开云体育

course of HOWTO build models like (sanken) STR-A6100

[email protected] All Messages By This Member #89714 Hi, : If someone wants to build his own model (without the help from vendor), for example: (sanken) STR-A6100. What kind of background knowledge should be prepared, or just view it as a sub-circuit like the (ST) L6561. If so, is the information mentioned in the datasheet good enough to finish the sub-circuit model ? Is this the domain of IC design ? The IC design engineer use spice to estimate in the first development stage ? Best regards.
Jim Wagner All Messages By This Member #89716 开云体育 To answer your last question: yes, at least at Linear Technology. That is one of the reasons why there are proprietary (encrypted) models - to keep from revealing the inner workings of their products. Mike Engelhard, the author, provides additional tools for the IC folks and it is the way it is, in part, because of having to work with other tools. Jim James Wagner Oregon Research Electronics toggle quoted message Show quoted text On Jul 25, 2016, at 6:12 PM, ericsson.sunshine@... [LTspice] <LTspice@...> wrote: Hi, : If someone wants to build his own model (without the help from vendor), for example: (sanken) STR-A6100. What kind of background knowledge should be prepared, or just view it as a sub-circuit like the (ST) L6561. If so, is the information mentioned in the datasheet good enough to finish the sub-circuit model ? Is this the domain of IC design ? The IC design engineer use spice to estimate in the first development stage ? Best regards.
Vlad All Messages By This Member #89722 Hello Ericsson > If someone wants to build his own model (without the help from vendor), for example: (sanken) STR-A6100. > What kind of background knowledge should be prepared, or just view it as a sub-circuit like the (ST) L6561. > If so, is the information mentioned in the datasheet good enough to finish the sub-circuit model ? That depends on the datasheet. If the information it provides is "good enough", then you could try to approximate the device using A-devices, ideal sources, etc, such that, overall, the response is close enough to reality that the model can be used. One very good example is Linear's frequency-independent RMS-to-DC convertec (LTC1967). You could model the sigma-delta and whatever extra circuitry there is inside, but, at its minimum, all you really need is this: B1 out 0 I=V(in)**2/V(out) C1 out 0 {0.5/pi/Fmin} Rpar=1 Where Fmin is the minimum frequency for the single pole at the output. This is just an example. Vlad ______________________ -- holding, among others: a universal analog/digital filter, block-level models for power electronics (and not only), math blocks with a more stream-lined approach, some digital ADC, DAC, (synchronous-)counter, JKflop, etc.
[email protected] All Messages By This Member #89728 Hi, Jim, Vlad: Thank you all. I think I don't need the model must be very precisely. I just need the behavior of it in the simulation rather than think "it" only in brain. For those power loss, efficiency, provided by the vendor would be better. Best regards.
jtanalog All Messages By This Member #89732 At 11:32 PM (-0700) 7/25/2016, ericsson.sunshine wrote: ---------- Original Message ---------- Hi, : If someone wants to build his own model (without the help from vendor), for example: (sanken) STR-A6100. What kind of background knowledge should be prepared, or just view it as a sub-circuit like the (ST) L6561. It helps to have a background in integrated circuit design so that you understand how all those blocks really work. Then you need a good strong math background if you want to create behavioral-level models. If so, is the information mentioned in the datasheet good enough to finish the sub-circuit model ? Yes... usually. Though often the datasheet doesn't include important data like quiescent current as the power supplies are ramped up. Is this the domain of IC design ? The IC design engineer use spice to estimate in the first development stage ? I create a lot of behavioral models commercially. The best approach is when the IC house will provide me a transistor-level netlist of the device to be modeled... then I can create a behavioral model that matches the chip BUT doesn't divulge any proprietary information. Encryption is then unnecessary, and the model can be used on any Berkeley-compliant Spice simulator. (I charge extra to develop models that will work on LTspice due to LTspice's quirky convergence and time-step-too-small issues... a royal pain in the... >:-} Best regards. ---------- End of Original Message ---------- ...Jim Thompson Web Site: <>
[email protected] All Messages By This Member #89761 Hi, Jim: Thank you. I am trying. But some info like "discharge" mechanism, not mentioned in the datasheet, I'm not sure if it's important or not. And the current feedback to control the PWM on\off (through a comparator), datasheet said that there is a amplifier for the shunt res's current before doing this PWM comparator, that's not clear too. As you said, it depends on the datasheet's content. (Vlad said it too.) Best regards.
jtanalog All Messages By This Member #89774 At 10:04 AM (-0700) 7/27/2016, ericsson.sunshine wrote: ---------- Original Message ---------- Hi, Jim: Thank you. I am trying. But some info like "discharge" mechanism, not mentioned in the datasheet, I'm not sure if it's important or not. It's a switch, dumping the timing ramp... download the AppNote: <> And the current feedback to control the PWM on\off (through a comparator), datasheet said that there is a amplifier for the shunt res's current before doing this PWM comparator, that's not clear too. As you said, it depends on the datasheet's content. (Vlad said it too.) Best regards. ---------- End of Original Message ---------- And years of experience helps as well... I was just musing that I designed my first OpAmp (MC1530/31) some 53 years ago! And it's still being sold! After all those years of IC design I tend to know exactly what's in a chip just from the datasheet/block diagram without ever seeing the schematic >:-} ...Jim Thompson Web Site: <>
[email protected] All Messages By This Member #89807 Hi, : I have done the (sanken) STR-A6100 sub-circuit for test, but this is the first time I do this thing (make spice model of some vendor by myself). I am not sure (not confident) that it's good. file is at: Can someone (familiar with this stuff (power IC, flyback)) provide some suggestions ? In fact, I have many questions that can't be answered ,almostly is the parameters like, charging current (Ix), and FB comparator's divider res, Ishunt gain, etc. But the most significant one is that I had been aware of the waveform of FB comparator's +,- (inverting & non-inverting) , these named "Vocpm" & "Vsc" in the datasheet at page.15. The picture in page.15 shown that the Vsc should be a vertical straight line, but my simulation shows it's rippled, though the output seems a little bit well. I just not sure whether it's right or not, I don't have the answer. Any opinion is appreciating. ? Best regards.
Andy All Messages By This Member #89819 ericsson.sunshine?wrote: ? ?"But the most significant one is that I had been aware of the waveform of FB comparator's ... ? ? The picture in page.15 shown that the Vsc should be a vertical straight line ..." I think you meant to write horizontal line. ? ?"...?but my simulation shows it's rippled ..." You did say that ANY opinion is appreciated, so here is mine.? This is based on knowing nothing about the device, the datasheet, or the SPICE model. I'm guessing the waveforms in that plot on page 15 are idealistic, not realistic. It is an internal voltage inside the IC, right?? Chances are it won't be steady, not a straight flat line.? I don't know what drives it, because the figure in the datasheet shows an empty box going to that node.? Is the voltage on the FB/OLP pin steady?? I guess it must be rippled too. Your uploaded file doesn't have the "TL431(TI)" (both symbol and model). Andy
[email protected] All Messages By This Member #89841 Hi, Andy: Thank you for the opinion. Yes, it's horizontal. (It's late, and I felt sleepy.) I have fixed the TL431/TI with LT1431 (LTspice built-in lib). The file is at here: I had been aware of the Ix current (constant current source to feed FB) would affect the Vsc ripple, but it's not mentioned the absolute value. I guess that's secret parameters of the vendor. I don't have the ability to reveal any secrets of the vendor, it's just for fun and some simple circuit topology test. Any opinion is appreciated. Best regards. ?
[email protected] All Messages By This Member #89944 ericsson: よろしく、お愿いします Good day. It's excited here to see someone who is interested in the (sanken) product STR-A6100 series. The Japan's companies seems not good at promotion of their products though they have good qualities. I personally think they should release more SPICE tool & support in simulation. About the STR-A6100, may I ask a question, ?could the example really be used in the real world ? Is that an exact design instance ? Thank you とも、ありがとう、ございます田村健太 Tamura Kentai
[email protected] All Messages By This Member #89945 Hi, Tamura Kentai: The example is NOT suitable to used in the real world, it's just a demo to see if the STR-A6100 could behave well. It's absolutely not suggested to use it in real world. Especially, the transformer was just roughly put parameters, it's meaningless and absolutely NOT proper-tunned. And the model of STR-A6100 is just simple sub-circuit for behavior reference. I really hope to see the day, those companies could release those models of SPICE to their customers to spread their application of those products. For example, if I never use the STR-A6100 IC, I would never know there is a company called (sanken) exists in the world. Thank you for your opinions. Best regards.?

Previous Topic Next Topic