Well, that's a bit like saying Joe Pass is better than George Benson etc.
I personally use, obviously, my own SS, but I also use Cadence extensively. Each one has specific features that are not in the other, which I find indispensable, so I need both. Although, as a side note, I don't use LTSpice, as it has no features that I can't do without, and a GUI that I personally, find unusable for my purposes. This is despite the fact that LTSpice, is probably the fastest and has the best convergence of any non i.c. targeted spice.
Kevin Aylward... Analogue IC designer and very part time flogger of cheap simulators
So.. the latest Cadence Spectre version has a switch. If you enable it, it runs 3 to 10 times as fast, and this is as a true standard spice simulator, not a "fastspice". Oh... I said, to the Cadence man, then why not have this switch enabled all the time...err..well it then uses 2 licences rather than the 1 he said.
Unfortunately, I have recently been using Simetrix as well, but soon to change, if all goes to plan. Its GUI, for 2008, is also truly dreadful. It can't compete with either PSpice Schematics, or SS. Actually plotting waveforms is a true plain in the arse. Its still so amazing that people are so clueless on what is really required when you are doing professional, err.. 37 1/2 hours a week simulations.
Really... what don't you like about it? I would agree the GUI is a bit unusual, but never found it particularly "dreadful."
I suspect that a lot of money Catena makes off of SIMetrix comes from licensing the simulator itself rather than selling the complete standalone package. (Pulsonix, among others, use the SIMetrix "engine" but the regular old Pulsonix GUI. They did a reasonably good job of letting the GUI exploit most of the simulator's power as well, although you still have have to put up with graphs in separate windows.)
I found the ability to specify plots on the schematic (or through a script) quite handy, actually.
This is unfortunately true of most CAD tools today -- very few of them are being written by people who actually use them. Better companies keep close tabs on feedback from the actual users, though!
Its, essentially, unusable for at the current time. See below.
When I have my simulation hat on, why on earth would I want to write scripts just so I can click on wire, have the old trace *replaced* with the new trace, and stay in that mode where I can again click on a signal and then have that then displayed? Why should I have to chose different probes to probe a current or a voltage? I can move the same test point to a wire or pin in SS, which I stole from PSpice Schematics. Why can't they also steal all the good features from all the spices as I do?
Why should I have to hold the mouse key down just to get a locked on readout value of a trace. In SS I stole the cadence, mouse locks to waveform trace when the mouse pointer is near the trace. Its absolutely indispensable to me, and one feature that would cause to reject any spice that doesn't have it.
When I am using spice, I want the key features all to be there, instantly. I have no idea how to write Simetrix scripts, nor do I have the desire to lean yet another coding language, nor the time. The whole idea of scripting is really bogus. Like, Simetrix, don't come with worst case corner reruns, do you want to go and write one? (see
formatting link
There are many, lacking features in the Simetrix GUI, for example, in SS you can just mouse over wires and pins to get DC operating conditions, and over the component to get dc and transient power. If I want to get mosfet Vgst, Vdsat, Vds, gm, etc, I just have to click on the signal list in the docked toolbar.
For me, I suspect that my complaints are because I am a genuine power user. I can easily spend the full working day, 5 days a week, for months at a time running simulations. I just don't have the time to piss about pressing numerous buttons, to do something that should be right there from the start. I suspect, many just use spice now and again, and have not really experienced the frustration of needless aggravation.
Thanks Kevin, that's a good list. I agree with your assessment that most of the annoyances are likely due to using it on a daily basis (not something that I do) and thus you expect it to work more "fluidly" than it does.
I have problems convincing some people that maybe, just maybe, it'd be possible to improve on ORCAD Capture... :-(
It really needs to be an adjustable option (e.g., regular clicking replaces old trace, shift-clicking adds a new trace), since there are plenty of times when I want to keep an old trace around to compare with what I'm clicking on next.
I expect because some people want to click on wires even when they're after voltages rather than currents... but this would be a nice option ("if I click on a wire, assume I'm after current, if I click on a pin, I'm after voltage").
Last I heard they were looking for good programmers if you're interested. :-)
My guess would be because it's considered more reliable when there are multiple closely-spaced traces (as with a Monte Carlo run)... but I agree with you that "snap to trace" is awfully nice.
Didn't you have to learn some HSpice scripting though? I haven't used it in years, but I seem to recall that the waveform viewer was quite barebones, and doing something as simply as measuring a rise-time required a bit of scripting (unless of course you wanted to manually do some arithmetic each run).
Cadence ADE has a very nice "calculator". I would say, the majority of extra functions (as you as note above) are available through that GUI interface, with no scripting.
Well, there you go -- clearly I'd try to run a simulator at least once differently than Kevin would, so it's probably safe for me to explicitly tell the simulator what I want to probe rather than have it assume I kow what I'm doing. :-)
I was thinking of what most people do in the "real world," apparently -- most often you probe/scope around for voltages on pins whereas you solder in shunt wires to measure currents.
But you haven't gotten inside the spice method. Connecting wires in spice are not paths but points (nodes), regardless of their length or number of branches on the schematic representation. That node must exhibit conservation of charge, regardless of how many currents enter or leave it and a single instantaneous voltage everywhere that it connects.
The only way to define a current entering or leaving a node is to work with some device connected to that node. Wires with actual physical length must to be modeled as components connecting with two ends connecting two different nodes, and approximating the properties of an actual wire.
Then you could measure the voltage drop across wires and the current entering or leaving either end (pin) and those two currents would not be instantaneously the same. You could define that wire model as two wires in series and then you could probe the current in the middle of that pair of wires in series. But none of that is possible with net list nodes (normally shown as wires on spice schematic editors).
Yes, I understand this is the case in terms of how SPICE actually operates.
I had a microwaves professor who was fond of saying, "In my class, there is no such thing as a wire." (I.e., everything was a transmission line to him. :-) )
Some (many?) SPICES will happily let you probe the current in a wire (it just finds the nodes connected to the wire and does the arithmetic) or, of course, the voltage on a pin, though. And as I say, for me I'd prefer to tell SPICE "probe V" or "probe I" and then not worry about whether I'm clocking on a node or a wire, whereas I have no problem with Kevin wanting a single "probe V or I based on what's clicked" tool either.
Exactly. In the old days, you put a zero volt voltage source in line to measure the current, sort of a perfect voltage shunt. Then, you realized that each device has current flowing into or out of the pins, so you could measure there.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.