does anyone has exp designing sawtooth generator?

Has the topic topic says, I would like to know if there is aomeone familiar with the design if sawtooth generator.

I havê design the BASIC sawtooth generator working até 500Mhz but it gails in corners.

I would appreciate any help.

What slew rate do you expect? for rising and falling edges?
Avalanche effect devices are usually needed.

What you mean by slew rate? Of what? Ota por sawtooth?

It is a challenge to create a sawtooth at a fast frequency. You need to allow relatively large current pulses, for a fast discharge rate on the capacitor.

A unijunction transistor is a typical choice to make a simple sawtooth waveform. These may be difficult to find.

Or try a configuration of two transistors, referred to as a programmable unijunction transistor.

**broken link removed**

This is not necessarily the optimum design. You may have better luck with an oscillator known as a ring oscillator. It operates from the innate capacitance in mosfet gates.

You can make a simple experiment with the Falstad fourier series applet and visualize the effect of a 2.5 or 5 GHz bandwidth limitation and a 500 MHz sawtooth.

This should clarify a basic problem of a hugh frequency sawtooth generator. Then specify your requirements, or better explain what you want to achieve.

Any BJT transistor circuits based on saturated switching can be definitely excluded.

I have enough experience, to have come to prefer a triangle
wave & comparators (LM555-style, quite often) for generating
a PWM master ramp signal. Benefits here can include a fair
degree of process / temperature insensitivity, and 0% - 100%
realizable duty cycle, and no fast discharge requirement at
all.

Now, 500MHz -is- fairly sporty and I would not expect much
purity out of the circuit. Consider that 2nS full cycle and a
desire to make a <=10% duty cycle position clean means
a 200pS high phase and in turn a desire for 20pS rise / fall.
That's tough territory for even the highest speed logic.
You're probably feeding this to a comparator somewhere,
which is going to be slower (and more variably slow) at low
overdrive than any bang-bang logic with a well driven input.
And you want an analog waveform to be all tight and pretty
across corners?

I think you might want to step back and think about what
you want for the real outcome - I'd assume this sawtooth
has a small part to play in a larger function, and maybe the
function can be had cleaner with some alternative approach.

Thanks for the reply. I forgot to tell that this will be implemented on a IC.