Topics: General purpose software tool for designing multiple output flyback converters
on Power Supply
General purpose software tool for designing multiple output flyback converters
Does anyone know of a good general purpose software tool for designing multiple output flyback converters? I know of the webBench and some of the off-line switcher programs from the various manufacturers but they seem to fall short. Thanks
09-13-2013 06:57 AM
Yes, they all fall short. A general purpose tool for flybacks and other topologies is POWER 4-5-6, it's the only one not tied to a specific vendor and part. All the other tools are designed to sell a part to you, and not to really probe into the design to try and find the flaws.
09-13-2013 09:54 AM
Hi Frank, There is no universal remedy for all diseases or general purpose vaccine. Same thing, if you want to do a good design, make it custom.
09-13-2013 12:32 PM
One thing that concerns me these days is an over-reliance on simulation.
To paraphrase my good friend and colleague John Beecroft, "If you do too much simulation, you start to believe that it's the real thing".
I often see design reviews with more simulation data than real waveform data.
I see university papers coming out that propose a new theory for modeling, then confirm it with a simulation, not hardware.
Don't get me wrong, I enjoy a good simulation. It's just that in reality, for me, it is a small part of the real design cycle, and I like to get it out of the way as fast as possible. I don't have time to try and force Spice to converge, especially when I know it's going to miss half of the waveforms details anyway.
09-13-2013 03:18 PM
i think simulation will help direct me to validated the designed circuit,and find some questions. it will be good to the people which don't have much experience.
09-13-2013 05:56 PM
Simulation is very useful for looking at basic waveforms of converters, choosing topologies, and assessing the main losses in the power components. It is good for educating someone without experience, as you say. It's also useful for sizing components and exploring a lot of what-if questions. But which program will you use for that?
Most people will gravitate towards LTSpice, because it's free, or some other version of Spice. You immediately run into the problem that if you simulate a converter open loop it takes for ever to settle down since it is usually a high-Q circuit. You can't wait that long. So you have to put a control loop around it, and Spice doesn't like that much at all.
Programs like Transim and POWER 4-5-6 were designed to get around this problem - they are quite happy with instantaneous transitions. Transim, however, doesn't design the loop components for you, so you have to get involved in loop design when you really only want to know which topology is going to work for you. 456 automates all of that for you, saving a huge amount of time.
Regardless, at this phase of a project, I have no issues with people doing simulation as long as they don't let the time spent on this detract from building the hardware.
Later on in the project, as we have seen in the thread on this group about failures, you WILL have circuits blow up on you. The only way you are going to find that is by building the hardware and catching the second- and third order effects that give us problems.
I don't think I have ever had a situation in may career where a simulation has predicted a failure before I have encountered it in the hardware. I may be able to reproduce it afterwards in the software, but rarely is there time to build a sophisticated enough model.
The problem I see these days is that there is an implication from component vendors that IF you use their part and IF you use their software, your circuit is going to work first time and you can skip the due diligence of proper design verification testing on the boards. That assumption gets a lot of people into trouble with their designs.
09-13-2013 07:58 PM
Nowadays I see a lot of the students who do their internship in our company working with simulation software as if it is the only way. I've even seen one tweek his hardware to make the signals look more like his simulation!
Most of the trainees I get in my office are Chinese. They are really good at math and simulating and think they can solve every problem by calculating and simulating. But if you want to let them build a prototype you first have to show them which side of the soldering iron to hold in their hand. Practical education is really limited, not just here in the Netherlands but also in Asia.
I agree that it's useful to simulate to see what the signals should (or could) look like. But most of my customers still want hardware to put in their system to test with their hardware asap. They never ask for simulation results. The only virtual thing most customers ask for is a mechanical model of the PSU to see if it will fit in their appliaction.
My way of working is still:
choose a topology that suits the needs of that particular requirement,
build a prototype to verify if you can get it to fullfil the requirements in real life
if it does start from there to make a final product.
And what if it doesn't: try harder, or start over with another topology.
This still requires a lot of experience and a good discussion among colleagues. So I hope that all the young and upcoming power engineers have (and see) the benefit of working with experienced engineers. That's the way we learned it too.
09-13-2013 10:06 PM
I'll give POWER 4-5-6 a try. Thanks for the tip.
09-14-2013 12:08 AM
Simetrix has a strong support for power supply design & simulation - it came with a penalty of heavy cost , about 5000 USD
Also there is an electromagnetic PCB analyze tool called EMIStream to help you pass EMI certification the easy way (the penalty here is even higher.....)
09-14-2013 02:10 AM
I've come to love LTSpice (but I am an FAE working for LTC). There is a large yahoo users forum and lots of 3rd party models to import so you're not necessarily locked in to only LTC parts. I never used it prior to joining Linear and generally scorned simulation. I was instead married to MathCAD and the bench. But I've come to appreciate it because of the abililty it gives to quickly generate a "starting point", a place to begin your circuit development with a high confidence level. That said, it is just another tool in your tool box like your scope, dvm, or dremel tool. An experienced designer should know how to use each one efficiently as well as their respective limitations.
09-14-2013 06:34 AM
There is a demo version available at http://www.ridleyengineering.com/power-4-5-6-demo-software.html
Like the full version it requires Microsoft Excel.
Please note that the 64-bit version of Excel doesn't work properly, it is missing many features. Microsoft always provides the option of installing either the 32-bit or 64-bit version and naturally, as engineers, we like to go for 64-bit. Microsoft don't exactly tell you at the time that there are features missing in the 64-bit version.
09-14-2013 11:53 AM
Get the tools for the job like any other profession.
If you want to extract the Bode plot of a closed loop switching regulator with all the important parasitics included. Then SIMetrix SIMPLIS. This runs far quicker than spice. I beleive it partitions up the circuit applies linear state equations. FFT algorithms is used to extract the gain and phase any where around the loop.
They have some PWM control IC models like UC3843 and PFC chip UC1845.
It needs a probe to be applied at a suitable node to tell the software where the transitions occure and mean value. I finf there is usually a good position to apply this probe. I used it to view the onsett of the double pole magnitude at the half sample frequency in a recent worse case analysis of a current mode buck pre-regulator followed by P/P converter confifiguration.
09-14-2013 02:06 PM
from this topic ,i know some new simulated software, some software don't be used ,for example ,power 4-5-6,SIMetrix SIMPLIS .in CHINA, saber and pespice is more popular.
i always use the saber to simulate. but sometime we can't get the suitable model
09-14-2013 04:58 PM
Power 4-5-6 is a great tool for simulating actual operation of a switching power supply circuit. I use it for determining the dissipation of the switching transistors within an IC to keep the dissipation below the maximum package dissipation. I also use it as a critical sanity check. I have known Ray for 28 years and he has my utmost respect as a truly knowledgeable academic and a truly practical power supply designer. He is one of the vital resources in the industry.
I had also taken a flyback design, which I had designed and compared Dr Rays results against my "15 minute" "Cookbook" manual calculations and found the values within 5% of my own. This not only reaffirmed my confidence, but reaffirmed the basic understanding of feedback loop compensation after many years of obscure derivations by previous authors.
Ray and I have both been attacked by the Chinese and for me, also the Russians by publishing our works free on websites. For me, it is not worth my time to revise my Cookbook again. Ray has a way to further protect his work. To my dismay, it has shut down information sharing. One never gets rich from technical publications, especially when one purposely publishes it in paperback to keep the price down for the buyers. I am proud to have contributed to the body of knowledge of power supply designers. The game has changed. Enjoy the endless challenge of switching power supply design.