------------- Do i need to simulate the whole dc/dc converter ? Well, we made few prototypes and the last one was working perfectly again until we hook up the transformer. Fuse was triggered and there was no damage to the pcb. btw. soft start is also implemented into prototype.
Because this is our fourth prototype i was thinking to simulate all this and maybe to detect what will happen when we connect the transformer again. this is the only problem, everything else is working fine.
In my opinion, and in this case, simulation should prevent such situations.
Webbench? That's IMHO a joke. I have tried it a few times and on several occasions it responded something like "the circuit cannot be designed with the data entered" or something like that. Well, some of them roll off the production lines for years now so I guess that software can't be very good.
They better come out with SPICE-compatible models. It's good for business, but that insight first has to sink in with upper management.
You have a basic troubleshooting situation that is best resolved by checking the physical parts present. This is Sesame St. 'One of these parts is not like the other..........'
You couldn't model a defect, without knowing what the defect is, anyways. Many faults could cause the symptom described, without telling you anything about the physical circumstances that actually exist.
This looks like a very useful part, but the model doesn't seem to work on LTspice.
formatting link
Thanks
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
One thing that slows most Spice, including LTspice and Berkeley 3f, to a crawl is trying to simulate a big (say 6-stage) CW multiplier. You need oodles of memory for it to complete at all.
--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
(Richard Feynman)
I've had numerous instances where Acrobat Reader makes a mess of printing documents, even those produced by Acrobat, whereas Ghostscript produces perfect results.
The Keithley 610C manual from their site is like that (scanned typescript).
--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
(Richard Feynman)
Well, I am not sure yet. TI has the tendency to dump support questions into "peer forums" or what they call E2E. They've done that to me twice. The first time I asked my question again there, no answers, so I walked. Response time is of the essence during designs. The competition's chip got designed in, meaning they lost business. Recently it was about a National part. This time a TI employee came online which was good. He tried his best and gave me all the info he had but he apparently was not an IC designer there. I could not obtain an answer that was exhaustive enough and walked over to the product of a competitor. Linear Technology, because they are hands-down the number one when it comes to applications support plus there is always a good simulator model. Too late for TI now, the project is done. In the old days of National I could contact switcher chip gurus directly, and it was usually the same guys. That has resulted in design-ins of National parts.
The topper at National was when I asked a question, and mine are usually tough ones, and their guru of PLL chips (Dean Banerjee) responded. Now that is what I call stellar service. It does happen at TI as well, for example when they helped me get in touch with an old group in Germany that did WDF filters for their MSP430. But this only happens once you are in contact with the real movers and shakers in the respective group. Since they started to defer to E2E that's not easy anymore.
But no matter what, they really do need to realize that TINA is not the ultimate cat's meouw. LTSpice is. Got to know when to fold'em :-)
Everything is relative. We get great service from the TI FAE and I'd get laughed out of the design review if I used anything from LT (well, I do use an LT I2C accelerator, but it's always an OPEN footprint). ;-)
When we ask the question, we get whoever knows the answer to answer. It doesn't take a second call. At the PPoE, it was a little different but the real problems were years ago. That was mostly fixed recently.
They know (they're not stupid). They're just not admitting it to you. ;-)
LTC is expensive, that's the downside. But their stuff works and when it comes to performance hardly anyone can touch them. When I design switchers it is usually about top notch performance.
Sometimes it's an economy switcher and then usually none of the big semi mfgs are used, then it's companies such as Richtek. Or I roll my own from scratch with jelly-bean parts if the 30c for the chip are still too much.
If they think that they can concentrate their support on large accounts then that's a mistake.
Well, I knew this several years ago. If they also know, then what takes them so long?
Others have similar performance. LTC excels in service to the engineers. Big companies get that level of service from everyone and can't afford LTC's overhead. Different strokes...
The 3A TI switcher I've been using lately is about $.30. The inductor is about twice the price of the silicon. LTC is twice to four times that.
They *have* to. It's also more cost effective going after the M piece order than the .1k order. OTOH, even at the PPoE we got good service from TI. Most of the others supported their stuff through disties, which wasn't all bad. Their support for ADC and Altera was very good.
They obviously don't agree that it's important (to them).
Sub-1ohm gate driver outputs at full standard level swing?
Can they beat this?
formatting link
Less than 20c a pop.
Distributors can be good. But my questions usually require the IC designers to get involved.
That kind of attitude has cost big companies their bread and butter. Just like it did for the big three automakers. Next in line might be the big test equipment places, which would be very sad. But I see the writing on the wall becoming bigger and bigger.
I'm more concerned with 16V (40V peak) inputs, but that's not a bad price.
5.5V input limit would severely limit my interest, even for a point of load regulator. My normal mode is to switch down to 5.8V (and 3.8V, though generally less current) as the bulk voltage and then down from there.
The disties' FAEs can get them involved. It's generally another level of bureaucracy, but it can happen. It even takes a while with direct FAEs, though they have a more direct financial interest. ;-)
Everyone has to decide where their priorities are. Only Democrats believe that money is free.
This market is a very tough one for our domestic companies, it has almost become like a commodity market. Lunches get eaten by others a lot.
Most
That's what is so great about LTC. The hand-off to the IC guys and usually the response all happen same day. When in the middle of a design I usually just don't have 2-3 "bureaucracy anticipation days".
Very true. It just makes me sad every time a large company screws up the priorities and in consequence hemorrhages big time.
For example, much of the exodus of technology in the lower end oscilloscope market and people like me buying Asian scopes (because they are better, not for price) could have clearly been avoided. This costs profits and real jobs.
Transformers are one of the most difficult components to simulate. This = is=20 especially true if it is a custom transformer, or a hand-wound = prototype.=20 There are just too many variables to take into account. But simulation=20 should get you to about 90%, even before you build the first prototype. = And=20 then when you have actual hardware you should use every piece of good = test=20 equipment available to determine what matches the simulation and what = does=20 not. Then you might be able to tweak the simulator models to approach = real=20 world performance, or you can tweak the components to make the prototype =
closely match what you see in the simulator (assuming it is what you = want).
One way to progress in the case of a transformer or inductor causing=20 catastrophic failure (including blown fuses), is to add some resistance = in=20 series, and maybe even a lamp which provides a good non-linear = resistance=20 and a "brilliant" indicator when there is an excess current situation. = It=20 would help if you could post more details of your schematic and overall=20 design criteria. Four prototypes and still far from success is pretty = bad.=20 Especially if it is something like this, where it still essentially = "doesn't=20 work". You should be able to get the first prototype to work, even if it =
involves ugly patches and glommed-on components. Then you commit to a = PCB,=20 and if it does not work then you need to look at layout. By the third=20 prototype you should be tweaking efficiency and reducing EMI and stuff = like=20 that.
I doubt even the best commercial programs will help in this case.
No, it needs at least a 40V no-smoke limit. The absolute maximum on the part is 26V and that's is a little on the skimpy side for a continuous spec, too.
The other requirement, that I missed before (200K to 2MHz is good) is an external sync pin.
Yeah, LDOs too. Under $.10 per is a little ridiculous. Can't even buy a decent PFET for that. ;-)
the
Most
It's nice if you can afford it. I can't, anymore. I've said it here before, I think, but LTC seems to have the opposite business model from them-that-can't-be-mentioned-here. LTC is all about service, and the price tag that goes with it. That's great, if you're not buying millions.
;-)
But they're not.
It doesn't help that "US scopes" are also "Asian scopes". Again, blame the Democrats for that one.
Take a look in the rearview: Why is it that Philips Semiconductor couldn't make a go of it anymore until it was also sold off to KKR? IIRC there was a $4B debt towards the end according to a Business Week article many years ago.
Or this:
formatting link
Key account thinking typically marries them to one or more of these markets: Automotive, consumer, PC. That's where you get "lopezed", where they squeeze the last dime of profit out of you.
For the loss of production activities? Yes, definitely. But technical or marketing blunders can squarely be blamed on individual companies. In this day and age, how can they seriously think a 2.5k or 4k sample memory for a DSO is going to fly? I thought that was a typo until I called them and found out it wasn't. That's when I walked.
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.