Ripple Cancellation

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Tue Nov 10, 2009 1:35 pm

I am using the following DC to DC converter.

http://www.newark.com/artesyn/bxa3-24s0 ... xa3-24s05j

I am learning about EN55022...

I have 3 different dc to dc converters the one above is giving me the best results.

I still have a noise issue but its not as bad. Playing with the capacitance on the input side is giving me some interesting results.

THANK YOU ALL FOR YOUR INPUT!!

I am real busy fixing this issue and can post more info later.

Tony

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Tue Nov 10, 2009 4:19 pm

Whoever can solve this gets a cookie...

This is my output of my Isolated DC to DC converter.
Image

I have the converter wrapped in M3 tape.

I am not having much luck killing this noise. I need to to go for my amplifier circuits on board to work properly.

Tony
:? :? :? :? :? :?

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Ripple Cancellation

Post by Robert Reed » Tue Nov 10, 2009 5:29 pm

67Mhz-74mv p-p ripple. What is this supply driving-a high speed clock?If so, what is the period between these transients? Does that period relate to any thing in the load circuit? Have you swapped loads with one of a pure resistive nature? Have you tried powering the circuit with a linear bench supply?
Yea, I know- a lot of questions :grin:

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Wed Nov 11, 2009 5:23 am

I will answer what I can and get more info later...
Robert Reed wrote:67Mhz-74mv p-p ripple. What is this supply driving-a high speed clock?
The load drives several processors running at 30,25,8Mz
*** I can put a probe on the converter and see the same AC signal
Robert Reed wrote:Does that period relate to any thing in the load circuit?
Nothing that I see... YET!
Robert Reed wrote:Have you swapped loads with one of a pure resistive nature?
I can do this next.
Robert Reed wrote:Have you tried powering the circuit with a linear bench supply?
Yes I have AND my board works GREAT!
This is what kills me! Jumping straight to the DC2DC out and when I look at the 5V from the supply to my board... it looks fine. No noise!


I am guessing I will need to somehow filter this. But how? I was assuming that I needed to do this at the output stage but playing with the input caps also gives me some interesting results.

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Wed Nov 11, 2009 7:49 am

Hooking up a external load to the DC2DC I see lots of noise on the output. More than what you see in the photo above.

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Ripple Cancellation

Post by Robert Reed » Wed Nov 11, 2009 8:21 am

Well it definitely looks like your noise source is the power supply. Your scope trace shows the actual transient ringing, but it would still be nice to know the period of those transients - I am suspecting a much lower frequency. Also, do you know the bandwidth of the scope and the probe you are using. 74 mv isn't ungodly high and normally wouldn't bother attached circuits - short of infiltrating sensitive inputs and references. Now, if in fact the scope/probe bandwidth was quite low, that 74 mv could actually be much higher!And that would be problematic.

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Wed Nov 11, 2009 9:13 am

Well this power output IS a problem that I will have to fix.

I'll check the period...

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Ripple Cancellation

Post by MrAl » Wed Nov 11, 2009 10:47 am

Hi there,


In some converters when the diode stops conducting the diode's internal capacitance rings with the coil at
fairly high frequency. This could be happening here. Is this a buck converter?

One way to get rid of it is to stop the diode from NOT conducting, which requires a bigger series
inductor in the power supply itself.

Another way is again, to go with a pi filter. That will knock the heck out of this if it truely
is coming from the power supply itself and not the load.
If you cant modify the power supply then use a pi filter...it doesnt take much to build one
as i said in a previous post. You'll be surprised how well it works.
The problem with capacitors alone sometimes is that they dont have any series impedance to
work with, and that's how they filter to begin with. By using the series inductor as well as
two caps for the pi filter it attenuates the noise quite a bit.
LEDs vs Bulbs, LEDs are winning.

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Wed Nov 11, 2009 12:43 pm

Hello Mr. Al
Yes it is a Buck.
How can it tell if its the load. The board works with a external power supply.
I am noticing noise in the Mhz range...
what do you think of this calculation...
http://www.wa4dsy.net/cgi-bin/lc_filter ... =MHZ&Z=466
those caps seam VERY low!

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Ripple Cancellation

Post by Robert Reed » Wed Nov 11, 2009 5:43 pm

Sam
Don't know why you would want to pass all those frequencys (Up t0 200MHz). As a general rule for your transients of interest - try a pi with 10 to 20 uH and 1000 pf caps for the values. Use short leads and connect right across the DC/DC output. Also run the loads ground directly to that same ground point.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Ripple Cancellation

Post by MrAl » Thu Nov 12, 2009 2:17 am

Sambuchi wrote:Hello Mr. Al
Yes it is a Buck.
How can it tell if its the load. The board works with a external power supply.
I am noticing noise in the Mhz range...
what do you think of this calculation...
http://www.wa4dsy.net/cgi-bin/lc_filter ... =MHZ&Z=466
those caps seam VERY low!

Hi again,


Well you can pretty much ignore the values on that page as it looks like it is meant for impedance matching,
something you really dont have to do here. What you are doing is trying to create a ripple (or noise)
attenuation filter and that is different than impedance matching. The configuration is the first
circuit on that page with the two caps and the one inductor however, so it's the same config but using
much bigger values for the components. Sometimes as little as 2uH can work though, so you can wind
an air core coil if you would like to try that. Sometimes a small ferrite bead with as little as three turns
through the center can also work as the inductor...something else you may wish to try. Heck it might
even work with one pass through the center like how many ferrite beads are used anyway.


As we have said before try a small inductor with a cap (say 0.1uf) on the input side and a cap on the
output side (say 0.1uf). 0.1uf caps are good for noise but 0.01uf works too sometimes.
Sometimes you also need a larger cap on the output like 100uf electrolytic.
LEDs vs Bulbs, LEDs are winning.

sghioto
Posts: 322
Joined: Wed Dec 26, 2007 5:16 pm
Location: Colorado
Contact:

Re: Ripple Cancellation

Post by sghioto » Thu Nov 12, 2009 5:48 am

Found this on the web, might be useful.
http://www.murata-ps.com/data/apnotes/dcan-31.pdf

Steve G.

Robert Reed
Posts: 2276
Joined: Wed Nov 24, 2004 1:01 am
Location: ASHTABULA,OHIO
Contact:

Re: Ripple Cancellation

Post by Robert Reed » Thu Nov 12, 2009 8:23 am

Sam,MrAl
When choosing bypass capacitors for a specific band of high frequencys, usually "less is more"
Almost all caps have a broad resonant frequency at where they perform best (for bypass). Below and above that frequency, the series impedance rises. The standard 0.1 is a general value chosen for lower Mhz range. when we start working our way towards 100 Mhz and above, 1000 pf is a superior bypass value than 0.1 is. In working at higher VHF frequencys, I have gone as low as 300 pf for bypass and it performed better than much larger values. In stubborn situations, I have had to calculate lead inductance of the cap and cut accordingly to make a series resonant package out of just the capacitor. When confronted with a broad band of noise (say 1-100 MHz) I usually "digitize" the bypass capacitor. That would be a 10 mf tantalum, 0.1, 0.001 in parrellel. This covers all bases and has always worked well for me. Also ceramics are your best choice above 10 MHz.

User avatar
Sambuchi
Posts: 366
Joined: Tue Jan 18, 2005 1:01 am
Location: Orlando FL
Contact:

Re: Ripple Cancellation

Post by Sambuchi » Fri Nov 13, 2009 7:38 am

Steve!
sghioto wrote:Found this on the web, might be useful.
http://www.murata-ps.com/data/apnotes/dcan-31.pdf
Steve G.
Image

This solved my problem! Thou I tried this before I read your post... :D

My board was experiencing ground bounce. Using a 100V 1uf cap going to both grounds fixed my VERY sensitive amp circuit!

Thanks sooo much to everyone that contributed here.
Being the only EE/SE/tech/component engineer at my company... its healthy for me to vent frustration onto you guys! :lol:

Till next time!

-Tony

sghioto
Posts: 322
Joined: Wed Dec 26, 2007 5:16 pm
Location: Colorado
Contact:

Re: Ripple Cancellation

Post by sghioto » Fri Nov 13, 2009 7:03 pm

Your welcome. Going to download my cookie now :grin:

Steve G.

Post Reply

Who is online

Users browsing this forum: No registered users and 40 guests