Shop OBEX P1 Docs P2 Docs Learn Events
Decoupling capacitors — Parallax Forums

Decoupling capacitors

LeonLeon Posts: 7,620
edited 2007-04-10 18:03 in Propeller 1
I've noticed that the Parallax boards only have a single 1 uF decoupling capacitor for the two Vdd and Vss pins, whereas one 100 nF capacitor for each Vdd - Vss pair is more usual with other MCUs. What is the reason for the single 1 uF capacitor?

Leon

Comments

  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2007-04-09 08:53
    Firstly, the capacitor range for decoupling capacitors may vary quite a bit as one cannot predict the actual frequency of incoming noise. A range from 1 mf to .005 uf is not uncommon. Most people stick with .1 mf or .01 mf.

    Additionally, you may be overlooking surface mounted capacitors. In many of the surfure mount boards, there are 4 small de-coupling capacitors [noparse][[/noparse]one on each side of the square chip]. Guenther explained some time ago that this was a superior method of noise reduction when compared with one capacitor on a DIP and the radial placement was quite important.

    Finally, with 4 layer board construction the two inner layers act to further noise reduction. One is usually Vcc and one is Vss.

    So...
    You just may have not noticed the capacitors and noise reduction scheme.

    In any event, the DIP boards are the only ones that have 'one decoupling capacitor'. For sure,that is the SX-28 board and I believe it is also 4 layer.

    I haven't examined the PropStick or Propeller 40-pin DIP boards, but the lower clock frequency might be a reason for the 1uf and the clock is the prime internal source of noise emitted onto the power lines.

    This is an area where testing and not pure calculation must provide the best answer.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    "If you want more fiber, eat the package.· Not enough?· Eat the manual."········
    ···················· Tropical regards,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan

    Post Edited (Kramer) : 4/9/2007 9:02:27 AM GMT
  • LeonLeon Posts: 7,620
    edited 2007-04-09 09:18
    Only one 1 uF capacitor is shown by the Propeller chip on the schematic for the demo board, hence the confusion. The example circuits in the documentation don't show any! Anyway, I've used my usual 100nF capacitors for each Vss-Vdd pair on a Propeller board I've just designed.

    Leon

    Post Edited (Leon) : 4/9/2007 9:24:13 AM GMT
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2007-04-09 17:18
    It wouldn't hurt to put in two values in parallel to cover a wider range of spikes. The only thing that I have ever read about this developing problems is mixing tantalum and ceramic. According to the 'The Art of Electronics', occasional parasitic oscillations might develop due to there difference in construction. This is a somewhat exotic explanation.

    I would have to dig to give you an exact page reference.

    Frankly, that text seems a bit frustrated about oscillations in general and I am not sure it is 100% objective. As the authors are university faculty, they may have just had a lot of trouble with students doing sloppy work and having to try to ferret out their problems with such.

    You seem to be following mainstream standard philosopy. But, I keep in mind that I should have a way to swap out the decoupling capacitor if a particular value just doesn't seem to work in a given situation.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    "If you want more fiber, eat the package.· Not enough?· Eat the manual."········
    ···················· Tropical regards,····· G. Herzog [noparse][[/noparse]·黃鶴 ]·in Taiwan
  • Paul BakerPaul Baker Posts: 6,351
    edited 2007-04-09 21:12
    Leon the major reason most of our designs have a single bypass cap is space constraints and the delaterious effects of 1 vs 2 bypass caps·in our experience are negligible. A robust design would have at least two located on opposite sides, and the voltage regulator would have two of different values to reduce the ripple of low and high frequency noise.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Paul Baker
    Propeller Applications Engineer

    Parallax, Inc.
  • TransistorToasterTransistorToaster Posts: 149
    edited 2007-04-09 23:30
    Leon,
    You need to be very careful with the capacitor on the output of a LDO regulator. If you don't fall within the proper parasitic ESR range supported by the LDO regulator, it will oscillate.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2007-04-10 00:51
    Parasitic in this context refers to what is not desired and becomes a nuisance. The LDO regulators are not as stable as the standard regulators and require a low Equivalent Series Resistance (and also inductance) capacitor of around 10uf or they may oscillate (depending upon load etc). The decoupling capacitor on the other hand provides that same low ESR very close to the load. Mind you, the decoupling capacitor is not as critical if your regulator practically sits right next to the load.

    The CPU like all switching logic can draw high currents as they switch for a variety of reasons. One of which is charging/discharging "parasitic" capacitance, this is the capacitance of connections and gate capacitance. Another reason is simply that as one output transistor is turning on the other is still turning off, it might not be much of a delay but there is a period where the sourcing transistor and the sinking transistor of the output are both on or partially on. There is a resultant current spike which if not properly regulated will result in a voltage drop which may affect other parts of the circuit.

    Since large currents are required for very brief periods the easiest way to achieve this is with a capacitor which can supply that current right there where it is needed. From the laymans perspective if this capacitor were too far away then the inductance of the connections would prevent that current being dumped when it is needed (later is not an option). That is why the capacitor needs to be of a certain "capacity" to supply that charge, however the larger capacitors exhibit too much ESR and inductance to be of much use. A value of around 10 to 100nf (usually monolithic ceramic) will work well if and only if you have them as physically/electrically close as possible to the supply pins.

    It may work without decoupling capacitors but don't be surprised if you have mysterious glitches in the operation of the unit. I've seen many a novice and even uni grads scratch their heads trying to figure out what is wrong with their design rolleyes.gif

    If a picture paints a thousand words then I have hardly drawn a blade of grass.

    *Peter*

    Post Edited (Peter Jakacki) : 4/10/2007 12:56:16 AM GMT
  • bambinobambino Posts: 789
    edited 2007-04-10 16:56
    Leon,

    Check out the anyCap LDO modules from Analog Devices.
    Their claim I don't agree with, but they do make finding a capacitor that will work a lot eaisier.
  • LeonLeon Posts: 7,620
    edited 2007-04-10 18:03
    I don't have any problems with my design, I used a LM317 regulator with the usual 100nF on the input and a 10uF on the output. I used a 100nF capacitor on each of the two Propeller supply pins.

    Leon
Sign In or Register to comment.