I measured some 1206 SMD COG caps, 1nF 50V, over temperature, Very stable -40 to +90C, well within spec of 5%, I got about 1%. The caps were slodered to thin flying leads.
Next test was on a board, things are a lot worse, most made expansion the spec but only just. The problem we had already identified as board expansion stressing the caps.
My question is how to use caps on real boards, who knows how many other tiny smd parts are experiencing the same problems?
Thinner boards Strategically drilled holes, Raisinig the caps slightly to reduce leverage
It's pretty hard for change in the PCB strays to effect a 5% change in such a large (1000pF) capacitor.
Is this being measured with a proper R-L-C bridge that would indicate any possible significant leakage on the PCB surface? Who is the cap mfr? How many samples were tested?
Best regards, Spehro Pefhany
--
"it\'s the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
Stange that you would get that much variation. Do you have other components connected to the circuit that could be affecting the measurement? Your test leads?
I'm using 0402 220pF NPO caps in an active bandpass filter with a Q of
Going from 25 to 70 deg C doesn't affect the filter noticeably. The phase response doesn't drift more than a couple degrees. This is on a FR4 board.
It looks like you have the problem nailed; if/when the parts manufacturers want to sell their parts as higher-precision, they'll solve some of these issues, but anything you do at your end will just be a stopgap until the manufacturer takes their next manufacturing-change step.
Many ceramics with good high dielectric values are piezoelectric, will respond to stress. Mounting them to a special substrate that has less thermal mismatch, then soldering to the substrate, can work (it's how Hall sensors are made, for similar reasons), but it's costly. Strategic holes, or slots, in thin boards is a good trick, also costly.
The only thing you can do that's not extra expense, is to change the mounting pads for the surface-mount artwork; it's possible that a bit of reshaping can change the local board flexure or buckling enough to improve the performance. Your fabricator, and the capacitor manufacturer, will then insist that any problems that arise are your "fault", of course.
Bell Labs/Western Electric was a conglomerate that could investigate and solve such problems; NASA could do it, too. Little outfits solving problems on a this-week-or-never basis just have to put up with the irritation.
Are you sure it was stress? FR4 has capacitance of its own, and the TC is ghastly... I measured 900 PPM/K on one sample, positive as I seem to recall.
Can't say it was stress, seemed the most likely culprit though. The caps are quite large valued and the FR4 effect is relatively small
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D It's pretty hard for change in the PCB strays to effect a 5% change in such a large (1000pF) capacitor.
Is this being measured with a proper R-L-C bridge that would indicate any possible significant leakage on the PCB surface? Who is the cap mfr? How many samples were tested?
It is measured with a HP impedance analyser and all measurements use the same equipment. I also measured 5 samples every time.
Of interest is that on the first cycle the results are different to subsequent cycles. After being temperature cycled once the C vs T curve is repeatable. The first cycle is slightly different. The change from forst to subsequent cycles is different for the different caps
It is easy enough to see that stress affects the cap by simply bending the board slightly, I can't say that the amount is the same as the haeting and cooling effect though.
Not all caps had the same trend after mounting on a board but it was reasonably close.
In all the cases, the same board layout and location was used for the caps
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D I'm using 0402 220pF NPO caps in an active bandpass filter with a Q of
Going from 25 to 70 deg C doesn't affect the filter noticeably. The phase response doesn't drift more than a couple degrees. This is on a FR4 board.
I'm not sure how much effect you would expect to see in a LC filter, obviously the square root of the capacitances so 220pf to 230pf should move the peak by only 2%. At a q of 10 this wouldn't register unless you were on the edge. Also, 0402 might be a key difference, I use 1206
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D The only thing you can do that's not extra expense, is to change the mounting pads for the surface-mount artwork; it's possible that a bit of reshaping can change the local board flexure or buckling enough to improve the performance. Your fabricator, and the capacitor manufacturer, will then insist that any problems that arise are your "fault", of course.
I have seen some solutions but not commercially. # One was a little metal clip that raises the cap up off the board to reduce the leverage that the board has on the cap. # One patent had clips attached to the caps # One user suggested a thick solder mask in the centre of the cap to raise it slightly # One designer said to drill stress relief holes adjacent to the cap but this was later said to exacerbate the problem # I saw a 'solution' where a large IC was relocated on the opposite side of the board to the cap (underneath the cap) but no results were published.
I am curious to know what happens when you measure this capacitance, mounted on the PCB, and then try to flex the board. Does it change when you bend the PCB ?? I will now have to try this myself, too. I know that caps can break when mounted near the edge of the board. I got some caps recently that supposedly handle stress better, but I think that mainly had to do with breakage when PCB is bent.
Yes, they change, the amount depends on their size, location, type etc. They can definitely crack as can resistors. The change may be permanent. A nice thick board may seem strong and robust but the parts on them don't like it when the board flexes.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.