I am generating 10 MHz PWM signal from esp32. Its peak voltage reduces to 2.5 volt at 10 MHz. I need to amplify this signal to 12 volt. Duty cycle is 15 percent.

 

Hi M37,
Welcome to AAC.
What Load is the 10MHz square wave driving into.?
E

 

Hi M37,
Welcome to AAC.
What Load is the 10MHz square wave driving into.?
E

its a 10 pico farad capacitive load

 

One option will be a high frequency op-amp,another could be a fast comparator, except that most of them seem to be open collector devices, so the leading edge would be a bit distorted. I did not find a fast enough digital device with a 12 volt rating.
What I did find that was fast enough, but will require A split power supply offset from the common is an LM733/ MC1733 Video amp IC That is good over 100MHz, but the max output is +-8 volts. And most logic families that fast only go to 5 or 6 volts supply maximum. So no 12 volt 10 MHz square wave logic devices that I see. But check the Analog Devices website.

 

its a 10 pico farad capacitive load

Presumably a piezo disc. If so, why would you drive it with PWM? A sine wave would be far better.

 

Indeed, what sort of application?? Driving a piezo disc with PWM makes no sense at all. So it might be an LED or a laser diode.
If the load could use a non-zero based signal, then a pair of LM733 amplifiers could easily deliver a 12 volt peak to peak drive. BUT it will need a split voltage power supply.

 

LEDs are not a 10pF capacitive load.

 

Would it be too much trouble for the TS to explain the problem rather than a hypothetical solution?

 

Would it be too much trouble for the TS to explain the problem rather than a hypothetical solution?

That seems to be the case more often than not.

 

That seems to be the case more often than not.

Indeed, I am really curious about why there is a need to do RF engineering with non-RF type signals (PWM for instance). Can anybody explain what that might be all about? I'm completely at sea on this one. The period of a 10 MHz square wave is 100 nanoseconds. A meaningful PWM signal at that frequency would require edge times, (rising and falling) in the neighborhood of 750 picoseconds. Using a standard rule of thumb that would require devices with a bandwidth of:

\( BW\;=\;0.35/t_r\;=\;0.35/(750\times10^{-12})\;\approx\;467\text{ MHz.} \)

If one would be willing to sacrifice 1.5% of the period for rise and fall times. That kind of design is far outside the capability of most of those who do not specialize in RF engineering. Laying out a PC board to work with such frequencies is no easy task either.

 

Think about a Gate Driver IC. The 10mhz is fast for them but...
Here is an example of a fast Gate Driver built to drive capacitors. Rise and Fall of 4nS.
The schematic is for driving gates, so remove Q1 etc. and attach your capacitor.
Supply voltage 4.5 to 20V.

 

4ns rise and fall times are impressive, but then 8% of your PWM signal is consumed. Are you sure this will have minimal effect on the result?

 

Designing even ten megahertz logic PCBs takes a bit of RF engineering, if it is important for the boards to perform very well.

 

Designing even ten megahertz logic PCBs takes a bit of RF engineering, if it is important for the boards to perform very well.

You'll get no argument from me on that score.

 

4ns rise and fall times are impressive, but then 8% of your PWM signal is consumed. Are you sure this will have minimal effect on the result?

Here are two waveforms. One has 0nS rise and fall time. The green trace has a slow edge. The average energy of each is the same. (within reason)

 

The PWM might possibly be a means of hiding a data signal under a radio transmission that is modulated in a more common manner. That is one possible application, and that could use a fairly minor shift in duty cycle that would be a real challenge to even detect. Shifting from 40% to 60% would be a serious hiding scheme if it were done on an FM broadcast signal, or in the audio portion of an NTSC modulated TV signal broadcast.

 

The PWM might possibly be a means of hiding a data signal under a radio transmission

Or, it might just be that he has no idea about what he is doing.

 

Here are two waveforms. One has 0nS rise and fall time. The green trace has a slow edge. The average energy of each is the same. (within reason)
View attachment 313766

If the actual value of the duty cycle has an effect in the overall system, as it would in a DC-DC converter, how would you describe the duty cycle of the green trace with the sloped edges. I'm not sure average energy is what you care about.

 

how would you describe the duty cycle of the green trace with the sloped edges.

The green trace has delay, without a change in duty cycle. But that depends on where you slice the trace and 100 other things.

We do not know what the requirements are. I am moving on to something more productive.

 

The green trace has delay, without a change in duty cycle. But that depends on where you slice the trace and 100 other things.

We do not know what the requirements are. I am moving on to something more productive.

OK - me too.