one-transistor full wave rectifier (link)

[Paul Perry (Frostwave)]

www.elecdesign.com/Articles/ArticleID/6379/6379.html#

I'm sure I've seen a simpler (but rougher) one.. anyone?

[Joe Davisson]

FWIW, I drove one with a gain stage and eliminated the feedback cap/resistor, and picking lightly makes an faint octave-up. Needs something to filter out the lower frequency and amplify the octave part, I guess.

[GFR]

Very interesting circuit!

I simulated it in spice, here's my findings:

It seems not very suited for small level signal rectification, as when the input signal goes below ~400mV peak to peak it looks no more like a FWR "comb" waveform but like a "mismatched sinusoid"

Also, the output impedance of your source matters. Above ~2k the output gets very assymetric and the octave up goes through the window. So driving it from the guitar or from a common emitter (source) stage is not a good idea.

The input Z of the following stage is not so critical, assymetry only gets bad below ~20k.

Also the RX in the text can have a big effect in matching both halves of the FWR.

So it seems that for an octave up effect you need to drive it with loads of gain and a low Z, perhaps an opamp stage is OK.

I have seem another "simple" FWR in the "IC Opamp Cookbook" (Jung), it uses a single opamp, a single diode, 3 resistors, single supply. The opamp needs to handle inputs near and below (V-) for it to work (the CA3140 will do it). It's fig 5.21 in my edition.

The trick is when the input is negative the gain is (-0.5), when the input is positive the opamp turns "off" and the 3 resistors make a resistive divider with gain (+0.5).

Disdvantages: The gain is 0.5. The Zout of the preceding stage and the Zin of the following stage matter a lot. Advantage: It does work OK for small signals (couple of mV).

[stm]

Initially I was very excited about this circuit.  "A single transistor octaver!" I thought.

I run several SPICE simulations with 2N3904, 2N2222 and 2N5089 transistor models, but in any case the circuit performance is far from ideal.  I used Rx=open or infinity, since I wanted the full-wave rectifier effect.

As GFR said, the output is assymetrical.  At very low input levels (50mV peak sinusoid) I managed to obtain good frequency doubling by doing a fine tuning on the 13k resistor.  With the 2N3904 transistor model the optimum value appeared around 13.25k in my simulation.  Under this condition the output waveform looked like nearly perfect 2 kHz sinusoid.

So this suggests using a trimmer on the biasing network.  The problem is that I thought OK, what will happen when the battery voltage drops from due to battery aging.  I changed the 12V supply to 11V and of course the trim is no longer good, so this circuit is not very practical as it is for low level signals.

Mmmhhh... not excited anymore...

Gotta try tweaking the feedback Rx resistor to see if some improvement is achievable.

Just remeber the "there's no free lunch" philosophy.

Regards,

STM

[GFR]

At very low input levels (50mV peak sinusoid) I managed to obtain good frequency doubling by doing a fine tuning on the 13k resistor.  With the 2N3904 transistor model the optimum value appeared around 13.25k in my simulation.  Under this condition the output waveform looked like nearly perfect 2 kHz sinusoid.

When the circuit is working as expected, it does not  look like a perfect sinusoid, but as a rectified sinusoid (like a "comb" - round tops, sharp "teeth" on the bottom). It doesn't happen until the input is very hot. And then the tops are not of the same "heigth".

[stm]

When the circuit is working as expected, it does not  look like a perfect sinusoid, but as a rectified sinusoid (like a "comb" - round tops, sharp "teeth" on the bottom). It doesn't happen until the input is very hot. And then the tops are not of the same "heigth".

I agree, the actual circuit behaves far from ideal!  Anyway I mentioned the "sinusoidal" frequency doubling characteristic at low signal levels because it appeared more attractive to me due to the reduced fizzines it may have.

I am starting to think the article was published on a rush without much verification on its actual performance, apart from a couple of particular cases.

Could you point to a schematic on the FWR op-amp circuit you mentioned above?

[GFR]

The preamp is saturated at idle. When the input swings it goes out of saturation for a gain of -1 or gets "reverse biased" for a gain of 1. The problem is, you need some voltage to drive it out of saturation, and some voltage to make the BC junction conduct (reverse bias). So there's a big "dead-zone" where the transistor is "off".

The sinusoidal characteristic may not be as attractive as it seems - it can be too "dull" sounding. Listen to the samples at

http://www.geocities.com/gfr.geo/octave.html

and judge for yourself

The opamp FWR I mentioned is in the "IC opamp cookbook" (Jung). I highly recommend buying the book, that's good stuff there from beginners to advanced level.

[Lazy_Jarl]

There is a one op-amp design on the intersil CA3140 datasheet.  It's not a general design, as I think you need one of the CA3140/30/60 op-amps to do it (page 17):

http://www.intersil.com/data/fn/fn957.pdf

peace

[GFR]

That's exactly the same circuit as in the book. Ooops, in the book he says he took it from the datasheet

It's not general, you need an opamp like the CA3140 that can handle inputs lower than V-.

It does work, I breadborded it.

Other opamp types that (may) work (from the book): 358, 324, 759, Lm-10, OP-20.

[stm]

Thanks for the info.  Interesting indeed.  I think also other rail-to-rail op-amps like the TLC2262 should work fine.

Did you notice the odd tonecontrols that appear one or two pages before the FWR on the datasheet?

[Tim Escobedo]

I use a similar one-op amp single diode FWR in the TMK, and find it works adequately. I use a TL062 without problems. I don't know how well it will work as a stand alone FWR, in the TMK it's driven by a low impedance gain stage (op amp).