The local oscillator and input filter lie in separate boxes and won't be discussed.
Direct Conversion Receive System (Video Supplement)
I've built a lot of stages for this video -- but wanted to test a few of the better 50 Ω input Z audio preamps in an actual DC receiver. A functioning direct conversion receiver provides a good way to test for voltage and power amp instability & noises. I also wanted to get a feel for how much gain we really need despite all the mythos about this topic. With this supplement, I won't have to go into too much detail about the test receiver on the video.
My main receiver goal hasn't changed for 25 years — lift desired RF signals out of the ether with a decent signal to noise ratio while listening to a speaker at comfortable room loudness.
Above — DC receive system. My chassis contains 2 BNC mixer input ports, an SBL-1 diode ring product detector, a post-product detector network and then 3 audio amplifiers plus a speaker jack. The AF amps = a 50 Ω impedance voltage amp, a second voltage amp with an active gain control - and finally a PA. I'll cover each AF stage separately, but first I'll show the complete receive system from inputs to output. Each stage lies on a separate piece of copper clad board. The active stages are numbered 1, 2 and 3. I soldered 2N4401 or 2N4403 for the BJTs.
* I build with Ugly Construction from DC to ~ 2.2 GHz and employ AF or RF bypass capacitors for DC line standoffs as they feed DC to each stage. E.g. I prefer to not use 1 to 10 megohm resistors as standoffs. Ideal voltage sources are low impedance -- I avoid any resistor unless it's needed & if needed, I strive to keep that resistor value as low as possible for sake of noise hygiene.
No receiver active low pass filtration gets applied as I want to hear all the noise, warts and quonks in my reference receiver. The logical place to insert a low-pass seems the input of the stage marked number 2. The shunt 470 pF cap serves as a simple low-pass filter in my receiver.
My RF bypass capacitors are suited towards lower HF reception. I've got some suggested RF bypass caps from 6.6 to 220 MHz covered here If you use SMT parts, this chart will be off. I have measured the series resonant frequency of every RF capacitor in my lab -- and have it written in a notebook, or on that part's drawer or envelope.
Above — Schematic of the PA with DC measures. At this point, I did not add the 22 Ω DC decoupling resistor shown in the completed receiver. Few homebrew PA's will oscillate when tested stand-alone with a signal generator into a resistive load. I also took AC measures with my DSO.
A single voltage divider network feeds each of the 2 current sources. A 220 µF capacitor shunts voltage divider noise to ground.
My readers/audience asked me to make a single DC supply audio PA only using TO-92 transistors as finals.This is it. I worked hard to find a solution where the maximum output transfer function would compare with the venerable LM386 -- and bonus -- this final transistor pair tend to not suffer thermal runaway and smoke up your lab.
The key design features to get those goals included serious degenerative feedback [ 68 + 39 + 1 Ω resistors ], plus current sources to drive the input pair and VAS/final base bias stack. I also set the voltage gain to 21. By increasing the the 2K7 feedback resistor, higher gain lies on tap (a voltage gain of 80 or more arises with a higher feedback resistor); however, this is a power amp and not a voltage amp. Low noise best practices suggest you build up your AF signal voltages with low-noise voltage amps and not within the PA stage.
Above — A close up of the PA in-situ. A temporary orange coloured 1 µF metalized poly film cap lies at ~ 6 o'clock. I connected either a 1 KHz tone or a CD player to the PA via this capacitor. I listened to this PA with my CD player for 4 nights and it sounded lovely & crisp. I built 2 separate PA stages to ensure my design worked. Although preferable, I did not match the input emitter-coupled pair.
Perhaps foolishly -- I did not place heat sinks on my final complimentary pair. All the base drive current comes from the current source and not the usual complimentary pair that drives the finals. Thus they do not run as hot as any other decently designed TO-92 stages I've built.
Above — An FFT of the PA driven to 250 mW with a ~ 1 KHz tone.
Above — An FFT of the PA driven to 697 mW with a ~ 1 KHz tone. Pretty good results from a single 2N4401/2N4403 emitter follower pair.
PA Instability
Once you connect all your AF stages together in a DC receiver, unwanted audio oscillations may occur.
This might be motor boating — a pulsed, typically low frequency oscillation that may even vary in amplitude and cause squegging. In addition to motor boating -- a steady, higher frequency oscillation tone that sounds hollow "or howls" may arise -- this usually occurs at loud volume.
I learned to think of your DC supply line as a highway connecting various stage inputs to outputs throughout your audio chain. Decoupling the DC line with series resistors and bypassing with AF and sometimes RF capacitors shunt to ground helps to stop AC signals from travelling along this highway. The ultimate way might be to use a capacitive multiplier BJT as shown on the first preamp labelled "one". The capacitor value connected to the base gets multiplied by the Beta of the transistor which sets a long time constant for very low frequency oscillations and those above this low corner frequency.
For motor boating, I normally place a 10-22 Ω decoupling resistor and
both a RF and AF bypass capacitor on the PA DC line. I suggest a 470-1000 µF for
the DC audio bypass capacitor as a minimum starting value. Each stage in
your DC receiver should get some low pass filtration with such an RC
network to keep AC signals from travelling down the DC highway.
Further, AF & HF oscillations may also occur in your PA voltage amp called the "VAS".
AF/RF oscillations also require low-pass filtration, but often just a local bypass capacitor alone will do the job. My PA emitted a ~ 800 Hz howling sound when the volume was turned up loudly. I tamed this by soldering a 270 pF MLCC RF cap from the emitter shunt to ground. For my guitar amps, I've had to apply other strategies.
Above — Ways to tame audio oscillations in a PA. The emitter degeneration resistor R1 could be increased to lower VAS gain. For example, from 39 to 47 Ω . And/or the C1 value could be increased to get the best result at high drive into the PA. This testing will annoy your family if you listen through a speaker like me! The VAS serves as the main PA voltage amp and offers a big source of instability in some PA designs.
Above — Additional circuitry I've used to tame a 40W guitar amp PA that oscillated at higher drive levels from AF to RF: 100 pF cap from collector to ground. then a 10 Ω resistor on the VAS collector with a RF bypass cap on each base followed by a series R at the BJT base terminal.
Interestingly in this DC receiver PA build, adding a Zobel network did nothing measurable, so I left it off. I've also connected the VAS base to ground via a series RC network. Sometimes, it's trial and error.
Voltage Amplifier with Active Gain Control
Above — The first version of the inverting active gain control stage. In my reference receiver, I employed the other half of the NE5532 as a follower/buffer. Technically, you do not have to use a buffer, but it helps isolate the active gain stage from the PA input. I've built other active gain control circuits that offer a better log response of the volume control, but this version seems simpler. If you need more maximum gain, drop R1 to 560 or 470 Ω etc..
However, the voltage amp always runs at full gain; potentially reducing headroom -- and perhaps more significantly, amplifies the noise by that maximal voltage gain. With an active gain control, noise is amplified only by the exact amount of gain needed to hear a signal.
Above — The sublime FFT of the active gain control with a NE5532. It's very difficult (but not impossible) to get distortion this low with discrete transistors. Further, the device noise performance is much better if you apply a low noise op-amp over discrete BJTs. I'll likely offer a discrete BJT version(s) of this amp in the future. I've designed and built several.
I prefer to follow the common base with a emitter follower: as aforementioned, this proves essential if you plan to use a active gain control with its inverting input. Further, you get maximum gain since the collector is less likely to get loaded down by any stage that follows. With a voltage gain of 107, this stage pretty much sets the noise figure for the receiver.
