307 needs 9V supply, and about -0.7 volt negative rail, total current about 50 mA.

Sawtooth generator needs a 18 Volt positive rail, 28mA current, and negative 9V, 28 mA.

The JXD-11 had a +- 24V transformer. convert 24V to 18V via DC seems the straightforward choice, 9V could get from 7809.

First version: LM2576 DC-DC, Target 18V

the goal of the DC-DC is provide the 18V positive voltage.

testing version DC-DC convert output 16.5V, but while i connect a 50 ohm load ( 300mA ), output drop to 15 Volt, and rectify output drop to 17.5V.  there is something wrong.  while load disconnected, rectify output 28 Volt, why output drop so much?

the transformer AC terminal output is 21.6V without load,  dropped to 18.3 Volt loaded.  I try  to check transformer's secondary DC resistance,  it's 11 ohm, MY GOD, it's huge.  For transformer, it's designed to drop about 10% while fully loaded !  if want it output 5 Watt,  you had to take the voltage falling into account. 

Rectify output is 28V without load,  calculated transformer peak voltage: 21.6/0.707 ~= 30V.  I finally realized the AC output readed from my multimeter is RMS value  :(-

A realistic power supply for this signal source

Testing the transformer and DC convert indicate provide 18 Volte from DC/DC is wrong way.  better solution is  convert the 21 Vrms voltage to 9.5V via DC-DC,  then the supply could provided at least 200mA for 9.5V supply. the 18 Volt positive rail supply for sawtooth generator could use linear voltage regulator, for some margin 17.4V is moderate.   

NOTE: 

        7905 need a capacitor parallel with 1.5k/2k POT, otherwise, 7905 became a oscillator.

Power Meter Ready

Reinstall power meter does run into trouble, while i connect the alligator clips testing the the 5V voltage, everything is ok, but while remove the clips, the output is crazy...carefully check found there is a ground isolation area in the double side PCB, that's why clips testing is ok, clip connect that isolated area together (GND).

Nothing else.

System block diagram

Debugging the saw-tooth generator

May/5/2016

Sweeper Oscillator Draft Design

May/13 2016

the idea is simple, get one VCO, maybe 2. 

one VCO running from 25Mhz to 100Mhz.   Another VCO mix with crystal control oscillator get an 0-30/40 Mhz single span.  with 2 band, cover 100 khz to VHF range. 

the PIN switch is intend to use 2 1N4007, poor man solution.  only one of the 2 signal will power on,  so the PIN switch actually not that critical for it's leaking performance. 

VCO

May/15 2016

the finished experimental VCO.

From 60 Mhz to 150Mhz, amplitude change less than 2dB.  

NOTE: the signal get from a 1k resistor connected to the oscillator.

Buffer the VCO

May/17 2016

 llced (llced@sohu.com) introduce me a good wind-band buffer, the price is -6 dB loss. for the sweeper, this is acceptable, the oscillator output about 7 dBm, very strong. loss -6dB (voltage loss) not a big thing. 

Ltspice simulation indicate the buffer flat response extended to 300 Mhz. the L2, 100nH base inductor is key element to extend wide band beyond 150 Mhz.  the compensate inductor accepted value range is from 100 nH to 150 nH , means this buffer is easy to alignment actually.

In the Ltspice schematic, L2, L3, L4 is the compensate inductor.

Debugging process:

the BJT need get 10 mA bias, the degeneration resistor value range is 100 ohm to 200 ohm.

while finished the buffer, i found the VCO stop while varactor voltage raise above 8V.  checking the draft layout, varactor  voltage actually DC couple to the JFET, holly shit, this mean JFET will be forwarding bias and became as a diode, i lost one JFET.... 

insert a couple capacitor, 224 ceramic,  again, there is problem, the amplitude bumped 10 dB  around 50 Mhz,  wired thing is if i manually sweep the VCO , the output level quite stable.

The reason: sweep voltage apply to BB152 via 10 k resistor, 10k combine 0.22uF capacitor, form a low frequency filter, while the sweep voltage jump from 16 V to 0 V,  need to discharge the capacitor via 10k resistor,  make to the VCO crazy.  replace the 224 with 100pF couple capacitor make everything perfect.

Crystal oscillator LO (Version A)

I ever made several crystal controlled oscillator, my first idea  is the Pierce overtone oscillator, get a 12.2Mhz crystal work on it's 5th overtone, get an 61 Mhz oscillator. 

bad things always happen, i trying to get it work, but seems only working overtone is 36.6Mhz. this oscillator does not has enough gain? trying to bypass R3, does not work.

how about miller? ground base one? 

still does not starting oscillate.  but this time when i trying to bypass the R3, it works:

There is an important thing should be address well: while the tank tune away from crystal's overtone frequency, it will work as a free oscillator. So, tune the tank to the correct frequency is the key to get a crystal controlled oscillator from this beast.  

(NOTE @9/02/2016: butter oscillator don't have this problem,  this oscillator is dropped)

Shield! shield !

I built the 2 oscillator on the perfboard.  while i check the VCO output, some strange thing shock me. the waveform trace on the oscilloscope is so thick, seems like parasitic FM component. there is many thing will present such a waveform, i.e, the FM component, a high frequency parasitic frequency, might be stand-wave if not properly terminated the coax.

what do I do to solve this problem? 

nothing, just look around. touch something on the board to see if some part problem, lucky, while i touch the buffer of the crystal controlled oscillator, problem gone.

so suspect the real reason is the radiation  crystal oscillator beating the VCO, generate some new frequency component.  On specific frequency this phenomenon get stronger , so this is a proof to confirm this root cause. i fixed by shield. 

A failed buffer for the crystal oscillator

This buffer simulation by Ltspice show the bandwidth is about 400 Mhz, unbelievable. but the prototype circuits result is awfully , severely distortion, and buffer insert loss is much more than simulation (@60 Mhz) . 

Crystal oscillator buffer amplifier

An 800 Mhz ft BJT working in the 60 Mhz, i did not expect it have too much gain,  if lucky it could provide some power to driven the DBM.  a simple design of buffer amplifier

the amplifier i actually built is much simple than i design.

i have some low Ft Old CAN package transistor, just for interesting, the result still surprise me. a better waveform, might because of the low Ft, but output level is almost not changed.

the 4:1 transformer built from a TV core, work very well on 60 MHz (4T+4T primary, about 2 uH).

DBM mixer

CLP-303 is suitable for this sweep generator. check the IF output, seems something wrong, i expect DBM capable to do more of rejecting RF/LO.

obliviously(left bottom Conner the 1M led not glorying), i use 1:1 probe but put the input of oscilloscope to 50 ohm...this is the opportunity to checking DBM time domain signal waveform.

50 ohm input(well terminated the DBM), high frequency component is much more than 1:1 probe could sense, 1:1 probe is kind of low pass filter. 10:1 probe with properly terminated the DBM, get the similar image as the 50 ohm does.

Design a 45 Mhz low pass filter

May/26  2016

for good attenuation the 60 Mhz LO signal, we need compromised the sweep range, reduce it from 0 - 60 Mhz to 0 -  45 Mhz.  Design the filter with LowHi and adjust with real component value by GPLA, the filter design is so easy. thanks EMRFD.

NOTE: the 1000 is typo, real ideal value is 100 pF, the inductor unit is nH.

Reference data if the coil Q is 200:

capacitor from left to right: 89, 158, 158,89

inductor from left to right: 268, 297, 268

on whole sweep range the amplitude change less than 2 dB (confirmed by manual sweep).

starting from 45 Mhz the amplitude reduce very quickly, work as expected, even through there is no any shield.

post diode mixer amplifier is key for achieve good performance from DBM ,emphasized by many people, books. the IF port should not reflect the wave back, so a isolation amplifier is required and it's input impendance should good match to 50 ohm( EMRFD suggest at least SWR > 10 dB ).

what i expected is wide band amplifier. to get a workable, relatively simple amplifier, i run several circuits in Ltspice ( Run into Wide-Band Buffer/Amplifiers  ).  and choose this CB-CC amplifier.

CB stage have a good isolation and good input impendance stability and surprisingly wide band.

Input impedance 50 ohm keep up to several hundreds Mhz, 60@400Mhz, quite flat.

the positive supply wire is one of sources to cause this problem, it run over the strong VCO's tank, another problem is sometime, not stable.  there should be a feedback from output to the C4, because there are near and 2 sold pad is vertical and close, form a small capacitor, cause too much gain at high frequency, separated the 2 solder pad, problem gone.

obviously, there is distortion. first thing come up into mind is the bias of the amplifier, but after trying to change bias,  nothing changed. especially at low frequency, typical, 200 khz. 

Output level:  -4dBm

Amplitude flatness:  good, but there is about 1dB bump round 20Mhz. VCO output changed around 20Mhz, that's not the amplifier's bad.

Distortion:  obviously.

Kill the distortion

June/2 2016

the following photo showing a ALT mode CH1, CH2.  CH1 is get from the filter output and terminated with 50 Ohm resistor. CH2 take signal directly from mixer output,  CH2 use the oscilloscope as the low pass filter.  20 Mhz BW LIMIT button is pushed, and 20 Mhz LED is glowing.

the signal being tested is around 200 khz, they can not well match after pass through the amplifier/filter.  indicate there is distortion in the amplifier, i guess.  

USE VCO driven the diode DBM

first thing i can improve is which oscillator driven the Mixer.  i have some strange feeling, while a fixed frequency oscillator driven the Mixer as LO, and VCO as RF, subconsciously, i think this is the thing should be. but that's not true.  LO is not that sensitive to the driven level change, most important, this LO is more powerful than the crystal oscillator.  Use the VCO to driven the DBM, will make the output level more stable at pass band.

YES, this works!

But distortion still there. remove the filter, use the oscilloscope's  BW limit function as the filter, i found the waveform output form the post mixer amplifier is quite good. recall there is suggestion everywhere, insert a attenuation pad before the filter.  i choose the 3 dB pad (GND -- 300---in--18---300--GND ).

Insert this 3 dB pad before the filter is the biggest improvement ,  stable the amplifier's load impendance, reduce distortion. filter reflect energy( impendance changed rapidly in passband),  impact the emit follower stage, this pad isolated the filter from the amplifier.  amplifier gain only 10dB, the 3-dB pad seems fairly acceptable.

there is still several point could be improved.  post mixer amplifier bias, the crystal oscillator's bias etc. insert a emitter degeneration resistor to crystal oscillator amplifier make the waveform a little bit improved. 

following is improved topology for the  post mixer amplifier.

finally, i got a quite good output @200 khz. there is still visible distortion, crystal oscillator itself is not that good,  this waveform almost same as the crystal oscillator, that's seems where the distortion came from.

Why high residue frequency component

weekly ago @June/7 2016

Another issue of the filter (actually not because the filter), high frequency (>100 Mhz) attenuation is not good. simulation there should better than -30dBc @60Mhz.   considering the quite bad topology and shield issue of the filter, might acceptable.

sweep from 100khz - 45 Mhz,  flatness  is quite good. ( 5dB per vertical div)

aggressive sweep showing the stop band is abnormally flat, just because seep voltage reach to positive rail and stop to going up.

sweep from zero beat to stop band 

then i trying  to shield the VCO. finished work include quite good shield mixer, and a VCO box. 

after re power, i got a pretty good output, and found output dropped  too much. now -30 dBm output. 

low output, but beautiful waveform, did i completely tame the beast?  NO... finally,  the amplitude reduce because crystal oscillator' amplifier actually disconnected, signal is parasitic coupled to mixer.

after fix everything, the residue frequency component come back. now i need research the residue component.while press the osciloscope 20 Mhz bandwidth limit, i get a 10 Mhz  low frequency , 5  mV,  and a 60 Mhz 5 mV component.

where the 10 Mhz  frequency come from?   I have got the perfect output by accident, right?  seem the crystal oscillator have a parasitic oscillation!!

which impact the output, and tune the oscillator, the 10 Mhz residue frequency vary with that. 

so, that might be why i got following output....

Why high residue frequency component II

several days ago @June/7 2016

cause i suspect the crystal oscillator have some problem,  the Butter topology will be the crystal oscillator instead.  new oscillator.

while i trying to make the oscillator work, i got a UHF 480 Mhz signal, and quite stable, mess construction cause some problem, and bad coil, especially does not design the coil, no calculation, just connect a little bit larger variable air capacitor. 

though i finally find the oscillation point, the capacitor and inductor configuration is not recommended.

sadly, he annoying low frequency residiu still there. playing with interchange LO/RF, insert pad, change oscillator coil. trying to replace the broadband amplifier with tuned circuits, but failed to find suitable resonant tank.

Why high residue frequency component III

today @June/7 2016

frustrated. carefully thinking the frequency conversion, a 61 Mhz crystal, and a VCO tune from 61 Mhz to 114 Mhz, the residue impact much to the high frequency, and its frequency changed while tune the VCO.

this remind me, the actually reason, might be the second harmonics of the crystal oscillator(LO). 2*61 - 144 = 8 Mhz, and this frequency within the pass-band of filter. nothing could help to eliminate this.

oh, this might be the ROOT CAUSE(refer JUMP TO CONCLUSION this is not the all/actually  happens here)!

think about the third harmonics, 3*61 - 114 = 69 Mhz, another generated frequency and this also explain why the LO suppression is so bad !!!, the 3rd harmonics mixer with high frequency of VCO, generate a frequency near LO, for the several week i think it's the LO leak from DBM.

Today, i have a feeling, i can cure the problem.new oscillator with DTC filter.

without the DTC, oscillator output is bad.

while finish tune the DTC, the output from filter is far more good than before!!! but still there is residue frequency. checking with oscilloscope, the output from buffer is messed up by driven the mixer LO. i think a 3-db pad will isolating the buffer, it does! output waveform is pretty now!  but only -10 dBm.

and latter, i suspect serial a 47 ohm resistor with the buffer amplifier will do the same thing.   and i works again!

i don't know what happens. i did notice the reflected energy from mixer, impact the buffer, even the oscillator, the resistor isolate mixer well. series with 47 ohm resistor does not make the output buffer amplifier's output impendance nearing the 50 ohm, but it works just like a attenuation pad. (this just because the RF level is reduced!!!)

Tank :

5T TDK parallel with 22pf( should be 16.8pf), with 22pf, 43pf feedback. output 61.2Mhz, distortion.

DTC

16.8 pf with 6T TDK white coil form.

Get a better configration

back to that day, I did believe this configuration would cure everything, but it's does not.  nightmare continue:

Improving:

Signal: Improve The Log Detector Sweeper 

Conclusion:

DBM Frequency mix generator