De-coupling capacitors in a board

[cameo_2007]

Hi

I am here to get an expert help.
I have a board designed for embedded applications,consisting of a processor,RAm,Rom etc.And i could see there are lot of de-coupling capacitors around those chips.
The problm is my board is not stable.Doesnt work sometimes.Does these decoupling capacitors got to do something with that.I figured a small error,i have assembled a 1uF/6.3V smd caps( i ahv only that value) instead of a 1uF/16V smd cap(told in the AN) which is for the RAM IC.Will that voltage difference create problms..?
Definitely need an expert opinion on this.

Regards,

 

[pico]

De-coupling caps

The voltage is depended on the operating voltage of the board which should not exceed the capacitor specified voltage. Please check that the 1uF has correct polarity connection.

 

[grittinjames]

De-coupling caps

usually decopling copacitor is of tantalum cap

 

[som_raparty]

Re: De-coupling caps

hi
as u r mentioning the dcaps will be tant or ceramic mostly placed as close as possible to the ic Vcc pin as to supress the noise making a shortest return path
and reg the voltage specification, it is usually preferable tohave the dcap rated for double the operating voltage of the ic u r using, its just to supress theovervoltage without damaging the cap..
i use mostly .01uf/50v ceramic(of coure the IC operating voltages is 3.3v and 5v) 10uf/10v tant

thnx
somanadh

 

[glucamaf]

Re: De-coupling caps

The AN said that RAM IC needs 1uF/16V. In effect it seems a littlle bit high value as for the capacitor as for the voltage (I suppose that the RAM supply isn't more than 5V).
I usually provide with 100nF/10V maximum for decoupling ceramic capacitors to avoid voltage bouncing problems.
However it could be probably an error in PCB trace design or even the voltage supply is not so stable at the start up....

 

[neoaspilet11]

De-coupling caps

Yes decoupling caps can turn into a problem if you put them into the RAM data and address lines.

If we recall, charging a RC circuit will take some time before it can be 66% of its settling value which we call time constant.

External RAMs usually have CAS latencies, which should not be exceeded so that data transfer between the MCU and RAM are not erroneous.

And based on your values 1uF is a very big value to decouple a signal. Remember

Time Constant = RC. The resistance is taken from teh resistance of teh board traces from the RAm to MCU multiplied to your large decoupling cap may exceed the rated CAS latency of your RAM.

I also worked in board designs, usually we dont put decoupling caps on RAM lines but resistors, to improve signal quality. you can then put decoupling caps to tune the rise time and fall time of your signals.

Hope this help a bit.

 

[sivakumar_tumma]

Re: De-coupling caps

I think there may be two possibilities for not working the board:

1. Frequency of operation: May be the capacitor working in the frequency range where, cap wont work as cap but as inductor.
So check the operating frequency

2. anti resonance between two caps: When two caps are placed at one pin there may be chance of anti resonance which have high resistance and don't allow the bypass current.

Hope this will be useful.....

 

[cameo_2007]

Re: De-coupling caps

hi,
Thankyou.your reply strikes me.

can you brieflyt explain your two points.i didnt understand it clearly.

working frequency of my processor is 396MHz and the RAM is almost abt in that range.How could you relate with the decoupling capacitor and what has to be doen next to avoid next.
Similarily with the anti resonance issue you mentioned.Could you explain it and the solution for that

Thanks n Regds,

 

[sivakumar_tumma]

Re: De-coupling caps

hi,
Thank you.your reply strikes me.

can you briefly explain your two points.i didn't understand it clearly.

working frequency of my processor is 396MHz and the RAM is almost about in that range.How could you relate with the decoupling capacitor and what has to be doen next to avoid next.
Similarily with the anti resonance issue you mentioned.Could you explain it and the solution for that

Thanks n Regds,

Hi,
In real sense cap in not only cap but also include esr and esl in series with cap. If you take frequency response for impedance then it will be in the shape of V. The lower point occurs when cap and esl are in resonance. only resistance will be affective at that point. Before that point capacitor will be dominant and after that point inductance(ESL) will be prominent. So the operating frequency should be before min impedance point frequency..

about anti resonance :
if ur power noise is in broader frequncy range, then one cap is not sufficient. So we need to keep one cap for low frequency and one for high frequency. Because cap is cap+esl+esr, the two capacitor combination can at certain frequency offer high impedance( You write down the equations and solve for that frequency, it is easy simple circuit analysis), which deride the basic requirement that cap to provide decoupling( providing charge to supply pin of chip). Because of this antiresonance,the impedance don't allow the charge requirement of the ic power supply pin...


Hope this is clear..

You can find some documents in the forum itself by name decoupling....

 

[popoyboys]

De-coupling caps

what is the difference between a BYPASS and DECOUPLING capacitor?

 

[shwoo]

Re: De-coupling caps

popoyboys,
You raised a good question, as some people got confused and use them interchangeably. To be strict, bypass capacitors are those used in parallel with emitter bias resistor, for example. Emitter bias resistor is usually part of the dc bias scheme and provides strong negative feedback to stabilize dc operating point. This negative feedback is sometimes undesirable at AC frequency, and you use bypass capacitor across the resistor.
Decoupling capacitors are used on the power bus, for example, to make each stage more isolated (or less coupled to each other).
Again, some people call both of them bypass capacitors.
S. H.

 

[jeyaraman]

Re: De-coupling caps

Decoupling - used along power supply pins to store charge temporarily, desirably for the time it takes for charge to arrive from the large PSU unit. To improve timing, large capacitances are placed locally for few blocks/sections.

small value capacitors, respond better to very high speed discharge due to their reduced ESR, but can provide very limited amount of charge. successively larger capacitances, will recharge the smaller ones, while also catering to the demand from the sink.

Check this out:

https://en.wikipedia.org/wiki/Decoupling_capacitor

All the while i have detailing about only one application.

- jeyaraman

 

[vsmGuy]

De-coupling caps

If the operating voltage of the RAM IC is less than 6.3v then there should not be any problem.

Designers usually specify 16v caps in their BOM because of bulk purchase.

I am yet to hear that decouping caps are producing glitches as they are tied close to GND and thheir only purpose is to "decouple" high swithing currents of the ICs from reaching the rails.

Howvere, it seems the design had been done by a 3rd party.. and hence the way the caps have been placed matter a lot.

Incorrectly placed caps do not help in transient protection.. furthur the caps from the IC, less the protection (the curve is semi-hyperbolic)

 

[popoyboys]

De-coupling caps

i have read that to have an effective decoupling capacitor setup, a set of capacitors with low capacitance should be used.. just to illustrate would be, 5pcs - 0.01uF, 5pcs - 0.1uF and 5pcs - 1uF...

may i know the reason why is here NO available capacitor array is out in the market for those commonly used capacitance values for decoupling?
it would be a space-saver on the pcb designer's point of view if you'll agree with me..

if there's a resistor network/array or a diode array, why can there be no capacitor array??