Hi all, I am new to this board. I have an associates degree in EE. I graduated back in 1983 but have not done much design work since. Please excuse my ignorance as some of my questions may seem remedial. My love for audio and designing has given me renewed motivation to pursue design work.
I recently purchased "Electronics Work Bench" to aid in design and development. I have not really used spice modeling in the past, back when I was in school we had to bread board all our circuits. I read a few threads concerning spice modeling. How does EWB stack up to other Spice model software? Just how reliable is Spice modeling for power amplifier designing? I have to assume you still need to prototype at some point correct?
P.S. What does DIY stand for? I assume it is not "Do it yourself".
I recently purchased "Electronics Work Bench" to aid in design and development. I have not really used spice modeling in the past, back when I was in school we had to bread board all our circuits. I read a few threads concerning spice modeling. How does EWB stack up to other Spice model software? Just how reliable is Spice modeling for power amplifier designing? I have to assume you still need to prototype at some point correct?
P.S. What does DIY stand for? I assume it is not "Do it yourself".
Most simulation programs use the pSPice engine to generate the outputs. the difference in the various softwares are the user interface and the device models. The accuracy is completely dependent on the device model. I use a program called Cadence (for IC design only) which has hyper accurate models. It has models built to match measured real transistors. It has different models for various DC bias condition, so the model changes depending on your DC conditions. If your DC condition doesnt match on of the models it takes the two closet and extrapolates the new model. My advice is to get the spice models for the transistors you want to use from the manufacutre and not use the default ideal BJTs. Regardless of how accurate the simulations are it still doesnt tell you how the amp will sound, that will require the old method of building and leasoning.
First a question: Which version of EWB and what price -- on their website I only saw pricing for the student versions? I'm just curious.
I agree with the comments above that the accuracy of the individual models is very important.
As for how accurate Spice is for power amps, the answer is complicated. Sometimes very accurate and sometimes not even close. A few examples from my experience, which may not be consistent with that of more skilled (intelligent?) users.
-If doesn't work in simulation, it will rarely work in real life.
-If it works in simulation, chances are good but not certain it will work in real life. With regard to active devices, Spice models seem to do a better job predicting voltages than currents - perhaps this is because the min-max range of beta for real-world BJT's is so broad.
-Spice treats power rails as having unlimited current available.
-If you have a functioning amp and want to predict the effect of incrementally changing a component value Spice can be a time saver then hand calculating is tediouis or error prone. A good example is finding the most appropriate resistor value(s) in a VI-limiting circuit.
-In Spice you can exceed SOA limits with out penaty. Convenient in the above case but a trap for the unwary. [ If only sme game designer would switch their interest to Spice engines we could have animated flames smoke
]
-Spice is helpful for measurments beyond the ability of you test equipment. I'm skeptical about the accuracy of the values in such cases, but the direction of changes "beeter vs. worse" are fairly reliable.
-It is clueless about most oscillation problems in amps. It seems to predict only the most extreme cases.
Anyway, I use to see if an idea is basicly workable and from that point on I build it and see what happens. I suspect that most of the shortcomming I've described can be solved by a used who reawlly understand the Spice engines and the inner workings of the models. Unfortunately for me my time is not limitless, if I got into that aspect it would most likely divert me from ever actually building anything.
I agree with the comments above that the accuracy of the individual models is very important.
As for how accurate Spice is for power amps, the answer is complicated. Sometimes very accurate and sometimes not even close. A few examples from my experience, which may not be consistent with that of more skilled (intelligent?) users.
-If doesn't work in simulation, it will rarely work in real life.
-If it works in simulation, chances are good but not certain it will work in real life. With regard to active devices, Spice models seem to do a better job predicting voltages than currents - perhaps this is because the min-max range of beta for real-world BJT's is so broad.
-Spice treats power rails as having unlimited current available.
-If you have a functioning amp and want to predict the effect of incrementally changing a component value Spice can be a time saver then hand calculating is tediouis or error prone. A good example is finding the most appropriate resistor value(s) in a VI-limiting circuit.
-In Spice you can exceed SOA limits with out penaty. Convenient in the above case but a trap for the unwary. [ If only sme game designer would switch their interest to Spice engines we could have animated flames smoke
]-Spice is helpful for measurments beyond the ability of you test equipment. I'm skeptical about the accuracy of the values in such cases, but the direction of changes "beeter vs. worse" are fairly reliable.
-It is clueless about most oscillation problems in amps. It seems to predict only the most extreme cases.
Anyway, I use to see if an idea is basicly workable and from that point on I build it and see what happens. I suspect that most of the shortcomming I've described can be solved by a used who reawlly understand the Spice engines and the inner workings of the models. Unfortunately for me my time is not limitless, if I got into that aspect it would most likely divert me from ever actually building anything.
Ewd $
Thanks for the replies fellas. I recently upgraded to the newest release of EWB, version 8 that is. I inherited EWB form a job I had at a research institute. I was their lab tech and designer. I used to build and repair a lot of custom equipment for them. I do not recall what they paid for the original version but I paid $1200 for each subsequent upgrade, this upgrade was actually my second one.
Their are three products to the EWB suite MultiSim, UltiRoute, and UltiBoard. I own all three, so my upgrade price was for all three. This is of course is the Pro version. They do have an advanced version which retails for some where around $5000. It boasts professional "real" instruments and I believe it comes with VHDL. VHDL is an add on for me. I really bought the package initially to create and finish art work. Back in the day I was sick and tired of using proto boards for design, so when they introduced EWD I ordered my boss to buy it. I now intend to put it to good use. Let their be sound!
They claim that their extensive library of spice models is accurate, it was of course upgraded with this latest version. I believe they actually use the manufactures spice models, and of course they have generic ones as well. They have detailed spice model design tools that allow you to create special devices. In theory you can emulate a speakers output impedance and even model your power source so that it emulates real world conditions, such as current source and impedances etc…..
Hope that helps.
Thanks for the replies fellas. I recently upgraded to the newest release of EWB, version 8 that is. I inherited EWB form a job I had at a research institute. I was their lab tech and designer. I used to build and repair a lot of custom equipment for them. I do not recall what they paid for the original version but I paid $1200 for each subsequent upgrade, this upgrade was actually my second one.
Their are three products to the EWB suite MultiSim, UltiRoute, and UltiBoard. I own all three, so my upgrade price was for all three. This is of course is the Pro version. They do have an advanced version which retails for some where around $5000. It boasts professional "real" instruments and I believe it comes with VHDL. VHDL is an add on for me. I really bought the package initially to create and finish art work. Back in the day I was sick and tired of using proto boards for design, so when they introduced EWD I ordered my boss to buy it. I now intend to put it to good use. Let their be sound!
They claim that their extensive library of spice models is accurate, it was of course upgraded with this latest version. I believe they actually use the manufactures spice models, and of course they have generic ones as well. They have detailed spice model design tools that allow you to create special devices. In theory you can emulate a speakers output impedance and even model your power source so that it emulates real world conditions, such as current source and impedances etc…..
Hope that helps.
I use EWB -- they have now turned the product into a "razors and blades business" -- you really do need the updates because there are some annoying bugs in the program --
here's one I found which (I brelieve) has since been patched -- the bode plotter in Multisim8 would go blank when you switched from gain to phase -- this never ocurred in Multisim 7.
the "Component Wizard" in Multisim is inelegant to say the least.
it is quite easy to get Ultiboard to lock up, or to be unable to move components.
one benefit to the suite -- if you buy the service packs they will help you out with problems.
have you found the 2n3055/2n2955 model yet in multisim ? -- it's everywhere on the web, easy to import -- but why do you have to build a "User" component for this ubiquitous part? there are no real good optocoupler models, etc.
don't bother with "Select All" to cut and paste a schematic to a word-processing program -- you are better off taking a screenshot and manipulating it -- Eagle is much, much better in this regard.
since National Instrument has purchased EWB I think it will get even more expensive.
An EWB users group, or users forum would be useful.
here's one I found which (I brelieve) has since been patched -- the bode plotter in Multisim8 would go blank when you switched from gain to phase -- this never ocurred in Multisim 7.
the "Component Wizard" in Multisim is inelegant to say the least.
it is quite easy to get Ultiboard to lock up, or to be unable to move components.
one benefit to the suite -- if you buy the service packs they will help you out with problems.
have you found the 2n3055/2n2955 model yet in multisim ? -- it's everywhere on the web, easy to import -- but why do you have to build a "User" component for this ubiquitous part? there are no real good optocoupler models, etc.
don't bother with "Select All" to cut and paste a schematic to a word-processing program -- you are better off taking a screenshot and manipulating it -- Eagle is much, much better in this regard.
since National Instrument has purchased EWB I think it will get even more expensive.
An EWB users group, or users forum would be useful.
The first version of Multisim (Vs6) released in 1999 was soooo bug ridden it was almost completely unuseable......
Asking folk to pay cash for that tissue of excrement was nothing short of daylight robbery by Electronics Workbench.... 😡
The fact that customers are now required to pay through the nose for obvious bugs that EWB should have banished in the first place reeks of organised corporate fleecing....
I suggest folk avoid purchasing early versions of EWB programs as a matter of routine....
..as a rule of thumb, EWB only seem to be capable of clearing the majority of obvious bugs from their product after at least two years....
...and they now charge existing customers for the previlage!!!!...🙄
You've been warned.... !
Asking folk to pay cash for that tissue of excrement was nothing short of daylight robbery by Electronics Workbench.... 😡
The fact that customers are now required to pay through the nose for obvious bugs that EWB should have banished in the first place reeks of organised corporate fleecing....
I suggest folk avoid purchasing early versions of EWB programs as a matter of routine....
..as a rule of thumb, EWB only seem to be capable of clearing the majority of obvious bugs from their product after at least two years....
...and they now charge existing customers for the previlage!!!!...🙄
You've been warned.... !
Free...!
...otherwise, i suggest folk use LTSpice, which is gratis, and frequently updated....
http://www.linear.com/designtools/softwareRegistration.jsp
...otherwise, i suggest folk use LTSpice, which is gratis, and frequently updated....
http://www.linear.com/designtools/softwareRegistration.jsp
Re: Free...!
Not to mention free support from the program's developer and some very knowledgeable users in the Yahoo users' group. There's some great stuff in the files section of that group.
For example, there's a loop gain probe that implements Tian's loop gain calculation and doesn't require subcircuit replication. See the file LoopGainProbe.zip in the "adventures with analog" part of the files section. BTW, I just found out today when catching up on that group that Tian's paper on this is now online at http://www.thekunderts.net/ken/docs/c&d2001-01.pdf
Also, Helmut Sennewald has taken the schematics of the TL431 and made a transistor-level TL431 model available that predicts things like stability vs. load capacitance. Very cool!
There are still some primitive aspects of the program, but hey, it's free.
Now all I need is a freeware that allows me to draw system block diagrams. I've tried this with Word/PowerPoint and it's really, really painful. Anybody know of something that does the job?
mikeks said:...otherwise, i suggest folk use LTSpice, which is gratis, and frequently updated....
Not to mention free support from the program's developer and some very knowledgeable users in the Yahoo users' group. There's some great stuff in the files section of that group.
For example, there's a loop gain probe that implements Tian's loop gain calculation and doesn't require subcircuit replication. See the file LoopGainProbe.zip in the "adventures with analog" part of the files section. BTW, I just found out today when catching up on that group that Tian's paper on this is now online at http://www.thekunderts.net/ken/docs/c&d2001-01.pdf
Also, Helmut Sennewald has taken the schematics of the TL431 and made a transistor-level TL431 model available that predicts things like stability vs. load capacitance. Very cool!
There are still some primitive aspects of the program, but hey, it's free.
Now all I need is a freeware that allows me to draw system block diagrams. I've tried this with Word/PowerPoint and it's really, really painful. Anybody know of something that does the job?
This is what I do. The price is right! Not all the bells and whistles of some of the big-time packages but it does the job for me. No obvious show stopping bugs that I've run up on.
Nonetheless, I yearn for an integrated package where you can go from Spice to PCB layout with just the obvious intermediate step of replacing "fictional" devices like current sources with real devices. At one time I used IVEX but they didn't have a Spice module at the time and it was just a DOS program with a WIN95 era GUI pasted on top - didn't like XP and visa versa.
Has anyone had experience with BeigeBag, either alone or bundled with Eagle?
one of the really great things about LTSpice/SwitcherCad is that they have models for almost all of the Linear Tech switch mode controllers.
But, it is very, very painful to use one modeling program and then manually create all the nets for the PCB.
But, it is very, very painful to use one modeling program and then manually create all the nets for the PCB.
Beige Bag.
Hi Sam9 ,
You can download their trail version and check it out. It's quite nice except getting spice models that work for it might be a problem. The trial is valid for 30 days.
Better you try the older version 4.2 because the new one seems to require a longer learning curve.
They have some pretty affordable prices for students etc. So I emailed them and asked if they have any special offer for people out of work.
They never replied , not even to say they couldn't make a special offer ! Or are they still deciding ? Really felt disappointed about not getting any kind of reply. I used to think they were a very professional group of guys.
Cheers.
Hi Sam9 ,
You can download their trail version and check it out. It's quite nice except getting spice models that work for it might be a problem. The trial is valid for 30 days.
Better you try the older version 4.2 because the new one seems to require a longer learning curve.
They have some pretty affordable prices for students etc. So I emailed them and asked if they have any special offer for people out of work.
They never replied , not even to say they couldn't make a special offer ! Or are they still deciding ? Really felt disappointed about not getting any kind of reply. I used to think they were a very professional group of guys.
Cheers.
Electronics Workbench 5.12 - Full version
http://edl.atw.hu/ewb512.zip
(7Megabytes!)
Maybe the file is not avaliable to download from abroad.
You can download transistor and diode models from Onsemi's homepage.
http://edl.atw.hu/ewb512.zip
(7Megabytes!)
Maybe the file is not avaliable to download from abroad.
You can download transistor and diode models from Onsemi's homepage.
Get yourself a domain name and a home page. All it has to say is "under construction". Then tell them you are a start up. All detailed questions can be answered with "sorry, that's proprietary info". It's kind of decieptive but at least people will talk to you politely.
andy_c said:
Use EWB vs. 5.12...below:
http://edl.atw.hu/ewb512.zip
Virtually bug-free...apart from the Fourier analysis...Which is excreble...
- Status
- Not open for further replies.