Obscure Electronics Topics

[D from BC]

D from BC [email protected] posted to sci.electronics.design:


I don't suppose that you have tried looking for three or four winding
SMD transformers?

Not yet...
But would that SMD transformer be 1206 size?
Tapped 100uH.
Small signal current.


D from BC

 

[Tim Williams]

I am pretty curious about the physical understanding / application of
the 3D FFT.

Hmm, sounds like the kind of integral you'd do in quantum mechanics!

Tim

 

[Vladimir Vassilevsky]

JosephKK wrote:

I am pretty curious about the physical understanding / application of
the 3D FFT.

The understanding is pretty simple: imagine a cube where the FFT is
computed along X, Y, Z. This works very good for the applications like
video compression, where the third dimension is time. The 3d FFT
approach is capable of better compression then the traditional schemes
like MPEG, and the computing workload is significantly lower. However in
MPEG the decompressor requires somewhat 100 times less of processing,
then the compressor, whereas in the 3d FFT approach the decompressor
demands are comparable to that of the compressor. So the decoders for
MPEG are cheap and simple, and all of the complexity is on the encoder
side. This is better for the typical consumer video applications.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[JosephKK]

D from BC [email protected] posted to sci.electronics.design:

Not yet...
But would that SMD transformer be 1206 size?
Tapped 100uH.
Small signal current.


D from BC

Nobody else will know until someone looks and posts back here.

Spehro Pefhany found something that may be similar.

 

[JosephKK]

Tim Williams [email protected] posted to sci.electronics.design:

Hmm, sounds like the kind of integral you'd do in quantum mechanics!

Tim

Hmmm. Possibly in analyzing crystal properties.

 

[JosephKK]

Vladimir Vassilevsky [email protected] posted to
sci.electronics.design:

JosephKK wrote:



The understanding is pretty simple: imagine a cube where the FFT is
computed along X, Y, Z. This works very good for the applications
like video compression, where the third dimension is time. The 3d
FFT approach is capable of better compression then the traditional
schemes like MPEG, and the computing workload is significantly
lower. However in MPEG the decompressor requires somewhat 100 times
less of processing, then the compressor, whereas in the 3d FFT
approach the decompressor demands are comparable to that of the
compressor. So the decoders for MPEG are cheap and simple, and all
of the complexity is on the encoder side. This is better for the
typical consumer video applications.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Nice. I was thinking also along 3D raster image manufacturing.

 

[foobar]

I haven't looked very much...but suspect it's a nonexistent animal.
I have to add one more detail...
That would be a tapped SMD chip inductor sharing the same core..
It's like a chip autotransformer.
Not 2 individual inductors in series.


D from BC

Well, there are SMT common mode chokes intended for removing common mode
noise on differential lines. If you connected the dot side of one coil to
the anti-dot side of the other coil, you would have a center tapped
autotransformer.

Depending on what you are trying to do, those might serve. They are quite
small.

Example:
Murata DLW21HN670SQ2L

Digikey has them.

 

[D from BC]

Well, there are SMT common mode chokes intended for removing common mode
noise on differential lines. If you connected the dot side of one coil to
the anti-dot side of the other coil, you would have a center tapped
autotransformer.

Depending on what you are trying to do, those might serve. They are quite
small.

Example:
Murata DLW21HN670SQ2L

Digikey has them.

Well whatdayahknow....
0805 size

But I'm not sure how to go from common mode impedance to inductance...
The common mode impedance is 67 ohm at 100Mhz.
I'll guess 67/2 = 2*Pi*100Mhz*L

If so...that's ~53nH...Perhaps that's why it's so small.

I'll look for similar CM xformers at 100uH.

Thanks..
D from BC

 

[foobar]

Well whatdayahknow....
0805 size

But I'm not sure how to go from common mode impedance to inductance...
The common mode impedance is 67 ohm at 100Mhz.
I'll guess 67/2 = 2*Pi*100Mhz*L

If so...that's ~53nH...Perhaps that's why it's so small.

I'll look for similar CM xformers at 100uH.

Thanks..
D from BC

Yeah, it is dicey to assume that the impedance is purely inductive.

Murata also has parts in a bigger package up to 2200 Ohms at 100 MHz. I
still don't think that will get you where you need to be.

I think TDK also makes them.

In the end, I suspect Win's advice is the best. ;-)

--Mac