Multisim Live going EOL in September - building free alternative

[BrittEspinosa]

Hi everyone,

Former NIer here. Worked on their cloud software for several years, Multisim Live included. Was a bit surprised when they announced the tool's upcoming EOL in September. I'd talked with literally hundreds of professors using it throughout their courseware. NI's recommended migration path to Multisim Desktop (starting at $900/year, plus installs) is going to be a non-starter for a lot of them.

My buddy and I are building an alternative called CircuitSim. Schematic editor, SPICE editor, simulations, charting, and a custom component editor. We just launched an open beta, and we're aiming to have it solid for educators by late summer so they can use it with their classes in the fall.

We're both W2s working on this nights and weekends, so any feedback or ripping apart of the simulator, schematics, or component creator really helps us prioritize.
It's mainly aimed at students and hobbyists, not trying to replace professional tooling.

https://circuitsim.com/

[Hawaka]

UI looks nice and refine, but I fail to see the advantages against LTSpice. Maybe can you elaborate on why would someone / a school choose your product against a fully free solution?

[voltsandjolts]

As excellent as LTSpice is, as an engineering tool, its interface is very dated and quirky. We users are used to it, but young students would be quite horrified I'm sure. Just having a modern, intuitive UI could be reason enough for the success of this venture.

[SteveThackery]

Ah, dammit! Does it have to be browser-based? Makes it a non-starter for me.

[BrittEspinosa]

UI looks nice and refine, but I fail to see the advantages against LTSpice. Maybe can you elaborate on why would someone / a school choose your product against a fully free solution?

Fair question. Same reason so many folks used Multisim Live. No install needed is a big point of friction for classroom use. IT departments, students on Chromebooks, etc. As someone else pointed out here, LTspice's UI/UX is pretty dated. Being able to learn on an interface more in step with what people use in their everyday lives is a nice point of friction to remove. Browser-based also makes things like groups, classroom grading, and circuit sharing a lot easier. Like I mentioned in the OP, no delusions here of replacing professional tooling...targeted at students and hobbyists who are about to lose a tool that was doing it well.

Ah, dammit! Does it have to be browser-based? Makes it a non-starter for me.

Ha! Right now it does, we were just discussing what it would look like to make a local version. We'll probably do it after we feel the browser based version is stable.

--

Saw abunch of you signed up! Thank you!

[Hawaka]

Any idea on pricing ? Will there be a free version at all? I do sometimes use Falstad, which is not terrible, but not great. This could have the potential to replace it.

[harerod]

As excellent as LTSpice is, as an engineering tool, its interface is very dated and quirky. We users are used to it, but young students would be quite horrified I'm sure. Just having a modern, intuitive UI could be reason enough for the success of this venture.

Heck, I was horrified back in 1992, when I had to learn that UI. No use to go soft on the modern engineering student. 
At that time my main machine was still a souped-up Atari ST.

For LTspice I had bought a used 486SX laptop and added a 487SX. Dunno, has the user interface evolved since then?
I also ran Matlab and Simulink on that laptop. I can't remember if LTspice even needed the math-coprocessor?

[harerod]

Ah, dammit! Does it have to be browser-based? Makes it a non-starter for me.

It's alright, we simply may not be among the target group. No hard feelings. Look at us silly old farts, insisting on local installations that run offline indefinitely...

[free_electron]

qspice , or Multi-Cap (freeware now, was top-notch commercial stuff. owner retired).

[voltsandjolts]

qspice , or multisim (freeware now, was top-notch commercial stuff. owner retired).

You meant Micro-Cap
https://en.wikipedia.org/wiki/Micro-Cap
also this
https://en.wikipedia.org/wiki/List_of_free_electronics_circuit_simulators

[Zero999]

As excellent as LTSpice is, as an engineering tool, its interface is very dated and quirky. We users are used to it, but young students would be quite horrified I'm sure. Just having a modern, intuitive UI could be reason enough for the success of this venture.

I love the LTSpice UI. I see it as an advantage over the other software I was forced to use at college, such as Electronics Workbench which I believe was bought out by Multisim.

If LTSpice's UI is not your thing, then how about making an alternative UI for LTSpice? LTSpice can be run without the GUI. It shouldn't be difficult to make a program which generates a netlist, simulates it using LTSpice, then displays the results.

[tooki]

Ah, dammit! Does it have to be browser-based? Makes it a non-starter for me.

Trust me, in education being browser-based is a huge competitive advantage. Education IT departments are loathe to install anything — if it were up to them, everyone would have a Chromebook and that’s it.

And I say this as a former IT person who holds nothing but contempt for IT nazis dictators who mistakenly think it is their job to proscribe how the actual users need to do their work.

When I was at vocational school a few years ago, the lab PCs — the ones we had to use — had to run Visual Studio through a goddamned VirtualBox VM because IT considered it too insecure to run locally.  Basically everything that wasn’t Firefox and MS Office was so locked down it barely worked. TINA (the circuit simulator) would sometimes just break. Same with Keil for MCU programming.

In fact, I started using a web-based compiler instead of the slow, buggy Visual Studio hack.

I should reach out to my old teacher and see if he is still using TINA in class or if he’s been able to switch to TINA Web. (It was still missing too many features a few years ago.)

P.S. As much as I am impressed by what the modern Web can do, I now pine for the olden days of simple static Web pages that didn’t download frameworks larger than the entire hard disk of the first Mac I bought back in 1993-94 — the first computer I used to surf the web on. I hate the pig the Web has been turned into…

[Unixon]

My buddy and I are building an alternative called CircuitSim.

If it's not GPL etc. and not on Github then it's not an alternative.
Another proprietary (cloud?) product... 

[BrittEspinosa]

Any idea on pricing ? Will there be a free version at all? I do sometimes use Falstad, which is not terrible, but not great. This could have the potential to replace it.

We haven't figured out pricing yet. There will always be a free tier, and I'm pretty set on making sure whatever we charge is better than what Multisim Live was ($9.99/month).

[free_electron]

My biggest gripe with all these simulators is finding accurate models that don't take shortcuts. I'd like to see transistor-level models. not a bunch of u/i sources...
And digital gates with the power and ground brought out, so i can tie those to the rails i want.

[BrittEspinosa]

My biggest gripe with all these simulators is finding accurate models that don't take shortcuts. I'd like to see transistor-level models. not a bunch of u/i sources...
And digital gates with the power and ground brought out, so i can tie those to the rails i want.

We expose the SPICE netlist in both the circuit editor and the custom component editor. The built-in components we've added so far are behavioral, but there's nothing stopping you from writing a full transistor-level model directly... If you build some, make sure to make them public so others can benefit

[free_electron]

you from writing a full transistor-level model directly... If you build some, make sure to make them public so others can benefit

you misunderstand ... i want the makers of the simulator to provide that !

And in that lies the big problem, it is very difficult to get accurate models because component makers will not share those. If they do it will be under NDA or in an encrypted format (like LTspice does) so you won't be able to use them.

[voltsandjolts]

And in that lies the big problem, it is very difficult to get accurate models because component makers will not share those. If they do it will be under NDA or in an encrypted format (like LTspice does) so you won't be able to use them.

LTSpice encyption implementation was fubar so it's been bypassed, e.g.
https://github.com/jtsylve/spice-crypt

[Unixon]

And in that lies the big problem, it is very difficult to get accurate models because component makers will not share those. If they do it will be under NDA or in an encrypted format (like LTspice does) so you won't be able to use them.

Are we talking about models themselves (read the system(s) of equations) or parameters of these models?
If it is about the parameters, is it possible to recover all of them from measurements?

[free_electron]

And in that lies the big problem, it is very difficult to get accurate models because component makers will not share those. If they do it will be under NDA or in an encrypted format (like LTspice does) so you won't be able to use them.

Are we talking about models themselves (read the system(s) of equations) or parameters of these models?
If it is about the parameters, is it possible to recover all of them from measurements?

what i mean with an accurate model is :
- A complete netlist , as implemented
- all parameters for every component used in the design.

Take a simple opamp or a 555. it contains a handful of transistors and some resistors. An accurate model will define each of those transistors with all their associated parameters and then a netlist that connects them.
When i was working in semiconductors we had full models of every single transistor. Those simulations were spot-on and behaved like the real thing. The spice netlist was created from the actual schematics.

Last time i tried to simulate something with a simple 555 timer it didn't behave like on the bench. i was using a constant current source with an lm334 to create a linear sawtooth. There was some trickery where the 555 was powered using an adjustable positive and negative rail to be able to offset the sawtooth. I was not using the 555 as a classic timer. Only the two comparators and the flipflop. The lm334 charged the capacitor until the top comparator tripped the flipflop , discharging the capacitor very fast until the bottom comparator tripped and the process restarted. i was not using the discharge pin but drove an external transistor from the output. The capacitor voltage was buffered with a simple opamp.

Drifting the supply allowed me to adjust the offset to have good symmetry. The ide was to create a ramp going from -2v to +2v in 10mS and then back to -2v in about 10microseconds.
I Never got the simulation to work right. I tried Tina and MicroCap. A model for an lm334 was already a bit of a problem.

i breadboarded it. worked perfectly on first attempt. i spent a whole day trying to get the simulation to work. the breadboarding took me less than an hour.

Simulators stand or fall with the availability of accurate models. And that is the big problem. Micro-cap has a relatively large library. LTspice is much more limited and limited to Linear Tech / Analog Devices parts.
Those vendor specific simulators are problematic in that aspect. And the "general purpose" are limited in models.

just my 2 cents.

[Zero999]

Regarding models:

Why is the writer of the simulator's job to create models?

Accurate models are good, but are much slower and can increase the risk of convergence errors, under certain conditions. More often than not a behavioural model is all that's required. I certainly wouldn't want to use a component level model for a 555 or logic gate, except on rare occasions.

[Nominal Animal]

Right now it does, we were just discussing what it would look like to make a local version.

Local and browser-based are not mutually exclusive, like I've tried to explain here a few times.  There are only a few quirks you cannot work around, like having to pop up a file save dialog every time the user wants to save user data, but otherwise the required interfaces are already widely supported.