On Wednesday, April 6, 2011 10:49:48 PM UTC-4, Vladimir Vassilevsky wrote:> Clay wrote: > > > > Most definitely optical processing has and is being done. > > Before the era of DSPs, the synthetic aperture radars used to do the > processing by optical means; that is recording the radar data to > photographic film and then exposing it like a hologram.There were some additional tricks. One of them was a conical lens that was part of the processing that converted the side-looking viewpoint to an as-seen-from-above map. Jerry -- Engineering is the art of making what you want from things you can get.
hey, have you guys ever heard of this?
Started by ●April 6, 2011
Reply by ●April 7, 20112011-04-07
Reply by ●April 7, 20112011-04-07
Clay wrote:> Most definitely optical processing has and is being done.No question about it. The earliest analog FFT patent was 1971: http://www.freepatentsonline.com/3581078.html Quite a few processors have been built, for instance: http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA057057 I seem to recall an analog chip in the early 80's which was designed for analog FFT in mind, but I don't recall the part number or manufacturer offhand. Kevin McGee
Reply by ●April 7, 20112011-04-07
On Apr 6, 10:49=A0pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:> Clay wrote: > > Most definitely optical processing has and is being done. > > Before the era of DSPs, the synthetic aperture radars used to do the > processing by optical means; that is recording the radar data to > photographic film and then exposing it like a hologram. > > Vladimir Vassilevsky > DSP and Mixed Signal Design Consultanthttp://www.abvolt.comIn fact when Gabor invented holography in about 1947, his object and reference beams were coaxial. It wasn't until 1963 when Lieth and Upatnieks applied the idea of side looking radar to the holography problem and created off axis holography. This has a great advantage that the image is no longer corrupted by the reference illumination light. Clay
Reply by ●April 7, 20112011-04-07
On Thursday, April 7, 2011 12:51:42 AM UTC-4, kevin wrote: ...> I seem to recall an analog chip in the early 80's which was designed > for analog FFT in mind, but I don't recall the part number or > manufacturer offhand.FFT, or FT? Jerry -- Engineering is the art of making what you want from things you can get.
Reply by ●April 7, 20112011-04-07
On Thu, 7 Apr 2011 06:58:03 -0700 (PDT), Clay <clay@claysturner.com> wrote:>On Apr 6, 10:49=A0pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote: >> Clay wrote: >> > Most definitely optical processing has and is being done. >> >> Before the era of DSPs, the synthetic aperture radars used to do the >> processing by optical means; that is recording the radar data to >> photographic film and then exposing it like a hologram. >> >> Vladimir Vassilevsky >> DSP and Mixed Signal Design Consultanthttp://www.abvolt.com > >In fact when Gabor invented holography in about 1947, his object and >reference beams were coaxial. It wasn't until 1963 when Lieth and >Upatnieks applied the idea of side looking radar to the holography >problem and created off axis holography. This has a great advantage >that the image is no longer corrupted by the reference illumination >light. > >ClaySAR is not the same thing as holography, though. SAR can be (and has been) processed optically, but a signal flow diagram of the optical processor is essentially the same as it is for a digital processor doing 2-D fast correlation. FT -> multiply by mask -> IFT then repeat in the other dimension. Eric Jacobsen http://www.ericjacobsen.org http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php
Reply by ●April 7, 20112011-04-07
On 4/7/2011 8:00 AM, Jerry Avins wrote:> On Thursday, April 7, 2011 12:51:42 AM UTC-4, kevin wrote: > > ... > >> I seem to recall an analog chip in the early 80's which was designed >> for analog FFT in mind, but I don't recall the part number or >> manufacturer offhand. > > FFT, or FT? > > Jerryor SA? Fred
Reply by ●April 7, 20112011-04-07
On Apr 6, 2:46=A0pm, Fred Marshall <fmarshallxremove_th...@acm.org> wrote:> ... > > I thought that the Spectral Dynamics spectrum analyzers of the 70's were > analog but the recent web page cites an "FFT analyzer" of 1968 - which > is likely the device I'm referring to.Some of the early analog spectrum analyzers from Spectral Dynamics were analog sampled data systems that used CCDs to record samples and allow multiple faster than real time playback for the direct calculation of the convolutions in the chirp-Z transform which was used for the spectrum analysis.> > The FFT was published in 1965. =A0Microprocessors started being generally > useful in the early 70's. =A0So, I rather doubt there was an FFT box as > early as 1968. =A0These things were a box just a bit bigger than the CRT > display that they used.In the 1970's high performance dynamic signal analyzers, microprocessors were not up to speed for the signal processing core. They were used for operator interface functions like reading switch positions and setting bits in hardware control registers based on switch positions. For a typical DSP implementation see the block diagrams in: http://www.freepatentsonline.com/3638004.pdf> > Maybe Dale or someone else remembers.... > FredDale B. Dalrymple
Reply by ●April 8, 20112011-04-08
On Thursday, April 7, 2011 12:51:42 AM UTC-4, kevin wrote: ...>> I seem to recall an analog chip in the early 80's which was designed >> for analog FFT in mind, but I don't recall the part number or >> manufacturer offhand.>FFT, or FT?>JerryFFT. Basically, the chip was an analog equivalent of a Programmable Logic Device (PLD). Instead of gates and registers, they had op amps that could be configured to add/subtract/multiply. I don't remember how they stored things, but they had a delay line like capability and ways of switching them in or out. They had a couple of design examples with it, and one of them was an analog version of a small FFT. (I hope this anwers Fred's question too) Kevin McGee
Reply by ●April 8, 20112011-04-08
On Thursday, April 7, 2011 11:56:04 PM UTC-4, kevin wrote:> On Thursday, April 7, 2011 12:51:42 AM UTC-4, kevin wrote: >=20 > ... >=20 > >> I seem to recall an analog chip in the early 80's which was designed > >> for analog FFT in mind, but I don't recall the part number or > >> manufacturer offhand. >=20 >=20 > >FFT, or FT? >=20 > >Jerry >=20 > FFT. Basically, the chip was an analog equivalent of a Programmable > Logic Device (PLD). Instead of gates and registers, they had op amps > that could be configured to add/subtract/multiply. I don't remember > how they stored things, but they had a delay line like capability and > ways of switching them in or out. They had a couple of design > examples with it, and one of them was an analog version of a small > FFT. >=20 > (I hope this anwers Fred's question too)Thanks. I can see how to implement the butterflies with op-amps, too. The t= widdle-factor gain can be hard wired with resistors, while the sums and dif= ferences are natural operations for op-amps. Bit-reversed order is not a p= roblem if the bits are routed with wires. Intermediate storage isn't needed= . Just apply the voltage samples to the inputs, wait for all stages to sett= le, and read the outputs. With care, one might do a 16-point FFT with accep= table accuracy. For more stages, I would turn to DDAs. Analog differential analyzers, some = of which incorporate ball-and-disk integrators, predate Lord Kelvin. A digi= tal implementation can be used as a digital op-amp. It ought to be possible= to build a DDA-based dedicated FFT chip that is faster than a computer usi= ng the same technology. Something to think about! Jerry --=20 Engineering is the art of making what you want from things you can get.
Reply by ●April 8, 20112011-04-08
Jerry Avins <jya@ieee.org> wrote: (snip)> Thanks. I can see how to implement the butterflies with op-amps, > too. The twiddle-factor gain can be hard wired with resistors, > while the sums and differences are natural operations for op-amps. > Bit-reversed order is not a problem if the bits are routed with wires. > Intermediate storage isn't needed. Just apply the voltage samples to > the inputs, wait for all stages to settle, and read the outputs. > With care, one might do a 16-point FFT with acceptable accuracy.You could also arrange some analog shift registers (CCD devices) to move the values around, to get them to the OP amp inputs at the right time, and take the outputs away. Also, that way you can get the output serially, such as for a display. (snip) -- glen
