I needed a 45MHz filter yesterday, but a few MHz wide, wider than an MCF. So, I was looking at doing an LC filter, but then found some off-the-shelf SAWs that would do the trick better than anything I could put together. BUT, then I noticed the insertion loss -- like 21dB!! ..so anyway, I'm doing an LC filter after all. But I am wondering, how on earth do you design with something like this?? For a tiny SMT device, 21db of insertion loss just seems insane. Doesn't MORE of the signal couple across the device exterior parasitically..? Am I missing something? How are these filters supposed to be used? Thanks.
SAW insertion loss
Started by ●July 19, 2007
Reply by ●July 19, 20072007-07-19
Jon Mcleod wrote:> I needed a 45MHz filter yesterday, but a few MHz wide, wider than an > MCF. So, I was looking at doing an LC filter, but then found some > off-the-shelf SAWs that would do the trick better than anything I could > put together. BUT, then I noticed the insertion loss -- like 21dB!! > > ..so anyway, I'm doing an LC filter after all. > > But I am wondering, how on earth do you design with something like > this?? For a tiny SMT device, 21db of insertion loss just seems insane. > Doesn't MORE of the signal couple across the device exterior > parasitically..? Am I missing something? How are these filters > supposed to be used?Jon, A factor of eleven in gain doesn't seem like too much to pay for good characteristics. Crystal filters, though once widely used, are similar. As for parasitic coupling around the device, good layout should pass much less than 9%. One pair of my sunglasses are ND 1. That's -20 dB. I'm curious about your asking this in a digital signal processing forum. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●July 19, 20072007-07-19
Jerry Avins wrote:> I'm curious about your asking this in a digital signal processing forum.I'm working with DDS's on this project, so I thought of this group. Usually a good samaritan like yourself will answer here even if the question is not absolutely dead on topic. But in my own defense, a SAW filter is really an FIR filter, isn't it? Thanks.
Reply by ●July 19, 20072007-07-19
Jerry Avins wrote:> Jon Mcleod wrote: > >> I needed a 45MHz filter yesterday, but a few MHz wide, wider than an >> MCF. So, I was looking at doing an LC filter, but then found some >> off-the-shelf SAWs that would do the trick better than anything I >> could put together. BUT, then I noticed the insertion loss -- like >> 21dB!!BTW I don't know what kind of filter OP is looking for, however speaking of several MHz bandwidth a TV IF filter comes to my mind.>> ..so anyway, I'm doing an LC filter after all. >> >> But I am wondering, how on earth do you design with something like >> this?? For a tiny SMT device, 21db of insertion loss just seems >> insane. Doesn't MORE of the signal couple across the device exterior >> parasitically..? Am I missing something? How are these filters >> supposed to be used? > > > Jon, > > A factor of eleven in gain doesn't seem like too much to pay for good > characteristics. Crystal filters, though once widely used, are similar. > As for parasitic coupling around the device, good layout should pass > much less than 9%. One pair of my sunglasses are ND 1. That's -20 dB.Agreed. Somewhat -10dB would be typical, and -20dB is not very unusual.> I'm curious about your asking this in a digital signal processing forum.No wonder. This is one of not so many EE places in usenet which is not totally ocupied with 555 timer problems and bullshit. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●July 19, 20072007-07-19
Jon Mcleod wrote:> Jerry Avins wrote: > >> I'm curious about your asking this in a digital signal processing forum. > > I'm working with DDS's on this project, so I thought of this group. > Usually a good samaritan like yourself will answer here even if the > question is not absolutely dead on topic. > > But in my own defense, a SAW filter is really an FIR filter, isn't it?SAW stands for "surface acoustic wave". A SAW filter is an analog device that consists of (usually two) patterned electrodes on a piezo-electric substrate. Calculating the dimensions and shapes of the launcher and catcher electrodes -- each is usually interdigitated -- is related at a deep level to calculating the values of FIR coefficients, but it is a separate art. SAWs, like FIRs, can convert impulses to chirps and vice versa. That could lead one to think that they are more alike than they really are. A quick web search turned up http://www3.sympatico.ca/colin.kydd.campbell/ You may fine something more apt. An important part of the art of SAW design is a treatment for the edge of the piezo (usually ceramic) chip to suppress reflections. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●July 20, 20072007-07-20
On Jul 19, 5:52 pm, Jon Mcleod <jonmcleod2...@yahoo.com> wrote:> Jerry Avins wrote: > > I'm curious about your asking this in a digital signal processing forum. > > I'm working with DDS's on this project, so I thought of this group. > Usually a good samaritan like yourself will answer here even if the > question is not absolutely dead on topic. > > But in my own defense, a SAW filter is really an FIR filter, isn't it? > > Thanks.SAW filters behave like windowed bandpass filters. SAW practitioners have made extensions to window design by generalizing Hann and Hamming windows (1) and improving upon the approximations made in the design of the Blackman family of windows(2). (1) The Classical Truncated Cosine Series Functions with Applications to SAW Filters Malocha, D.C. Bishop, C.D. Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions on Jan 1987, Volume: 34, Issue: 1, p75- 85 (2) Improved sidelobe performance of cosine series functions Kulkarni, R.G. Lahiri, S.K. Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions on Mar 1999, Volume: 46, Issue: 2, p464-466 Dale B. Dalrymple http://dbdimages.com
Reply by ●July 20, 20072007-07-20
Hi Jerry, Jerry Avins wrote:> Jon Mcleod wrote: >> I needed a 45MHz filter yesterday, but a few MHz wide, wider than an >> MCF. So, I was looking at doing an LC filter, but then found some >> off-the-shelf SAWs that would do the trick better than anything I >> could put together. BUT, then I noticed the insertion loss -- like >> 21dB!! >> >> ..so anyway, I'm doing an LC filter after all. >> >> But I am wondering, how on earth do you design with something like >> this?? For a tiny SMT device, 21db of insertion loss just seems >> insane. Doesn't MORE of the signal couple across the device exterior >> parasitically..? Am I missing something? How are these filters >> supposed to be used? > > Jon, > > A factor of eleven in gain doesn't seem like too much to pay for good > characteristics. Crystal filters, though once widely used, are similar. > As for parasitic coupling around the device, good layout should pass > much less than 9%. One pair of my sunglasses are ND 1. That's -20 dB. > > I'm curious about your asking this in a digital signal processing forum.Do you know of a surface acoustic signal processing forum in which to ask? :-) Steve
Reply by ●July 20, 20072007-07-20
Jerry Avins wrote:> Jon Mcleod wrote: >> Jerry Avins wrote: >> >>> I'm curious about your asking this in a digital signal processing forum. >> >> I'm working with DDS's on this project, so I thought of this group. >> Usually a good samaritan like yourself will answer here even if the >> question is not absolutely dead on topic. >> >> But in my own defense, a SAW filter is really an FIR filter, isn't it? > > SAW stands for "surface acoustic wave". A SAW filter is an analog device > that consists of (usually two) patterned electrodes on a piezo-electric > substrate. Calculating the dimensions and shapes of the launcher and > catcher electrodes -- each is usually interdigitated -- is related at a > deep level to calculating the values of FIR coefficients, but it is a > separate art. > > SAWs, like FIRs, can convert impulses to chirps and vice versa. That > could lead one to think that they are more alike than they really are. A > quick web search turned up http://www3.sympatico.ca/colin.kydd.campbell/ > You may fine something more apt. An important part of the art of SAW > design is a treatment for the edge of the piezo (usually ceramic) chip > to suppress reflections.SAWs are not "like" FIRs. They *are* FIRs. They are just implemented in an analogue way. Sure there are fudging requirements to optimise them, because the ends of the fingers don't cut off dead, but that isn't a whole lot different to fudging a digital FIR to allow for truncation. Steve
Reply by ●July 20, 20072007-07-20
Steve Underwood wrote:> Hi Jerry,...> Do you know of a surface acoustic signal processing forum in which to > ask? :-)That's a good point, Steve. besides, as Jon so kindly observed, we're a bunch of nice guys with a pretty good pool of knowledge. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●July 20, 20072007-07-20
Steve Underwood wrote: ...> SAWs are not "like" FIRs. They *are* FIRs. They are just implemented in > an analogue way. Sure there are fudging requirements to optimise them, > because the ends of the fingers don't cut off dead, but that isn't a > whole lot different to fudging a digital FIR to allow for truncation.Not only do end and edge effects need to be dealt with in SAWs, the finger spacing needn't be uniform. the freedom to alter the spacing opens design possibilities not possible with clocked FIRs. In a way, each type is a subset of the other. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
