simple ramp generator

[electronics man]

i am creating a circuit that requires a ramp generator, it needs to be fairly high speed (4Mhz to about 20Mhz), it doesn't require a 555 timer as there are squire waves available in the circuit. can anyone help?

it does seem like a really dumbs question i know because i have been doing electronics for years

[dannyf]

CCS charging up a capacitor.

20Mhz however is kind of high.

[c4757p]

Yes, even at 20 MHz a current source charging a capacitor is likely going to be your best bet. However, everything needs to move fast, and you're going to have to put a lot of consideration into things like switching (when you switch on/off the current source, you'll couple in charge through the capacitance of the switches...)

There were some old function generators that could take their triangle wave/ramp this fast - not many, but a few. Wavetek made a couple, IIRC. Look for service manuals for ideas.

[dannyf]

Think of a voltage source + resistor as a (less-than-ideal) ccs. So you just need a bjt/mosfet over the capacitor to discharge it at the desired frequency. The voltage over the capacitor would be your output.

[tautech]

i am creating a circuit that requires a ramp generator

Any active devices in the circuit already that could possibly be used?
Micro, ADC's etc?

[c4757p]

Think of a voltage source + resistor as a (less-than-ideal) ccs. So you just need a bjt/mosfet over the capacitor to discharge it at the desired frequency. The voltage over the capacitor would be your output.

...or, you could put absolutely no thought into charge injection at all.

Why don't you try it, and get back to us?

I look forward to the "el cheapo ramp generator, ghetto style" thread.

Here's mine. More in there than "just" a BJT over a capacitor - I'll let you figure out what it is. And that's not even a very good ramp.

[w2aew]

..as stated above, 20MHz is a bit fast, but with careful design choices, a constant current source into a cap is a good way to go.   Here's a video I did that uses this technique along with some diode switching.  This particular circuit won't likely reach 20MHz, but maybe it will spawn some thoughts and ideas for you:

[dannyf]

Here is what I meant:

5Mhz ramp, 1k/1n network, discharged via a bjt.

You can vary a few factors to get the desired results.

Adding a jfet will help linearity.

[c4757p]

Is that by any chance a "perfect" NPN model? Try sticking a realistic one that has all its capacitances in there.

[Marco]

RF transistors have sub-pf range capacitances, doubt it matters much ... what the simulator won't simulate is that transistor blowing up, cause I'm pretty sure the discharge current is going to be in the Ampere range

[electronics man]

i like w2aew's circuit, when it comes to speed i will try to get it to go as fast as i reasonably can

[Marco]

You're going to need low capacitance Schottky diodes and a much smaller capacitor though (100 pf range, at which point you will get some feed through through the bridge).

Temperature stability will of course not be great (capacitance and current source change the slope, Vos on the comparator changes the thresholds).

[dannyf]

The r/c idea mentioned earlier is nothing more than an r/c (relaxation) oscillator used on mcus or dedicated oscillators -> that's another way to build it.

You can also approach this digitally: a counter + weighing resistor network gives you a precision ramp generator. It has the advantage of consistent waveform amplitude and good tuning range -> change the input clock and your output will be perfectly synchronized.

[IconicPCB]

This is nothing new... this kind of approach ( current source into a capacitor is used by modern flying probe bare board testers to measure capacitance down in the femto farad region ( lower limit approx 50 fF ) which is the capacitance of the a pad in the order of a couple of square mm.

[electronics man]

The r/c idea mentioned earlier is nothing more than an r/c (relaxation) oscillator used on mcus or dedicated oscillators -> that's another way to build it.

You can also approach this digitally: a counter + weighing resistor network gives you a precision ramp generator. It has the advantage of consistent waveform amplitude and good tuning range -> change the input clock and your output will be perfectly synchronized.

that sounds quit good, there is already a high speed counter and clock in the circuit