Electronics > Beginners

Filtering PWM to smooth DC

<< < (7/12) > >>

Zero999:
 
--- Quote from: Rapsey on September 24, 2018, 02:36:48 pm ---Wow, thank you so much Hero999, you are really going above and beyond!

I've spent a few hours playing around with different RC vaues and studying the results. Getting a good feel for how it all fits together now. It seems that for the specific scenario of feeding PWM through an RC LPF you could say that:

* Capacitor value primarily affects ripple (smaller cap will drop more in voltage during off-portion of PWM cycles)
* Resistor value primarily affects control range in two ways:

* Limits the top end of your PWM control (smaller resistor will drop less voltage and thus get closer to the 12V max)
* Shifts your control curve (smaller resistor will charge the capacitor faster, e.g. 10% pulses could be enough to fully charge the capacitor and if it is big enough to carry through the remaining 90% this would shift your effective control curve to ~0-10% PWM)Therefore in practice it seems best to do a balancing act choosing your resistor (as low as possible for the upper limit but not so low that it destroys your control curve) and then choose an appropriate capacitor based on how much ripple you can tolerate. Of course all of the above depends on how great the load is. Greater loads will require smaller resistors and larger capacitors to maintain similar results.

A few questions about your RC circuit:

* Is that LTSpice?
* How do you arrive at a load resistance of 80R?
Do you simply use Ohm's law and divide the voltage by the rated current? If so, then judging by my volt/amp measurements this remains fairly constant at lower speeds.
Can you measure the ESR of a component like this by simply measuring resistance with a multimeter?
* You mention 830mA current spikes but the graph shows the spike going as high as 14A and eventually settling at 830mA?And the LC circuit:

* Would, say, a B82791H2251N020 be a good inductor here? (47mH, max 250mA / 2.4R, 30% tolerance)
* How would the characteristics differ from an RC circuit? Why is this better?
If these big inductors have resistance values in the same ballpark as the resistors we used for RC, is there still a point in using them?As for the rectifier circuit... I think I have some more studying to do. :)

--- End quote ---
Yes, I simulated it in LTSpice.

80R was calculated using the load's current and voltage ratings and applying Ohm's law. I assumed it was a resistive load, which could be incorrect. If the current doesn't reduce much, when the voltage drops, use a constant current load.

The 14A spikes, during start-up don't matter, as they don't last for long. The peak current can be calculated from the supply voltage and resistor value.

Yes, tyhe ESRT of an inductor can be measured with a meter.

As mentioned above, it makes more sense to use a higher frequency and smaller inductor, but if you can't change the frequency, then a large inductor is unavoidable. The losses won't be that high at the low frequency of 500Hz.

wraper:
As I said before, I see barely any sense going resistive PWM as well. Just use something like LM1117-ADJ and put smoothed PWM on it's ADJ pin. No current spikes, no large cap needed. As a bonus, output voltage will be independent from the load.

David Hess:
The buck regulator structure using a capacitor, inductor, and diode is the best way.  Some old motherboards worked this way to control the speed of non-PWM capable fans.

Essentially it becomes a buck regulator without feedback control so the output voltage becomes the supply voltage * duty cycle.

Zero999:

--- Quote from: wraper on September 24, 2018, 10:16:07 pm ---As I said before, I see barely any sense going resistive PWM as well. Just use something like LM1117-ADJ and put smoothed PWM on it's ADJ pin. No current spikes, no large cap needed. As a bonus, output voltage will be independent from the load.

--- End quote ---
Assuming there's a permanently live conductor, then that would work perfectly. When I designed the previous circuit, I made the assumption that it was just a two wire connection: one 0V and one PWM.

I wouldn't even bother with the regulator, just use an emitter follower. Remove D1 and C1 from the circuit I previously posted and connect 13.8V directly to Q1's collector.

Rapsey:

--- Quote from: Hero999 on September 24, 2018, 10:02:48 pm ---The peak current can be calculated from the supply voltage and resistor value.

--- End quote ---
How would you calculate that?


--- Quote from: Hero999 on September 25, 2018, 07:45:07 am ---Assuming there's a permanently live conductor, then that would work perfectly. When I designed the previous circuit, I made the assumption that it was just a two wire connection: one 0V and one PWM.

--- End quote ---
That's correct, at present there's only those two wires going to the fan. I could always lay an additional 12V line from the PSU but the simpler the solution the better.


Now I'd like to try an LC filter but I'm getting completely lost in the massive spectrum of inductors that are available. My local hardware store has 2 radial 47mH inductors (both shielded), one with an RDC of 82 (Q=70) and one with an RDC of 52 (Q=100). Both have an IDC of 8mA.

Is the RDC value really the series resistance I can expect in a DC circuit? If so, then values this high render it completely useless here.
Is the IDC value really the max DC current it can handle? If so, then values this low are equally useless here.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod