sawtooth wave generator

[C Bradley]

Mr. Vex posted this many moons ago. Can't remember if it was here or on Ampage, but here it is:

my favorite sawtooth source is the negistor. take a 2n2222 and clip off the base lead. connect the collector to ground,
put a .1 to 1uF cap across the collector and emitter, and connect a variable resistor (perhaps 10k to 50k) from the
emitter to your positive power supply (9 to 18 volts.) you may wish to add a 470 ohm resistor in series so you don't melt
the transistor down when you set the variable resistor to 0 ohms.

there's an avalanche property, little explored in typical circuits, that causes a wonderful sawtooth to appear at the
emitter of the transistor. use a 10 uF cap to block the d.c. from the emitter and voila, you have a sawtooth source.
with a little tweeking, it can be set up to range the full width of human hearing using only 5 components. 8^)

zachary vex

I've tried this and can't get it to work. I haven't clipped the base lead, but it's not connected to anything so I figured that I didn't need to do that when bread-boarding it. I'm using a 2N2222 and the other components are all in the range Zach specifies except for the pot which is 250k, but even that should work at one end of the range. Any ideas where I might be going wrong?

I'd like to get it working and then possibly use it as an LFO to drive a LED/LDR for my opamp wah.

Chris B

[zachary vex]

9 volts is on the hairy edge, so you may need to try several different transistors.  12 volts is much better.

[C Bradley]

I've also got some 2N3906 transistors. Will the extra gain help? I guess I need to build that 15V power source for my bench now.  :wink:

Chris B

[zachary vex]

first off, you can only find negistor action in certain NPN transistors, and never in PNP.

the only common transistor i've found that works is the 2N2222A.  i've found another transistor from a surplus store that works well, but it's a really odd part number that i can't find anywhere, and the 2N2222A works just about as well.  once you get it working, you can try every single NPN you can get your hands on, and you might find another kind that works, but it will probably be a relatively uncommon part.

remember to put the collector to ground, and the emitter to the + lead of the capacitor (if it's electrolytic).

[C Bradley]

Thanks Zach!

I guess I just need more power. The 9V battery I'm trying to run this with measures 8.4V out of circuit. Also, the transistor is a Radio Shack NPN "switching transistor" (supposedly a 2N2222 type) so that might be a problem too. I'll keep trying...

[R.G.]

It's far more consistent to use a UJT or a PUT.

The UniJunction Transistor and Programmable Unijunction Transistors were designed for this, so they'll all work, and at reasonbly low voltages.

But then, you might clip off some transistor leads and get lucky.

An NE555 does a good job if you accept a 1/3 power supply level height on the sawtooth and use a diode to make discharge FAST.

A schmitt-trigger of any stripe does a good job under the same conditions as a 555, and you can get a bigger sawtooth out of them, sometimes the full power supply almost. That can be two transistors, opamp, comparator, CMOS schmitt trigger gate, etc. Under some conditions, you can get this from a single CMOS gate.

Finally, you can use a CMOS counter and a resistor network to count up, then reset to zero. PITA? Sure, but you have absolute control of the waveform and speed and can control it from DC to daylight with the speed of the clock you feed it.

The real question here is - how much control of the waveform do you need?  Synthesizer guys have elaborate sawtooth generators made from several ICs with temperature compensating parts so that they get exponential response and the *same* waveform for any control voltage.

The tradeoff is control. Using the side effect of a single transistor's action means letting it determine what you get. That may be fine, but it usually means doing what I refer to as "kissing frogs" - you know: if you're trying to find a single and accomodating Prince, you have to kiss a lot of frogs. This is a common maxim in the femail world, by the way.

[Vsat]

Accidentally made an "avalanche" sawtooth oscillator many years ago - using the 2N5449 NPN (IIRC it was a Fairchild part). It was used as a reset switch for an op amp integrator used in a linear VCO running on +/-15V - was very surprised to see it continue to work when the NPN base was disconnected from the comparator part of the circuit. Voltage control provided by the input resistor for the op amp integrator. Tried it with several 2N5449, there was quite a bit of variation between units. Made a nice buffered sawtooth - but not something that will work with a 9V battery.
Mike

[Ry]

I've actually been doing a bit of playing around with this circuit and haven't had a problem with any of the 2N2222's that I have (purchased from ALL Electronics about 5 years back).

I have one nagging question about this oscillator, though.  Do I need to buffer the sawtooth output to get optimal performance?  I'm concerned about it being loaded down and I'm not sure what the drive capabilities of a transistor working in this manner are.

Ry

[Vsat]

If you want a nice linear sawtooth the capacitor should be charged by a current source - bipolar transistor or FET.  Some buffering (op amp or emitter follower) will be needed unless the load is very high impedance. Otherwise, use large capacitors and a large charging current to provide a low enough impedance to drive the existing load...  being careful that you don't burn out the avalanche transistor.
Mike

[puretube]

how about the usual 2-opamp Schmitt-trigger/integrator LFO,
with the "diode-trick"...

[Ry]

What's the "diode-trick"?  

I was actually looking at the Schmitt-trigger/integrator LFO yesterday, but wanted the smaller footprint of the negistor(even with an emitter follower after it - thanks, Mike). Plus, I was just interested in playing around with it.  I don't know that I would design the negistor into a production design unless I had a very large bucket of 2N2222's that I knew would work.

Sorry to hijack the thread, Chris.

[puretube]

http://diystompboxes.com/sboxforum/viewtopic.php?t=30082&highlight=diode+trick
http://diystompboxes.com/sboxforum/viewtopic.php?t=21197&highlight=diode+trick

[puretube]

I think they have a diode-trick mod:
http://www.commonsound.org/tremulus.html

[jmusser]

Is this saw tooth generator something you use for diagnostics, or is it something you tune to different frequencies for synth use? Is this something you can trigger with guitar, and have sawtooth fuzz? I know that The Simple Square Wave Shaper does a sort of saw tooth wave form, that tends to have a trombone sort of tone to it. I'm just trying to find out what it's normal application is.

[jmusser]

Just Bumping here. I'm still wondering what the common use of the Saw tooth wave form is?

[puretube]

to chase a cathode-ray across the screen...   :lol:

[gez]

Is this saw tooth generator something you use for diagnostics, or is it something you tune to different frequencies for synth use? Is this something you can trigger with guitar, and have sawtooth fuzz? I know that The Simple Square Wave Shaper does a sort of saw tooth wave form, that tends to have a trombone sort of tone to it. I'm just trying to find out what it's normal application is.

To me, a saw tooth is a waveform that ramps up linearly then plummets to earth only to start its ascent all over again.  Some people use it to describe a triangle waveform though.

If used for a LFO either wave form may be used to sweep a filter.  Triangle waveforms are pretty easy to generate and are used in many effects to sweep phasers and filters etc or to modulate gain, as in a tremolo.

A guitar signal may also be shaped into either waveform, but that wouldn't usually involve an oscillator.

An sawtooth oscillator of fixed pitch might also be used in a synth circuit...

[jmusser]

Gez, Thanks for the response. That would make sense, since the wave form ramps up and then dies at precise entervals to use it for sweeping.