Greetings and welcome!
I’ve blogged about
the LM386 over 2 decades. I’ve received many 10s of emails where builders
told me their problems with the LM386 IC ( and other AF power amps -
including my designs ). Many seem bog standard concerns – such as
motor boating, or excessive noise when the LM386 gets operated in
high gain mode.
For other builders, their problems seemed to arise
when they combine RF, plus high gain AF stages, and a PA like the
LM386 within their radio receiver.
For sure, a solitary LM386 with
reasonable low pass filtering on Pin 6 will generally behave nicely
when tested on your bench. It usually sounds lovely if you bother to
audio test it with recorded music through a speaker.
However, when you place that ol' LM386 in a direct conversion receiver with >= 80 dB AF gain, things might change. Noises such as “howling”, “hissing”, “wild oscillations”, “weird modulations”, “harsh noise emanation” – “especially when the audio gain is turned up” have all been described in emails.
Some builders run the receiver AF + RF stages on a home-designed, fabricated single PC board. Then they may connect a digital VFO to the LO port of the mixer ( and also perhaps use the same DDS as the BFO in superheterodyne receivers). These builders now have DC, digital, AF and RF circuits working with the same 0 volt ‘ground’ return path on their main receiver board.
Ground
Ground (as we mostly think of it) = a 0 volt reference and does not imply earth ground. Combining DC, RF, digital, AF voltage gain stages -- plus an AF power amp with the same return path on a single board may lead to disaster unless you are skilled at designing PC boards. People go to school to learn this stuff – especially in today’s digital electronics world where RF, DC, and high speed logic gates & oscillators all coexist.
Star Grounding
Some believe that star grounding ( defined as the parallel connection of isolated circuit board ground planes to a single 0 volt reference point ) will work when combining DC, RF, an AF power amp +/ a digital VFO. It might help in some cases, however, often it won’t.
Parallel single point “star” grounding generally works for signals less than 100 KHz; however @ RF – ‘ground wires & traces’ may cause a high inductive reactance RF ground return path. As you move up in frequency, PC board traces & wires become transmission lines. At RF, these transmission lines may inductively or capacitively couple to other transmission lines, or nearby circuity, in what we often call cross talk.
Also at RF, grounding wires may exhibit capacitance and often enough, parasitic capacitance.
Sadly, some
builders may also learn that high bandwidth noisy circuits may
corrupt nearby low noise circuitry with all this unwanted coupling
going on in their circuitry.
DC Supply Filtration or Battery?
Poorly filtered DC circuit paths may serve as a ‘freeway’ allowing noise, RF, AF and digital pulses to modulate other AC signals, or perhaps provide the needed phase shift from output to input to create an oscillator around amplifier stage(s) -- and really frustrate the home brew experience. I won’t even delve into the world of ground loops and hum – another nail biter.
Back to the single PC board builders who were not experienced in PC board design. To their credit, usually, their VFO was located in a separate box, albeit they seemingly rarely employed RF tight enclosures.
To solve their AF power amp woes, some of these builders – opted to power their PA such as the LM386 with a 9 volt battery cell (or other battery pack) and keeping the audio PA circuit a reasonable distance from their main receiver. Others went with a powered speaker. These might be the only practical solution for builders who made a single PC board receiver and could not get a stable audio power amp despite trying to patch the problem.
Yay 1970s
It’s so great to learn from the past and happily builders in the 1970’s knew how manage the LM386 -- but sadly, they didn’t make videos back then. In short, place your audio PA on a separate copper clad or printed circuit board with the ground wire returning to your single point 0 volt node on your receiver DC power buss. Filter the DC voltage going to pin 6 very well. E.g. solder a 470 to 1000 µF capacitor right on pin 6. You may also have to insert a series resistor such as 10 ohms between pin 6 and the DC supply to lower the RC low pass 3 dB cutoff frequency.
Keep the input circuitry away from the output circuitry. Watch that you don’t push too much low frequency or high frequency audio into the LM386 since these may increase instability, clipping and noise.
A well designed audio chain attenuates the low and high frequencies that we don’t need in our receive audio chain. Case in point – some builders prefer to make all discrete BJT audio stages – for example, with the 2N3904. This is OK, however, consider that the 2N3904 has an fT of 300 MHz and that we use this BJT as an RF amplifier @ HF.
At audio, a 2N3904 amp may offer the same gain at 1-2 KHz as it does at 40-50 KHz. Do you need high gain at 40-50 KHz in a communications receiver audio chain? Ideally, the audio gain should taper off as we move up in audio frequency to reduce the noise bandwidth, and to promote stability in the audio chain which includes our LM386. Added noise is always unwanted and in simple terms -- higher bandwidth means higher noise.
Some Hams and SWLs have normalized receiver wide band AF noise since even commercial gear makers may cut costs on their audio chains while focusing on the highly marketable, high dynamic range bleeding edge RF circuitry -- and beautiful, colorful waterfall displays.
For home brew – local and global amplifier negative feedback, AF low pass + some high pass filtration, careful layout with respect to inputs and outputs, strategies to reduce AF ground loops -- plus careful DC filtration will boost your chances of making stable, great sounding, receiver audio.
While HF construction is common, very few builders venture into VHF and UHF -- and some never acquire the hard-fought skills needed to successful lay out, shield and manage real and parasitic interfering currents with common RF return path circuitry. Once you’ve done work + measurement above 30 MHz, you tend to build every circuit like it operates at 1 GHz.
I encourage builders to venture into VHF and beyond. It’s great fun!
A 7.X MHz receiver
may be quite forgiving to sloppy construction, layout and DC
filtration – great for new builders; but perhaps less than ideal
if you seek to truly learn radio craft and move beyond cook book / copyist
status.
Whether you make digital, analog, or mixed-circuit radios that drive a speaker or headphones – basic rules
apply & the laws of physicals don’t change.
Above — An LM386 with negative feedback ( 10K to 1 nF) to reduce audio bandwidth. Pin 6 features perhaps over-the-top DC filtration - an active low pass filter via the BD139.
I plan to experiment with this circuit further -- and to measure the differences with distinct Pin 6 filter circuits. Meanwhile, I'm enjoying listening to monaural music through it.
Above — DSO screen shot with LM386 driven to maximum clean output power @ 1 KHz
Above — FFT with LM386 driven to 1 KHz maximum clean output power
Above — I tested many aspects of the LM386. For example: what happens when you bypass Pin 8 to ground. This drops the voltage gain by ~ 60% @ 1 KHz! I also tested it with different speakers with different speaker efficiencies to measure the impacts on volume, stability and average current draw.
I still love the LM386 despite it's ancient pedigree, input noise, and its many other warts. Very few analog ICs enjoy such an amazing legacy. It's simply a fun part to experiment with -- and on which to learn our craft.
Best to you!
Amen
ReplyDeleteThank you Vasily. Many builders seem to think that, "if only Dean and Bill had allowed the use of ICs, we wouldn't have all these feedback problems in the AF amp of the SolderSmoke Direct Conversion receiver." You excellent report should disabuse them of this notion. As you note, LM386 chips can and do oscillate and motorboat. 73 Bill Hi7/N2CQR
ReplyDelete