Decoupling Capacitors

Looking over the pololu boards that are posted here ( [objects.reprap.org] ) and I see that the decoupling capacitors are significantly larger capacitance than I was expecting.

I normally use small .1uF capacitors in noise isolation like this.

-Is the 100uF the right size for this application? Does it matter?
-Would there be any adverse effects if it is sized too high (inconsistent power supply to drives), or too low (inability to isolate EM feedback)?

---

-----------------------------------------------
[mfsamuel.blogspot.com]

The datasheet recommends both high and low frequency decoupling. I think the high frequency decouplers are on the pololu boards as they need to be close to the chip.

---

[www.hydraraptor.blogspot.com]

i might have had this wrong for a while then. my understanding was a smaller capacitor provides decoupling from lower frequency?

also, i though 10uF-100uf was just for power supply decoupling at the board, in which case one for the whole supply would probably be sufficient.

Edited 1 time(s). Last edit at 03/19/2010 02:54PM by mfsamuel.

---

-----------------------------------------------
[mfsamuel.blogspot.com]

I recently read an article about decoupling capacitors. It recommends, as a rule of thumb, one .1uF monolithic ceramic capacitor for each IC, and possibly others for long wires.

---

---

Darwin clone, Gen 2 electronics, Arduino Duemilanove w/ AtMega328, 5D Firmware, Pinchwheel extruder
[www.codeerrors.com]

mfsamuel,
a few hints about capacitors. The digital circuits in the microcontrollers, computers, stepper drivers, etc, have very high peak current when the clock switches and every internal device change state at once. But since they have low average current draw, it drops quickly to almost zero. This is the perfect environment for the small, ceramic capacitors. The brief surge is small enough, and most especially SHORT enough, that there small capacity to enough to supply all the current, and they have very fast response to high frequency effects. The bigger capacitors have extremely power high frequency response, so they only help on slower changing current needs. But that is exactly what motor drivers, and stepper motors in particular need. To move a stepper one step you need a very high inrush current that lasts for many milliseconds. This would totally drain the small ceramic capacitors, because their capacitance is just way to small to store that much energy. But the big electrolytics handle this well. They provide the boast of power needed when the MOSFETs switch on, and short the motor to the power bus. The initial drain of electricity could drop the voltage on the microcontroller if it were run over all the wires, leads, traces, connectors back to the power supply. But with the electrolytics right there, the current comes for them long enough to satisfy the motors, and everything is good.

Moral, small ceramic caps for digital electronics => Good
Large electrolytic caps for each motor => Good
trusting small ceramics to handle motor inrush => BAD


Mike

When you say small capacitors, does this refer to capacitors with a large value, like 100 uF, with 0.01 uF being a small capacitor?

---

---

Darwin clone, Gen 2 electronics, Arduino Duemilanove w/ AtMega328, 5D Firmware, Pinchwheel extruder
[www.codeerrors.com]

Decoupling capacitors are usually in the 0.1 to 1.0 uF range. Once you get to 100uF, it tends to be either big and expensive, or made of some other material. The electrolytics for the motors should be more in the 100+ uF range.

Mike

I'd vote that 0.1uF is the largest I'd use for digital logic- especially modern chips. the critical properties are parasitic inductance and ESR- as rocket_scientist says, their job is to supply fairly large currents for very short periods of time.

1uF is usually physically large or electrolytic, both factors which massively increase parasitic inductance, so they're out. Tantalums are frequently recommended for decoupling, but they tend to degrade or explode when their ripple current is exceeded, so they're actually quite a poor choice for this task.

I put both 0.1uF and a 680uF electrolytic on my motor supply. The inductance merely in the traces from a large capacitor to the motor driver FETs can allow significant radiation of noise. An 0.1uF physically close to the driver FETs shorts out a lot of the radio-frequency noise and gives the larger capacitor time to catch up with the sudden change in current.

If you look at the bottom of my motherboard (see blog), you'll see a liberal scattering of decoupling capacitors, on logic supply and motor supply, and a number of other locations such as off-board power connections and at the end of long(ish) runs. I do this to short out noise as close to its source as possible, so an absolute minimum is radiated or transmitted to other sections.

When switched high currents and logic are anywhere near each other, decoupling is absolutely critical for reliable operation, as noise is profoundly insidious in its ability to travel from section to section and cause subtle mischief.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

so, the purpose of the 100uF capacitors is basically distributed power not decoupling.

anyone have guidance on how to calculate the proper size for this distributed power application based on the stepper motor used in the application? or alternatively a way to measure using the actual drive.

Triffid_Hunter's use of capacitor's is what I was expecting. A large capacitor at the supply and smaller .1uF close to the drivers.

Edited 1 time(s). Last edit at 03/20/2010 09:56AM by mfsamuel.

---

-----------------------------------------------
[mfsamuel.blogspot.com]

current in a capacitor is given by I=C.dv/dt (yep that's calculus)

so if you need 2A, can suffer a 500mV drop and the main supply takes 1ms to catch up, you end up with 2A=C.0.5v/1ms, 2A*1ms/0.5v=C = 4 millifarads.

My board only has 680uF, so I would drop 1v in 340uS at 2A. maybe I need a bigger reservoir capacitor?


So, in order to calculate the size of your reservoir capacitor, you need to know how long it takes your supply to respond to transients, and how much voltage drop you can handle.

Alternatively, you can just stick in a big-ish one and hope it's enough.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

I have a question sort of.

I have a Creality CR10 and I recently replaced the caps on all the a9488 drivers to low esr Panasonic 16v 220uf smd’s. the original smd’s were 35v 100uf. I did this as I thought it would improve the motherboard as the early cr10 mbs had 220uf 16v on them.

But having looked at a few other 3D printer boards they all seem to be using 35v 100uf caps.

So have I made things worse or better?

100uF is sufficient for most stepper driver chips, unless you make a habit of moving the motors by hand very quickly. Using 35V capacitors instead of 16V makes that aspect of the electronics compatible with using 24V power. If you only intend to use 12V power, then 220uF 16V is OK.

Edited 1 time(s). Last edit at 06/16/2018 07:06AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].