Fifty years ago analog computers were faster (i.e. shorter turnaround time) than digital computers. Later digital computers won out based on increased precision. I've always had a bias towards analog computers. They were called *ANALOG* computers for a reason. [For the youngsters, resistance, capacitance, inductance, voltage and current were analogs of the physical parameters of the system being simulated.] For the models I'm currently interested in SPICE or ECAP of the 70's would be overly complex overkill. The versions I've had contact with put much effort into transient analysis and piecewise linear approximations of active devices. My world is much simpler. All passive components are linear and described by a positive scalar. My only active component is an ideal op-amp. When simulation begins, all inputs have been 0 since -infinity. A quick Google search shows that academics and hobbyists have been doing related things. e.g. https://courses.cit.cornell.edu/ece576/DDA/ http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf I doubt I'm looking for a commercial product as I'm on a hobby budget and my model would be too simple to be useful to a commercial audience. Comments? Suggestions of appropriate keywords to search? TIA
Digital emulation of "Analog Computers"
Started by ●June 17, 2013
Reply by ●June 17, 20132013-06-17
On Mon, 17 Jun 2013 12:10:01 -0500, Richard Owlett <rowlett@pcnetinc.com> wrote:>Fifty years ago analog computers were faster (i.e. shorter >turnaround time) than digital computers. Later digital computers >won out based on increased precision. > >I've always had a bias towards analog computers. They were called >*ANALOG* computers for a reason. [For the youngsters, resistance, >capacitance, inductance, voltage and current were analogs of the >physical parameters of the system being simulated.] > >For the models I'm currently interested in SPICE or ECAP of the >70's would be overly complex overkill. The versions I've had >contact with put much effort into transient analysis and >piecewise linear approximations of active devices. > >My world is much simpler. >All passive components are linear and described by a positive scalar. >My only active component is an ideal op-amp. >When simulation begins, all inputs have been 0 since -infinity. > >A quick Google search shows that academics and hobbyists have >been doing related things. >e.g. https://courses.cit.cornell.edu/ece576/DDA/ > >http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf > >I doubt I'm looking for a commercial product as I'm on a hobby >budget and my model would be too simple to be useful to a >commercial audience. > >Comments? Suggestions of appropriate keywords to search? > >TIAAre you looking for analog simulation software? The most straightforward analog simulator that I'm aware of is MultiSim, and there's a free downloadable version available at Analog Devices website: http://www.analog.com/en/content/glp_Amps_NIMultisim/fca.html?gclid=CM73mbfM67cCFRHhQgodpg8ArQ It's pretty easy to use and has all the basic stuff, and a lot of Analog Devices parts in the part library. For some things I've found the models surprisingly accurate even for subtle effects in corner cases. It's pretty straightfoward to hook a bunch of op-amps together in this tool and build the equivalent of an analog computer. I think you can still get working analog computers for not extravagant amounts of money on ebay and elsewhere. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●June 17, 20132013-06-17
>On Mon, 17 Jun 2013 12:10:01 -0500, Richard Owlett ><rowlett@pcnetinc.com> wrote: > >>Fifty years ago analog computers were faster (i.e. shorter >>turnaround time) than digital computers. Later digital computers >>won out based on increased precision. >> >>I've always had a bias towards analog computers. They were called >>*ANALOG* computers for a reason. [For the youngsters, resistance, >>capacitance, inductance, voltage and current were analogs of the >>physical parameters of the system being simulated.] >> >>For the models I'm currently interested in SPICE or ECAP of the >>70's would be overly complex overkill. The versions I've had >>contact with put much effort into transient analysis and >>piecewise linear approximations of active devices. >> >>My world is much simpler. >>All passive components are linear and described by a positive scalar. >>My only active component is an ideal op-amp. >>When simulation begins, all inputs have been 0 since -infinity. >> >>A quick Google search shows that academics and hobbyists have >>been doing related things. >>e.g. https://courses.cit.cornell.edu/ece576/DDA/ >> >>http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf >> >>I doubt I'm looking for a commercial product as I'm on a hobby >>budget and my model would be too simple to be useful to a >>commercial audience. >> >>Comments? Suggestions of appropriate keywords to search? >> >>TIA > >Are you looking for analog simulation software? > >The most straightforward analog simulator that I'm aware of is >MultiSim, and there's a free downloadable version available at Analog >Devices website: > >http://www.analog.com/en/content/glp_Amps_NIMultisim/fca.html?gclid=CM73mbfM67cCFRHhQgodpg8ArQ > >It's pretty easy to use and has all the basic stuff, and a lot of >Analog Devices parts in the part library. For some things I've found >the models surprisingly accurate even for subtle effects in corner >cases. > >It's pretty straightfoward to hook a bunch of op-amps together in this >tool and build the equivalent of an analog computer. > >I think you can still get working analog computers for not extravagant >amounts of money on ebay and elsewhere. > > >Eric Jacobsen >Anchor Hill Communications >http://www.anchorhill.com >Also check out Linear Technologies, LTSpice, its free from their website and works good and has what you need. _____________________________ Posted through www.DSPRelated.com
Reply by ●June 17, 20132013-06-17
jacobfenton wrote:>> On Mon, 17 Jun 2013 12:10:01 -0500, Richard Owlett >> <rowlett@pcnetinc.com> wrote: >> >>> Fifty years ago analog computers were faster (i.e. shorter >>> turnaround time) than digital computers. Later digital computers >>> won out based on increased precision. >>> >>> I've always had a bias towards analog computers. They were called >>> *ANALOG* computers for a reason. [For the youngsters, resistance, >>> capacitance, inductance, voltage and current were analogs of the >>> physical parameters of the system being simulated.] >>> >>> For the models I'm currently interested in SPICE or ECAP of the >>> 70's would be overly complex overkill. The versions I've had >>> contact with put much effort into transient analysis and >>> piecewise linear approximations of active devices. >>> >>> My world is much simpler. >>> All passive components are linear and described by a positive scalar. >>> My only active component is an ideal op-amp. >>> When simulation begins, all inputs have been 0 since -infinity. >>> >>> A quick Google search shows that academics and hobbyists have >>> been doing related things. >>> e.g. https://courses.cit.cornell.edu/ece576/DDA/ >>> >>> http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf >>> >>> I doubt I'm looking for a commercial product as I'm on a hobby >>> budget and my model would be too simple to be useful to a >>> commercial audience. >>> >>> Comments? Suggestions of appropriate keywords to search? >>> >>> TIA >> >> Are you looking for analog simulation software? >> >> The most straightforward analog simulator that I'm aware of is >> MultiSim, and there's a free downloadable version available at Analog >> Devices website: >> >> http://www.analog.com/en/content/glp_Amps_NIMultisim/fca.html?gclid=CM73mbfM67cCFRHhQgodpg8ArQ >> >> It's pretty easy to use and has all the basic stuff, and a lot of >> Analog Devices parts in the part library. For some things I've found >> the models surprisingly accurate even for subtle effects in corner >> cases. >> >> It's pretty straightfoward to hook a bunch of op-amps together in this >> tool and build the equivalent of an analog computer. >> >> I think you can still get working analog computers for not extravagant >> amounts of money on ebay and elsewhere. >> >> >> Eric Jacobsen >> Anchor Hill Communications >> http://www.anchorhill.com >> > > Also check out Linear Technologies, LTSpice, its free from their website > and works good and has what you need. >They both look like excessive overkill. Analog devices wants to put me on a sales pitch mailing list - not specified devices for ~30 years. It appears Linear Technologies is invites the hobby market to download - I'll download the next time I go to the library (I'm on dial up). One of the sites gave starting point of chain of references leading to http://sourceforge.net/projects/ngspice whose Windows version evidently has a Tcl/Tk interface which could be convenient.
Reply by ●June 18, 20132013-06-18
I've wondered why there aren't more analog computers in certain situations, especially where power consumption is an issue. You can make a multiplier with a couple of transistors instead of hundreds or thousands. I guess it always comes down to precision. Many years ago I read an interview with Carver Mead in which he claimed processing would move back toward analog, but I don't think this has happened anywhere. I saw a neat analog computer in a museum once. It was meant for integrating curves. You traced a pen over a function drawn/printed on paper and it charged up a capacitor and gave you a readout (probably on a needle) of the integral.
Reply by ●June 18, 20132013-06-18
On Tue, 18 Jun 2013 09:54:46 -0700 (PDT), Kevin Neilson <kevin.neilson@xilinx.com> wrote:>I've wondered why there aren't more analog computers in certain situations,= > especially where power consumption is an issue. You can make a multiplier= > with a couple of transistors instead of hundreds or thousands. I guess it= > always comes down to precision. Many years ago I read an interview with C= >arver Mead in which he claimed processing would move back toward analog, bu= >t I don't think this has happened anywhere. =20 > >I saw a neat analog computer in a museum once. It was meant for integratin= >g curves. You traced a pen over a function drawn/printed on paper and it c= >harged up a capacitor and gave you a readout (probably on a needle) of the = >integral.I've noticed a bit of a rennaisance lately for analog computers. As you mention, power consumption can be a Big Deal in many applications, as well as latency, so analog processing can have some critical advantages. I think flexibility requirements usually lead one to prefer a digital system, though. Here's a hotrod analog processor that can do Fourier Transforms: http://www.nasa.gov/topics/technology/features/analog-microchip.html Things like noise-cancelling headphones are almost always done analog, and there's sort of a philosophical war in the hearing aid area of whether digital or analog is "superior". Analog computing still has its place, I think, and that place may be growing. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Reply by ●June 18, 20132013-06-18
Richard Owlett wrote:> Fifty years ago analog computers were faster (i.e. shorter > turnaround time) than digital computers. Later digital computers > won out based on increased precision. > > I've always had a bias towards analog computers. They were called > *ANALOG* computers for a reason. [For the youngsters, resistance, > capacitance, inductance, voltage and current were analogs of the > physical parameters of the system being simulated.] > > For the models I'm currently interested in SPICE or ECAP of the > 70's would be overly complex overkill. The versions I've had > contact with put much effort into transient analysis and > piecewise linear approximations of active devices. > > My world is much simpler. > All passive components are linear and described by a positive scalar. > My only active component is an ideal op-amp. > When simulation begins, all inputs have been 0 since -infinity. > > A quick Google search shows that academics and hobbyists have > been doing related things. > e.g. https://courses.cit.cornell.edu/ece576/DDA/ > > http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf > > I doubt I'm looking for a commercial product as I'm on a hobby > budget and my model would be too simple to be useful to a > commercial audience. > > Comments? Suggestions of appropriate keywords to search? > > TIAOne of the other analog simulators from the 60's was Pactolus written in Fortran. It was simple and worked quite well. The modern simulators are capable of delivering a lot more detail and much better results. w..
Reply by ●June 19, 20132013-06-19
On Tuesday, June 18, 2013 5:10:01 AM UTC+12, Richard Owlett wrote:> Fifty years ago analog computers were faster (i.e. shorter > > turnaround time) than digital computers. Later digital computers > > won out based on increased precision. > > > > I've always had a bias towards analog computers. They were called > > *ANALOG* computers for a reason. [For the youngsters, resistance, > > capacitance, inductance, voltage and current were analogs of the > > physical parameters of the system being simulated.] > > > > For the models I'm currently interested in SPICE or ECAP of the > > 70's would be overly complex overkill. The versions I've had > > contact with put much effort into transient analysis and > > piecewise linear approximations of active devices. > > > > My world is much simpler. > > All passive components are linear and described by a positive scalar. > > My only active component is an ideal op-amp. > > When simulation begins, all inputs have been 0 since -infinity. > > > > A quick Google search shows that academics and hobbyists have > > been doing related things. > > e.g. https://courses.cit.cornell.edu/ece576/DDA/ > > > > http://code.google.com/p/propforth/downloads/detail?name=PropFacs.f.rtf > > > > I doubt I'm looking for a commercial product as I'm on a hobby > > budget and my model would be too simple to be useful to a > > commercial audience. > > > > Comments? Suggestions of appropriate keywords to search? > > > > TIAYou can emulate an analogue computer with packages such as SimuLink or by using say LabView data-flow language. If you understand signal-flow graphs it is dead easy. of course it is not really analogue but near enough provided you sample high enough. I do this with Trapedzoidal integration and it works great.
Reply by ●June 19, 20132013-06-19
Eric Jacobsen wrote:> [snip] > > Here's a hotrod analog processor that can do Fourier Transforms: > > http://www.nasa.gov/topics/technology/features/analog-microchip.html >Two questions 1. How would an "FFT" be done in analog domain except as a bank of band pass filters? [I suspect I'm missing something ;] 2. Any "preliminary data sheet" on the NASA device?
Reply by ●June 19, 20132013-06-19
gyansorova@gmail.com wrote:> [snip] > > You can emulate an analogue computer with packages such as SimuLink or by using say LabView data-flow language. If you understand signal-flow graphs it is dead easy. of course it is not really analogue but near enough provided you sample high enough. I do this with Trapedzoidal integration and it works great. >As I'm on a hobby budget, they would be out of reach. However your post reminded me of Scilab which I have. I'd have to rethink my problem definition, but that is likely to be beneficial in and of itself.
