EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Agony on May 09, 2013, 02:50:09 pm
-
So i thought of gettign one to test and play around with circuits etc before it gets to the real deal.
A nice interface (drag n drop?) would be one major advantage with a decent simulation options.
https://www.circuitlab.com/ (https://www.circuitlab.com/) has nice interface but lack of components, simulation etc.
national instruments designer had extremely bad interface.
Any other around with a Good interface but with some decent simulation aswell?
-
LTSpice IV.
www.linear.com (http://www.linear.com)
-
Everyone seems to be a fan of LTspice.
I like to use TINA-TI (http://www.ti.com/tool/tina-ti) when I can. I'm using it right now, as a matter of fact.
I prefer its user interface much more than LTspice's.
-
This one is free: http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite (http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite)
I like Electronics Workbench but it is now Multisim and they screwed it up in my opinion. I use the above a lot now!
-
This one is free: http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite (http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite)
KiCad does not simulate.
-
If you want the best, it's either Mentor Graphics or Cadence, however neither are in budget for a hobbiest, and are usually out of reach for buisness use unless you're doing IC design where the cost of another set of artwork is £100k plus.
-
https://www.circuitlab.com/ (https://www.circuitlab.com/) has nice interface but lack of components, simulation etc.
What do you mean when you say it lacks simulation? It simulated my little op-amp circuit (https://www.eevblog.com/forum/projects/keithley-2000-gain-repair/) well enough for me.
-
the simulator is only as good as the models provided .... and sadly, that is where a lot of them are seriously lacking.
ltspice or tina are your best bet. at least there you have accurate models of the included devices.
simulator are NOT a replacement for a brain , nor a replacement for the actual circuit. meaning ytou need to know what you are doing.
if the simulator claims the output of an opamp is at 1000 volts ... ( 1 volt in , 1k // 1 meg to get a 1000 gain ) and driving 50 ohms.... you ned to be clever enough to see that it is bullshitting you. it's not the simulators fault. that thing is simply solving an equation... its the input that is wrong and or the constraints that are wrong. you will have to set those yourself.
look my power supply simulation says i can an draw 100 ampere from my lm317 power supply with dinky transformer. try it and you will get nothing.
-
the simulator is only as good as the models provided .... and sadly, that is where a lot of them are seriously lacking.
ltspice or tina are your best bet. at least there you have accurate models of the included devices.
simulator are NOT a replacement for a brain , nor a replacement for the actual circuit. meaning ytou need to know what you are doing.
if the simulator claims the output of an opamp is at 1000 volts ... ( 1 volt in , 1k // 1 meg to get a 1000 gain ) and driving 50 ohms.... you ned to be clever enough to see that it is bullshitting you. it's not the simulators fault. that thing is simply solving an equation... its the input that is wrong and or the constraints that are wrong. you will have to set those yourself.
look my power supply simulation says i can an draw 100 ampere from my lm317 power supply with dinky transformer. try it and you will get nothing.
Multisim does the same thing so it doesn't simulate either? If you think about it all programs solve an equation. Only way to completely simulate a schematic is build it.
-
if multisim does not take into account that the circuit only has 5 volts to work with and tells you the output is 1000 volt : fling it through the window.
a LOT of simulators use the voltage controlled voltage sources or current controlled source ( there is a whole range of thes 4 terminal devices ) to 'emulate' the behavior.
a real model will have a transistor level description of the opamp. in other words : the real netlist as opposed to an 'ideal' model. but that comes with a price ... and calculation time.
That is why these 'big boy' simulators like ELDO cost a million dollar and need a compute farm to run... you can send the actual design , at transistor level in there. it even looks at the layout and extracts all the parasitics from the semiconductor chemistry and takes those into account. Those simulation outputs are spot on , but only if they are driven by an angineer that knows what he is doing. blindly clicking parts together and hitting 'run' gives only gives you lies ...
simulators are good to sweep a filter and get an idea : if i pull here , that point will go that way. if i push there that point will go there...
you cant just plonk an entire system in there. it wont work. this is a fundamental problem. things like multisim come with libraries of parts but most theses libraries are optimized parts , emulated parts or incomplete parts. they are not transistor level description. for the simple fact that for many parts there is no transistor level description. manufacturers don't hand those out , unless in very special circumstances and then NDA's will need signing...
So that is why LTspice is so great. it has the actual real models for the linear tech parts.
The same for Tina. it has the real models.
now, a word of caution : don;t make the mistake of designing something with ltspice using an lt1006 opamp and in the end result drop in a lm741. it aint gonna work ! you need the REAL parts.
-
That is why these 'big boy' simulators like ELDO cost a million dollar and need a compute farm to run... you can send the actual design , at transistor level in there. it even looks at the layout and extracts all the parasitics from the semiconductor chemistry and takes those into account. Those simulation outputs are spot on , but only if they are driven by an angineer that knows what he is doing. blindly clicking parts together and hitting 'run' gives only gives you lies ...
Wow, really? That's actually really cool.
-
If you're just messing around with the basics, try this one: http://www.falstad.com/circuit/ (http://www.falstad.com/circuit/)
If you're doing more complex stuff, like looking at parasitics, modelling specific components for real product design, then LTSpice or TINA-TI should be your starting point. But, as was mentioned, you do need to know what you're doing with it to get meaningful results.
Some EDA packages, like Altium, will do things like signal integrity simulations and such.
-
if multisim does not take into account that the circuit only has 5 volts to work with and tells you the output is 1000 volt : fling it through the window.
a LOT of simulators use the voltage controlled voltage sources or current controlled source ( there is a whole range of thes 4 terminal devices ) to 'emulate' the behavior.
a real model will have a transistor level description of the opamp. in other words : the real netlist as opposed to an 'ideal' model. but that comes with a price ... and calculation time.
That is why these 'big boy' simulators like ELDO cost a million dollar and need a compute farm to run... you can send the actual design , at transistor level in there. it even looks at the layout and extracts all the parasitics from the semiconductor chemistry and takes those into account. Those simulation outputs are spot on , but only if they are driven by an angineer that knows what he is doing. blindly clicking parts together and hitting 'run' gives only gives you lies ...
simulators are good to sweep a filter and get an idea : if i pull here , that point will go that way. if i push there that point will go there...
you cant just plonk an entire system in there. it wont work. this is a fundamental problem. things like multisim come with libraries of parts but most theses libraries are optimized parts , emulated parts or incomplete parts. they are not transistor level description. for the simple fact that for many parts there is no transistor level description. manufacturers don't hand those out , unless in very special circumstances and then NDA's will need signing...
So that is why LTspice is so great. it has the actual real models for the linear tech parts.
The same for Tina. it has the real models.
now, a word of caution : don;t make the mistake of designing something with ltspice using an lt1006 opamp and in the end result drop in a lm741. it aint gonna work ! you need the REAL parts.
the old Electronics Workbench worked great. I still have a copy but I don't know how long it will last. Mulltisim is lacking some of the good stuff about Electronics Workbench. When they renamed it, they ruined it in my opinion
-
the old Electronics Workbench worked great. I still have a copy but I don't know how long it will last. Mulltisim is lacking some of the good stuff about Electronics Workbench. When they renamed it, they ruined it in my opinion
What have they changed since it was renamed? I'm fairly new to using simulation tools and have been going back and forth between Multisim and LTspice. They each seem to have their own good and bad parts. Just curious what your thoughts on multisim are as someone with more experience I guess.
-
That is been my grief on simulators. They are happy to keep on simulating even in situations where real circuit would blowup on your face.
Put thats more of my personal preference of having that magic smoke moments as feedback of catastrophic failures. OOps, overloaded that component..
-
Well so what? Acording to Ohms law and my Casio, this little 0.33 ohm resistor I have here will dissipate 174545.45W if I connect it to 240V mains. How is the calculator supposed to magically quess the power rating of my components?
These lines of argumentation from the anti-SPICE brigade are so daft they're hardly worth bothering with.
-
That is why these 'big boy' simulators like ELDO cost a million dollar and need a compute farm to run... you can send the actual design , at transistor level in there. it even looks at the layout and extracts all the parasitics from the semiconductor chemistry and takes those into account. Those simulation outputs are spot on , but only if they are driven by an angineer that knows what he is doing. blindly clicking parts together and hitting 'run' gives only gives you lies ...
I agree with your sentiment, but I used to work on modelling FETs (at the physical level) for simulation in circuits and there are inherent flaws in even very expensive simulators that mean they can only approximate reality. For example a very common technique used is Harmonic Balance and built into this technique is the assumption that the time delay through a FET is constant, but in a physical FET the time delay depends on how far the gate depletion region extends towards the drain so is voltage dependent.
Additionally, most models are inherently too crude to properly model high frequency behaviour. Typically they assume that the I(V) characteristics are those measured at dc whilst the presence of charge traps within the device (deep levels) mean that there is frequency dependent behaviour (at dc there is sufficient time for charge to leave the traps whilst under microwave frequencies the charge is stuck - this can have a dramatic effect on some devices).
Even very detailed physical simulations would often produce wrong results because though they could model the bulk semiconductor behaviour they would fail to model the surface effects and in many devices surface effects can be significant (i.e. the effects of charge trapped in surface states).
I am referring to very high frequency analogue circuits. I am also a bit out of date now (its been eight years since I left the industry) so improvements must have been made but I would always treat simulations with a degree of cynicism even if the simulator costs tens of thousands or more.
-
That is why these 'big boy' simulators like ELDO cost a million dollar and need a compute farm to run... you can send the actual design , at transistor level in there. it even looks at the layout and extracts all the parasitics from the semiconductor chemistry and takes those into account. Those simulation outputs are spot on , but only if they are driven by an angineer that knows what he is doing. blindly clicking parts together and hitting 'run' gives only gives you lies ...
I agree with your sentiment, but I used to work on modelling FETs (at the physical level) for simulation in circuits and there are inherent flaws in even very expensive simulators that mean they can only approximate reality. For example a very common technique used is Harmonic Balance and built into this technique is the assumption that the time delay through a FET is constant, but in a physical FET the time delay depends on how far the gate depletion region extends towards the drain so is voltage dependent.
Additionally, most models are inherently too crude to properly model high frequency behaviour. Typically they assume that the I(V) characteristics are those measured at dc whilst the presence of charge traps within the device (deep levels) mean that there is frequency dependent behaviour (at dc there is sufficient time for charge to leave the traps whilst under microwave frequencies the charge is stuck - this can have a dramatic effect on some devices).
Even very detailed physical simulations would often produce wrong results because though they could model the bulk semiconductor behaviour they would fail to model the surface effects and in many devices surface effects can be significant (i.e. the effects of charge trapped in surface states).
I am referring to very high frequency analogue circuits. I am also a bit out of date now (its been eight years since I left the industry) so improvements must have been made but I would always treat simulations with a degree of cynicism even if the simulator costs tens of thousands or more.
Having recently worked at a major supplier of high frequency high power FETs, the models are still a long way from being complete.
-
Well so what? Acording to Ohms law and my Casio, this little 0.33 ohm resistor I have here will dissipate 174545.45W if I connect it to 240V mains. How is the calculator supposed to magically quess the power rating of my components?
These lines of argumentation from the anti-SPICE brigade are so daft they're hardly worth bothering with.
:-+
-
For school usage EWB 5.12 is quite good.
(http://i.imgur.com/M2x9oxq.gif)
What is the Mentor Graphics package for simulation?
Alexander.
-
That is been my grief on simulators. They are happy to keep on simulating even in situations where real circuit would blowup on your face.
Put thats more of my personal preference of having that magic smoke moments as feedback of catastrophic failures. OOps, overloaded that component..
They can already calculate power... who else thinks simulators should have the option to simulate the release of magic smoke too? (Maybe complete with realistic animation and audio!)
-
Well so what? Acording to Ohms law and my Casio, this little 0.33 ohm resistor I have here will dissipate 174545.45W if I connect it to 240V mains. How is the calculator supposed to magically quess the power rating of my components?
These lines of argumentation from the anti-SPICE brigade are so daft they're hardly worth bothering with.
:-+
For the record, it is not SPICE that is annoying. It is the people who use "But it worked in the simulation!" as argument when their stuff fails in the real world.
-
That is been my grief on simulators. They are happy to keep on simulating even in situations where real circuit would blowup on your face.
Put thats more of my personal preference of having that magic smoke moments as feedback of catastrophic failures. OOps, overloaded that component..
Now there's an idea for a kickstarter project. A little 'USB simulation realizer' ... A little box connected to the computer. It has a 3 pin socket on top and a bag of BC547 transistors. And a 'start simulation' button. To start the simulation you must put in a working bc547 in the socket and hit start. If at the end the bc547 still works the simulation is good. If the bc547 blows up , releases smoke or dies a fiery death .. Obviously there is something wrong.
You will get the real lighteffects, noise , real smoke and the real smell of burining devices !
And only for 50$. Sign up now ...
No seriously. Simulators like eldo wont go on simulating. You can set stop criteria like max power dissipation in an element. The simulator will abort when you surpass stuff like that. But like i said. Compute farms and lots of cash... And you don't buy licences for that.. You lease them on yearly basis...
-
Well so what? Acording to Ohms law and my Casio, this little 0.33 ohm resistor I have here will dissipate 174545.45W if I connect it to 240V mains. How is the calculator supposed to magically quess the power rating of my components?
These lines of argumentation from the anti-SPICE brigade are so daft they're hardly worth bothering with.
:-+
For the record, it is not SPICE that is annoying. It is the people who use "But it worked in the simulation!" as argument when their stuff fails in the real world.
Yes, very true. Still, those same people use something quite similar to "But I want it to work!" when a real-world experiment that they haven't simulated fails. |O I've found the simulators to be quite accurate if you know what you're doing. Yeah, it'll happily tell you that a 2N2222 can dissipate a megawatt and keep working, but if you have any experience with it at all you'll know to check for that. It's very, very useful to be able to (e.g.) test expensive parts without risking damaging them, as long as you can tell whether or not to trust the results you get.
I've also found it to be a good learning experience. Trying to figure out how to make my simulations more accurate has given be a very good feel for layout parasitics and such that I would have a hard time getting otherwise.
-
I've also found it to be a good learning experience. Trying to figure out how to make my simulations more accurate has given be a very good feel for layout parasitics and such that I would have a hard time getting otherwise.
This is what I get from simulations, at least the SPICE ones. SPICE has been around for a long time, even when I was in school, but we barely used it much, I hated punching the cards, submitting the job through the operator, waiting for the results printout in the rack down the hall... etc... it was better and safer to develop instincts, and learn by doing, learn what works, what the tricks are, what to do, what not to do, where are the parasitics, etc. After a while you just know.
But these days having a SPICE simulator on my PC ( I use LTSPICE and TINA-TI because they are free ) is pretty damn cool :)
I am blown away that the design engineers can simulate the physical wafer process from the input files for some new chip, fed into Silvaco ATHENA, and it will simulate the process steps, trenching, oxidation, deposition, planarization, building the wells, the gate, polysi, halo and source drain implants, then it can extract the SPICE parameters from the simulation results of the virtual chip (!)... and can be used in a virtual circuit, if it fails, go back to some process parameters, tweak it and try again. They never have to actually make a die from the masks until they are really really sure.
-
Was "Circuit Maker" the name given to the software suite that was swallowed up by Protel? I remember evaluating a trial version of the simulation part of the package years ago as an apprentice. In addition to the standard selection of components you could incorporate animated rockets, race cars and blinking LEDs into you schematic. The rockets and race cars could be configured to launch at logic thresholds when the simulation was run and the LEDs would light up when conducting current. I remember simulating a ripple counter that successively launched and array of rockets to demonstrate the capabilities of the software to my boss. Unfortunately he poo-pooed the idea of purchasing the software on grounds that "we don't need to launch rockets or race cars", so we were stuck with Multisim instead (which was pretty crap, IMO).
-
Yes, that was what I used going through school.
That and traxmaker, for laying out boards designed in circuitmaker.
I still have them installed on my old computer.
-
You should try ViaDesigner, it's Mentor Graphics System Vision which is a mixed signal tool based on Eldo and ModelSim. The have free evaluations for Amp Hour listeners.
ViaDesigner.com/theamphour