emm harmony synthesizer analog project

[pinkjimiphoton]

squaring up the input, preamping it and even filtering all the harmonics out doesn't help. its just a shit circuit.

or maybe i just say that cuz i couldn't get it to work worth a fart in a blender. i recycled its box already, i may have actually circle filed the pcb, cuz it sucked so bad

[Mark Hammer]

I have an etched and populated board sitting in a ziplock bag, awaiting debugging.  Never got mine working either, though I have to say my "debuggng" efforts amounted to powering it up, plugging it in  and not hearing anything.

[Mr. Lime]

I will try Mark's suggestion anyway in the next view days on my HG pedal. At the moment Jimi's right, it's just not really usable as it is although I like the idea so much.

The path for best tracking usually leads to the adaptive peak detectors in the EHX synth and Boss octave pedals.
Or you slink off guiltily into a dark corner and play with a POG!

Haha, no a POG will never come to my mind. Even not in the darkest hour.

Well the Micro Synth's envelope detector really seems advanced with different "squelch" trimmers. Even Harry Bissll's high praised circuit seems a bit over the top for DIY as well.
I took a closer look at MFOS's Sub Commaner and guitar to gate circuit which is somewhat similar to Boss. The Sub Commander seem to trigger or gate an AR circuit.
http://musicfromouterspace.com/analogsynth_new/GUITARSYNTHAUG2007/GUITARSYNTHAUG2007.html
http://musicfromouterspace.com/analogsynth_new/HOT_TIPS/images/guitartogateschematic.gif

With such a gate and trigger circuit a Druid's ENVGEN 8 VCADSR could be controlled that shapes a simple envelope that could be tracked more easily. Actually that's what I think it would do.. Please correct me if I'm wrong.

https://electricdruid.net/product/envgen8/

If I would build such a synthesizer from scratch but incorporate the PLL and the extension circuit I would start with this:



Anything wrong here?
It basically mimics the real guitar envelope but should be more synthesizer friendly? Still beside of the "guitar to gate" circuit, not many parts are needed..

[anotherjim]

One of the pitfalls in the front end is to make it too "kind" to the guitar input.
We don't want all of the original harmonics
We don't need to care about maintaining signal polarity
Circuit noise is the least of its problems - so there's no need to discount cheap op-amps or CMOS inverters on that basis.

A non-inverting op-amp always passes x1 clean signal no matter what you put in the feedback loop. That's great in a TubeScreamer but the harmonics all want cutting with a low-pass filter. So an inverting amp works best. You can use a non-inverting clean buffer at the immediate input if you want to make sure the following conditioning circuit always has consistent operating conditions - but also to make a full signal available to an envelope detector. It's a mistake to low-pass before this detector as it usually means the guitars higher pitches are too quiet and you get a gated or intermittent note detection.

[Ben N]

You can use a non-inverting clean buffer at the immediate input if you want to make sure the following conditioning circuit always has consistent operating conditions - but also to make a full signal available to an envelope detector. It's a mistake to low-pass before this detector as it usually means the guitars higher pitches are too quiet and you get a gated or intermittent note detection.

Makes sense--an envelope detector cares about dynamics, not pitch. But at the same time, I guess anything that is well outside the instrument's range--say under 70 hz and over 8-10khz--can still be profitably filtered, no?

[anotherjim]

Guitar generally finishes over 5kHz and fundamentals over 1.5Khz. You might accept a lower maximum for pitch tracking if the synth can produce octave up, but you'll probably still want to get the full signal for the dynamics.

[Mr. Lime]

Having received my EH Micro Synth and compared it to Freppo's Into the Unknown here are my observations:

The EH tracks beautifully and offers great clean wah sounds with the VCF. The ItU lacks tracking and the gate sounds interrupted after a view seconds but I love the additional synthy octaves the PLL offers and the modulation is awesome as well.

One of the pitfalls in the front end is to make it too "kind" to the guitar input.
We don't want all of the original harmonics
We don't need to care about maintaining signal polarity
Circuit noise is the least of its problems - so there's no need to discount cheap op-amps or CMOS inverters on that basis.

A non-inverting op-amp always passes x1 clean signal no matter what you put in the feedback loop. That's great in a TubeScreamer but the harmonics all want cutting with a low-pass filter. So an inverting amp works best. You can use a non-inverting clean buffer at the immediate input if you want to make sure the following conditioning circuit always has consistent operating conditions - but also to make a full signal available to an envelope detector. It's a mistake to low-pass before this detector as it usually means the guitars higher pitches are too quiet and you get a gated or intermittent note detection.

Not sure if I understood all but here's what I'm thinking about to do:

Merlin B's U-Boat Sub Octaver adresses those tracking problems I think: http://www.valvewizard.co.uk/uboat.html
According to Merlin the tracking is improved by the zero crossing detector. Altough there's a low-pass in front of the detector and comparator he claims that for the best tracking the audio needs to be as well filtered as possible, to remove harmonics that cause the comparator to switch at the wrong times.
How about the zero crossing detector's noise catching? Does it keep quite when no notes are played?

I still like the idea of an AD circuit. Druid's One Shot Event Generator can be activated by a simple trigger pulse.
The trigger thrashold has to be set that only really strong strokes of the guitar strings are triggering the AD/LFO.

[Mark Hammer]

I brought my little One-Shot Generator box to a guitar show in Montreal two weekends back, along with a cable for plugging into expression pedal jacks.  Folks from Chase Bliss, Empress, and Red Panda played with it for a bit and were tickled by some of the possibilities for short-lived bursts of modulation-on-demand.  Mind you, that little box is actuated by a momentary footswitch, although I included a mini-jack for externally generated triggers, using the datasheet circuit that Tom W provided.

[anotherjim]

Zero crossing detectors can be hypersensitive. Some seem to rely on luck (or more likely offset errors) to keep the input DC levels just enough apart. You can probably do a little better by giving one input a deliberately different DC bias which in Merlins circuit might be to change Vb going into the Inverter op-amp. At any rate, you'd want the offset to be reasonably consitent despite supply voltage changes and you only need very little offset in the order of millivolts. What you get anyway from opamp offset errors is probably good enough 99.9% of the time.

[Mr. Lime]

I brought my little One-Shot Generator box to a guitar show in Montreal two weekends back, along with a cable for plugging into expression pedal jacks.  Folks from Chase Bliss, Empress, and Red Panda played with it for a bit and were tickled by some of the possibilities for short-lived bursts of modulation-on-demand.  Mind you, that little box is actuated by a momentary footswitch, although I included a mini-jack for externally generated triggers, using the datasheet circuit that Tom W provided.

I can imagine that it's a funny circuit. Am I right that with the Repeat control at max you basically get a LFO?

Zero crossing detectors can be hypersensitive. Some seem to rely on luck (or more likely offset errors) to keep the input DC levels just enough apart. You can probably do a little better by giving one input a deliberately different DC bias which in Merlins circuit might be to change Vb going into the Inverter op-amp. At any rate, you'd want the offset to be reasonably consitent despite supply voltage changes and you only need very little offset in the order of millivolts. What you get anyway from opamp offset errors is probably good enough 99.9% of the time.

So it might work as it is?
I should add an extra path from the envelope detector to a VCA to keep things quite on the back. Shouldn't be too difficult with the given circuit.
I have to draw a schematic of the whole circuit in Eagle.

[ElectricDruid]

I can imagine that it's a funny circuit.

It is!! Anything that makes your filters go BOING!BOING!Boing!Boing!Boing!Boing! is funny!

Am I right that with the Repeat control at max you basically get a LFO?

No, not quite. Maximum repeats is 36, and they decay away linearly. Also the "Delay" control puts a gap between one event and the next, unlike an LFO where one wave cycle is always right next to the last one! The other big difference is all the output is unipolar, so it modulates *up* from a baseline, not *up and down* around a baseline like an LFO.

Of course, there are similarities too. With the Delay at zero, you can get some stuff which is pretty much like an enveloped LFO output.

It's a bit of a half-way house, neither entirely envelope generator nor LFO, and all the better for it.

Tom

[Mr. Lime]

Well I did some drawing.
The One Shot Event Generator part isn't there yet because I think about a daughterboard.
Anyway here's the circuit of "my" improved emm harmony synth.
Not sure about the VCA setup and the mixing with the clean path.
Is there anything terribly wrong on the first sight?

There are 5 controls for mixing the octaves and the unison squarewave together plus a trigger control and a clean blend.

[Mr. Lime]

mark, thanks bro, but i got like, sooooooooo many things ahead of this

that said... i assume you've seen this?

http://www.muzines.co.uk/articles/circuit-maker/4310



Harmony Generator Extension


Paul Williams

The Harmony Generator published in E&MM October 1981, although very useful, is only capable of generating square waves. Also, only one octave output is available at a time. The circuit shown in Figure 1 not only provides the Harmony Generator with triangular and sine waveforms on a selected octave output, but also provides facilities for mixing these and all seven octave outputs simultaneously.

The additional waveforms are produced by a voltage controlled function generator, IC 11, which is used in a phase locked loop (PLL) in conjunction with IC10. The PLL IC10 is used only for its phase comparison and error voltage generation functions, its internal VCO not being used. Instead, IC10 controls the frequency of oscillation of IC11 by a control voltage applied to pin 8 via an inverting and level shifting stage, TR2. The CV is adjusted such that the loop frequency is the same as the input frequency presented to pin 14 of IC10. IC10 and IC11 thus constitute a square wave to sine wave (and triangular wave) converter.

Since waveforms other than square waves are involved in the mixing, the envelope shaping must be linear rather than by means of the very elegant chopper system used in the original Harmony Generator design. An operational transconductance amplifier, IC12, is used both as a virtual earth mixer and a current controlled amplifier to perform the envelope shaping function. The mix is set up on RV4-12 and summed to the virtual earth node at IC12 pin 3 by R31-39. The rectified input voltage from IC2 in the Harmony Generator is converted into a control current for IC12 by TR3 operating in the common base mode. The final output mix between direct and harmony signals is affected by RV13 which is in a rather odd configuration best suited to the current output of IC12.

This circuit is not suitable for operation on the original +9V battery derived supply, so a +12V mains operated supply is shown. The duty cycle of the square wave, which should be monitored at IC11 pin 9 is adjusted by RV3. The 50% duty cycle setting can be clearly detected by ear since its harmonic structure is quite markedly different from any other duty cycle setting. After the duty cycle has been trimmed, the sine wave output will be of reasonably low distortion. However, if very low distortion is required then R29 can be replaced by a 100k preset which is adjusted for minimum sine wave THD.

Although it would add considerable complexity, there is no reason why the circuitry of IC10 and IC11 should not be duplicated for all octave outputs, allowing sine and triangular waveforms to be produced on all seven octave outputs simultaneously.

I just can't get my head around how IC12 (VCA and summing amp) does work. It's a rather elegant solution having the diffrent outputs summed up and an envelope controlled amplifier within just one IC.

As a non-inverting summing amp isn't R4 needed?

[anotherjim]

The summing input isn't a virtual ground, so R1 and R2 form a voltage divider when the inputs are different. If R3=R4, then the basic gain is x2 which I suppose should compensate for the voltage divider, assuming R1=R2.

[Mr. Lime]

Thanks for the reply.

Using a LM13700 and a single +9V supply the conversion would look like this and still sums up the different inputs coming from the binary counter, right?
An additional summing amp wouldn't be necessary and the darlington buffers the output so the clean buffered signal can be mixed at the output.

[anotherjim]

Ah, sorry, I answered for the op-amp circuit. The OTA I think has current differencing inputs. R4 and R3 will be needed to set the operating reference.

[Mr. Lime]

Here's my updated VCA /summing amp with a clean blend at the output.

I think the clean path might have some volume loss. More important, did I draw the VCA part and mixing part right?
It's basically the output section of the extension circuit.