Super Full Wave Octaver

[Eb7+9]

I recently came across an interesting Full-Wave rectifier circuit while reversing a compressor circuit with a fellow forumite ... this FW circuit works along the same line as a super-diode op-amp ciruit - that is, it produces an absolute value function f(x)=|x| but without the deadband around the origin normally associated with diode rectifiers ...

the circuit here is an adaptation/variation of that circuit suitable for instrument processing ... having near-ideal transition characteristics means the circuit produces a sharp noise spike at the bottom of the waveform ... so, by nature, it isn't as clean as a multiplying octaver ...

the one thing that I found interesting with this circuit is what happens when you vary the feedback resistor in the second stage (see simulation deck below) ... when it is zero you get a super half-rectified output, when it is equal to the resistor going between the two inverting inputs you get balanced half-wave and when larger the other half of the waveform grows larger ...

the sound goes from a splatty octave to a smooth ringy octave to a louder fuzz-like buzz ... I added a volume control but it isn't absolutely necessary since the slope is unity in both quadrants in balanced mode ...

... there ya go, another Octaver  with balance tuning

~jc

Code Select Expand


Super(FW)Rectifier Test

vin 1 0 0 ac 1
vcc 99 0 9
vee 98 0 -9

xfilter1 1 2 99 98 3 LM741NS
xfilter2 4 5 99 98 6 LM741NS

d1 2 3 1N4148
d2 3 4 1N4148

R1 2 5 100k
R2 5 6 100k
R3 4 0 100k
*   alter R2 for unbalanced modes

*//////////////////////////////////////////////////////////
*LM741 OPERATIONAL AMPLIFIER MACRO-MODEL
*//////////////////////////////////////////////////////////
*
* connections:      non-inverting input
*                   |   inverting input
*                   |   |   positive power supply
*                   |   |   |   negative power supply
*                   |   |   |   |   output
*                   |   |   |   |   |
*                   |   |   |   |   |
.SUBCKT LM741NS     1   2  99  50  28
*
*Features:
*Improved performance over industry standards
*Plug-in replacement for LM709,LM201,MC1439,748
*Input and output overload protection
*
****************INPUT STAGE**************
*
IOS 2 1 20N
*^Input offset current
R1 1 3 250K
R2 3 2 250K
I1 4 50 100U
R3 5 99 517
R4 6 99 517
Q1 5 2 4 QX
Q2 6 7 4 QX
*Fp2=2.55 MHz
C4 5 6 60.3614P
*
***********COMMON MODE EFFECT***********
*
I2 99 50 1.6MA
*^Quiescent supply current
EOS 7 1 POLY(1) 16 49 1E-3 1
*Input offset voltage.^
R8 99 49 40K
R9 49 50 40K
*
*********OUTPUT VOLTAGE LIMITING********
V2 99 8 1.63
D1 9 8 DX
D2 10 9 DX
V3 10 50 1.63
*
**************SECOND STAGE**************
*
EH 99 98 99 49 1
G1 98 9 5 6 2.1E-3
*Fp1=5 Hz
R5 98 9 95.493MEG
C3 98 9 333.33P
*
***************POLE STAGE***************
*
*Fp=30 MHz
G3 98 15 9 49 1E-6
R12 98 15 1MEG
C5 98 15 5.3052E-15
*
*********COMMON-MODE ZERO STAGE*********
*
*Fpcm=300 Hz
G4 98 16 3 49 3.1623E-8
L2 98 17 530.5M
R13 17 16 1K
*
**************OUTPUT STAGE**************
*
F6 50 99 POLY(1) V6 450U 1
E1 99 23 99 15 1
R16 24 23 25
D5 26 24 DX
V6 26 22 0.65V
R17 23 25 25
D6 25 27 DX
V7 22 27 0.65V
V5 22 21 0.18V
D4 21 15 DX
V4 20 22 0.18V
D3 15 20 DX
L3 22 28 100P
RL3 22 28 100K
*
***************MODELS USED**************
*
.MODEL DX D(IS=1E-15)
.MODEL QX NPN(BF=625)
*
.ENDS

.MODEL 1N4148  D(Is=0.1p Rs=16 CJO=2p Tt=12n Bv=100 Ibv=0.1p)


[soggybag]

This is very interesting, does it run on +9V or does it require two batteries for +-9V?

[Processaurus]

Check the supply pins on the first opamp in the schematic.  Looks like it works on the old 9v.  Thanks for the circuit, JC.  The Scrambler too has a pot that goes between half rectified and full, which is nice too have.

[Mark Hammer]

Well THAT sure is simple enough!

Elektor had an op-amp FWR octaver way back when.  I have the English article somewhere in a binder, but I'm fairly confident it was posted by someone from Brazil (the Portugese version that is; same layout, just more soul).

[Arn C.]

~JC,
   So, you used a 741?  Have you tried any other opamps?   Just curious

Peace!
Arn C.

[puzzle87]

Try LM358,  100k instead 15k and 100k pot, and Ge diodes. Good result.

[jmusser]

Did you see my review on this circuit? Puzzle87 are you saying that the suggested mods would bring out the fuzz and octave? Have you tried them yourself?

[puzzle87]

I`m  tried this as full wave rectifier in my compressor circuit 

[gez]

How is this circuit a full-wave rectifier?  I only see a half-wave rectifier (going by the schematic, I haven't breadboarded this).  Am I missing something??

[Paul Perry (Frostwave)]

How is this circuit a full-wave rectifier?  I only see a half-wave rectifier (going by the schematic, I haven't breadboarded this).  Am I missing something??

It might work like this: Imagine adding together a signal, plus TWICE the inverted negative half cycle. So you have the + part of the cycle go straight thru, but the negative half cycle is detected (half wacve rectifier) then inverted & doubled so it cancels the original negative cycle 7 replaces it wiht an inverted waveform.

[gez]

Paul, you’ve described the classic full-wave ‘ideal-diode’ thingy, but I don’t see it here. 

On +ve input swings the first amp's output goes +ve.  The diode in the first amp’s feedback loop will cut off so the only possible feedback comes via the second amp.  The threshold of the diode connecting the two amps is accounted for so you have the ‘ideal’ diode bit going on here.  But, on -ve input swings the output of amp 1 goes -ve and the diode in the feedback loop clamps the output to a diode drop below Vref (before tracking input signal) and provides feedback to the first amp.  The other diode is reverse biased so no signal appears at the output of amp 2 - only half-wave rectified then.

The only way I can see this working is if the output of the first amp somehow pulls its -ve input down and this gets amplified by the second amp acting in inverting mode.  Seems unlikely though, and it would be by a miniscule amount surely?!

I ask as I had a PM and this circuit came up (not being funny here JC, just wondering if I’m missing something).

[gez]

The only way I can see this working is if the output of the first amp somehow pulls its -ve input down and this gets amplified by the second amp acting in inverting mode. 

Ahh, got it.  I think this is what's happening - negative input follows signal on +ve input once once diode in feedback loop is accounted for by the output, then this signal is amplified by second amp in inverting mode.  Yes?

[gez]

OK, read through the original post again and I understand it now.  Was forgetting that 1st amp was non-inverting and acts as a follower. 

Thanks anyway Paul, you made me think about it a little more

[Paul Perry (Frostwave)]

OK Gez, you can explain it to ME now.....
reminds me of the joke about the maths lecturer who is asked to explain something, he says "It's obvious", then he thinks a bit more, goes away to his room & comes back half an hour later saying, "Yes, I was right, it IS obvious".

[gez]

OK Gez, you can explain it to ME now.....

I take it you were pulling my leg Paul? 

Actually, this is a really cool circuit.  The method you described Paul is the one I normally use, but when I actually scoped the thing a few years ago I discovered it wasn't that accurate -  I needed a trimpot in there as there was some discrepancy between the amplitudes of each half-cycle (thanks Ton for that one!).  I should think the above circuit is more accurate.  I'll have to breadboard it some time and give it a work out! 

[Eb7+9]

there's a 741 (1558/4558) macro-model in the sim but I used a TL072 dual op-amp in my proto / I don't see the op-amp type as being crucial  ... good to hear various comments, Ge diodes, etc ...

someone asked me if there's a way to get rid of the "fizz" in this circuit - like I posted above these harmonics are built into the design ... they are caused by the ideal behavior around the origin - by approximating f(x)=|x| to a very high degree around x = 0 ... in frequency domain sense this is responsible for producing a higher number of Fourier terms than a high-threshold FW recitifier (0.2v Ge, 0.7v Si) a la Green Ringer - it therefore puts out a greater number of higher even harmonics aside from its dominant second ...

reason why I posted is because it gives the opportunity to hear what f(x)=|x| really sounds like ... in theory this is probably the dirtiest possible octaver circuit of the FW type - short of adding clipping or other wave-shapping ... as supremum it can be used in reverse to provide a basis for defining the concept of a clean-er analogue octaver ...

~jc

[brett]

Hi.
RE: fizz extraction
Try 2 1-pole low-pass filters (ie 2 resistors in the signal path with a cap to ground following each).  Really simple stuff, but can make all the difference.  A cutoff frequency of 5 to 8 kHZ is usually good.  Frequencies around 10 to 15kHz really bug me.  If you're younger than me, you might notice some loss in tone.
PS If you can hear the 14kHZ whistle from TV sets, you've got young ears or good hearing.  I haven't heard it since I was 30. 

[gez]

Would be interesting to hear what it's like with a diode in parallel with the pot (cathode to - input, anode to amp output) plus a resistor and cap in series from the - input of second amp to ground.  This would clip the whole shebang...

[Mark Hammer]

If you can hear the 14kHZ whistle from TV sets, you've got young ears or good hearing.  I haven't heard it since I was 30. 

1) It's 15.75khz
2) Most youth can't hear it, and have absolutely no idea what I'm talking about when I mention it.  In fairness, its a lot less noticeable in today's picture-tube based products (and obviously nonexistent in LCD screens) than it was 20 years ago, but in honesty the SPL's in evidence amongst most ear-bud toting youngsters is likely to be responsible for significant hearing loss, and perhaps even a cause of the need for such SPL's amongst such youth.

[MartyMart]

Well i'm glad to say that after 20 + years of "gigging" etc I CAN hear it !!
There's a couple of nice "dog ear splitting" tones in the studio, though not
coming from the flat screen monitors 

MM.