go_away

Author Topic: Power supply capacitors?  (Read 1615 times)

0 Members and 1 Guest are viewing this topic.

Offline TheBadgerTopic starter

  • Jr. Member
  • **
  • Posts: 33
  • Helpful? 0
Power supply capacitors?
« on: April 27, 2011, 11:25:33 AM »
I am getting myself confused, I kind of understand decoupling around chips, like you place a small ceramic capacitor on the supply to a chip as this acts like a low pass filter getting rid of any noise or spikes.

I generally see 0.1uF for this cap and then larger ones on the supply rails,

I just don't understand how you spec these supply caps, like what formulas do you use. For example how did the admin arrive at the value he has for the large caps on the axon?

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: Power supply capacitors?
« Reply #1 on: April 27, 2011, 01:33:30 PM »
Hi,

I am getting myself confused, I kind of understand decoupling around chips, like you place a small ceramic capacitor on the supply to a chip as this acts like a low pass filter getting rid of any noise or spikes.
I'd rather call its main action a buffer.


I generally see 0.1uF for this cap and then larger ones on the supply rails,

I just don't understand how you spec these supply caps, like what formulas do you use. For example how did the admin arrive at the value he has for the large caps on the axon?
When you see 100nF (0.1µF) they are chosen rather subconsciously and not spec'd as such.
It's simply going by rule of thumb and work most of the time.

In most cases, 100nF is not needed and sometimes as little as 1nF will do, but since choosing the right one would mean extra labour, it's not always worth it in a commercial setting.

If you wanna spec your own, you need to test with a scope while it makes the max amount and strength of transitions - we're talking digital circuits here - then you simply find the value that makes the spiking on the supply tolerably low.

The more current is being switched, the larger the cap has to be.

The larger supply caps is more as a general buffer for the supply and to lower the impedance of the supply lines.
The per-IC caps is buffers for the supply to the IC in question, to avoid its transitions getting onto the supply lines.
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline TheBadgerTopic starter

  • Jr. Member
  • **
  • Posts: 33
  • Helpful? 0
Re: Power supply capacitors?
« Reply #2 on: April 27, 2011, 01:53:37 PM »
There must be a mathematical way of deciding on what supply cap to use though even though no one does, it can't just be trail and error surely?

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: Power supply capacitors?
« Reply #3 on: April 27, 2011, 02:11:24 PM »
Hi,

There must be a mathematical way of deciding on what supply cap to use though even though no one does, it can't just be trail and error surely?
If you know the exact parameters (like magnitude, speed, current etc) of the transition spikes, you can calculate it of course (C [Farad] = I [Ampere] * t [seconds] / U [Voltage], but it's like relying on sims... You'll never get the full picture, as you'll never have all the parms.

Finding it empirically is comparatively fast and gives you a real life view of what goes on.
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline TheBadgerTopic starter

  • Jr. Member
  • **
  • Posts: 33
  • Helpful? 0
Re: Power supply capacitors?
« Reply #4 on: April 27, 2011, 02:18:48 PM »
So if I built say a microcontroller development board you suggest I build it and then max it out with software while checking it with a scope?

I'm a little confused on the process here, I need to build before I design...chicken and egg:(

thanks for the help though, I kind of get it now.

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: Power supply capacitors?
« Reply #5 on: April 27, 2011, 04:34:00 PM »
Hi,

So if I built say a microcontroller development board you suggest I build it and then max it out with software while checking it with a scope?

I'm a little confused on the process here, I need to build before I design...chicken and egg:(
No, you use eg. 100nF (or whatever your preference) caps in your design for starters (more if you are designing with eg. a 100..250 I/O lines controller with heavy I/O activity) and then you test your prototype and see if it passes your noise/supply requirements and decide whether you wanna bother finding the smallest cap that will (this only matters if you are making at least several hundred thousand units).

If you study computer boards from the last 30 to 40 years, you'll see that the major part of them has got a 100nF cap on any digital IC on the board. From the early boards heavily equipped with LSTTL chips to modern days various MOS technologies, with vastly different current requirements, speed etc.
Coincidence? No, simply a place where it's counterproductive to dabble too long (time is money) and you don't save much in reducing the caps a bit, so you grab something that you know will most likely work and if issues shows, you 'scope and change accordingly.

It's actually so with a large part of a design - some parts you choose some values for from experience and if in doubt, you calculate the given circuit, other parts you simply need to calculate - like the air gap in a switch mode core. The Rosetta Stone is knowing when to calc and when to go with your belly (which has a tendency to grow with age as well - ah, must be the experience pressing against my belt ;D)

Decoupling isn't really part of a design, it's more the basis of any circuit (like eg. a PCB) and quite often you won't see those in a schematic, as they're implicit in any circuit (and pretty boring compared to the circuit proper).


From the point where you build your first prototype to the point where you are ready to manufacture your product(or get it manufactured) is called the maturing period and you will probably go through several prototype iterations (just read about Admins secret new product, although I'd say that he's probably been more unlucky in that respect than what's usual, even without knowing its degree of complexity).

I recognize your youthful want for being able to calculate it all, but even if we have come far in this game, where we can take some iron, wind some copper around it and are able to tell pretty accurately how it will behave electrically under different circumstances, we still need all parameters to do so and it's not always possible to get each needed parm with a useable precision, so approximations, qualified guesses and gut feeling is still major players in design/prototyping, but that is of course a little hard to accept until you gain access to those "tools" - don't worry though - each new design you make is a step towards it :)
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline knossos

  • Robot Overlord
  • ****
  • Posts: 278
  • Helpful? 14
Re: Power supply capacitors?
« Reply #6 on: April 27, 2011, 05:11:58 PM »
This topic reminded me of this article (somewhat off topic since its about resistors).  It talks about using resistor triads to replace resistor values that have been selected in test.  It's worth a quick read and includes sample code for a program to select appropriate resistors that best fit the target value.
"Never regret thy fall,
O Icarus of the fearless flight
For the greatest tragedy of them all
Is never to feel the burning light."
 
— Oscar Wilde

Offline waltr

  • Supreme Robot
  • *****
  • Posts: 1,944
  • Helpful? 99
Re: Power supply capacitors?
« Reply #7 on: April 27, 2011, 07:10:45 PM »
Nice article and it does fit in with Soeren's gut feel designing.

I use the parallel/series resistor triad in designs where I do not want a pot but know there will be some value adjustment due to known parameters. One place I commonly use this is in the voltage feed-back divider for switcher power supplies. This way once the prototype PCB arrives the output voltage can be easily tweaked with common values.

Quote
So if I built say a microcontroller development board you suggest I build it and then max it out with software while checking it with a scope?

I'm a little confused on the process here, I need to build before I design...chicken and egg:(

If you were on the team that is developing a new microcontroller then you may load it with software and evaluate all the power and IO. If not then there usually are guide line published by the manufacture on how many and which caps to use.


Offline TrickyNekro

  • Contest Winner
  • Supreme Robot
  • ****
  • Posts: 1,208
  • Helpful? 15
  • 1.6L Peugeot 307 tuner
Re: Power supply capacitors?
« Reply #8 on: April 28, 2011, 02:58:43 AM »
I kinda fell for this trap in my current design.


I needed to calculate some resistors for an H-Bridge design and I did run sims, I did paper math etc etc etc...

Only to find out, when I finished the PCB that, the values I wanted would be hard to find and that they were really "on the limit"...

So I said, who cares, I'll go the gut way and place values I knew they would work...

And guess what... That H-Bridge works dump fine...

And believe it or not, after all those calculations, I didn't some other, well not errors, but I didn't take into consideration
things I should...
Things that would actually get my MOSFETs burned, without me even knowing why...

And something else... If every man is using 100nF... You won't cut budget getting a 10nF or something else... You'll get a bigger budget...

Know why? If everyone is using the 100nF, then it means that it's more mass produced than the 10nF and then it means it would have a lower price... simple...

Also not that caps have tolerances and self resistance... A MKT capacitor is generally better not noise suspension that a simple ceramic...
And many many more things...


But all these as Söeren said, don't come in a day's sleep... You gotta drink a lot of coffee and lose many hours of your beauty sleep to get to know that the ..... is going on...

Also don't assume that everything will work up to the specs... Most IC are tested in "Laboratory conditions" and that's bad the least...
You always gotta assume greater tolerances that the onces on the specs...

And if designing something professionally, then fail-safes will come into play...

So if I were you, I'd better stick to that 100nF right now  :P  :-* :-* :-*


Best Regards, Lefteris
Greece
For whom the interrupts toll...


P.S. I've been inactive for almost a year... Don't give promises but I'll try to complete my tutorials. I'll let you know when..

Cheers!

Offline The arctic wolf

  • Jr. Member
  • **
  • Posts: 18
  • Helpful? 0
Re: Power supply capacitors?
« Reply #9 on: April 28, 2011, 04:53:30 PM »
Hello everyone,
Soeren, you mentioned a buffering capacitor on every digital IC almost in every computer circuit, acting as a buffer.
 And here is something I don't understand ever since I started the practical part of electronics and runned into the same buffering caps, to this very day.
By what logic do we need them??
Those circuits already have a filtering capacitor between the supply lines that handles the spikes, adding any additional caps in paralles will just increse the overall capacitance between the supply lines..
Why for heavens sake buy all those additional caps and increase the total cost of manufacturing??


Andrei.

Offline knossos

  • Robot Overlord
  • ****
  • Posts: 278
  • Helpful? 14
Re: Power supply capacitors?
« Reply #10 on: April 28, 2011, 05:52:20 PM »
Just because the total capacitance is the same doesn't mean that the effect is the same.  For example large capacitors are better for filtering low frequency noise while small capacitors are better at filtering high frequency noise.  Also, positioning of the capacitors makes a difference as well.

I was trying to find a good article I had read a long time back on how digital circuits are all really just a subset of analog circuits, but I can't find it now.  This subsection of the Wikipedia digital electronics entry touches on some of what was in that article:

Quote
Analog issues in digital circuits

Digital circuits are made from analog components. The design must assure that the analog nature of the components doesn't dominate the desired digital behavior. Digital systems must manage noise and timing margins, parasitic inductances and capacitances, and filter power connections.

Bad designs have intermittent problems such as "glitches", vanishingly-fast pulses that may trigger some logic but not others, "runt pulses" that do not reach valid "threshold" voltages, or unexpected ("undecoded") combinations of logic states.

Additionally, where clocked digital systems interface to analogue systems or systems that are driven from a different clock, the digital system can be subject to metastability where a change to the input violates the set-up time for a digital input latch. This situation will self-resolve, but will take a random time, and while it persists can result in invalid signals being propagated within the digital system for a short time.

Since digital circuits are made from analog components, digital circuits calculate more slowly than low-precision analog circuits that use a similar amount of space and power. However, the digital circuit will calculate more repeatably, because of its high noise immunity. On the other hand, in the high-precision domain (for example, where 14 or more bits of precision are needed), analog circuits require much more power and area than digital equivalents.
"Never regret thy fall,
O Icarus of the fearless flight
For the greatest tragedy of them all
Is never to feel the burning light."
 
— Oscar Wilde

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: Power supply capacitors?
« Reply #11 on: April 28, 2011, 06:11:31 PM »
Hi,

This topic
 reminded me of this article (somewhat off topic since its about resistors).  It talks about using resistor triads to replace resistor values that have been selected in test.  It's worth a quick read and includes sample code for a program to select appropriate resistors that best fit the target value.
I've been using that technique on PCBs for decades for getting exact values, as I used to only stock the E-6 range (which is fine for most stuff), but it seems the article want to use them more like Robert Pease described back in the seventies in Linear Brief #46 (LB-46) from National Semiconductor, (also worth a read and a keeper) about trimming variable voltage regulators in production.

While a trimmer is easier for that, it's also more error prone and may get out of adjustment.

Too bad that most amateur builders are more concerned about the amount of components and the price, than what's best - it's a bit disappointing, when you have spent time cooking up a good (i.e. optimized) circuit for someone and the first question following is "can I make it without this, that and some other components?"   :-\
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline Soeren

  • Supreme Robot
  • *****
  • Posts: 4,672
  • Helpful? 227
  • Mind Reading: 0.0
Re: Power supply capacitors?
« Reply #12 on: April 28, 2011, 06:48:28 PM »
Hi,

Soeren, you mentioned a buffering capacitor on every digital IC almost in every computer circuit, acting as a buffer.
 And here is something I don't understand ever since I started the practical part of electronics and runned into the same buffering caps, to this very day.
By what logic do we need them??
Those circuits already have a filtering capacitor between the supply lines that handles the spikes,
Eh?
I never said anything on adding a cap over supply pins that already have a cap?
If they already have a cap, that's the cap (although in a design you're making, there is no caps to start with ;)), so I don't know what you're on about. You aren't talking about "stripline" caps or parasitic capacitance from the recommended running supply lines close to each other (for keeping the inductance low) by any chance?
Or is it the overall buffer caps on the main supply lines you are thinking of?


adding any additional caps in paralles will just increse the overall capacitance between the supply lines..
As knossos already mentioned, caps ain't just caps. I'd add that each type have different ESR, which is a very important parameter in buffering/filtering a supply and eg. ceramic caps are aces there.
Caps have self resonance as well and without going into the nitty gritty of why, this is the main reason you'll sometimes see a triplet of caps on a supply - It's like teamwork, wht one cap cannot handle, one of the other can and you get wide-band action as a result.

And I can only second his comment on analog influence.
Digital is analog driven to the extreme, sort of speaking and that's why you won't get far in digital circuits with a 2..10MHz 'scope or a 25MHz LSA - You need equipment that can show the shortest times that will influence the circuit in question.


Why for heavens sake buy all those additional caps and increase the total cost of manufacturing??
You can be damn sure of one thing about commercial products, at least when you reach the second iteration to hit the market, that the "money men" have been bashing the engineers to get the component count to the bare minimum and all calculated to last an average of a few years for consumer goods (aero, mil and health sectors do require better specs and get it).

So, if you see a second generation mass produced board with 137 caps, be certain that it wont run or will do so erratically with 136 caps.
That said, as I mentioned earlier, sometimes you just use something known to work, as it may be too expensive to fiddle with, compared to just use a standard solution.


Seems we got really OT on that one, but I hope we're forgiven, as this is some of the important stuff that is often overlooked (like power supplies in general). You might say that it's like a pair of shoes - shooting for cheap, fancy or whatever might enable you to work a couple of miles max, but getting something that fit's your feet ad supports them optimally will enable you to work all day and then take a half marathon... Well, sort of at least  ;D
Regards,
Søren

A rather fast and fairly heavy robot with quite large wheels needs what? A lot of power?
Please remember...
Engineering is based on numbers - not adjectives

Offline TheBadgerTopic starter

  • Jr. Member
  • **
  • Posts: 33
  • Helpful? 0
Re: Power supply capacitors?
« Reply #13 on: April 29, 2011, 03:36:04 AM »
Soeren you are a genius!

that is all.

Thanks for the help, I now need to look into ESR and resonance.

Offline The arctic wolf

  • Jr. Member
  • **
  • Posts: 18
  • Helpful? 0
Re: Power supply capacitors?
« Reply #14 on: May 04, 2011, 12:51:59 PM »
Ok, I understand the issue better now, thank you.

 


Get Your Ad Here

data_list