• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 63 subscribers
  • Views 496 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPS43061 Inadequate

CraigHB
Prodigy 215 points
Other Parts Discussed in Thread: TPS43061

I'm trying to work out a design with this part and I'm finding some shortcomings with it.  

Why does this controller have a liberal 65ns forced dead time between the switch drivers.  When running at 750kHz and 50% duty cycle it spends almost 25% of the pulse riding the body diode in the high side switch.  With the high output currents I require it wastes a ton of power and creates heating issues.

The TPS43061 has a limited range on duty cycle.  It forces pulse skipping under 10% and tops out at 75% with maximal switching frequency.  Why would anyone design a boost controller than can not run in CCM down to zero duty cycle?

There's no pin option to force CCM.  Why is that function not provided?  It's better to let inductor current go negative and avoid pulse skipping if the power savings are not needed.  Instead there's a "power good" pin.  That's a redundant function since startup time is programmable.

Current mode is nice because it's easier to compensate, but why is maximal sense input only 60mV?  A part with external switches is more suitable for a high output design.  That calls for inductor DCR sensing which makes such a low sense limit harder to design around.

Normally TI parts offer superior performance and features.  What happened with this one.  Is this one they gave to the interns?

I'm going to work out a design using this part anyway since boost controllers like this one are not exactly a dime a dozen, but it's discouraging to see these shortcomings end up in my design.  For a rather unique part such as this, I would have expected TI to put their usual high performance and liberal feature set into it.

  • 0
    TI__Mastermind 35185 points

    Hi Craig,

    Thank you for all of your feedback. There are many tradeoffs that come with choosing specifications for an IC so it's not possible to make a part with all of the perfect for every application. A few comments on some of your points.

    The dead time, minimum on time and minimum off time is comparable to other parts in the market. You will find the same duty cycle limitations on every boost when operating at high frequencies. The TPS43061 actually has the lowest dead time of the 3 parts I compared.

    When designing for high power applications typically a 750kHz switching frequency would not be used. There are a lot of losses associated with switching at a high frequency. Not only the loss in the body diode of the high-side MOSFET but also switching loss in the low-side MOSFET. To reduce the dead time losses in your application, as you probably know, you can add a schottky diode in parallel to the high-side body diode.

    I agree CCM option is definitely useful for certain applications and is nice to have. However the TPS43061 wasn't designed to target these specific applications. Combined with the low Iq the TPS43061 was designed to have good light load efficiency. Also the PGOOD is a useful feature for sequencing that many other customers are using. We have to consider if more customers want a PGOOD feature or forced CCM feature.

    A maximum sense voltage works much better for resistor current sensing. If this voltage were higher, there would be more power dissipated in the current sense resistor for the same current limit.  This does add some challenges to using DCR sensing, but again a tradeoff needs to be made.

    Thank you again for all of your comments.

    Best Regards,
    Anthony

  • 0 in reply to Anthony F
    Prodigy 215 points

    Anthony, that was a gracious response and thanks so much for addressing my concerns. I kind of thought I was asking for a flaming reply on that one.

    Anyway, my application is highly unusual.  I don't want to discuss it in a public forum, but you can email me personally if you want to know exactly.  The switching frequency is high, but there's a good reason for it.  I have very strict size limits and it's the only way I can get the outputs I need with the volume I have available.  The TPS43061 offers frequencies up to 1MHz so that should always be considered in the part's design anyway.

    Adding a bypass diode is not an option for me due to size constraints, or yes, I would do that.  Unfortunately, I have to rely on the high side switch's body diode.  If I can figure out a way to work one in there I'l do it.  I'd probably have to stack it somehow, it would be odd.  Even so, the bypass diode is then blowing off the power, albeit a little less than the switch does.

    I would not expect TI to compare themselves to "what else is out there" unless it's their own stuff.  I expect TI to be the innovators, not the followers.   The dead time is liberal.  Try simulating an output of 15 Amps with a switching frequency of 500kHz (which is not extreme) and you'll find that high side switch burning off a lot of power in the body diode.  If not the switch's body diode, a bypass diode will also burn off considerable power.  

    A controller that uses a PMOS high side switch has considerably less dead time.  Seems to me be the point of using an NMOS high side switch is to improve performance, but with that dead time, design is more convoluted and efficiency takes a big hit.  It pretty much defeats the point.

    In any case, if the TPS43061 offers the least dead time I can get, that's the least dead time I can get.  I just have to work with what is available.

    I suppose the lack of a forced CCM option is simply a design decision made in favor of light load efficiency, but why throttle the feature set.  TI normally doesn't do that.

    I'm not saying the maximal sense voltage should be a lot higher, 80mV would have been better and that's only 33% more.  The thing is, if you're not using DCR sensing, then you're probably not using a high output.  In that case, a higher value sense resistor is not an issue.  If using DCR sensing, you are not affected by the maximal sense voltage in terms of power loss so make it easier to design by using a higher limit.

    Well thanks a bunch for taking the time to address these issues I have with the TPS43061.  Really appreciate it.

  • 0 in reply to CraigHB
    TI__Mastermind 35185 points

    Thanks for the further comment. I like seeing high level of standard you hold for TI and our products.

    Another thing we have to consider on a 60 V controller, customers will have different layouts and different MOSFETS. Specs like this will need to be more robust for these unknowns and voltage levels. With our internal FET parts we can push the deadtimes a lot more as we have more of the design in our control.

    Also I sent you a friend request since I'm interested in seeing more on your application. This way we can use a conversation to share some details through the forums.

     

  • 0 in reply to Anthony F
    Prodigy 215 points

    Sounds good, I started a conversation.