MXR phase 100 - LFO hypertriangular or?

[mikeb]

Taking a look at the MXR phase 100 schematic:
http://www.generalguitargadgets.com/v2/diagrams/mxr_p100_sc.gif
... does the manipulation of the emitter-receiver resistance of Q1 result in a logarithmic LED voltage response? Yes, I could (and will!) build it and find out, just trying to understand beforehand. It seems the following needs to be considered when determining whether the final allpass filter response is logarithmic/hypertriangular in a LED/LDR-type phaser:
- LFO shape -> LED driving voltage
- LED voltage -> LED brightness
- LED brightness -> LDR resistance

MikeB

[Boofhead]

For get an exponentiated collector current the voltage changes across the emitter resistor must be small in comparison to the base-emitter voltage changes of the transistor.  This means a quite a small emitter resistor - and that affects the biasing circuits etc.

The emitter resistor in that circuit essential causes the collector current changes to be proportional to base voltage changes.   The presence of the emitter resistor also limits the maximum current through the LEDs.  If you make the emitter resistor too low you might damage the LEDs, because the current limiting is lost - especially when testing or adjusting the unit.  It would be safer to put say a 47ohm (maybe upto 100ohm) resistor in series with the LEDs to prevent this - if properly chosen the resistor shouldn't affect the VCO operation.

[mikeb]

Thanks .... so the answer is no then? Hmmm. What I'm leading towards is getting a hypertriangular response for a 9V based LFO. The 'standard' way of using the FET triangle->sine shaper with one diode removed needs a +/-5 input waveform to operate. Next thing is to look at the NNP/PNP transistor approach some synth circuits seem to use to do the exponential conversion I guess...

Mike

[R.G.]

There's a much simpler way to get hypertriangular. The current through a silicon junction is an exponential function of the voltage across it.  All you have to do is arrange to drive a silicon junction with a triangular voltage, then convert the current through it to a voltage. One way to do that is to use a silicon diode as the "input resistor" into an opamp, with a feedback resistor. Ground the + input. The - input is then held at virtual ground. If you feed the free end of the diode a voltage that's less than what drives the diode to huge currents, then the opamp forces the current through the resistor to be the same as the current through the diode. It does this by making a hypertriangular waveform on its output.

All of the "synth" npn/pnp circuits are really variations of this approach, but tinkered up for better accuracy.

It's even simpler if you make the active part of whatever you're trying to control be OTA based, then drive the Iabc junction with a triangular voltage. The triangle voltage forces the Iabc to be hypertriangular. (as a special prize, if you exceed 1ma, the OTA dies, but that's another story.)

[Boofhead]

I wasn't saying you can't do it,  you just need to address those issues I mentioned.

Take a look at tbe Boss flanger BF-2.  That's exponential and when it boils down to it it's a very similar circuit.

The JFET shaper can be used as well.

You can pretty much do anything you want.  There's always a few gotcha's when you come-up with "shaped" VCO circuits.  You have to get the the input swing to the shaper right, then you have to make sure the output swing is right to drive the following ckt.  If there's any DC levels involved (which there usually are)  you have to consider the DC levels in and out of the shaper.  It's often fiddly to get all these right, especially when you start putting device tolerances into the equation, putting pots in helps but it often turns out to be hard to set-up properly/reliably.

For ideas look at: muton Biphaser,  Electroharmonix Small Stone, Electroharmonix Bad Stone,  and the (Black) Ross Phasers.

[mikeb]

Thanks very much RG and Boof .... I should have mentioned I want to steer clear of OTA-based solutions (price and availability of parts). Forgive me for not completely understanding the first explanation - this is a new area for me, and although I've valguely familiar with some DIY synth modules I've built I've never really understood some of the core theory behing them. So time to get down and dirty with some experimental circuits....  :twisted:

Mike

[gez]

Mike, Isn't a hypertriangle a negative going rectified sine?  I've used simple diode shaping techniques to get a positive going rectified sine (sounds really nice in autowahs), but the whole network can be flipped to do what you're asking.  Although slightly angular under the scope (only a bit) you don't hear it in practice and it uses a bog-standard two op-amp triangle generator.

It needs to be run from a 5V reg though to keep things consistent, and I've used 7621 op-amps (they can be run at 3V which keeps current spikes to a minimum so you get tick-free operation).  However, you could use CMOS inverters run at this voltage (would probably be better actually as you could use a spare inverter as a virtual ground and use it to get more linear control for the rate pot).

Anyway, if you're interested PM me and I'll sort you out a schematic.  If you don't like it, fair enough, but it's a cheap and cheerful way of doing it and I liked the results.

[Paul Perry (Frostwave)]

Mike, Isn't a hypertriangle a negative going rectified sine?  

Well, that is one way to make it, but not the only way. Strictly speaking, a 'hypertriangle' is what you get if you "uncompress" a triangle, so the top is stretched out more as you go up. Futzing around with non-linear elements in op amp ckts is the way to go, as RG suggests. A good old book (might still be in some engineering libraries) for non-linear waveshaping is Wong & Ott, "Function Circuits Design & Applications". And of course anything by Graninio Korn.

[gez]

Well, that is one way to make it, but not the only way

I like the sound of the idea RG mentioned too, but I did say mine was the 'cheap and cheerful' approach!  

[Paul Perry (Frostwave)]

I did say mine was the 'cheap and cheerful' approach!  

Well...it's only cheap if you already have a sine wave

[puretube]

Well...it's only cheap if you already have a sine wave

ooops! thank you, Paul,
for this forehead-slapper!!!
   

[gez]

No Paul, I'm really cheap, this one shapes a triangle wave!!  

Actually, I've recently been messing with the simple sine wave LFO used in the Silicon Chip tremolo article someone recently posted (thanks!).  It can be simplified even further by using a comparitor to produce the square and getting rid of the LEDs (waste of current anyway).  The wave form is really impressive and is amplitude stable (no auto gain control needed).

It's the easiest way I've seen to produce a (decent) sine (only two op-amps and a T network).  Only drawback is the rate pot isn't very linear and everything is bunched up at one end.  A non-linear active used as the variable resistance would sort this out though (but I've been too lazy to get round to doing this).