How to add duty cycle on a PWM signal

Question

Here the situation. I have 2 fans controlled by 1 PWM signal. Works perfectly like that.

Now, I want to trick one to be always 40% ahead of the other (and mechanically, have a minimum speed of 40%).

so for instance :

• PWM signal is 20% : Fan 1 goes at 20% of the speed, Fan 2 goes at 60% of the speed
• PWM signal is 40% : Fan 1 goes at 40% of the speed, Fan 2 goes at 80% of the speed
• PWM signal is 60% : Fan 1 goes at 60% of the speed, Fan 2 goes at full speed
• PWM signal is 80% : Fan 1 goes at 80% of the speed, Fan 2 goes at full speed

Any idea of a circuitry that could modify the PWM signal for my fan 2?

I do not need it to be perfectly accurate on the percentage,even if the added value fluctuates between beginning of the scale and end of the scale, that would be OK.

• Simon Marcoux: I have no final solution here, but some random trails that popped into my mind: fan 1 could receive the initial PWM signal and fan 2 needs either to keep a stable output a fixed amount after the falling edge is detected or finding a way to add a properly time delayed signal. My guess would be that the former proposition might work better. Maybe adding a 555 monostable that will trigger on falling edge of your PWM for a fixed duration. Then you add up your initial PWM and your monostable together for the 2nd PWM signal. Quite similar to user's approach, but using different parts!
• Nick Alexeev: What is the frequency of the PWM?
• Simon Marcoux: No space left: if you have all parts available, would you mind doing both approaches? I'm quite curious to see if they could both work in a similar way.
• Memes: @NickAlexeev : 25kHz
• Memes: @SimonMarcoux : will try to do so

Answer 1

I think this circuit should be able to do the task

enter image description here

Vi should be the fan 1 signal and Vo the fan 2 signal. The low going edge of Vo is always that of Vi delayed by a constant time value which depends on R, C and the threshold voltages of the Schmitt trigger not gates (e.g. CD40106).

If my math is not failing the delayed time ("d" in the picture) should be d=RC*ln(Vh/(Vh-Vlth)) where Vh is the high output voltage value and Vlth is the low-to-high voltage threshold.

• Stack Exchange Broke The Law: Note that if fan 1 is totally off, so will be fan 2.
• Simon Marcoux: Could a off state for fan 1 be a ridiculously small PWM value that looks like a dirac? I assume that the PWM is programmable in some kind of way.
• Memes: Thanks, I'll test that and mark as answer if it is good.

Answer 2

Perhaps you need 2 PWM signals:one for Fan 1,the other for Fan 2.

• Simon Marcoux: Unfortunately, this is clearly not a proper solution if you take a look at how his question was constructed.
• Memes: Yes, that somewhat what I am trying to do... create a second signal out of the first one...