When are ferrite beads the most approproate component to use?
in [Design]
|
Prev: Idiots
Next: Looking for photo frame that does GIF...
From: John Larkin on 27 Dec 2009 12:42 On Sun, 27 Dec 2009 07:15:28 -0600, "mook johnson" <mook(a)mook.net> wrote: > >"John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message >news:sbndj5hvn62d2je8jo8tkvoi0scv9tvkb0(a)4ax.com... >> On Sat, 26 Dec 2009 21:33:16 -0600, "mook johnson" <mook(a)mook.net> >> wrote: >> >>>I've never found an application where a I felt a ferrite bead as head ahd >>>shoulders over a resistor or some other means filtering the signal. Most >>>of >>>the develpments Im involved with are power conversion and data acquisition >>>(ADCs and DSPS)from various sensors. >>> >>>For lowish frequency stuff (say sub-10MHz) in what applications would a >>>ferrite bead be the bees knees? >>> >>>Just want to make sure I'm not overlooking a useful electrical component. >>> >>>thanks guys. >>> >>> >> >> They're great for keeping high frequency stuff out of opamps and such. >> They have low DC resistance, so don't mess up low frequency circuits, >> and they have miserable Qs, so kill resonances. >> >> ftp://jjlarkin.lmi.net/Ferrite.JPG >> >> This bead reduced the RFI sensitivity of a thermocouple front-end by >> 20 dB or so, specifically by killing some wiring resonances in the >> 100-300 MHz range. >> >> Surface-mount beads are also good, cheap power supply decouplers. Most >> of them act like lossy inductors in roughly the 5 uH range. >> >> John > > >Thanks John, > >What processes do you use to predict the action of applying a ferrite bead. >Is is all cut and try? Frankly, it's more like wearing garlic to keep away vampires. When you suspect that a circuit will have noise coupling problems on Vcc rails, or that an opamp might rectify RF from the outside world, or emit RF and fail compliance testing, you design in some beads and some bypass caps. When it works, you congratulate yourself, and never know if it was really necessary. To be fair, most of us have found real situations where beads helped, and have applied them liberally since. Surface-mount beads are cheap insurance, a few cents each, and can be used to isolate and/or measure supply currents, an added bonus. Lots of opamps have ghastly supply-noise rejection at high frequencies. We recently got bit by one that had *gain* from V- supply to output, datasheet to the contrary. We generally use ferrites and big ceramic caps to isolate low-level stages like photodiode amps, and use beads or monolithic filters (essentially beads+caps in a block) between the outside world and low-level stages that might rectify RF. > >Can you spice it in a fairly straight forward way? >Any other analysis techniques? This doesn't analyze very well. Part models usually do a very bad job of simulation IC supply behavior, both current noise kicked out and supply rejection. And wiring/PCB resonances are major EMI issues, which Spice doesn't model well even if you can furnish the correct circuit. About all Spice can do is verify the gross rolloff and check for low-frequency resonances. We usually don't bother. John
From: Jim Thompson on 27 Dec 2009 12:49 On Sun, 27 Dec 2009 09:42:18 -0800, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >On Sun, 27 Dec 2009 07:15:28 -0600, "mook johnson" <mook(a)mook.net> >wrote: > >> [snip] >> >>What processes do you use to predict the action of applying a ferrite bead. >>Is is all cut and try? > >Frankly, it's more like wearing garlic to keep away vampires. [snip] > >Lots of opamps have ghastly supply-noise rejection at high >frequencies. We recently got bit by one that had *gain* from V- supply >to output, datasheet to the contrary. [snip] >John That "gain" is actually quite common... which is why PSRR, on data sheets, is always referred to input ;-) ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | Help save the environment! Please dispose of socialism responsibly!
From: krw on 27 Dec 2009 13:03 On Sun, 27 Dec 2009 09:42:18 -0800, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >On Sun, 27 Dec 2009 07:15:28 -0600, "mook johnson" <mook(a)mook.net> >wrote: > >> >>"John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message >>news:sbndj5hvn62d2je8jo8tkvoi0scv9tvkb0(a)4ax.com... >>> On Sat, 26 Dec 2009 21:33:16 -0600, "mook johnson" <mook(a)mook.net> >>> wrote: >>> >>>>I've never found an application where a I felt a ferrite bead as head ahd >>>>shoulders over a resistor or some other means filtering the signal. Most >>>>of >>>>the develpments Im involved with are power conversion and data acquisition >>>>(ADCs and DSPS)from various sensors. >>>> >>>>For lowish frequency stuff (say sub-10MHz) in what applications would a >>>>ferrite bead be the bees knees? >>>> >>>>Just want to make sure I'm not overlooking a useful electrical component. >>>> >>>>thanks guys. >>>> >>>> >>> >>> They're great for keeping high frequency stuff out of opamps and such. >>> They have low DC resistance, so don't mess up low frequency circuits, >>> and they have miserable Qs, so kill resonances. >>> >>> ftp://jjlarkin.lmi.net/Ferrite.JPG >>> >>> This bead reduced the RFI sensitivity of a thermocouple front-end by >>> 20 dB or so, specifically by killing some wiring resonances in the >>> 100-300 MHz range. >>> >>> Surface-mount beads are also good, cheap power supply decouplers. Most >>> of them act like lossy inductors in roughly the 5 uH range. >>> >>> John >> >> >>Thanks John, >> >>What processes do you use to predict the action of applying a ferrite bead. >>Is is all cut and try? > >Frankly, it's more like wearing garlic to keep away vampires. When you >suspect that a circuit will have noise coupling problems on Vcc rails, >or that an opamp might rectify RF from the outside world, or emit RF >and fail compliance testing, you design in some beads and some bypass >caps. When it works, you congratulate yourself, and never know if it >was really necessary. To be fair, most of us have found real >situations where beads helped, and have applied them liberally since. A few of us have found situations where they really hurt, too. I don't use them in supplies unless there is a reason to suspect problems. >Surface-mount beads are cheap insurance, a few cents each, and can be >used to isolate and/or measure supply currents, an added bonus. We use 0-ohm resistors in front of all regulators. We recently replaced a few of them with beads when one of the other regulators (a "brick") was spewing 200MHz everywhere. The beads at least got it out of the other supplies. >Lots of opamps have ghastly supply-noise rejection at high >frequencies. We recently got bit by one that had *gain* from V- supply >to output, datasheet to the contrary. You betcha. We're having problems finding a decent headphone driver now. The one we're using, a RoHS replacement, seems to detect 2.4GHz quite well. While we can't see anything, I suspect it's the input bipolars rectifying the 2.4GHz. It could also be on the power supplies, or both. Though we've put "y" supressors on the power supplies with limited success (sometimes it's there, sometimes not). >We generally use ferrites and big ceramic caps to isolate low-level >stages like photodiode amps, and use beads or monolithic filters >(essentially beads+caps in a block) between the outside world and >low-level stages that might rectify RF. > >> >>Can you spice it in a fairly straight forward way? >>Any other analysis techniques? > >This doesn't analyze very well. Part models usually do a very bad job >of simulation IC supply behavior, both current noise kicked out and >supply rejection. And wiring/PCB resonances are major EMI issues, >which Spice doesn't model well even if you can furnish the correct >circuit. About all Spice can do is verify the gross rolloff and check >for low-frequency resonances. We usually don't bother. Yes, it's basically "FM", particularly when the carrier can't be seen.
From: Paul Keinanen on 27 Dec 2009 14:46 On Sun, 27 Dec 2009 09:42:18 -0800, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >And wiring/PCB resonances are major EMI issues, >which Spice doesn't model well even if you can furnish the correct >circuit. About all Spice can do is verify the gross rolloff and check >for low-frequency resonances. We usually don't bother. As a rule of thumb anything longer than about 1/10 wavelength should be treated as a transmission line section. At 2.4 GHz, the free space wavelength is 13 cm, so 1/10 wavelength would be 1.3 cm and considering the velocity factor on conventional PCB materials, structures larger than 1 cm should be treated as transmission line sections. Even if the device is intended for DC/audio applications, the EMC filtering should be designed as an RF circuit.
From: Joerg on 27 Dec 2009 15:01
mook johnson wrote: > "John Larkin" <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote in message > news:sbndj5hvn62d2je8jo8tkvoi0scv9tvkb0(a)4ax.com... >> On Sat, 26 Dec 2009 21:33:16 -0600, "mook johnson" <mook(a)mook.net> >> wrote: >> >>> I've never found an application where a I felt a ferrite bead as head ahd >>> shoulders over a resistor or some other means filtering the signal. Most >>> of >>> the develpments Im involved with are power conversion and data acquisition >>> (ADCs and DSPS)from various sensors. >>> >>> For lowish frequency stuff (say sub-10MHz) in what applications would a >>> ferrite bead be the bees knees? >>> Essentially anything where you cannot have an added DC-resistance. Such as coil driver outputs, speaker outputs, DC power outputs and the like. >>> Just want to make sure I'm not overlooking a useful electrical component. >>> >>> thanks guys. >>> >>> >> They're great for keeping high frequency stuff out of opamps and such. >> They have low DC resistance, so don't mess up low frequency circuits, >> and they have miserable Qs, so kill resonances. >> >> ftp://jjlarkin.lmi.net/Ferrite.JPG >> >> This bead reduced the RFI sensitivity of a thermocouple front-end by >> 20 dB or so, specifically by killing some wiring resonances in the >> 100-300 MHz range. >> >> Surface-mount beads are also good, cheap power supply decouplers. Most >> of them act like lossy inductors in roughly the 5 uH range. >> >> John > > > Thanks John, > > What processes do you use to predict the action of applying a ferrite bead. > Is is all cut and try? > Look up the datasheet and there you will find a curve that lists equivalent impedance over frequency. Depending on the ferrite material this can be very different. > Can you spice it in a fairly straight forward way? > Any other analysis techniques? > Nah, just those datasheet curves and a serious lop-off for DC load because Z drops rather quickly with DC load. I always try to oversize for DC by a factor of two to four. I have salvaged many, many projects for clients that had capsized at the EMC lab, using mostly ferrite beads. Sometimes those 3c-parts can be like miracle medicine. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM. |