Comments for "Discrete Sawtooth Oscillators"

27th November 2013 00:34

SV3ORA wrote...


How could I make it reverse sawtooth, i.e the fast rising edge to uccur first, then the slow falling edge to occur?

16th August 2010 15:22

Alan Yates wrote...


Yes. An inverting follower is probably the easiest method, but I see no reason why you couldn't build the circuit "upside down", flipping all the device polarities... Honestly I haven't thought too much about it.

If you want a generic, easily tweaked circuit I'd suggest using op-amps instead. The discrete circuits here are mainly a study in minimalism.



30th July 2010 20:05

bk wrote ...

Dear Alan,

Is it possible to invert the sawtooth? I mean a sawtooth with a sharp rise and slow discharge.

Thanks anyway

15th June 2010 13:54

Andres Rada wrote ...

How to recalculate this circuit if I used a 5 VDC?

Great job.

23rd December 2009 09:48

EricG wrote ...

Thanks for being useful in explaining and posting practical circuits. I appreciate what you are doing.

18th December 2008 17:22

Alan Yates wrote...


I'll have a bit more of a look at the bootstrap over the weekend. The DC coupled version seems to work OK in software, but I haven't tried to build it yet.



18th December 2008 09:59

Russell wrote ...

Top of R1 could be a diode and bottom of R1=100k, with the bootstrap cap feeding the junction of them.

Of course, other resistors may need adjusting for proper operation.

17th December 2008 23:37

Russell wrote ...

What i meant to say was:

"Another way (my preferred) to obtain a linear ramp is to use a bootstrap capacitor. In fig.1, voltage on C1 drives the base of a BJT voltage follower. The low impedance output drives a large coupling capacitor connected to the center of R1."

17th December 2008 22:33

Alan Yates wrote...


OK, played with the bootstrapping idea. Works fine as long as the follower doesn't saturate. That can be ensured by sacrificing a bit of output amplitude by setting the threshold voltage lower. I'll attack it analytically when I get a chance.

It is annoying that the resistor needs to be split pretty close to evenly - making tracking for widely variable rates kinda annoying to implement. In that respect a simple current source is easier to control by just varying its base voltage with a pot.

I like the trick though, thanks for that. I am just learning all the ins and outs of discrete design. I think it teaches you a lot more about electronics than using black-box ICs.



17th December 2008 21:38

Alan Yates wrote...


Yep, understand the drawbacks of relying on diode/BJT drop matching and their temperature dependence. For a repeatable circuit I'd use more diodes and a resistor in the emitter. For my own uses the temperature stability is more of a problem, I can always trim a resistor value but can't ensure thermal bonding.

I hadn't considered the topology you mention - bootstrapping the charging resistor so it has a constant voltage across it. I'll have to play with that in spice. Sounds quite interesting.



17th December 2008 17:55

Russell wrote ...

Another way (my preferred) to obtain a linear ramp is to use a bootstrap capacitor. In fig.1, voltage on C1 drives the base of a BJT voltage follower. The low impedance output drives the center of R1.

17th December 2008 17:46

Russell wrote ...

Hi Allan,

I find your circuits quite creative and original.

Beware that Vbe(on) of 2N3904/3906 and many other transistors have a variation of up to +/-100mV on data sheets (gives an Ic ratio of 2200:1), and 1N4148 has +/-50mV on a data sheet.

Paralleling diode and transistor junctions is thus not amenable to consistent production performance, unless the components are matched beforehand, or resistors placed in series.

Also, if one junction becomes warmer than the other in a current mirror, the Vbe match will become dramatically mismatched.

Simple problems like this is often seen in magazines and even manufacturer application notes.

Leave a comment on this article.