Hacker News new | past | comments | ask | show | jobs | submit login
After 36 years as a paid product, the Micro-Cap Circuit Simulator is now free (spectrum-soft.com)
414 points by lightlyused 14 days ago | hide | past | web | favorite | 80 comments

The website says that the company was closed.

Does anyone have an insight on why Spectrum Software is now closed?

P.S. The man behind the company, Andy Thompson, seems to be an excessively humble person. He left a lot of details in the mist. However, all these details are priceless to us mere humans. Memoirs? A blog? 39 years in business is not a small feat.

Perhaps his humility is the most important lesson to learn, esp. in the dominant culture of tech today.

And maybe the true payments were the friends he made along the way...

There needs to be a name for this kind of dead end aphorism thinking.

Cliche? Cheese?

“The true payments,” haha.

After 39 years the authors might have just retired.

Maybe their customer base was too small and their reference market too saturated to convince anyone to buy it.

This is very exciting! Especially the SPICE model library. Microcap's libraries have lots of SPICE models that I haven't seen anywhere else. Being able to grab them will be awesome, even if I wind up using them with LTSPICE.

I’m not familiar with microcap, but I am familiar with LTSpice. If they’re both free SPICE GUIs, surely microcap must be better than LTSpice. I can’t imagine any interface worse than LTSpice.

More than just GUIs, rather SPICE compatible simulators. LTSpice has macro models for most LT parts, that would not be compatible with MC. Likewise MC probably has models that are not compatible with LTSpice.

Now I wish LTSpice would take MC GUI and features, and implement it with their simulator. They should have bought it.

LTSpice for OS X is a special kind of frustrating.

I really hope someone mirrors this on github or something, because there is no guarantee that a website for a closed company will be around for any length of time.

I lost some important records because yahoo shut down it's group archive last year (and banned the archive team from saving stuff wtf), and it's been on my mind lately.

We (ArchiveTeam) are still trying to archive portions of Yahoo Groups (our methods have changed a bit since Yahoo turned off the web archive in December, but the data isn't deleted until Jan 31).

What's the name of the group?

If you're a member of the group (or can join it now) it's only a few clicks to make a request for the data. Doesn't get absolutely everything but all the emails/posts should be included.


40 years isn’t bad for a small software shop. I hope Mr. Thompson is satisfied with how things turned out.

Anyone here use the software?

I used this in 1991 as an EE undergrad. I initially used it on my 8086, 4.77 MHz XT. It took several minutes to simulate a simple common emitter amp. I don’t remember if I had an 8087 FPU.

I then bought a 486DX, 50 MHz and the same simulation was finished before the mouse button lifted.

Fun times. Microcap was really cool at the time as it was the only GUI based SPICE, and the student version was under $50.

Nowadays I use LTSpice specifically since LT makes good switching regulators, and only develops those models for LTSpice. Otherwise I use Microwave Office.

MC still looks much more polished than LTSpice, at least the GUI. Schematic entry in LTSpice is still abysmal.

Yes also LTspice has some horrible clunk and bugs in it.

How does this compare to modern circuit simulators?

https://www.tinkercad.com/ (Web)

http://www.virtualbreadboard.com/ (Windows)

It doesn’t. Those are not professional level simulators.

MC is a modern simulator, and professional level. I assume the guy is retiring. You are not going to be simulating large ICs, but it’s still plenty good for other stuff.

There's so pretty much seemingly unique proprietary software. A while ago I found this awesome logic minimizer called "logic Friday". [1] I don't think there's a free or open source version of a tool like this.

I have an idea that the "espresso" algorithm could be used to minimize not only electronic circuits but general boolean expressions for any programming language... I think it would do for a useful refactoring tool.

1: https://web.archive.org/web/20131022051516/http://www.sontra...

I like Helmut Neemann's Digital: https://github.com/hneemann/Digital

It simulates logic, supports automated testing, simulates and analyses combinatorial and sequential logic, comes with a large library of components (generic stuff, specific 7400 logic, displays and memories, etc), it can output VHDL or Verilog, and it can export JEDEC files for GALs.

Looks great, I'll give it a try! Looks like it implements a different minimization algorithm [1].

1: https://en.wikipedia.org/wiki/Quine%E2%80%93McCluskey_algori...

Quine McCluskey is equivalent to using Karnaugh maps. Both are basic digital logic minimization workflows, and are taught in any course covering digital logic or CS-focused discrete math.

This seems more interesting as it is open sourced, support current hardware etc.

> I have an idea that the "espresso" algorithm could be used to minimize not only electronic circuits but general boolean expressions for any programming language... I think it would do for a useful refactoring tool.

I think the goal for most boolean expressions in code is for the reader to be able to conceptually grasp what's being checked for, not for the expression to be as short as possible. Distributive properties can make an expression shorter while simultaneously making the concepts behind the expression much more obscure.

It's useful in the compiler. For instance one optimization I haven't seen but would like to is going from

  if( value1 < 0 || value2 < 0 || value3 < 0 ) {

  temp = value1 | value2 | value3;
  if( temp < 0 ) {
Taking a sort of hardware, Boolean logic view makes this possible, as VHDL generally does this when synthesed.

It seems like Clang is capable of doing this optimization:

> https://godbolt.org/z/oCQNbq

Neat! It didn't a few years ago.

This is most of the inner loop of a barycentric coordinate rasterizer, so I was pretty concerned about micro-optimizations when I was checking.

As a compiler writer I can tell you that this is a complex optimisation with very limited use.

N.B. the || (logical or) operator is an "early out" operator whereas the | (binary or) operator is NOT

The compiler would need to look for special cases of value1, value2 and value3... such as: are any of them volatile, perhaps one refers to a memory mapped I/O device, perhaps one if actually a function call, or array dereference, or pointer dereference, what about value1 being a "char" which needs to be promoted before it can be or'd, what about any of the "values" being a constant < 0 which would short circuit the "if" statement and remove the need for any tests.

As I said, this is not as simple as it first appears.

Note, if the three comparisons are sufficiently predicable, then your transformation would be _less_ efficient than the simple branches.

A predicable branch is almost free in a modern processor.

Three comparison/branch pairs are not free.

Clang will optimize

  bool swap_if(bool c, 
      int& a, int& b) {
    int ta = a, tb = b;
    a = (-c & tb)|((c-1) & ta);
    b = (-c & ta)|((c-1) & tb);
    return c;
into cmov instructions. Used in a quicksort partition loop as

  l += swap_if(
    *r <= pv, *l, *r);
It makes quicksort fully 2x as fast as the usual branch.

That optimization fudges the difference between boolean or (comparing truth values) and bitwise or. It happens to work under the safe assumption that value1, value2, and value3 use sign bits (as all two's-complement integers do), but logically it's a disaster.

monocasa isn't saying this should be in source code, but rather the compiler should create machine code that does this. Compilers know the types, so they will know whether they're signed (if they're not signed, then < 0 can be optimized out as impossible). Compilers have no goal of creating assembly that would look good in source code.

Good point, although the idea would be you get to choose one for or the other. I think sometimes the simplified expression makes it easier to see the complement (else branch).

btw, I started to think about "boolean refactoring" when I saw J. Edwards' schematic tables video [1].

1: https://vimeo.com/140738254

+1 for Logic Friday.

I had fun with that, it is/was quite nice to use. I wonder what happened to it. The "sontrak.com" domain is bust now it seems.

The about page has this in the copyright though:

> *Espresso and misII are copyright © 1988-1993, Regents of the University of California.

You should take a look at Z Notation, which is a formal specification language based around mathematics & set theory.

I hated working in it and have done my best to forget everything about it.

I've used Logic Friday once to make a wired remote controller device that operated on 74-logic. I haven't really seen any tools that offer more user-friendly ergonomics for that type of problem, except maybe a curve-fitting tool called Eureqa which used to be open-source and went closed-source. Or perhaps Wolfram Mathematica could be used to spit out a few solutions, again closed-source.

> There's so pretty much seemingly unique proprietary software.

Indeed. Something to consider when choosing a license and distribution model for your own projects, including web-based software. Your life is finite, but the life of your software may not be.

I have never heard of this, is it appropriate to ask if anyone could describe it in one sentence?

See the "Product Info" section: http://www.spectrum-soft.com/demo.shtm

And there's a Feature Tour with screenshots: http://www.spectrum-soft.com/demo/schemati.shtm

Thank you for asking - I was a little afraid to ask as well.

It's software for simulating circuits.

At the moment I use Linux (Didn't see any Linux ports?) with ngSpice, I haven't found ngSpice that useful though

I found ngspice extremely useful. I have compared the results with Cadence Virtuoso and the results are usually identical or with negligeble difference. I am trying to use a combination of ngspice and magic for IC design.

Micro-Cap is Windows-only software (it runs on Linux via Wine though) and perhaps more accessible than your current (which probably is "make circuits somewhere, generate netlist, simulate with ngspice, analyze somewhere else") workflow as it has an integrated schematic editor. The most powerful Linux alternative to Micro-Cap et al is probably KiCad.

Thanks, I will stick to KiCad and ngspice (when I need it) I am a hobbyist and dont have complicated needs. Not a fan of Wine.

Free as in beer. It will have been ever better if it had been released as free software.

Is this good for analog stuff, at least as effective as LTSpice?

I've been using the latter on a Mac for simulating vacuum tube (UK: valve) circuits with some success and a very large amount of frustration. I would love something a little less actively user-hostile...

What makes LTSpice seem so "actively user-hostile" to you?

(Founder of https://www.circuitlab.com/ (YC W13), an analog circuit simulator that many universities have now started using.)

I didn't grow up with it, so coming at it from a perspective of someone who's gotten used to modern software, it's remarkably hard to learn.

The interface is weird. Placement of components isn't so bad but when you want to do stuff like move them or rotate, you fall into this strange mode system that's unlike any other software I've ever used.

Including models is done by writing arcane text commands on the diagram. And also by setting parameters in a hidden window on the component diagram itself.

Models != visual components.

Sometimes I have to set the model designator for a component in the UI twice before it'll "take". That's hostile!

Finickity pin alignment on custom/3rd party components sometimes leads to open circuits when they look closed.

The parameter/settings windows are cryptic.

The wire drawing tool is really nice though.

It's obviously better than writing a setlist in TextEdit but I've found the learning curve very steep, with all the underlying complexity of Spice exposed. The fairly prehistoric interface paradigm that means that any muscle memory and expectations from using any other graphical software just don't help. In fact, they hurt.

I tried your thing when I was looking for circuit design software. I liked it. It didn't have any vacuum tube models at the time (not that I blame you, it's niche) and so I couldn't use it for what I needed.

It’s not quite abandonware because Cadence acquired and killed it, but look for CircuitMaker 2000. It’s out there.

Best schematic editor I’ve found. It also has tubes.

Those all make sense. Thanks!

Happy to help. I'll keep an eye on Circuitlab too - thanks for the reminder of that.

Does your tool support custom key commands and macros (also can I use it out of a browser)?

Last time I played with it that was what got me to leave. A drag and drop wysiwyg is wholly inappropriate for serious work. Also the search feature doesn’t work well, it stalled looking for “op amp”

And by “serious” work I mean a small circuit with two dozen elements.

Open source possible?


Any chance of it becoming FOSS?

Wow, used this extensively in my undergrad. Always used the free version. I remember occasionally running into the "too many nodes" error. Thanks to the creator for releasing the full version for free. I'm sure many EE undergrads will be grateful.

It seems to be working with Wine under GNU/Linux.

I'll try it better when I get home.

free as in free beer

I pinched the title from https://news.ycombinator.com/item?id=20495077 since it's more informative. If this is inaccurate, please let us know. (Submitted title was "Micro-Cap User Downloads – Now Free".)

I'd love to try it out, but I'm not downloading an executable from an `http` URL.

Edit: I hope the downvotes are because it supports `https` (just not by default, thanks to progman32 for correcting me) and not because you think that not wanting to download an executable over an open connection is an unreasonable thing.

How is downloading a executable over https safer? It mitigates man-in-the-middle attacks that could modify the compressed artefact on the fly, but that's possibly the least likely attack on a download imaginable. It's far more likely that an attacker would try to replace the file on the server (that way they can also change and documentation around the file, like the reported MD5/SHA hash you might use to check the file is correct). Why would you happily download that over https?

If you're security conscious enough to not download random executables over http then you really should be aware that it's 99% as dangerous to download them over any link.

On most networks, anyone else on the same subnet as your IP can hijack your connection using ARP spoofing and send you whatever they want if you're connecting over Http instead of HTTPS. That's usually a lot easier to pull off than compromising the server and replacing the content there.

Something to think about for anyone still putting Google and Amazon devices on the same WiFi network as their phone and other PCs.

Ok. Don't.

Looking a gift horse in the mouth?

You would download an executable over an unencrypted connection? Whether it's free or not?

That’s what digital signatures are for. So yes.

yes. it's amusing how many people confuse transport layer security with application layer security.

I've published signed debian packages to PPAs, so I'm well aware.

My question was in relation to a unsigned executable on an unencrypted http site, as the OP site loads by default. Would you download and run one?

I'd wager that a large number of the system integrators, firmware engineers and random developers scattered throughout the production chains of most of the hardware and some of the software that you're currently using have run completely untested code in scenarios orders of magnitude more hair-raising than this.

The same is probably equivalently applicable to the hardware that was used to compile the compiler that compiled the compiler that compiled the compiler that you're using.

While it's theoretically possible that a hostile party could modify the package in-transit over HTTP/TCP, it seems far more likely that the hacked package would be placed on the "legit-looking" webserver (by hacking the webserver). In this case, looking to the SSL transport layer as any kind of credibility enhancement to the unsigned binary is worse than ignoring it.

Therefore, an unsigned binary is an unsigned binary, no matter the transport mechanism. I agree that distributing unsigned binaries is poor security practice, but I also think that it is dangerous to think that the transport of an unsigned binary over an SSL connection gives it any credibility.

So.. you haven't heard of automated MITM executable-patching (infecting) solutions? I believe I read about them a couple of years ago.

Once implemented, it's much easier than hacking servers and more convenient to do targeted, semi-targeted, local network/cafe script-kiddie attacks, without it being easily detected. Unfortunately for attackers, these days people don't download and run unverified executables as often, especially over http, so you may need lots of patience if you want do infect a specific person.

I covered that in the portion of the comment where I mentioned http/tcp

I would really love some data (or good reasoning) on how server attacks are overwhelmingly more likely, so much so that the false security impression increases risk.

MITM executable patching attacks are not theoretical. AFAIU, the first hit on "mitm executable infection" [1] and an interceptor (ARP/wifi/whatever) is all a script kiddie needs.

[1] https://n0where.net/mitm-pe-file-infector-peinjector

It's a good point.

I got: http://www.spectrum-soft.com/download/mc12cd.zip sha3-256: 1e9c7d1ec04019446fa448fec74af36f53eaf6508def75068eb32ae0d7f5109a

Not sure what the drama is about... The installer is clean.


Haha, good one!

How do you obtain an trusted signature over http?

Fair enough, I gladly use apt for the same reason. But I mean download unsigned executables over an unencrypted connection, like this site is by default. You do it?

Site supports https.

You could spin up a virtual machine and try it out on there.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact