Webench and Quickstart models for LM2121x yield significantly different results

Hi,
Regarding the relationship between the Quickstart and WEBENCH Power Designer, the foundation calculations of both are the same. However the implementation is different due to the different environments, so the resulting component values may be a bit different. The inductor target value appears to be almost identical.

Note that WEBENCH Designer also includes Compensation Designer, which allows the user to target a specific crossover frequency and phase margin, and obtain values for the compensation components that will give that result.

Regarding versions or saving WEBENCH designs - Each design you create is saved in your "MyDesigns" listing. You can modify components, run simulations, modify more components, etc. on an individual design. The last state it was in is what's saved in MyDesigns.

If you are working in a design, and change component selection (including selecting custom parts), those components are saved until you change them again. The exception here is the compensation components, which can get over-ridden by Compensation Designer, when you click the button to "Apply changes to design."

Here's something to watch for, though - If you are working in the design, and want to see the operating values for a different load current or input voltage, it's best to go to the "OpVals" view and change them there. That keeps the design with the components you have selected. If you instead use the "Change Design Inputs" control, that will over-write your design with a fresh design matching your new requirements. Do this cautiously! If you want to keep the previous design, and create a new one for comparison, then click either the Visualizer or the New button (at the top of the page) and enter your requirements there, then click Recalculate. When you click Open Design for a specific solution, that creates a new design for you.

If these notes do not describe what you have experienced, please provide additional information and examples.

Hi Wanda,

Thanks very much for the helpful tips - I will try them out.

AFA switcherpro and webench models, if I understand you correctly, you are saying that I would be safe in using either one?  After I wrote the original posting, I surfed through the site some more, and several posts seemed to indicate that AFA TI advice was concerned, the consensus was that although switcherpro had features that webench lacked, it was better to use webench because it benefits from ongoing development, while switcherpro is frozen.  Since I am using the LM21215A-1, that's not a brand new part anyway.

I don't mean to debate this endlessly, I am just curious.

I have been tinkering with the design and checking for differences with the transient response simulator.  One thing I don't understand is why it seems that the transient response (AFA as the variations in Vout at the current edges) doesn't seem to be very sensitive, if at all, to the output C.  I expected that it would. 

I looked at both the LM21215A-1 and TPS56121, since I am using both of them in different parts of my design.  AFAICT, the LM21215A-1 has a terrible transient response, with transitions of well over 100mV.  The TPS is a lot better.  But, if I look at switcherpro for the LM21215A-1, the transient, although not graphed, is WAY better, and is in the range of a 30mV.

One further thing - by default, switcherpro doesn't do a full load transient the way webench does.  And, you can change the transient step size.  It doesn't appear that you can do that with webench.  So, with webench, I could be seeing bad transient response because the chip is in DCM. There isn't enough info to tell.

Any opinion on this ?


When I did the original design, I did use Webench.  At that point, either the SIM function wasn't in Webench or I didn't try it.  I don't remember. 

It makes me wonder if I have chosen badly in my application, since I am using the LM21215A-1 for several supplies.  Mostly I chose the LM21215A-1 because it accepted an external sync.  I have 4 supplies and run each on a different phase of and LT 4 phase clock generator to minimize ripple on the input rail. It appeared to be a good part and in two spins of the design I can't say I have made any changes to the initial comp values.  I just tweaked the voltages a little.  I have seen no problems with these supplies.

So, this is one reason why I am concerned.  What is real?

I am not sure what to do.  Should I just leave the circuit values as they have been since I did the original designs and not worry about? 

I was going to use this same supply setup in a derivative design, and the GPU in that design is more picky about transients.

Any advice, comments, etc would be much appreciated.

Thanks!

Victor







At 07:02 PM 3/27/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi,
Regarding the relationship between the Quickstart and WEBENCH Power Designer, the foundation calculations of both are the same. However the implementation is different due to the different environments, so the resulting component values may be a bit different. The inductor target value appears to be almost identical.

Note that WEBENCH Designer also includes Compensation Designer, which allows the user to target a specific crossover frequency and phase margin, and obtain values for the compensation components that will give that result.

Regarding versions or saving WEBENCH designs - Each design you create is saved in your "MyDesigns" listing. You can modify components, run simulations, modify more components, etc. on an individual design. The last state it was in is what's saved in MyDesigns.

If you are working in a design, and change component selection (including selecting custom parts), those components are saved until you change them again. The exception here is the compensation components, which can get over-ridden by Compensation Designer, when you click the button to "Apply changes to design."

Here's something to watch for, though - If you are working in the design, and want to see the operating values for a different load current or input voltage, it's best to go to the "OpVals" view and change them there. That keeps the design with the components you have selected. If you instead use the "Change Design Inputs" control, that will over-write your design with a fresh design matching your new requirements. Do this cautiously! If you want to keep the previous design, and create a new one for comparison, then click either the Visualizer or the New button (at the top of the page) and enter your requirements there, then click Recalculate. When you click Open Design for a specific solution, that creates a new design for you.

If these notes do not describe what you have experienced, please provide additional information and examples.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Victor,
Switcherpro has some similarities to WEBENCH, but you are correct in that it is not in development, while WEBENCH is very much in ongoing development & expansion. I don't expect you were able to run a design for LM21215 in Switcherpro, because that device was not supported by that tool. You may be thinking of the quickstart spreadsheet.

In WEBENCH Designer, when running transient simulations you do have the ability to change the input or load step. The LM21215 uses synchronous rectification, so it will not be in DCM. But comparing responses for different step sizes will give different results. (To change the load step size, click the pulse load symbol, and change the Nominal values for I1 and/or I2. If you see a table full of values for Component Simulation Parameters, then you are looking at the values for the previous simulation. Click the button "Show Latest Design Version" to prepare for the next sim.)

The LM21215 is a nice high-current regulator, especially when well-compensated. The TPS56121 looks like it would have similar performance. From what I can see, either would be a good choice.

At 09:07 PM 3/27/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,
Switcherpro has some similarities to WEBENCH, but you are correct in that it is not in development, while WEBENCH is very much in ongoing development & expansion. I don't expect you were able to run a design for LM21215 in Switcherpro, because that device was not supported by that tool. You may be thinking of the quickstart spreadsheet.

Oh, sorry, right.  I meant quickstart :)   I was running the TPS part on that.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Wanda,

Is it also the case that quickstart is "frozen"?

Thanks for your help.  It is hard to know where to turn to try to solve these mysteries :)

Victor



At 09:07 PM 3/27/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,
Switcherpro has some similarities to WEBENCH, but you are correct in that it is not in development, while WEBENCH is very much in ongoing development & expansion. I don't expect you were able to run a design for LM21215 in Switcherpro, because that device was not supported by that tool. You may be thinking of the quickstart spreadsheet.

In WEBENCH Designer, when running transient simulations you do have the ability to change the input or load step. The LM21215 uses synchronous rectification, so it will not be in DCM. But comparing responses for different step sizes will give different results. (To change the load step size, click the pulse load symbol, and change the Nominal values for I1 and/or I2. If you see a table full of values for Component Simulation Parameters, then you are looking at the values for the previous simulation. Click the button "Show Latest Design Version" to prepare for the next sim.)

The LM21215 is a nice high-current regulator, especially when well-compensated. The TPS56121 looks like it would have similar performance. From what I can see, either would be a good choice.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Victor,

I'm not aware of any continuing development on the quickstart spreadsheet for this device.

Hi Wanda,

I have attached the quickstart and webench equivalents files. 

You can see that the phase and gain plots look very similar but, for example, the zero crossing point for the gain is very different between the two programs.

I have made the input conditions as identical as I can and I don't see there the remaining differences are significant


I also "shared" the design with you with a couple of plots, including the most recent SIM.

Please let me know if you get all this.

Thanks,

Victor





At 01:58 PM 3/30/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,

I'm not aware of any continuing development on the quickstart spreadsheet for this device.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Wanda,

I know this email is going to sound like whining, but really, I am not quite sure how anyone can get reliable results out of webench.

I have spent more time than I care to admit trying to rationalize the results between quick-start and webench for the LM21215A-1. It is hard to believe that they share the same model

(well, on quickstart, I have to use LM21215 because they don't have the A-1, but that doesn't matter as the only difference is that one has a current limit pin and the other has externals sync).

I have entered identical components into both and get seriously different results for the phase and crossover calculations, not to mention ripple and transient dip, although these latter two are not so far apart.  It is not unusual for the phase margin and crossover freq to be a factor of 2 or 3 or even more apart.  And in some cases, webench will fail convergence altogether, hang, or otherwise fail to run.  When that happens, it says it can't talk to the webench server and wipes out all of my modifications and custom parts and plugs something else in.

And, why can't I can't get webench to startup with ceramics as the default caps?  And, why can't I specify transition peak and dip limits for the transient?  As it is, you have to go through this crazy iterative process that is a huge time waster and offers little hope of converging on the desired solution.

Plus, webench seems to have a bug in that if you have 6x output caps, it fails with a "connection" failure.

If I use re-comp to try to tune the results manually, the 4th parameter (gain margin) sometimes shows up and sometimes not. 

And, if I run the auto re-comp, every time I tell it to do an auto compensate it provides different values.

And, sigh, I have found that the only way to get reasonable results in manual is to first set the 3rd pole to multi-megahertz number. That puts the B parameter (whatever that is for) off the scale and will then generate values the sim will then actually complete with.  BUT, in auto, after one or two auto cycles it pulls B into range, sometimes very close to A, and this provides unusable results.

And, why the heck can't I truly save a working instance?  What is wrong with having a real save operation.

Sorry to dump on you - I have spent 2 weeks on this thing and have gotten no where with it.

And yet, when I used webench about 3 or 4 years ago, it gave me good results, at least the circuits have been working.  If I plug in the values I got before using re-comp now, it basically says my designs shouldn't be working.  I wish I could run the webench versions from 3 and 4 years ago.


Thanks!

Victor








At 01:58 PM 3/30/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,

I'm not aware of any continuing development on the quickstart spreadsheet for this device.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Victor,
Thanks for sharing your experiences in creating designs with us. I can offer a couple of techniques that can make things go a little easier, plus additional info on the LM21215 quickstart spreadsheet.

First, on differences between the quickstart spreadsheet and WEBENCH design calculations - the WEBENCH calculations use a slightly different value for the ramp voltage, which gives better matching to bench data. The result is that the crossover frequency may be a bit different between the two tools, but the phase margin is usually pretty close.

Now, some tips:
1) Saving versions of designs - When you have a WEBENCH design in a state you want to keep, click on the MyDesigns/Projects button. This brings up a listing of your WEBENCH designs, most recent one on top. Identify the one you want to create a new version of, then note the buttons on the right side of that row, in the Actions field. The 3rd button is for "Copy Design." Click that, change the design name or description, and click "Copy This Design." It will appear in your designs listing. You can now edit the new design - or keep it as a preserved version and edit the original.

2) Creating design with ceramic output capacitor - In the Visualizer tool (which shows the solutions matching your design requirements), in the Change Inputs panel, there is a link for "Use Advanced Options." Click that, and see on the right a section for Component Selection. The third line is for Cap Selection. Click the box for "Ceramic Only," then click the green Recalculate button at the bottom. The resulting designs will be using ceramic output capacitors.

3) Auto-Recompensation yields varying solutions - To get to a recompensated solution quickly that meets the targets, the Compensation Designer incorporates a bit of randomness. The result is compensation components that may be different each time the utility is run.

4) Simulation convergence - The LM21215 model can be a bit finicky. If the design's phase margin is low, under about 25deg, it can cause problems with convergence.

We'd be glad to talk with you about your experience with the Compensation Designer and with this design in general. If you like, email me at wanda.garrett@ti.com and I'll set up a phone meeting for next week.

Hi Wanda,

I tried the Advanced Options and it actually doesn't work correctly. It thinks that the LM21215A-1 doesn't support soft start, 1% ripple, or external sync.  Of course, it does all of these things.  And then, I told it ceramic only and it didn't observe that.  Sigh.

Victor






At 03:57 PM 4/2/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,
Thanks for sharing your experiences in creating designs with us. I can offer a couple of techniques that can make things go a little easier, plus additional info on the LM21215 quickstart spreadsheet.

First, on differences between the quickstart spreadsheet and WEBENCH design calculations - the WEBENCH calculations use a slightly different value for the ramp voltage, which gives better matching to bench data. The result is that the crossover frequency may be a bit different between the two tools, but the phase margin is usually pretty close.

Now, some tips:
1) Saving versions of designs - When you have a WEBENCH design in a state you want to keep, click on the MyDesigns/Projects button. This brings up a listing of your WEBENCH designs, most recent one on top. Identify the one you want to create a new version of, then note the buttons on the right side of that row, in the Actions field. The 3rd button is for "Copy Design." Click that, change the design name or description, and click "Copy This Design." It will appear in your designs listing. You can now edit the new design - or keep it as a preserved version and edit the original.

2) Creating design with ceramic output capacitor - In the Visualizer tool (which shows the solutions matching your design requirements), in the Change Inputs panel, there is a link for "Use Advanced Options." Click that, and see on the right a section for Component Selection. The third line is for Cap Selection. Click the box for "Ceramic Only," then click the green Recalculate button at the bottom. The resulting designs will be using ceramic output capacitors.

3) Auto-Recompensation yields varying solutions - To get to a recompensated solution quickly that meets the targets, the Compensation Designer incorporates a bit of randomness. The result is compensation components that may be different each time the utility is run.

4) Simulation convergence - The LM21215 model can be a bit finicky. If the design's phase margin is low, under about 25deg, it can cause problems with convergence.

We'd be glad to talk with you about your experience with the Compensation Designer and with this design in general. If you like, email me at wanda.garrett@ti.com and I'll set up a phone meeting for next week.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Wanda,

Another weirdness I have found with webench for the LM21215A-1 is that I can get a nice well-damped transient response waveform with a particular set of CC1, etc components and then, if I try to run the exact same circuit again a couple of minutes later, and with no changes the SIM won't run  - it just times out.
Or, if I start all over again and recreate the circuit, it won't run at all.

I have to do some more experimenting, always painful and time consuming, but I think there may be some dependence on the SS (smooth start), if I use something other than the default 0.5ms setting.

I will let you know.

Victor











At 03:57 PM 4/2/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,
Thanks for sharing your experiences in creating designs with us. I can offer a couple of techniques that can make things go a little easier, plus additional info on the LM21215 quickstart spreadsheet.

First, on differences between the quickstart spreadsheet and WEBENCH design calculations - the WEBENCH calculations use a slightly different value for the ramp voltage, which gives better matching to bench data. The result is that the crossover frequency may be a bit different between the two tools, but the phase margin is usually pretty close.

Now, some tips:
1) Saving versions of designs - When you have a WEBENCH design in a state you want to keep, click on the MyDesigns/Projects button. This brings up a listing of your WEBENCH designs, most recent one on top. Identify the one you want to create a new version of, then note the buttons on the right side of that row, in the Actions field. The 3rd button is for "Copy Design." Click that, change the design name or description, and click "Copy This Design." It will appear in your designs listing. You can now edit the new design - or keep it as a preserved version and edit the original.

2) Creating design with ceramic output capacitor - In the Visualizer tool (which shows the solutions matching your design requirements), in the Change Inputs panel, there is a link for "Use Advanced Options." Click that, and see on the right a section for Component Selection. The third line is for Cap Selection. Click the box for "Ceramic Only," then click the green Recalculate button at the bottom. The resulting designs will be using ceramic output capacitors.

3) Auto-Recompensation yields varying solutions - To get to a recompensated solution quickly that meets the targets, the Compensation Designer incorporates a bit of randomness. The result is compensation components that may be different each time the utility is run.

4) Simulation convergence - The LM21215 model can be a bit finicky. If the design's phase margin is low, under about 25deg, it can cause problems with convergence.

We'd be glad to talk with you about your experience with the Compensation Designer and with this design in general. If you like, email me at wanda.garrett@ti.com and I'll set up a phone meeting for next week.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Wanda,

I was looking at the TPS544C20 this evening, mostly because I have been thinking how nice it would be to be able to have intelligent control over the supply.  I2C is a good way to do that and since that is what PMBus is, I thought I would look at it.  Also, it has a lot more margin.

The webench is WAY better behaved for the TPS544C20, and the load transient is 2 orders of magnitude better that the LM21215A-1.  Is that really possible?

The TPS544C20 is a much better match AFA transient and ripple for the very picky GPU that I have to support.  But, it does seem like the LM21215A-1 works fine now, but if the SIM is to be believed, I have just been really lucky.

So, can you please tell me: is the TPS544C20 really 100 times better than the LM21215A-1?

Thanks!!

Victor










At 03:57 PM 4/2/2015, Wanda Garrett - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Wanda Garrett - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,
Thanks for sharing your experiences in creating designs with us. I can offer a couple of techniques that can make things go a little easier, plus additional info on the LM21215 quickstart spreadsheet.

First, on differences between the quickstart spreadsheet and WEBENCH design calculations - the WEBENCH calculations use a slightly different value for the ramp voltage, which gives better matching to bench data. The result is that the crossover frequency may be a bit different between the two tools, but the phase margin is usually pretty close.

Now, some tips:
1) Saving versions of designs - When you have a WEBENCH design in a state you want to keep, click on the MyDesigns/Projects button. This brings up a listing of your WEBENCH designs, most recent one on top. Identify the one you want to create a new version of, then note the buttons on the right side of that row, in the Actions field. The 3rd button is for "Copy Design." Click that, change the design name or description, and click "Copy This Design." It will appear in your designs listing. You can now edit the new design - or keep it as a preserved version and edit the original.

2) Creating design with ceramic output capacitor - In the Visualizer tool (which shows the solutions matching your design requirements), in the Change Inputs panel, there is a link for "Use Advanced Options." Click that, and see on the right a section for Component Selection. The third line is for Cap Selection. Click the box for "Ceramic Only," then click the green Recalculate button at the bottom. The resulting designs will be using ceramic output capacitors.

3) Auto-Recompensation yields varying solutions - To get to a recompensated solution quickly that meets the targets, the Compensation Designer incorporates a bit of randomness. The result is compensation components that may be different each time the utility is run.

4) Simulation convergence - The LM21215 model can be a bit finicky. If the design's phase margin is low, under about 25deg, it can cause problems with convergence.

We'd be glad to talk with you about your experience with the Compensation Designer and with this design in general. If you like, email me at wanda.garrett@ti.com and I'll set up a phone meeting for next week.

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com

Hi Victor,

TPS544C20 is a Hysteretic mode control topology (DCAP, DCAP2) whereas LM21215A is Voltage Mode Control Topology. Hysteretic topology  inherently have a faster transient response when compared to Voltage mode topology. As far as the exact specifics of the 100x improvement, it is very hard to say without looking at your exact designs.

This document has some more details on DCAP topology and why is it faster than Voltage Mode

Thanks,

Hi Nikhil,

Thanks very much for this information.  I think unless one is an expert in switchers it is hard to know what the best think is to use.

My application is an AMD graphics chip and some other chips. Because of severely limited board space, integrated switchers were an absolute requirement.

Given all the devices on board, I have to supply:

1.8V @6A
1.055 @4A
1.15@6A
0.9 to 1.1V @14A (using switched Rs on the Rfb to change Vout)

I have implemented all of this with the LM21215A-1. 

Plus, since Vin can be either 5V or 12V, I have a 12V to 5V converter that can be switched in in front of the LM21215A-1, since they can only take 5V in.  For that, I am using a TPS56121, Vin 11 to 13 (typically on the low side) Vout 5V@15A.

I will admit that this not as clean a solution as using parts that can run from 5 to 12 on their own, like the TPS56121 itself.  I struggled for some time to rationalize the above-mentioned approach. I was concerned about the large amounts of noise on the input being fed back to the source and also into local devices that run on the 5V rail.  For that reason, I took an approach that we had used in an earlier design, namely slaving the 4 LM21215s from a 4 phase clock generator.  This was done to make sure that the supplies never peaked at the same time.

In two prototype runs, we have had no problems, but I can't say that that is the same thing as having users beat on them. I have gotten into this present pickle because I had done the designs for the LM and TPS supplies a few years ago and thought I should check in with Webench to make sure the values were OK.  That was about 3 weeks ago and I have had no end of problems with webench trying justify the values it gave me before, or even to get it to reliably (esp the LM21215A-1) run SIMs. 


Now, back to the app note. The problem with the D-CAP regulators is that they can't be run  synchronously (external clock), and in fact, there are very few non-module high current switchers that will accept an external clock.  That is how I got to the LM21215 in the first place and it doesn't appear that in 3 or 4 years, things have changed very much.

If webench is to be believed, the transient load response for the LM21215A-1 is really terrible - more than 100mv delta for a 0.9v supply won't fly.  I don't have a good way of measuring what we are really getting on the board, but I don't think the chip will run with a supply like that, so I don't believe the SIM.

Any ideas?

Thanks!

Victor





 




At 01:00 PM 4/6/2015, Nikhil Gupta - WEBENCH Design Center wrote:

 
 
A Message from the TI E2E™ Community
 
 

 

Nikhil Gupta - WEBENCH Design Center replied to Webench and Quickstart models for LM2121x yield significantly different results.

Hi Victor,

TPS544C20 is a Hysteretic mode control topology (DCAP, DCAP2) whereas LM21215A is Voltage Mode Control Topology. Hysteretic topology  inherently have a faster response to Voltage mode topology. As far as the exact specifics of the 100x improvement, it is very hard to say without looking at your exact designs.

This document has some more details on DCAP topology and why is it faster than Voltage Mode

Thanks,

 
 

You received this notification because you subscribed to the forum.  To unsubscribe from only this thread, go here.

Flag this post as spam/abuse.

Victor Gold

Rastergraf

7145 Marlborough Terrace
Berkeley, CA  94705
(510) 849-4801
email: victor@rastergraf.com