EEVblog #767 - Super Regenerative Receiver Problems

[EEVblog]

Dave shows a potential problem with Super Regenerative Receiver circuits, and looks at the issue of interference with his LED light remote control mains switches.
Some fun with two spectrum analysers, a scope, a H-Field probe, and yes, of course a teardown.
The Tekbox EMC near field probes: http://amzn.to/1Ma20zh

[ElectroIrradiator]

Saw this episode coming from a mile away.

[EEVblog]

Saw this episode coming from a mile away.

Yeah, didn't take Nostradamus!

[ElectroIrradiator]

Some additional comments:

One needs to differentiate between a regular regenerative detector/receiver, and a superregenerative ditto. Their schematic may at first blush look eerily similar, yet their method of operation is completely different.

A classic regenerative RX generates a continuous sinewave oscillation at or very near the intended receiving frequency. The RF input signal from the antenna is mixed with the LO signal and an audio or supersonic low frequency signal is the result. This is where the frequently seen audio transformer comes into play in your MW broadcast regenerative RX. So with a regenerative detector you'd expect to see a single frequency being radiated by the detector.

A superregenerative detector is a completely different beast. In this the oscillator is continuously driven into and out of oscillation, usually at a supersonic rate, called the quench rate. Thus the name. The trick is that when the oscillator is turned on, the gain of the RF transistor is 'slowly' increased up until the onset of oscillation. This is done by using some form of positive ramp on the base current.

When an oscillator starts up, it is the noise in the circuit, which provides the initial 'seed' for the amplifier in the circuit. This is the effect, which is (ab)used in a superregenerative detector. How this works is, that at the point of onset of oscillation, the time it takes for the oscillations to exponentially ramp up from the local noise level hugely depends on whether - by coincidence or design - a (very) low level signal is injected into the oscillator from the outside. This effect is ridiculously sensitive, so a supersonic detector is quite a bit more sensitive than your regular regenerative detector with continuous oscillations.

The oscillator thus converts RF input level from the antenna/ambient into time it spends at full oscillation amplitude. Higher signal level means it ramps up much quicker. IE. the DC power input to the oscillator, averaged over many quench cycles, indicates the ambient RF level around the RX frequency. If, like in this case, the quench rate is several hundred kHz, then you can easily detect data bit rates from the transmitter of several thousands. Just measure the voltage drop across a fixed resistor, or, like what is maybe used here, use a simple diode detector of the RF oscillator signal.

So the oscillator ramp up waveform depends heavily on the ambient RF noise/signal level. Which means the ambient RF noise around the RX frequency modulates the oscillator, which in turn is the main reason for the ridiculously wide noise piedestal you see on the SA.

[EEVblog]

So the oscillator ramp up waveform depends heavily on the ambient RF noise/signal level. Which means the ambient RF noise around the RX frequency modulates the oscillator, which in turn is the main reason for the ridiculously wide noise piedestal you see on the SA.

Yes. Excellent additional detail, thanks.
David2 and I discussed the possibility of a Fundamentals Friday video on this.

[vk3yedotcom]

David2 and I discussed the possibility of a Fundamentals Friday video on this.

Great idea!  Superregen receivers can be bad tempered (with all sort of squeals and hash) yet still grudgingly likable for what it can do.

A single MPF102 & 5 turns of wire wrapped around a pencil is good for the FM broadcast band.  There may be some interesting effects when tuned to a station - definitely recommended as part of a FF demo.  Sensitivity is incredibly high - even with no external antenna connected.  Another thing that's magic is the tiny current required - eg about 1mA for a working receiver, eg

[rs20]

How does the build quality of the units feel? I've been meaning to build a bluetooth-based mains timer, but got stuck on the mechanicals/case for it. For $13.30, this looks like a great compact empty box with Aussie mains plug and socket ready to use!

[rollatorwieltje]

How would you shield this thing if the noise is actually emitted by the "antenna"? You obviously can't shield the antenna itself?

[bktemp]

You can add a preamplfier for decoupling the antenne from the oscillator and shield the LC tank circuit.

[N2IXK]

How about a Fundamentals Friday discussing the basics of RF receivers and detection schemes?  You know, Regen/Superregen/TRF/Superhet and the strengths/weaknesses of each topology. Same for diode detector/synchronous detector/BFO/ratio detector/FM discriminator/etc.

The basics of RF don't need to be such a "black art" as they are often made out to be, particularly at lower frequencies, where components like inductors and caps  are still recognizable as such.

[G7PSK]

You could try giving the receiver that wont work while close to the other tweak on the coil core, that sometimes helps by reducing the gain in the circuit.

[PA0PBZ]

I wonder how much of the spikes you got on the SA just came from the mains and not from the receiver, you just placed the antenna on top of an AC fed box.
Also, increase the RBW of the SA and decrease the span to get a faster update, there's no need to look at 5 MHz or 500Hz bandwidth when looking for noise produced by the receiver.

[AF6LJ]

Well Dave welcome to the world of pure garbage engineering. Only the lowest of the low would use a regenerative receiver. I suppose you got what you paid for.
And of course the transmitter looks much worse.

Pure garbage as its lowest.

[AF6LJ]

Just as post script on this...
I paused the video and did a little quick reverse engineering and it looks like the extra lead length on coil L2 may be the actual antenna....
I could be wrong but I don't think so.

Give me an HP-8569 any day over that Rigol SA.
or for that matter my old HP-8551 / 851. (it is almost older than I).

[TheSteve]

Dave, this was a great video to see. Many years back in the numeric and alphanumeric pager days I came across a problem with a VHF pager made by NEC. It used a 17.9 MHz IF that leaked something fierce. You could pickup the LO 20 feet away, any other pager with a 17.9 MHz IF would lose reception if within 10 feet of the NEC pager. It was just a dream come true to track down, lol.

[G0HZU]

I guess you could try putting a small sheet of RF absorber in between the two units but the absorber would probably cost much more than both units 

Well Dave welcome to the world of pure garbage engineering. Only the lowest of the low would use a regenerative receiver.

I think that regen and super regen receivers are actually quite elegant in terms of their performance vs simplicity .

[vk3yedotcom]

I think that regen and super regen receivers are actually quite elegant in terms of their performance vs simplicity .

Especially if you can add an electret microphone to the superregen and exploit its radiation to make it a 1 transistor transceiver!   

Eg http://www.vk2zay.net/article/156

Two for the price of one.

[f4eru]

what a heap of crap, these receivers.
Seriously, a superhet IC with XTAL and PLL costs 85cent these days, and it works, even when there is another ISM receiver around.
People building receivers that don't work should be shot at.

[apis]

Nice screwdriver hack. I wonder why they bother with those screws, it won't stop anyone who has decided to open one and I imagine it cost extra. But maybe that's where the budget for an extra transistor went.

David2 and I discussed the possibility of a Fundamentals Friday video on this.

That would be great!

The basics of RF don't need to be such a "black art" as they are often made out to be, particularly at lower frequencies, where components like inductors and caps  are still recognizable as such.

I suspect it has a lot to do with most hobyists not having access to GHz bandwith equipment (and then there is all the regulations surrounding transmitters).

You can add a preamplfier for decoupling the antenne from the oscillator and shield the LC tank circuit.

But why add another transistor when it almost works without it!?

[AF6LJ]

what a heap of crap, these receivers.
Seriously, a superhet IC with XTAL and PLL costs 85cent these days, and it works, even when there is another ISM receiver around.
People building receivers that don't work should be shot at.

You don't even need a crystal with a superhet...
The garage door receivers here in the states are superhets and the ones I have worked on are not crystal controlled. Just an LC oscillator at half the frequency of that remote lighting junk.
 

[AF6LJ]

There is no black art to RF and it gets easier as you approach 1GHZ.

[RupertGo]

I've got a couple of commercial super-regen receivers here in actual radios. One is a super-crappy 1970s single channel CB walkie-talkie, which has (if I remember correctly) a whole three transistors - a 27.125 MHz crystal oscillator, one doing AF things, and one being a super-wide super-regen receiver. You can tell when it's on because anything else in the place that's tuned between 26-30 MHz makes a rude hissing noise; it doesn't have that much more range when you actually transmit. But a fine example of what you can do with very little. When conditions were good, I remember hearing Radio South Africa on the 26 MHz band come thundering in...

The other is more interesting, it's the B set in my Wireless Sets 19, the classic British/Allied tank radio from WWII. The A set is a proper superhet transceiver on 2-8 MHz (actually, I think it's the first ever transceiver, in that the RX and TX are tuned by the same circuit) that puts out about 10 watts. The B set, though, is for 230-250 MHz (ish), also tuned by one knob, and I think that produces around 10mW. That was used for tank-to-tank comms. Most people who used the 19 set after the war pulled out the B set gubbins, because it produced enough interference to mess up VHF TV over quite a wide area and thus made a chap unpopular. I think that's around three valves (the A set has ten or so), which was really important in the days when valves were in critically short supply. It was also a very high frequency for the time, which made for a very short range and thus fewer security worries.

I would love to see Dave do a teardown on a 19 Set - but it won't be mine, first because it's a prized piece of my collection, but also it weighs about 40 kilos all-in and I think I could afford a holiday in Oz for the cost of shipping the thing from Blighty.

(odd radio fact: the German tank radios used 27 MHz, much like my little walkie-talkie, which was also short range enough to avoid interception nearby, but which during the excellent propagation of the time of the African desert war meant that receivers in the UK could pick up Radio Rommel just fine for large parts of the day.)

[AF6LJ]

I have a good teardown for Dave but it costs too much to ship all the way across the pond.
MY HP-5245L Universal Counter, there ain't no chips in that puppy and it works great all the way to 3GHZ.

[apis]

Found this 1998 article from arrl if anyone wants to build one:
http://www.arrl.org/files/file/Technology/tis/info/pdf/9811qex026.pdf

[AF6LJ]

Found this 1998 article from arrl if anyone wants to build one:
http://www.arrl.org/files/file/Technology/tis/info/pdf/9811qex026.pdf

That is a good article, the old school regen receivers had one or two stages of tuned RF amplification ahead of the detector. For a time I owned a LF/VLF regenerative receiver from the 1930s. (14KHZ to 600KHZ coverage.) It worked like a bomb and was well shielded.