Heterodyne: Build A Traditional Radio
May 22 2017
A superheterodyne is my standard build target in each year. this time, i hope there will be a design consideration journey, re-discovery trade-off and problems Hams ever made.
LO
every amateur builder will seeking unavailable/rear parts/components, maybe just too lazy to build a coil. especially i already have a bunch of stock IF transformer and LO coil (10x10 mm Tokyo style IF transformer ).
the Harley oscillator, equipped with JFET, will get so many fancy feature, frequency stability, amplitude stability, wide frequency coverage.. all of these achieved with so simple circuits.
first shot with a SZZ11 AM specific coil. 165 uH +- 25 uH, put the tracking problem away, the waveform is beautiful, easy starting, stable, simple as expected ( i have built this VCO: AM Hartley LO )
based on tracking calculation googled from google, the LO tank should had 117 uH-137 uH coil( 520 Khz-1650 Khz). i had the closest components is this SZZ11 traditional AM oscillator coil. which inductance in range 140~180 uH. (some of them could reach to minimal 135uH )
i had to build another coil, which feedback and signal output via 2 separate winding
this coil give out a very distortion waveform at higher frequency while get signal from a separated winding. however the tank waveform is obviously good. i guess might because separated winding is loose couple to tank, while extracting signal, the feedback winding energy been pulled out, leading to distortion. but just guess.
Buffer
This buffers is famous.
recommended by ARRL.
there is several limitation should keep in mind:
1. The Bandwidth is very limited. roll off @4Mhz even through the 9018 Ft> 800 Mhz. lower 33k to 10k extended the BW to around 8 Mhz.
2. Drive power is limited around 1 Vpp (500 mV peak) in this configuration. refer to AMP: Simple RF Amplifer buffer stage, increase it's drive power lower it's efficiency very soon.
what if C2 is omitted? ok, if R6 is DC ground with IFT1, simulation indicate quiescent current will increase a little bit depends R6 value. everything other is same.
2n2222， from 1Mhz to 2 Mhz output is reduce 4 dB. the BW is a problem. i'd to change 2N2222 to 2sc9018 which Ft is 800 Mhz to flat the response.
Tracking
because worry about my 150uH IFT track problem, i revised a ancient coil.
what already mentioned, this coil seems to ringing distorted waveform if adding the separate winding for feedback and signal extract. how bad if use that IFT?
to answer this question, i need a tracking calculator. this calculator should provide possibility to change the exactly tracking points(IF is equal to LO-Signal). and should be efficents to change parameters, the command line tool is best choice, of course, there should have tracking graph.
(NOTE: the Ct is always assume to be zero)
So, i wired up some python code, and get a tool for this purpose.
Heterodyne tracking calculator
evaluate parameter one by one, i discovery some good news. different calculator might given different result. if allow choose the 3 exactly tracking points, the total tracking error and the LO inductance could be tuned to suitable value.
lower the low end single frequency from 520 khz change to 510 khz, the most of SZZ1 suitable (SZZ1 inductance value spread from 135uH - 190uH) with no further compromised. even without tune the signal coverage range, this calculator's indicate the best LO inductance is 140uH, and max tracking error 5.5 KHz. SZZ1 is usable.
Conclusion:
if the Cz range is (12.7- 344) or with some additional capacitance (12.7-355), even 365, the minimal Lz is 141uH. in this condition, pad capacitor, trimmer capacitance changed in several pF, max tracking error will in 5 KHz- 7 kHz, all acceptable.
the SZZ1 inductance could meet this requirements.
band cover estimate
python track.py --IF 455 --Cmax=355 --Cmin=12.7 --Fmin=0.515 --Fmax=1.650
Signal Pad Capacitor CL: 24.2460143859 pF
Signal tank inductance: 251.825500838 uH
LO Capacitor Cm: 32.2597670667 pF
LO Pad Capacitor Cp: 352.640350996 pF
Lo Inductance Lz: 144.12 uH
============Tracking graph===============
Max tracing error: 5.95362123329 KHz
Exactly tracking points : 0.577425 1.00305 1.5592 Mhz
Traditional BJT Mixer and IF amplifier
this kind of typical old radio if amplifier, nothing new. one thing is rarely used, floating signal ground, i don't know why yet.
all facility used here is barely simple and traditional. it's so normal seems definitely work, but like always, it won't.
fist time connect all together, sound totally mass, big noise, no station at all, even with a 12 meter long wire as antenna. such a strong signal should over driven the simple Het which will be expectation.
there is strong LO leaking on T3, seem 455 kHz tank had no signal selection function at all. use the desktop audio amplifier, with the long antenna, i got several strong station, which mean the sensitivity is so bad.
another problem might be IF strip itself is not stable. but after disconnect mixer i got a very stable IF amplifier, no any phenomenon indicate IF stage is unstable.so definitely mixer's problem.
i suspect mixer driver level is too high, 4 dBm to 50 ohm load, which is definitely over the range of suggestion LO level (around 200 mVpp - 400 mVpp). for cure the mixer's huge noise, i had to set quiescent Current to zero, Q3's base on 0.4 volt. in this condition, with long antenna, i got several station. the sensitivity is so low.
the tracking alignment won't fix anything, does not take the responsibility of this. i also try add simple audio filter, alter the TDA2822M audio power amplifier, to isolation the problem.
i run the IF strip without AGC, but after add AGC, nothing changed at all.
Understanding the BJT mixer
i highly suspect the mixer drive level is too high. second day, i realized i had the step attenuator, why not try it now. sad, does not cure anything, but the dawn is coming.
what if i drive the mixer bypass the buffer, connect the C2 directly to the mixer? i can not believe it, the Het works damn well. now the real problem coming, how the buffer make everything mass? seems not a reasonable effect.
from now on, everything got right direction, i noticed every traditional AM receiver driven the bipolar transistor mixer via the transformer, this got to mean something i did not notice before. yeah, the key might be the frequencies generated by mixer, how them being suppressed? if driven mixer with transformer, the secondary wind present a low impendance to all signal, then the collector 455 KHz tank will amplifer the wanted signal, suppress others.
now if i bypass the R5, and driven with buffer, what will happen?
astonish! it works as well as driven by transformer. the first glance make a illusion, all signal will bypassed to ground include LO. but anything is not perfect, and the emitter input had a very low input impendance, so Lo could work in such a situation. even more, the bypassed capacitor make the 455 KHz signal get large gain in this station.
if there is no bypass, and does not driven by transformer, the 455 KHz gain of the mixer will very low, the suppress capability is loss, thus the S/N ratio is low.
compare to bypass capacitor(which is totally odd), driven by a coil, present low impenitence for IF and other frequency component is the best choice ever.
the collector load impedance in 1k - 2k range, use a lower resistance R10, should get it work(TO BE TEST).
Do we really need transformer driven
one thing left to try, if it need some gain in this stage to to improve S/N, how about use a low R10, ie, the 100R, then buffer the signal? Testing result from this change is failed. still extremely low S/N.
and how about adding a transformer? real circuits tell it's bad idea, exactly same failure pattern, Huge noise flooding the real signal. got to remove the buffer to kill the noise.
dead end? but, wait, i find this:
this time, driven the BJT with buffer works perfectly, even the noise is lower than driven directly by oscillator tank's secondary winding. Suddly i GOT IT. bipolar transistor's dynamic range is so low, driven hard leading to huge noise. i have nothing to say about it.
do we really need this transformer? no, we don't. the solution is light couple, buffering output couple to bipolar transistor via a 1K resistor or 50 pF(LOOK THE number!!), works as well as driven directly by LO, and this time pulling effect gone, we got virtue from buffer instead of noise!
We DO NOT NEED the TRANSFORMER！
and the last question, do we need this mixer get some gain? YES, we did, recover the R10 to 1k confirm we NEED THE GAIN ALSO. we had to AC ground the R10 via about 100 R resistor, just as the testing circuits does.
the most wired thing is resistor RF attenuation DOS NOT WORK i confirm this again!