Results 1 to 5 of 5

Thread: Recommended SMD decoupling capacitor

  1. #1
    Join Date
    Dec 2010
    Location
    SF Bay Area
    Posts
    759
    Post Thanks / Like

    Default Recommended SMD decoupling capacitor

    Searching DigiKey based on ESR, size, and material (25Ohm, 0805, tantalum, (0.1uF, 20V, 10%)) I found this:

    https://www.digikey.com/product-deta...86-2-ND/929918

    Does this look like a good choice for decoupling cap for drivers/counters in a high-speed 5v design? There are also Vishay (a brand I know and have used in the past) but they are not stocked and 5x the price.

    Bumping up in size to a 1206 I see a KEMET 10Ohm, 35V, 10%:

    https://www.digikey.com/product-deta...6-6-ND/3759514

    ...but the SMD advocates seem to consider this a humongous size.
    Last edited by ags0000; 02-28-2018 at 02:33 AM.

  2. #2
    Join Date
    Dec 2011
    Location
    Key West, FL
    Posts
    3,620
    Post Thanks / Like

    Default Re: Recommended SMD decoupling capacitor

    Tantalum is the wrong material for decoupling caps. You want Multilayer Ceramic Capacitors.

    Multilayer Ceramic Capacitors will cost you less than 9 cents each.

    I'm going to suggest a pair of caps at each decoupling point: A 1uf and a .1uf. You will have better frequency/transient response with the pair.

    If you have the time, ebay can provide pricing around 1 to 2 cents each.

    Depending on your board size, look into 0604 size SMD devices.

    When I first started doing home-grown SMD work, I started using 0805 because of my age and eye site.

    I quickly found I could easily work with 0604 and the PCB got a lot smaller.

    If you have not tried SMD at home yet, ebay has a couple of SMD test boards and parts for a dollar or two so you can see what your hand soldering or home re-flow capabilities are before committing your design.

    Hope that helps.

    Joe
    Link to my DownLoad Site: [B][COLOR=#ff0000][URL]http://www.joehinkle.com/HLS[/URL]

    [/COLOR][/B][IMG]http://joehinkle.com/HLS/HLS%20Logo%20Small.jpg[/IMG]

  3. Thanks ags0000 thanked for this post
  4. #3
    Join Date
    Dec 2010
    Location
    SF Bay Area
    Posts
    759
    Post Thanks / Like

    Default Re: Recommended SMD decoupling capacitor

    @JHinkle thanks for the input. It is helpful.

    1) I think I mis-remembered a previous design where I used tantalum for the regulator (along with an electrolytic).
    2) I presume the theory behind using the 1uF/0.1uF pair is to have both the ultra-low ESR/high frequency response (0.1uF) and a larger reservoir (1uF) at once? Or stated differently, the impedance/frequency curves are different between the two, so in parallel they offer a wider range of frequency response (noise decoupling). Here's a KEMET 0.1uF/25v/10% for $0.10: https://www.digikey.com/product-deta...68-1-ND/411443 and a KEMET 1uF/16v/10% for $0.12: https://www.digikey.com/product-deta...84-1-ND/416060 Do you recommend any particular brand(s) - or dis-recommend any?
    3) I will be driving this board with a small Meanwell 5v/25w switching supply (datasheet here: http://meanwellusa.com/productPdf.aspx?i=392#1 ). Seems this is pretty noisy... what should I do (if anything) to condition the overall board power - use a large electrolytic?
    4) I think I may use Oshstencils and try that method out. Seems like it is the easiest way - even for one or two boards - and only $5-10. If I do, I'll also need a heat gun; from what I've seen on YouTube, one meant for electronics (with temperature and airflow control and a small nozzle) may be worth considering, but I have no idea the cost for something that's not junk.
    Last edited by ags0000; 02-28-2018 at 05:21 PM. Reason: fixed links

  5. #4
    Join Date
    Dec 2011
    Location
    Key West, FL
    Posts
    3,620
    Post Thanks / Like

    Default Re: Recommended SMD decoupling capacitor

    Quote Originally Posted by ags0000 View Post
    @JHinkle thanks for the input. It is helpful.

    1) I think I mis-remembered a previous design where I used tantalum for the regulator (along with an electrolytic).
    2) I presume the theory behind using the 1uF/0.1uF pair is to have both the ultra-low ESR/high frequency response (0.1uF) and a larger reservoir (1uF) at once? Or stated differently, the impedance/frequency curves are different between the two, so in parallel they offer a wider range of frequency response (noise decoupling). Here's a KEMET 0.1uF/25v/10% for $0.10 and a KEMET 1uF/16v/10% for $0.12. Do you recommend any particular brand(s) - or dis-recommend any?
    3) I will be driving this board with a small Meanwell 5v/25w switching supply (datasheet here: http://meanwellusa.com/productPdf.aspx?i=392#1 ). Seems this is pretty noisy... what should I do (if anything) to condition the overall board power - use a large electrolytic?
    4) I think I may use Oshstencils and try that method out. Seems like it is the easiest way - even for one or two boards - and only $5-10. If I do, I'll also need a heat gun; from what I've seen on YouTube, one meant for electronics (with temperature and airflow control and a small nozzle) may be worth considering, but I have no idea the cost for something that's not junk.
    #2 - You are correct about the theory. Bypass caps that you are looking at are a jelly-bean product. Everyone make them and they are all pretty much the same from all suppliers given the same specifications. Let price be your guide - higher is not always better -- just more costly.

    #3 - It always good to have a filtered supply network on your board when being fed from an external source. Just watch the amount of bulk cap you use so your source does not go unstable. You might also look into using a ferrite bead to isolate parts of your design from having the hi frequency get into the low speed circuits. If you use a ferrite bead (I use them all the time) just DON'T use on in the supply that powers a packaged oscillator (if you use one). Depending on the frequency, the ferrite bead may prevent the oscillator from starting. I had a design that I isolated into 4 different power sections - each one isolated by a ferrite bead. I was using a 25mhz oscillator to feed my micro and the micro was one of the 4 power networks. I could not make the micro come alive - the oscillator would not start. After 4 days of research and hair pulling, I finally contacted the manufacture of the oscillator - Fox. I spoke with an FAE for 30 minutes with no insight into my problem. Just before hanging up I asked him if he had any out of the box ideas about my issue. He paused and said - "is there a ferrite bead in the supply feeding the oscillator". It's an issue that is not published much - just by word of mouth by those that have been bitten.

    #4 - Also look into Seed and Elcrow for you boards. They are both in China so the wait is about another 5 days over the US companies but the are much less expensive. I've had Elcrow actually do my boards and a 10 unit assembly build that I was happy with - both in quality and cost.

    Why do you need a heat gun? A nice soldering iron works just fine as does a frying pan if you want to reflow at home and don't have a toaster oven.

    Hope that helps.

    Joe
    Last edited by JHinkle; 03-02-2018 at 05:20 PM.
    Link to my DownLoad Site: [B][COLOR=#ff0000][URL]http://www.joehinkle.com/HLS[/URL]

    [/COLOR][/B][IMG]http://joehinkle.com/HLS/HLS%20Logo%20Small.jpg[/IMG]

  6. Thanks ags0000 thanked for this post
  7. #5
    Join Date
    Oct 2012
    Posts
    260
    Post Thanks / Like

    Default Re: Recommended SMD decoupling capacitor

    Good and proper advice by all that responded (we have a really great group of seasoned designers here). I usually just use a 1uF MLCC for bypass, for most devices. If you have a high speed processor, then I will range the cap values to cover the frequency range of transients. Proper layout as well will make a big difference. My preference is to place the cap near the pin of course, and then via to power and run the trace thru the cap landing and then to the pin. The current wants to travel to the pin, if you put the via or power trace between the cap and the pin, then you do not get the full effect of the capacitance.

    Also for board vendors, I think I posted this earlier on another question : https://pcbshopper.com

  8. Thanks ags0000 thanked for this post

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •