Opinions on PSIM or VisSim or the like

[MagicSmoker]

It seems inevitable that power electronics is moving ever more towards digital control and compensation (even when it doesn't provide any advantage a traditional analog controller). Up til now our software engineer has done a good job of translating my explanations of control loops and transfer functions into code, but we are bidding on a project to design a fault-tolerant, constant power arc/plasma power supply and I think that will be too much for the poor guy.

So I am contemplating going to the dark side and using power electronics simulation software to generate the code (either C for a uC or HDL for a FPGA) that implements the switching strategy and control loop compensation, leaving the UI and supervisory functions to our software engineer.

I haven't been terribly impressed with this type of software in the past, either because it tends to rely on ideal diodes, switches and transformers, and/or it has a ridiculously steep learning curve as a result of trying to maintain backwards compatibility over decades of software revisions, and, of course, because it is invariably very expensive.

However, I am open to the possibility that modeling/simulation/code generation software has gotten better and would like some anecdotal, real-world commentary from the peanut gallery here before I sign up for demos that inevitably turn into the kind of high pressure sales presentation that would make a used car salesman blush.

[MagicSmoker]

Clearly, this sub-forum is the red-headed stepchild of the bunch. Le sigh...

[KaanMutlu]

Hello MagicSmoker,

I also try to find the most useful and suitable software for myself. I've been working as an analog design engineer for 2 years and instead of writing C code alone, it seems it is better to use a model-based structure like Vissim or using PSIM that allows us to make also simulations and let us use its modules specialized on some certain DSPs.

Could you fine a good answer in 3 weeks (after posting this)? I still have question marks about which to choose

Kaan

[MagicSmoker]

Haven't made much progress on selecting a good simulation/code generation platform. VisSim mainly (only?) works with certain TI MCU/DSP but is about half the price of Matlab+Simulink (+other modules specific to power electronics and code generation) and seems pretty capable. PowerSim seems to assume ideal component behavior out of the box so requires quite a bit of fiddling before what it shows during a simulation looks anything like reality (similarly, I am not sure if its code generation module actually produces usable code), but it looks like it can handle more complex circuits (e.g. - more complex than a 3ph. motor inverter, which seems to be what most of these packages are optimized for).

I haven't really looked into Matlab+Simulink (etc.) because it is so expensive.

[Marco]

How about SIMPLIS?

No idea if it's any good, but even in the worst case you'll still have a license for Simetrix.

[MagicSmoker]

How about SIMPLIS?

No idea if it's any good, but even in the worst case you'll still have a license for Simetrix.

I very well could be wrong, but I think Simplis is only capable of generating HDL output, not C code. Thanks for the reply, though - I had all but given up on this thread!

[MagicSmoker]

Microchip has a new tool out. I stumbled on it while looking at PIC24s.

Interesting suggestion. It looks a bit rudimentary right now but something to keep an eye on. 

Personally I don't see any benifit in an all digital SMPS. I see a benifit in having analog blocks integrated in a uc to customize an SMPS ,combined with the digital part to adjust output power on the fly under varying condtions. You don't need a dspic or even a 24 series PIC for that though, their 8b portfolio covers that.

While it is true that a digital control loop (in hardware or software) will never outperform an analog one, there are all kinds of things you can do with a software-defined SMPS implementation that are either impossible or impractical with hardware alone. For example: minimum pulse width (on or off) plus randomized pulse skipping to extend effective PWM control range; FIR filters which are guaranteed stable and express functions with no analog equivalent; coordination and synchronization of multiphase converters; precise firing of auxiliary switches in quasi-resonant converters; guaranteeing overlap of main switches in current-fed converters; it is easy to add more protection functions, change all sorts of parameters from control loop to switching frequency later on; etc.