TZFM SAW VCO

[Luka]

It would be very cool to have 2 or 4 of these things on a single pcb

[Dego]

[frijitz]

Another update.

I've been mostly been working on the waveshape/waveshapers. The up ramp and down ramp sections of the core were not quite the same. It turned out that the problem was an unusually large amount of charge injection into the integrating cap during switching. The reasons for this were twofold: (1) To try to get the OTA current modulator stable and accurate, I set it up to use lower than usual current levels. This meant going to a smaller than usual integrating cap. (2) Resetting the positive and negative ramp sections requires two FETs in tandem and a large voltage swing on their gates. Since charge injection is proportional to the gate voltage swing, this lead to a larger than usual amount of injected charge.

The result of these two factors was that the zero points of the ramps were offset upward by ~0.4 V. This meant that the magnitudes of the up and down ramps differed by ~ 15%. I stewed for quite a while thinking about all the possible ways this problem could be fixed and finally decided to try the method of injecting a countercharge. This required just a transistor to invert the switching pulse and a small cap to inject the inverted pulse into the integrating cap. Surprisingly, it was real easy to get this to work. Of course, there are still small switching spikes, but the big problem is taken care of.

All this also made it easier to shift the positive and negative ramps into alignment (to produce a final waveshape exhibiting proper phase reversal).

Next I built a SAW-TRI shaper. This is just a full-wave-rectifier plus a level shifter / amplifier (although they don't usually tell you this). I found a FWR circuit that is a little simpler than what is usually employed. Its limitation is that its input impedance is different for positive and negative signals, but that doesn't matter for this application. (Jung, IC Op-Amp Cookbook, Fig. 5-13) It might be possible to save an opamp here with an even simpler circuit, which I may eventually try. This shaper produces nasty spikes at the switching point (as usual). I tried to figure out a simple way to use a countercharge injection pulse to eliminate the spikes but wasn't successful, because the input discontinuities are positive and negative, whereas the rectifier's spikes are all negative. So I ended up choosing to use fast opamps to keep the spikes narrow and some filtering caps to attenuate them. The resulting triangle is OK, with just small transient spikes.

A short noodley demo using the TRI output is attached. This was actually made before the charge-injection corrector was built, so the triangles have fair-sized steps on them. But it definitely sounds useful.

Since I have half an OTA let over, I'm going to try adding a SIN shaper following the TRI. It will be interesting to see if the waveforms have a small number of harmonics as you could get with a sinusoidal core.

So this project is still very much in the whiteboard stage. Probably a couple more weeks of testing before making a protoboard.



Ian

---

trichaosdemo.mp3
Description:		Download (listen)
Filename:	trichaosdemo.mp3
Filesize:	649.84 KB
Downloaded:	1954 Time(s)

---

[phdinfunk]

Wow, I appreciate the details in your message!

Funny thing, I always imagined that method for saw to tri conversion in my mind (I'm not a designer, but I'm learning. I've got lots of spare time so I'm studying harder right now). For a guy that's reading Randy Stone's Tab Guide to Electronics while building projects like a madman, your posts are totally cool!

--Jonathan

PS: I like the sounds this thing's been putting out so far, mate!

[Tim Servo]

Hey Ian,

Neat stuff! I'm glad you're doing this. One thought on the Saw-Tri waveshper: how about using a "crappy" low bandwidth op-amp like a 741 or 1458 to reduce or perhaps even eliminate the switching spikes? It's a common technique, and might just do the trick. Other than that, rock on Ian.


Tim (just used up my ONE good thought per day) Servo

[frijitz]

[frijitz]

Yesterday I put in the sine shaper and got it adjusted. The circuit I used follows the OTA-opamp with feedback approach. I showed previously that this can have very low distortion. The tradeoff is that adjustments for drive amplitude and symmetry must be included. But it seems worth it.

I recorded some deep FM tones using sine wave modulation and put them through a spectral analysis program. All the main peaks were where they belonged, and I didn't see any problems with unwanted extra features.

Here's an mp3 with some percussive sounds.

Next I have to figure out what to do about sync. It needs to be quite gentle so to not distort the sine waves and at the same time reliable enough to keep the phases locked.



Ian

---

sinedemo01.mp3
Description:		Download (listen)
Filename:	sinedemo01.mp3
Filesize:	396.37 KB
Downloaded:	2149 Time(s)

---

[bridechamber]

That's fantastic!

I thought I'd need ten, but let's see... If I make a six-operator FM with four voices... I guess I'll need 24!


Scott

[LektroiD]

I'd be interested in some of these too, not sure how many as yet

[phdinfunk]

Ian, yes, Please. As everyone else has said, please put more than one OSC per board. It's only natural!

--Jonathan

[frijitz]

[bridechamber]

I forgot about phase.
Besides sync, there's really no way to address that, is there?

For stacking multiple boards, you could add "power-through" pads for a power bus for them all. Then it's just a matter of extra standoffs, and the individual user could make one VCO or twenty.

I don't know a ton about FM and such, so let me know if I'm off on this:
For a typical FM patch, you'd still want the same controls as the "final" VCO, right? I.e., fine and coarse tune, fm attenuator(s).
If so (and I know it's early to be thinking about panel stuff), maybe a 1U tall-format panel could hold one VCO, with board-mounted pots. That would be simple, fast and inexpensive.
That would also give you maximum patching flexibility for different multi-op set-ups.
I'm sure a similar solution is there for Frac/ Euro.

What would it be like to insert a phaser between two FM operators? I can't wait to find out!

Cheers,
Scott

[fluxmonkey]

[frijitz]

[e-grad]

[frijitz]

This project marches on. Most of the past week was spent on developing a sync circuit. It turns out that two circuits were needed, one for the up ramp and another for the down. I also looked carefully at whether a fixed type of sync could be used.

I expected that a hard sync would always work reliably, but I found that some fm waveforms work better with weak sync pulses. I couldn't find a soft-sync scheme that always worked well, either. So I ended up using an adjustable soft-sync setup that can go from no sync to a strong sync that is quite similar to a standard hard sync.

I've been able to get locking at all kind of oddball frequency ratios, such as 5:7 and so on. Most of these need a fairly weak sync and will not track over a wide frequency range. But of course, that depends on how well the two oscillators track each other. This is one area where beating hard on the tracking problem pays off.

Here are a couple of demos. The first illustrates master/slave tracking over three octaves with modulation index in the 8-10 range. (This means that the change in frequency is up to ten times the modulation frequency.) Followed by a little doodle.

The second demo shows some results obtained by abusing the sync feature to produce dynamically changing waveforms. Lots of horrid sounds here for you noise freaks.

I'm wondering how to do the Initial Frequency (bias) control. Right now it is just a pot. Would a multiposition switch be preferable? The continuous variability doesn't really seem necessary, although it works fine. Fixed levels would make it easier to to explore different modulation depths and to get back to previously dialed-in sounds, but It would be an additional expense.

Opinions?



Ian

---

syncDemos01.mp3
Description:		Download (listen)
Filename:	syncDemos01.mp3
Filesize:	344.53 KB
Downloaded:	1992 Time(s)

---

---

anharsyncDemo.mp3
Description:		Download (listen)
Filename:	anharsyncDemo.mp3
Filesize:	481.27 KB
Downloaded:	1756 Time(s)

---

[machine.cuisine]

[frijitz]

[machine.cuisine]

[phdinfunk]

Ian,

I've used the Yamaha FS1r extensively for FM synthesis, so I'm pretty hip with how FM works, BUT...

...I'm not exactly sure what you mean by "bias"... is that the initial frequency of the osc, as in "frequency" or "tune" or is it the modulation amount? The way you describe it it sounds like a modulation depth control...

Sorry for the confusion -- email's not necessarily the best form of human communication

--Jonathan

[widdly]

I'd like to see some PWM control on there. Then it would really cover a lot of bases...FM, sync and PWM from one VCO.

[loss1234]

ian

i am assuming you would do this with a rotary switch..which isnt really too expensive in my experience

but maybe you meant something else. i like the idea of a stepped switch. there are some moments i ge a bit tired of always having to work really hard to get back to a certain sound due to all of my knobs covering such giant ranges

thnx for this awesome project
cant wait to add one to my synth (or three)

[frijitz]

[aerogramma]

[frijitz]