Home |
Search |
Today's Posts |
#1
![]() |
|||
|
|||
![]()
"MARTHA" is a unique circuit analysis software package that was
originally developed at MIT in the early 1970's. Rather than a "canned" analysis package that only performs pre-programmed circuit analysis operations, it is a circuit-description extension to the APL computer programming language. It will perform analysis of linear networks like other CAD packages, but by coupling it with APL, it can be used to perform custom analysis and synthesis of circuits that are difficult or impossible to do in other CAD environments. MARTHA was originally marketed as a mainframe time-share product. It was used at MIT for teaching circuit theory, and by engineers at MIT Lincoln Laboratory to design RF & microwave circuits for radar and communications satellites. Lincoln Laboratory used the capabilities of MARTHA to create circuit synthesis programs (mostly for microstrip and LC filters), in addition to expanding the analysis capabilities of MARTHA. MARTHA never caught on as a commercial product, but it continued to thrive at Lincoln Lab even after competing PC products like Touchstone became available. MARTHA and the other Lincoln software were ported to the PC, and additional features (like better graphics) were added. All of the Lincoln software was cleaned up and documented to create LLAMA (Lincoln Laboratory Advanced MARTHA Applications), which is still in use today. MIT Prof. Paul Penfield (MARTHA's original author) and Doug White (prime mover behind the LLAMA effort) have obtained permission to release MARTHA, LLAMA and the associated manuals into the public domain. The initial release is set up to run using a free DOS-based APL interpreter. The APL section of the web site includes this software, along with installation instructions and documentation on using APL. Additional information and all of the necessary files are available at the web site: http://www.marthallama.org. MARTHA is NOT for everyone. It doesn't have fancy schematic capture or PCB layout features. It also requires learning at least a little about the APL computer language (which could be considered a worthwhile exercise in its own right). There is no canned installation routine that will magically get it all up & running on your computer, but there are procedures & batch files to try to make it as painless as possible. Doug White |
#2
![]() |
|||
|
|||
![]()
On Dec 2, 6:22 pm, (Doug White) wrote:
"MARTHA" is a unique circuit analysis software package that was originally developed at MIT in the early 1970's. Rather than a "canned" analysis package that only performs pre-programmed circuit analysis operations, it is a circuit-description extension to the APL computer programming language. It will perform analysis of linear networks like other CAD packages, but by coupling it with APL, it can be used to perform custom analysis and synthesis of circuits that are difficult or impossible to do in other CAD environments. MARTHA was originally marketed as a mainframe time-share product. It was used at MIT for teaching circuit theory, and by engineers at MIT Lincoln Laboratory to design RF & microwave circuits for radar and communications satellites. Lincoln Laboratory used the capabilities of MARTHA to create circuit synthesis programs (mostly for microstrip and LC filters), in addition to expanding the analysis capabilities of MARTHA. MARTHA never caught on as a commercial product, but it continued to thrive at Lincoln Lab even after competing PC products like Touchstone became available. MARTHA and the other Lincoln software were ported to the PC, and additional features (like better graphics) were added. All of the Lincoln software was cleaned up and documented to create LLAMA (Lincoln Laboratory Advanced MARTHA Applications), which is still in use today. MIT Prof. Paul Penfield (MARTHA's original author) and Doug White (prime mover behind the LLAMA effort) have obtained permission to release MARTHA, LLAMA and the associated manuals into the public domain. The initial release is set up to run using a free DOS-based APL interpreter. The APL section of the web site includes this software, along with installation instructions and documentation on using APL. Additional information and all of the necessary files are available at the web site: http://www.marthallama.org. MARTHA is NOT for everyone. It doesn't have fancy schematic capture or PCB layout features. It also requires learning at least a little about the APL computer language (which could be considered a worthwhile exercise in its own right). There is no canned installation routine that will magically get it all up & running on your computer, but there are procedures & batch files to try to make it as painless as possible. Doug White Thanks to you and to Prof. Penfield for your work in making these programs available to us. I'll set aside time to do a proper installation (the web pages make it sound like I'll need to pay attention to what I'm doing...). This is kind of a "blast from the past." I recall using MARTHA a few times back around 1972 on our APL terminal. We also used APL for exactly what your web pages say it's good for: quick development of what in Fortran of the day would have been ugly, since we had problem statements already in matrix form. Cheers, Tom |
#4
![]() |
|||
|
|||
![]()
From: (Doug White) on Sun, Dec 2 2007 6:22 pm
MARTHA is NOT for everyone. It doesn't have fancy schematic capture or PCB layout features. It also requires learning at least a little about the APL computer language (which could be considered a worthwhile exercise in its own right). There is no canned installation routine that will magically get it all up & running on your computer, but there are procedures & batch files to try to make it as painless as possible. That's interesting that the 35-year-old A Programming Language (APL) is still around somewhere. I learned of it way back when in '72 while employed by RCA Corporation...and getting interested in programming after a very satisfactory and productive introduction to 'LECAP,' RCA's frequency-domain version of the original ECAP from IBM. However, today's desktop PCs have MORE speed, mass memory, RAM storage than any mainframe computer of 35 years ago at less than $1K US new off-the-shelf (today's Fry's Electronics ad had a PC with LCD monitor for $400). We don't really need Interpreter-based high-level languages now. My circa-1979 Apple ][+ needed those to work with its fantastic clock rate of a whole MegaHertz and 48 KBytes of RAM! My 4- year-old PC box with 1 GHz clock rate, 100 MHz RAM access rate, 2 MB RAM, and small 40 GB hard disk can handle 50-node SPICE circuits with perhaps 2000 time increments in a couple of seconds...and store the results in a file. shrug Three decades ago the UC Berkeley group came up with SPICE and made the core of that program FREE, no restrictions (source code available for the cost of paper reproduction and mailing). SPICE derivatives are the electronics industry standard today. Linear Technology Corporation has a working, auto-installable, schematic capture enhanced (or manual netlist entry for die-hards) absolutely FREE. "LTSpice" available at Linear Technology's website as a single download. Many commonly-used semiconductor component models are available as part of that. A tube-oriented audiophile website in the UK even has several vacuum tube SPICE models (they call tubes 'valves' - :-) for LTSpice. As a long-time homebrewer and electronics experimenter (since 1947), a knowledge of programming languages is not a priority to homebrewing; that is a separate subject, interesting in itself, but not germane. A full-blown, working, easier-to-use SPICE derivative that if FREE is more suited (I think) to homebrewers. I've used LTSpice myself, compared its calculations to actual, working hardware and find NO discrepancies. SPICE derivative programs all allow netlist descriptions of component characteristics in algebraic form plus a shorthand that is common to all SPICE forms for special Models. There is no need to learn FORTRAN (SPICE core's original language) or versions written in C++ (which are available now), or any other high-level language. For my programming needs today, Assembler-level source code for PIC or ATMEL microcontrollers is more germane to homebrewing. Some of those are also FREE. At least two PCB fabrication companies have FREE PCB layout programs available without ordering any boards. Transferring a schematic from LTSpice to theirs is just a schemtic redrawing task, usually a one-shot task. I don't mean to diminish any importance of programming languages or any different CADs or CAEs, but, let's face it, APL is an old high-level language. I still have the 100 W American Beauty soldering iron I got in 1948 but it is out-of-date for soldering parts on a PCB now. 73, Len AF6AY |
#5
![]() |
|||
|
|||
![]()
AF6AY wrote:
From: (Doug White) on Sun, Dec 2 2007 6:22 pm MARTHA is NOT for everyone. It doesn't have fancy schematic capture or PCB layout features. It also requires learning at least a little about the APL computer language (which could be considered a worthwhile exercise in its own right). There is no canned installation routine that will magically get it all up & running on your computer, but there are procedures & batch files to try to make it as painless as possible. That's interesting that the 35-year-old A Programming Language (APL) is still around somewhere. I learned of it way back when in '72 while employed by RCA Corporation...and getting interested in programming after a very satisfactory and productive introduction to 'LECAP,' RCA's frequency-domain version of the original ECAP from IBM. However, today's desktop PCs have MORE speed, mass memory, RAM storage than any mainframe computer of 35 years ago at less than $1K US new off-the-shelf (today's Fry's Electronics ad had a PC with LCD monitor for $400). We don't really need Interpreter-based high-level languages now. I guess the guys that program in perl, javascript, python, lisp, and apl, among many many others, should just quit? ;-) .... Three decades ago the UC Berkeley group came up with SPICE and made the core of that program FREE, no restrictions (source code available for the cost of paper reproduction and mailing). SPICE derivatives are the electronics industry standard today. Spice is an interpreter. .... I don't mean to diminish any importance of programming languages or any different CADs or CAEs, but, let's face it, APL is an old high-level language. Unix is an old operating system, yet it still seems to get a whole lot of use. APL is the ultimate programmable calculator. It is a beautiful language that can do beautiful things in a very succinct manner. Perhaps if you actually had used it, you might think more of APL. -Chuck |
#6
![]() |
|||
|
|||
![]()
On Dec 5, 6:16�pm, Chuck Harris wrote:
AF6AY wrote: From: (Doug White) on Sun, Dec 2 2007 6:22 pm Spice is an interpreter. Whatever. My point was as a homebrewer using a tool as an aid to building something. SPICE and its derivatives (all individual 'wrappings' on the Berkeley SPICE core) are VERY FAST. They work for me just as SPICE works for thousands of other circuit designers, both pro and amateur (I am both). I don't mean to diminish any importance of programming languages or any different CADs or CAEs, but, let's face it, APL is an old high-level language. Unix is an old operating system, yet it still seems to get a whole lot of use. �APL is the ultimate programmable calculator. �It is a beautiful language that can do beautiful things in a very succinct manner. Everyone has their 'favorite' high-level language, each one saying that Their language is the best, most beautiful, and other fancy sayings, complete with all sorts of academic praises and plaudits. My new HP-35S isn't as pretty as the HP-32S II which is also on my computer table and the programming commands aren't quite compatible. Esthetics aside, I wouldn't trade either one of them for SMALL programming tasks. What I want are the numbers from the results so that hardware can be completed. Both do that very nicely for what I want. Perhaps if you actually had used it, you might think more of APL. Perhaps if you had actually used LTSpice, a FREE download from Linear Technology and actually built some circuits using the LTSPice results you might think more of it. shrug Your message appeared in rec.radio.amateur.homebrew and also sci.electronics.cad. I am replying from homebrew. I'm not a programmer despite once having several years complementary membership in the ACM...or teaching myself FORTRAN IV from Dan McCracken's softcover book on the subject. I have MS FORTRAN 5.1 package, bought and paid for myself and have used it for a variety of different tasks...until MS dropped support of their product and also of similar products in later versions of Windows. shrug again If you want to get all arrogant about computer languages, please remove rec.radio.amateur.homebrew from your message routing. In the meanwhile I will continue to do my own homebrewing without going through even-more learning curves of old languages or old OSs just to be with the 'best' tool. Even the 'best' tools can make cruddy circuit calculations. 73, Len AF6AY |
#7
![]() |
|||
|
|||
![]()
AF6AY wrote:
Everyone has their 'favorite' high-level language, each one saying that Their language is the best, most beautiful, and other fancy sayings, complete with all sorts of academic praises and plaudits. My new HP-35S isn't as pretty as the HP-32S II which is also on my computer table and the programming commands aren't quite compatible. Esthetics aside, I wouldn't trade either one of them for SMALL programming tasks. What I want are the numbers from the results so that hardware can be completed. Both do that very nicely for what I want. Perhaps if you actually had used it, you might think more of APL. Perhaps if you had actually used LTSpice, a FREE download from Linear Technology and actually built some circuits using the LTSPice results you might think more of it. shrug I have, and I think it is a very nice tool. Perhaps the best of the available spices. But because it is not a free tool (eg. open source) I have to live with everything just the way that Mike Englehart wants it to be. That isn't a bad thing, but it is very limiting because one day Mike won't be there to support LTSpice anymore, and LT will decide that they haven't the funds to hire some new support, and it freeze. To cease being supported is to die in software land. MARTHA's source is open, and because anyone with the desire to support it can, it will live forever. Your message appeared in rec.radio.amateur.homebrew and also sci.electronics.cad. I am replying from homebrew. I'm not a programmer despite once having several years complementary membership in the ACM...or teaching myself FORTRAN IV from Dan McCracken's softcover book on the subject. I have MS FORTRAN 5.1 package, bought and paid for myself and have used it for a variety of different tasks...until MS dropped support of their product and also of similar products in later versions of Windows. shrug again So, instead of shrugging about how MS wronged you by dropping support for an old fortran package, wander over to linux, or BSD, and run the open source f77 program, along with all the other open source goodies that have been made available for everyone to use free of charge.... including some damn nice implementations of programs for hamradio use. If you want to get all arrogant about computer languages, please Begging your pardon, but you were the one that felt the need to slam the MARTHA cad program and apl. The announcement you saw was a simple announcement of a gift to everyone of this package. That bothered you for some reason, so you felt the need to tell everyone to ignore it because apl was old, and anything old couldn't be any good. 73, Chuck |
#8
![]() |
|||
|
|||
![]()
"Chuck Harris" wrote in message
... I have, and I think it is a very nice tool. Perhaps the best of the available spices. But because it is not a free tool (eg. open source) I have to live with everything just the way that Mike Englehart wants it to be. He's certainly open to input from users, Chuck -- it's part of his job. No guarantees he'd add something you'd want, of course, but in my opinion Mike is going to be a lot more responsive to the average user than, say, Synopysys would be if you asked them to add something to HSPICE. That isn't a bad thing, but it is very limiting because one day Mike won't be there to support LTSpice anymore, and LT will decide that they haven't the funds to hire some new support, and it freeze. That's a rather pessimistic viewpoint. Worst case, LTSpice simply isn't developed any more, but it'll then always still be just as good as it is the day that happens. To cease being supported is to die in software land. Everyone and everything dies at some point... MARTHA's source is open, and because anyone with the desire to support it can, it will live forever. Oh, come on... open-source software is, if anything, more likely to die than most commercial software because there's usually no profit motive behind keeping it alive. I realize that it's not quite the same in that open-source software, even if "dead," can be "resurrected" at any time whereas that's often not the case with commercial software... but there's plenty of open-source software that's been "buried" for so long now the chance of anyone resurrecting it rather than just coming up with a new "baby" from scratch is remote. ---Joel |
#9
![]() |
|||
|
|||
![]()
Joel Koltner wrote:
"Chuck Harris" wrote in message ... I have, and I think it is a very nice tool. Perhaps the best of the available spices. But because it is not a free tool (eg. open source) I have to live with everything just the way that Mike Englehart wants it to be. He's certainly open to input from users, Chuck -- it's part of his job. Mike is all aces as far as I am concerned. His level of responsiveness is very close to as good as what I typically get from open source authors. No guarantees he'd add something you'd want, of course, but in my opinion Mike is going to be a lot more responsive to the average user than, say, Synopysys would be if you asked them to add something to HSPICE. Agreed. I have for many years seen a connection between software's price and the responsiveness of the company towards the customer: Synopsys charges a boat load, and they are not very responsive. LT charges nothing for LTSPice, and is so responsive that I would not be at all surprised if Mike Englehardt jumps into this thread. That isn't a bad thing, but it is very limiting because one day Mike won't be there to support LTSpice anymore, and LT will decide that they haven't the funds to hire some new support, and it will freeze. That's a rather pessimistic viewpoint. No, it isn't! It is a realistic viewpoint. I have been in this industry long enough (37+ years) to have seen this happen over and over again. It *will* happen with every single piece of commercial software ever written at some point... guaranteed, unless the owner decides to commit it to the public domain, or open source, like MARTHA's owner so generously did. Worst case, LTSpice simply isn't developed any more, but it'll then always still be just as good as it is the day that happens. It sure will, and just like good old DOS Orcad, you will have some people who keep around old legacy DOS systems just so they can use it. I have rather a lot of software that was written for Windows 95, that is no longer usable with NT, XP, or Vista. Am I supposed to keep a '95 box around just to run it? With open source, I just relink to the latest library, and I am back in the game. And what if I need to change it? I won't be able to make LTSpice do anything that it cannot currently do. That is bound to be a problem if I need to simulate flux-gate capacitors. Or want a better matrix solving algorithm than Mike knew to choose. To cease being supported is to die in software land. Everyone and everything dies at some point... The only way open source software can die is if it gets lost so badly that nobody can find it. This is unlikely, given the wide distribution that most of these packages have had. MARTHA's source is open, and because anyone with the desire to support it can, it will live forever. Oh, come on... open-source software is, if anything, more likely to die than most commercial software because there's usually no profit motive behind keeping it alive. MARTHA has already outlived most any other commercial software that was written in the same time frame. As long as the source code doesn't get lost, and *anyone* is interested in it, it will continue to survive. I have been running a quaint little editor that Jonathan Payne, of Sun JAVA fame wrote when he was a wet behind the ears kid in high-school. It was designed to run under unix on a pdp-11 with 64K-I, and 64K-D. I have ported(or simply used) it to(on) every operating system, and platform that I have used since he wrote it. Before jove, I was enamored by a nice little editor called edix. It was proprietary, and ran only under DOS. It died 25 years ago. Sure, I can cart its mangy carcass off to linux and run it under DOSEMU, but if I want to change anything about it I am out of luck. You say there is no profit motive, but that is where you are completely wrong. *I* profit from the open source software that I use. As long as that is true, I will see to it that the software I use is available on the systems that I am currently using. I realize that it's not quite the same in that open-source software, even if "dead," can be "resurrected" at any time whereas that's often not the case with commercial software... but there's plenty of open-source software that's been "buried" for so long now the chance of anyone resurrecting it rather than just coming up with a new "baby" from scratch is remote. Odds are pretty good that that new "baby" will have in some way benefited from the program that came before it. You might not be able to see the connection, but it is very often there. -Chuck |
#10
![]() |
|||
|
|||
![]()
Chuck Harris wrote:
Spice is an interpreter. SPICE makes sense as an interpreter. Compilers are best for programs you write once and use several times; interpreters are best for programs you run once but write several of. If you consider every change to a circuit to be a separate program you run once (which makes perfect sense when you consider you would be recompiling it every time you made a change), and you consider that once the circuit works to your liking you probably won't be running that simulation often, then an interpreter is perfect for the job and a compiler would be slow and inconvenient. -- A staffer for Democrat Senator Maria Cantwell of Washington has been arrested for trying to arrange a sexual tryst with a 13-year old boy. http://www.thesmokinggun.com/archive...72senate1.html |
Reply |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
![]() |
||||
Thread | Forum | |||
"Sirius wins "Fastest Growing Company" in Deloitte's 2007 Technology Fast 500" | Shortwave | |||
(OT) : "MM" Requests Any Responses Containing Parts Or All Of My Posts Have The "X-No-Archive:" In The First Line To Avoid Permanent Archiving. | Shortwave | |||
"meltdown in progress"..."is amy fireproof"...The Actions Of A "Man" With Three College Degrees? | Policy | |||
"Fan" capacitors for microwave RF design | Homebrew |