Sorry if this is a topic that appears a lot but I have a question about smaller value bypass capacitors and I am finding conflicting information.
My project is a simple microphone preamplifier with an OP627 opamp. In the datasheet provided by Burr Brown, it is suggested that "applications with low impedance loads or capacitive loads with fast transient signals demand large
currents from the power supplies. Larger bypass capacitors such as 1µF solid tantalum capacitors may improve dynamic performance in these applications."
All well and good, but I have always used film capacitors, or ceramic in lower values, in this position-- as close as possible to the pins. I did this because they were cheap and I've seen other builders and designers using them so I figured it was acceptable practice. However, upon reading Analog Devices MT101 decoupling techniques tutorial, they do not recommend using film capacitors: "In general, film type capacitors are not useful in power supply decoupling applications because they are generally wound, which increases their inductance. They are more often used in audio applications where a very low capacitance vs. voltage coefficient is required."
So when users here and elsewhere (i.e. Rod Elliot for one) recommend using film bypass caps is this actually a misunderstanding on my part? Does it depend more on the application? And what would a cost effective alternative be to film be in 1uF value? (I've never actually used tantalum as it was more expensive and had read bad reports on the quality.)
My project is a simple microphone preamplifier with an OP627 opamp. In the datasheet provided by Burr Brown, it is suggested that "applications with low impedance loads or capacitive loads with fast transient signals demand large
currents from the power supplies. Larger bypass capacitors such as 1µF solid tantalum capacitors may improve dynamic performance in these applications."
All well and good, but I have always used film capacitors, or ceramic in lower values, in this position-- as close as possible to the pins. I did this because they were cheap and I've seen other builders and designers using them so I figured it was acceptable practice. However, upon reading Analog Devices MT101 decoupling techniques tutorial, they do not recommend using film capacitors: "In general, film type capacitors are not useful in power supply decoupling applications because they are generally wound, which increases their inductance. They are more often used in audio applications where a very low capacitance vs. voltage coefficient is required."
So when users here and elsewhere (i.e. Rod Elliot for one) recommend using film bypass caps is this actually a misunderstanding on my part? Does it depend more on the application? And what would a cost effective alternative be to film be in 1uF value? (I've never actually used tantalum as it was more expensive and had read bad reports on the quality.)
I use small caps as close as practical to the pins, followed by successively larger caps. I might start with a 1000pF ceramic 0402, followed by a 0.1uF ceramic 0603, and if it's a high-current situation as you mention, maybe a 2.2uF ceramic 0805. I also use small isolation resistors to the plus and minus supply rails, to decouple high frequency transients. If the op-amp draws too much current to allow more than a few ohm isolation resistor, I use an appropriate ferrite bead instead. The overall technique has worked on a lot of op-amps, even ones with 5GHz unity gain frequencies.
Film caps are wound but not necessarily very inductive. It depends on the type of construction. The better ones are wound with the two metal layers offset so that they protrude from the sides of the roll, slightly, one from one side of the roll and the other from the other side, with end caps that make contact with the entire length of the long edges of the rolled metal layers. The worse ones would connect the leads only at one end of the roll, to the short edges of the metal layers.
I believe from my researches that ESR is very important to decoupling effectiveness.
Film capacitors in general are very low ESR.
Electrolytics again in general are medium to high ESR.
HiK ceramics are a middle ground that have medium ESR.
These x7r seem to make good decoupling capacitors the highest frequencies.
Electrolytics seem to make good decoupling for medium frequencies.
Film capacitors tend to be very good at passing signals. That "very good" just about bans them from decoupling duty.
As I said at the beginning, this is what I believe, but there are certainly experts on the Forum who may have different views.
Film capacitors in general are very low ESR.
Electrolytics again in general are medium to high ESR.
HiK ceramics are a middle ground that have medium ESR.
These x7r seem to make good decoupling capacitors the highest frequencies.
Electrolytics seem to make good decoupling for medium frequencies.
Film capacitors tend to be very good at passing signals. That "very good" just about bans them from decoupling duty.
As I said at the beginning, this is what I believe, but there are certainly experts on the Forum who may have different views.
ceramic through hole from those plots show 1uF 0r011 @ 1M5Hz and 0.1uF 0r030 @ 5M2Hz
whereas film shows 1uF 0r018 @ 1M2Hz and 0.1uF 0r034 @ 5M1Hz.
That seems to indicate that these two examples are very similar, rather than very different.
Could the leads of the through hole have such high inherent inductance that this swamps the other parameters?
Would this indicate even more that short leads are important?
whereas film shows 1uF 0r018 @ 1M2Hz and 0.1uF 0r034 @ 5M1Hz.
That seems to indicate that these two examples are very similar, rather than very different.
Could the leads of the through hole have such high inherent inductance that this swamps the other parameters?
Would this indicate even more that short leads are important?
yes, and from what I understand, snubber is better for this (resistor/cap)
its now well known that what may appear like a simple regulator might tend to oscillating, if not careful
actually I thought it was better to use cap with not too low ESR, hence the snubber resistor on film caps
or maybe I have misunderstood what it is about
its now well known that what may appear like a simple regulator might tend to oscillating, if not careful
actually I thought it was better to use cap with not too low ESR, hence the snubber resistor on film caps
or maybe I have misunderstood what it is about
The problem that can result when using several scaled bypass caps is that, above resonance, the larger capacitor can look like an inductor, and can form a tank circuit with the smaller capacitor that is placed in parallel. So, the combination can ring at some magic frequency depending upon the effective inductance of the larger cap and the value of the smaller cap.
If the larger cap is somewhat lossy, as a tantalum or aluminum electrolytic would be, this resonance is damped. A film cap would offer nearly no losses at all, so it would tend to ring much more easily. A series resistor from the supply to the pair of caps will not damp the oscillation currents flowing between the caps, but will help to keep them out of the supply rail.
Again, as has been stated earlier, it really depends on what you're 'bypassing'… if it's a digital circuit with a fixed clock frequency, you're asking it to couple regular transient pulses to ground, so specific resonant frequencies could be a problem. If it's just a 'keep the rails low impedance so the circuit remains stable' bypass, then specific resonant frequencies may not be important.
If the larger cap is somewhat lossy, as a tantalum or aluminum electrolytic would be, this resonance is damped. A film cap would offer nearly no losses at all, so it would tend to ring much more easily. A series resistor from the supply to the pair of caps will not damp the oscillation currents flowing between the caps, but will help to keep them out of the supply rail.
Again, as has been stated earlier, it really depends on what you're 'bypassing'… if it's a digital circuit with a fixed clock frequency, you're asking it to couple regular transient pulses to ground, so specific resonant frequencies could be a problem. If it's just a 'keep the rails low impedance so the circuit remains stable' bypass, then specific resonant frequencies may not be important.
MLCC SMD small as possible case for a particular value, COG for the very small values X7R for larger values, inductance is the killer here.
Just laying out a design with some AD8620ARZ, LMH6553MR and ADS8471 amongst other devices and there used on this design and basically every other high spec design I have worked on uses mainly X7R's with occasional COGs for crystals and RF sections.
A decoupling capacitor provides the required power at switching, so low values nearest the pins larger values further away, via from power planes to small caps to IC pins, minimal track length, wide as possible track.
Through hole are not a lot of use these days lead length is to long.
In fact with MLCC you can get a low value and a 10u reservoir next to the op-amp pins great for high Mhz analogue layouts.
Just laying out a design with some AD8620ARZ, LMH6553MR and ADS8471 amongst other devices and there used on this design and basically every other high spec design I have worked on uses mainly X7R's with occasional COGs for crystals and RF sections.
A decoupling capacitor provides the required power at switching, so low values nearest the pins larger values further away, via from power planes to small caps to IC pins, minimal track length, wide as possible track.
Through hole are not a lot of use these days lead length is to long.
In fact with MLCC you can get a low value and a 10u reservoir next to the op-amp pins great for high Mhz analogue layouts.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.