Proprietary SMPS design tools, what for?
in [Design]
|
From: Joerg on 7 Sep 2009 15:05 JosephKK wrote: > On Thu, 03 Sep 2009 16:50:11 -0400, Spehro Pefhany > <speffSNIP(a)interlogDOTyou.knowwhat> wrote: > >> On Wed, 02 Sep 2009 09:54:28 -0700, Jim Thompson >> <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >> >>> On Wed, 02 Sep 2009 16:38:34 GMT, qrk <SpamTrap(a)spam.net> wrote: >>> >>>> On Tue, 01 Sep 2009 12:25:07 -0700, Joerg <invalid(a)invalid.invalid> >>>> wrote: >>>> >>>>> Tim Wescott wrote: >>>>>> On Tue, 01 Sep 2009 09:09:59 -0700, Joerg wrote: >>>>>> >>>>>>> Today I and probably others received the announcement of the umpteenth >>>>>>> SMPS "iteractive design simulator", EDesign Studio or whatever from ST. >>>>>>> I am not even going to try it. I've hung up on those things a long time >>>>>>> ago, after WebBench from National flagged the third of my designs as >>>>>>> "cannot be ..." (all in mass production now). The millisecond you try >>>>>>> something unorthodox which I always tend to do they fall off the cliff >>>>>>> anyhow. So I use LTSpice for everything. Since it seems that only LTC >>>>>>> furnishes good SPICE models for their PWM parts this has brought them >>>>>>> quite some business from my side. >>>>>>> >>>>>>> What's the point with all this proprietary stuff? >>>>>> Because most designers don't really have a clue about what they're doing, >>>>>> and don't want to. So spoon-feeding them super-safe circuits that are >>>>>> designed by a machine that's designed by some zit-faced kid in India will >>>>>> lead them to better success than trying to teach them basic principals. >>>>>> >>>>>> (Note that LTSpice _is_ proprietary, and part of the reason the models >>>>>> for Linear parts work better in it is because they use the proprietary >>>>>> features of LTSpice, not 'regular' spice models which LTSpice can't use >>>>>> as fast as it's 'own' stuff). >>>>>> >>>>> Sure, but: I can use other companies' stuff and just accept a slower >>>>> simulation speed because the sub-circuits become kind of hefty. Gets me >>>>> to the goal line while those online calculators never did. I have yet to >>>>> encounter a situation where LTSpice would reject a proper SPICE model of >>>>> a non-LTC part. >>>>> >>>>> >>>>>>> Why can't they invest >>>>>>> their time in much more useful activities like furnishing proper SPICE >>>>>>> subcircuits? >>>>>> Because in most companies marketing is an expense, not a profit center, >>>>>> and a proper SPICE circuit is 'too expensive'. >>>>>> >>>>>> Why did Zilog spend 20 years driving away any designer who wasn't going >>>>>> to order 20000 parts at a whack? Because they're stupid! After they >>>>>> taught all those kids to shun them, they had to go and convince them (me >>>>>> amongst them) to use Zilog after all -- and I still won't. >>>>>> >>>>> And here I thought they had already gone belly-up :-) >>>>> >>>>> >>>>>>> The most daft answer I got was along the lines of "We'll >>>>>>> only create a SPICE model if the business volume warrants it". As if I'd >>>>>>> be so stupid to promise them sales volume before test driving. Long >>>>>>> story short that business volume went to a competitor. >>>>>> When you say 'daft', your questions answer themselves. >>>>>> >>>>> :-) >>>> I'm amazed at the support the LT gives. Even for us little guys, we >>>> are given eval boards and lots of support from their field engineer. >>>> Plus, a decent way to simulate their switchers which I tend to use in >>>> unconventional ways that WebBench couldn't deal with. WebBench is so >>>> slow and limiting I gave up. The one National switcher controller I >>>> did use required doing a prototype board with a couple different >>>> configurations of the switcher. Fortunately, the overly simplified >>>> model of the NS part I created in LTspice was close enough to reality. >>>> >>>> Our LT rep said most of LT's business is smaller companies. >>> Part of the issues is CHEAP-CHEAP. A few years ago I was approached >>> by a major ANALOG company (you can guess who :-) to model a complex >>> device. I figured about a week of my time to do it proper justice. >>> Too expensive for them :-( >>> >>> ...Jim Thompson >> Seems like they (LT) decided to solve the problem of creating models >> efficiently, rather than create bespoke models one at a time-- which >> is why they can claim to have more SMPS chip models than all the rest. >> >> "I see no reason for the continuing existence of AMD " >> -- Mike E. > > I can see at least one good reason to keep AMD, a sufficient thorn in > Intel's side to keep them halfway honest. There may be more reasons. AMD can make darn good processors. My laptop has a 64-bit Turion in there and while older that thing still gives newer machines at clients a run for their money when doing sims. Best was a session where we did hands-on design on Cypress PSoC. About eight guys starting the compile at the same time. When I signaled mine was done a couple of guys across the table said "WHAT?". They had freaking expensive Thinkpads. -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: JosephKK on 7 Sep 2009 14:45 On Thu, 03 Sep 2009 16:50:11 -0400, Spehro Pefhany <speffSNIP(a)interlogDOTyou.knowwhat> wrote: >On Wed, 02 Sep 2009 09:54:28 -0700, Jim Thompson ><To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: > >>On Wed, 02 Sep 2009 16:38:34 GMT, qrk <SpamTrap(a)spam.net> wrote: >> >>>On Tue, 01 Sep 2009 12:25:07 -0700, Joerg <invalid(a)invalid.invalid> >>>wrote: >>> >>>>Tim Wescott wrote: >>>>> On Tue, 01 Sep 2009 09:09:59 -0700, Joerg wrote: >>>>> >>>>>> Today I and probably others received the announcement of the umpteenth >>>>>> SMPS "iteractive design simulator", EDesign Studio or whatever from ST. >>>>>> I am not even going to try it. I've hung up on those things a long time >>>>>> ago, after WebBench from National flagged the third of my designs as >>>>>> "cannot be ..." (all in mass production now). The millisecond you try >>>>>> something unorthodox which I always tend to do they fall off the cliff >>>>>> anyhow. So I use LTSpice for everything. Since it seems that only LTC >>>>>> furnishes good SPICE models for their PWM parts this has brought them >>>>>> quite some business from my side. >>>>>> >>>>>> What's the point with all this proprietary stuff? >>>>> >>>>> Because most designers don't really have a clue about what they're doing, >>>>> and don't want to. So spoon-feeding them super-safe circuits that are >>>>> designed by a machine that's designed by some zit-faced kid in India will >>>>> lead them to better success than trying to teach them basic principals. >>>>> >>>>> (Note that LTSpice _is_ proprietary, and part of the reason the models >>>>> for Linear parts work better in it is because they use the proprietary >>>>> features of LTSpice, not 'regular' spice models which LTSpice can't use >>>>> as fast as it's 'own' stuff). >>>>> >>>> >>>>Sure, but: I can use other companies' stuff and just accept a slower >>>>simulation speed because the sub-circuits become kind of hefty. Gets me >>>>to the goal line while those online calculators never did. I have yet to >>>>encounter a situation where LTSpice would reject a proper SPICE model of >>>>a non-LTC part. >>>> >>>> >>>>>> Why can't they invest >>>>>> their time in much more useful activities like furnishing proper SPICE >>>>>> subcircuits? >>>>> >>>>> Because in most companies marketing is an expense, not a profit center, >>>>> and a proper SPICE circuit is 'too expensive'. >>>>> >>>>> Why did Zilog spend 20 years driving away any designer who wasn't going >>>>> to order 20000 parts at a whack? Because they're stupid! After they >>>>> taught all those kids to shun them, they had to go and convince them (me >>>>> amongst them) to use Zilog after all -- and I still won't. >>>>> >>>> >>>>And here I thought they had already gone belly-up :-) >>>> >>>> >>>>>> The most daft answer I got was along the lines of "We'll >>>>>> only create a SPICE model if the business volume warrants it". As if I'd >>>>>> be so stupid to promise them sales volume before test driving. Long >>>>>> story short that business volume went to a competitor. >>>>> >>>>> When you say 'daft', your questions answer themselves. >>>>> >>>> >>>>:-) >>>I'm amazed at the support the LT gives. Even for us little guys, we >>>are given eval boards and lots of support from their field engineer. >>>Plus, a decent way to simulate their switchers which I tend to use in >>>unconventional ways that WebBench couldn't deal with. WebBench is so >>>slow and limiting I gave up. The one National switcher controller I >>>did use required doing a prototype board with a couple different >>>configurations of the switcher. Fortunately, the overly simplified >>>model of the NS part I created in LTspice was close enough to reality. >>> >>>Our LT rep said most of LT's business is smaller companies. >> >>Part of the issues is CHEAP-CHEAP. A few years ago I was approached >>by a major ANALOG company (you can guess who :-) to model a complex >>device. I figured about a week of my time to do it proper justice. >>Too expensive for them :-( >> >> ...Jim Thompson > >Seems like they (LT) decided to solve the problem of creating models >efficiently, rather than create bespoke models one at a time-- which >is why they can claim to have more SMPS chip models than all the rest. > >"I see no reason for the continuing existence of AMD " >-- Mike E. I can see at least one good reason to keep AMD, a sufficient thorn in Intel's side to keep them halfway honest. There may be more reasons.
From: Spehro Pefhany on 3 Sep 2009 16:50 On Wed, 02 Sep 2009 09:54:28 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >On Wed, 02 Sep 2009 16:38:34 GMT, qrk <SpamTrap(a)spam.net> wrote: > >>On Tue, 01 Sep 2009 12:25:07 -0700, Joerg <invalid(a)invalid.invalid> >>wrote: >> >>>Tim Wescott wrote: >>>> On Tue, 01 Sep 2009 09:09:59 -0700, Joerg wrote: >>>> >>>>> Today I and probably others received the announcement of the umpteenth >>>>> SMPS "iteractive design simulator", EDesign Studio or whatever from ST. >>>>> I am not even going to try it. I've hung up on those things a long time >>>>> ago, after WebBench from National flagged the third of my designs as >>>>> "cannot be ..." (all in mass production now). The millisecond you try >>>>> something unorthodox which I always tend to do they fall off the cliff >>>>> anyhow. So I use LTSpice for everything. Since it seems that only LTC >>>>> furnishes good SPICE models for their PWM parts this has brought them >>>>> quite some business from my side. >>>>> >>>>> What's the point with all this proprietary stuff? >>>> >>>> Because most designers don't really have a clue about what they're doing, >>>> and don't want to. So spoon-feeding them super-safe circuits that are >>>> designed by a machine that's designed by some zit-faced kid in India will >>>> lead them to better success than trying to teach them basic principals. >>>> >>>> (Note that LTSpice _is_ proprietary, and part of the reason the models >>>> for Linear parts work better in it is because they use the proprietary >>>> features of LTSpice, not 'regular' spice models which LTSpice can't use >>>> as fast as it's 'own' stuff). >>>> >>> >>>Sure, but: I can use other companies' stuff and just accept a slower >>>simulation speed because the sub-circuits become kind of hefty. Gets me >>>to the goal line while those online calculators never did. I have yet to >>>encounter a situation where LTSpice would reject a proper SPICE model of >>>a non-LTC part. >>> >>> >>>>> Why can't they invest >>>>> their time in much more useful activities like furnishing proper SPICE >>>>> subcircuits? >>>> >>>> Because in most companies marketing is an expense, not a profit center, >>>> and a proper SPICE circuit is 'too expensive'. >>>> >>>> Why did Zilog spend 20 years driving away any designer who wasn't going >>>> to order 20000 parts at a whack? Because they're stupid! After they >>>> taught all those kids to shun them, they had to go and convince them (me >>>> amongst them) to use Zilog after all -- and I still won't. >>>> >>> >>>And here I thought they had already gone belly-up :-) >>> >>> >>>>> The most daft answer I got was along the lines of "We'll >>>>> only create a SPICE model if the business volume warrants it". As if I'd >>>>> be so stupid to promise them sales volume before test driving. Long >>>>> story short that business volume went to a competitor. >>>> >>>> When you say 'daft', your questions answer themselves. >>>> >>> >>>:-) >>I'm amazed at the support the LT gives. Even for us little guys, we >>are given eval boards and lots of support from their field engineer. >>Plus, a decent way to simulate their switchers which I tend to use in >>unconventional ways that WebBench couldn't deal with. WebBench is so >>slow and limiting I gave up. The one National switcher controller I >>did use required doing a prototype board with a couple different >>configurations of the switcher. Fortunately, the overly simplified >>model of the NS part I created in LTspice was close enough to reality. >> >>Our LT rep said most of LT's business is smaller companies. > >Part of the issues is CHEAP-CHEAP. A few years ago I was approached >by a major ANALOG company (you can guess who :-) to model a complex >device. I figured about a week of my time to do it proper justice. >Too expensive for them :-( > > ...Jim Thompson Seems like they (LT) decided to solve the problem of creating models efficiently, rather than create bespoke models one at a time-- which is why they can claim to have more SMPS chip models than all the rest. "I see no reason for the continuing existence of AMD " -- Mike E.
From: Jim Thompson on 3 Sep 2009 16:59 On Thu, 03 Sep 2009 16:50:11 -0400, Spehro Pefhany <speffSNIP(a)interlogDOTyou.knowwhat> wrote: >On Wed, 02 Sep 2009 09:54:28 -0700, Jim Thompson ><To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: > >>On Wed, 02 Sep 2009 16:38:34 GMT, qrk <SpamTrap(a)spam.net> wrote: >> >>>On Tue, 01 Sep 2009 12:25:07 -0700, Joerg <invalid(a)invalid.invalid> >>>wrote: >>> >>>>Tim Wescott wrote: >>>>> On Tue, 01 Sep 2009 09:09:59 -0700, Joerg wrote: >>>>> >>>>>> Today I and probably others received the announcement of the umpteenth >>>>>> SMPS "iteractive design simulator", EDesign Studio or whatever from ST. >>>>>> I am not even going to try it. I've hung up on those things a long time >>>>>> ago, after WebBench from National flagged the third of my designs as >>>>>> "cannot be ..." (all in mass production now). The millisecond you try >>>>>> something unorthodox which I always tend to do they fall off the cliff >>>>>> anyhow. So I use LTSpice for everything. Since it seems that only LTC >>>>>> furnishes good SPICE models for their PWM parts this has brought them >>>>>> quite some business from my side. >>>>>> >>>>>> What's the point with all this proprietary stuff? >>>>> >>>>> Because most designers don't really have a clue about what they're doing, >>>>> and don't want to. So spoon-feeding them super-safe circuits that are >>>>> designed by a machine that's designed by some zit-faced kid in India will >>>>> lead them to better success than trying to teach them basic principals. >>>>> >>>>> (Note that LTSpice _is_ proprietary, and part of the reason the models >>>>> for Linear parts work better in it is because they use the proprietary >>>>> features of LTSpice, not 'regular' spice models which LTSpice can't use >>>>> as fast as it's 'own' stuff). >>>>> >>>> >>>>Sure, but: I can use other companies' stuff and just accept a slower >>>>simulation speed because the sub-circuits become kind of hefty. Gets me >>>>to the goal line while those online calculators never did. I have yet to >>>>encounter a situation where LTSpice would reject a proper SPICE model of >>>>a non-LTC part. >>>> >>>> >>>>>> Why can't they invest >>>>>> their time in much more useful activities like furnishing proper SPICE >>>>>> subcircuits? >>>>> >>>>> Because in most companies marketing is an expense, not a profit center, >>>>> and a proper SPICE circuit is 'too expensive'. >>>>> >>>>> Why did Zilog spend 20 years driving away any designer who wasn't going >>>>> to order 20000 parts at a whack? Because they're stupid! After they >>>>> taught all those kids to shun them, they had to go and convince them (me >>>>> amongst them) to use Zilog after all -- and I still won't. >>>>> >>>> >>>>And here I thought they had already gone belly-up :-) >>>> >>>> >>>>>> The most daft answer I got was along the lines of "We'll >>>>>> only create a SPICE model if the business volume warrants it". As if I'd >>>>>> be so stupid to promise them sales volume before test driving. Long >>>>>> story short that business volume went to a competitor. >>>>> >>>>> When you say 'daft', your questions answer themselves. >>>>> >>>> >>>>:-) >>>I'm amazed at the support the LT gives. Even for us little guys, we >>>are given eval boards and lots of support from their field engineer. >>>Plus, a decent way to simulate their switchers which I tend to use in >>>unconventional ways that WebBench couldn't deal with. WebBench is so >>>slow and limiting I gave up. The one National switcher controller I >>>did use required doing a prototype board with a couple different >>>configurations of the switcher. Fortunately, the overly simplified >>>model of the NS part I created in LTspice was close enough to reality. >>> >>>Our LT rep said most of LT's business is smaller companies. >> >>Part of the issues is CHEAP-CHEAP. A few years ago I was approached >>by a major ANALOG company (you can guess who :-) to model a complex >>device. I figured about a week of my time to do it proper justice. >>Too expensive for them :-( >> >> ...Jim Thompson > >Seems like they (LT) decided to solve the problem of creating models >efficiently, rather than create bespoke models one at a time-- which >is why they can claim to have more SMPS chip models than all the rest. > >"I see no reason for the continuing existence of AMD " >-- Mike E. I would imagine that the LT chip designer either creates the model himself, or works closely with the model creator. That's much easier than after the fact, being spoon-fed "declassified" data :-( ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | | http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.
From: Jim Thompson on 3 Sep 2009 17:19
On Thu, 03 Sep 2009 13:59:03 -0700, Jim Thompson <To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >On Thu, 03 Sep 2009 16:50:11 -0400, Spehro Pefhany ><speffSNIP(a)interlogDOTyou.knowwhat> wrote: > >>On Wed, 02 Sep 2009 09:54:28 -0700, Jim Thompson >><To-Email-Use-The-Envelope-Icon(a)My-Web-Site.com> wrote: >> >>>On Wed, 02 Sep 2009 16:38:34 GMT, qrk <SpamTrap(a)spam.net> wrote: >>> [snip] >>>>>:-) >>>>I'm amazed at the support the LT gives. Even for us little guys, we >>>>are given eval boards and lots of support from their field engineer. >>>>Plus, a decent way to simulate their switchers which I tend to use in >>>>unconventional ways that WebBench couldn't deal with. WebBench is so >>>>slow and limiting I gave up. The one National switcher controller I >>>>did use required doing a prototype board with a couple different >>>>configurations of the switcher. Fortunately, the overly simplified >>>>model of the NS part I created in LTspice was close enough to reality. >>>> >>>>Our LT rep said most of LT's business is smaller companies. >>> >>>Part of the issues is CHEAP-CHEAP. A few years ago I was approached >>>by a major ANALOG company (you can guess who :-) to model a complex >>>device. I figured about a week of my time to do it proper justice. >>>Too expensive for them :-( >>> >>> ...Jim Thompson >> >>Seems like they (LT) decided to solve the problem of creating models >>efficiently, rather than create bespoke models one at a time-- which >>is why they can claim to have more SMPS chip models than all the rest. >> >>"I see no reason for the continuing existence of AMD " >>-- Mike E. > >I would imagine that the LT chip designer either creates the model >himself, or works closely with the model creator. That's much easier >than after the fact, being spoon-fed "declassified" data :-( > > ...Jim Thompson Elaborating: The ideal way to make a _behavioral_ model is to start with a _device_level_ netlist, particularly if a back annotated (with strays) version can be had. Then it's just a matter of fitting behavioral functions. Unfortunately foundries are _very_ unlikely to give an outsider a true netlist :-( ...Jim Thompson -- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | Obama wee-wee'd on himself, that's no problem, But you should be worried that he poo-poo'd on your Constitution. |