I have boards with 1100 parts. 10-layer boards with several FPGAs and
22 power supplies. Boards full of Eclips Plus logic and SiGe comparators and power PHEMTS and RF transistors. Boards that sell for $11,000. If I can knock out a switcher design in an hour, using an LTC part, I can move on to the hard/fun stuff.
If I expect to build thousands of something relatively cheap (which I prefer to not do!) I might spend some time evaluating cheaper parts, which might not have a decent Spice model at all.
On that 22-supply board, I used mostly TPS54302 synchronous switchers, very cool little parts. I breadboarded five different voltage outputs with different inductors and caps and loop compensations. I have never got WebBench to load on my PC.
--
John Larkin Highland Technology, Inc
picosecond timing precision measurement
It's not a sales tool, IMO, but a support tool. Imagine that your job is to answer the phone when people call with questions about one of your company's hundreds of SMPS controllers. Being able to say "Send us a simulation" or "Check out this example" has got to be more or less priceless.
That'll be why they originally called it "SwitcherCAD." It wasn't originally meant to model opamps and 6L6s and stuff.
Wait a minute. You said competent engineers optimized the designs even for low volume production.
You can't even search TI for an SMPS in a half an hour. It takes their FAE longer than that to find the best fit for some applications (ones he hasn't already searched).
Exactly. For low volume, high margin, production it doesn't make sense to spend a lot of time on support circuits. High-volume, low margin (they go together) designs, one spends a *lot* longer than a half an hour optimizing things.
Boy, ain't that the truth! I don't know how many projects I've lost, saying, "You're not listening, so I'm outta here."
My personal burn is they come to me with a board heavily populated by OpAmps everywhere, and want me to simply put it all on a chip. When I ask for an overall system spec, they get all grousey. It's hard for me to convince these dorks that an OpAmp function, on-chip, can often be replaced with a couple of transistors... if any are needed at all.
Some of those very dorks lurk here >:-} ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
The LT switcher models are very good, in my limited experience. Mike says that that's because they're real transistor-level models and so have to be encrypted.
(I try to use LM2594s for everything, myself. No modelling required.) ;)
SPICE IC models are sufficiently crappy that if I'm relying on them for anything delicate I have to breadboard anyway. Discrete circuitry is the only place I'm prepared to actually trust it at all.
The point was about who was the biggest rock star, Mike or Me, not Spice. The spice bit was just a filter to avoid counting that Newfoundland minister...
I wasn't aware mike played guitar at all, if he does, he's probably shit :-)
Look, we are trying to estimate the worth of LTSpice to sales, of all sales. Sure, low production runs, of low part count might well have an incompetent engineer unable to design in parts other than what is in his simulator,and hence spend $1,000 on LT Parts for the whole run.
BUT low production runs, don't generate much money for LT. Because, they are dah... low production runs. Its the larger runs that will dominate profits. The larger the production run, and the larger the number of parts, the more money to LT, and the more incentive to avoid giving that money to LT. Dah....
You need to sum up the contributions from all sources. So, sure, LTSpice has some value, but those requiring larger number of parts, that generate the most money for LT, will have a larger probability of need to check out the optimum part from other sources.
As for "optimize", its a no brainer for me to check out in a tiny fraction of the time needed to layout a board, design the system, do the design reviews etc...etc.., whether or not another part would be a better fit. Its truly trivial. I did that for 15 years, before the internet existed. We used data books then.
So, yes, even for low production runs *I* would indeed "optimise" by looking for alternatives, which would take maybe, an hour or so.
If it is really true, that 99% of engineers designing products today, are so incompetent to not check out alternates, than I retract all my claims.
Your trying to claim that, its an amazing, staggering amount of work to do minimal checkout of alternatives. I don't.
Getting a design together, even for run of 100, is going to take way, way more time than checking out another parts
What size is the design? 100 chips, 1000 chips. How long does it taker to design a system with 1000 chips. If its a two op amp, 100 off guitar effects pedal, how much money will that get LT?
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.