TPS43061: can I use a schottky diode instead of M2?

we noticed your question. the engineer supporting this device will reply when back to office

Hi Vern,

Please see my comments below,

question:  it is nice to have uvlo (which happens if output is shorted), then the boost circuit shuts down. circuit is protected from output shorts?   is this correct?

The is current peak current limit detection on the TPS43061, this will protect the MOSFET from over current. I would not consider UVLO a short circuit protection, as a boost converter inherently has no short circuit protection as current can flow freely through the diode or the body diode of the synchronous FET. If short circuit protection is needed a disconnect switch will be needed. The LM5121 is a possible solution for this.

question:  I found a schottky diode with much lower power loss (1/2) than using best mosfet.  thus there is not efficiency gain using a mosfet.  it is ok to not connect HDRV?

I would recommend keeping the HDRV signal connected to the MOSFET. If a non-synchronous designs is desired please use a device such as the LM3481.

note:  read and searched 100 pdfs regarding mosfet vs. diode, about 2/3 of them end up with a schottky diode (smaller) anyway, to reduce the mosfet diode losses.  every single pdf uses a relatively high voltage drop diode for analysis.  I searched thoroughly in TI, ONsemi, vishay, and 5 others for a mosfet that has lower loss than the diode.  none.  so is synchronous rectification all blown out of proportion?  hype

As with everything it depends on the application. In some high current applications the diode losses will be quite high. Also in application like an automotive pre-boost that is not on all the time the diode losses will be high because even if a Schottky is used.

question2:  if I use a 2nd mosfet as shown in tps43061 schematic, then if my output has a short, will the 2nd mosfet burn out since it is directly connected to input?  many pdfs said the mosfet conducts both ways.  the tps43061 pdf does not mention what happens if output is shorted.  if I need to protect the mosfet from damage, and the mosfet is there to "prevent" damage, then I'm not using a 2nd mosfet for rectification.

Current will flow freely though the body diode of the synchronous MOSFET. This is the same in a non-synchronous device. As mentioned above a disconnect switch is need to to truely protect against short circuits. If this is desired the LM5121 can be used.

Please let me know if you have any questions.

Thanks,

Garrett

thank you very much for quick and nice replies. especially nice to have professional help.
I had studied LM3481, and won't rule it out. reread its pdf. gate drive is 6v, which is excellent compared to 7.5v of others thanks, now it's down to 5.5v vs. 6.0 drive. LM5121 I rule out because of high drive voltage wastes a lot of power compared to 5.5v drive.
Anthony, your associate, gave this wonderful link: www.how2power.com/.../H2PToday1403_design_TexasInstruments.pdf
I like the tps43061 because it is newer, and can sense current using the inductor--improves cost and efficiency.

thank you for telling me the 2nd mosfet does not help a short circuit, so I desiring a diode (mine has only .12v drop), instead of 2nd mosfet and it's issues.

my entire circuit uses cheap $10 ac-dc 24v out power adapter, located 5 feet away. It is short circuit protected. this ac supply drives my tps43061 boost, which is 34v 4.2A. tps43061 drives a motor/controller, which can have a short circuit "easily". I tested an actual circuit, and a motor short keeps the AC/DC output at 0, without heating. circuit seems to be self-protecting. my fall back is serial insertion of a 6A metal heat switch, like a circuit breaker. I say all this because a output short would otherwise be catastrophic.

SEPIC was invented to prevent blowouts, right? but my inductor research shows 2 inductors adds a lot of wasted power. every coupled inductor has relatively high resistance, so I gave that up 2 weeks ago.

Since mosfet 2 does not protect for short output, then I want to insert a passive .12v schottky diode, have half the wasted power loss, and leave tps43061 mosfet high drive flapping in the breeze, unconnected.

update on my tps43061 design...

am using a high side mosfet, maybe one with built-in schottky diode, or will add a schottky diode in parallel to reduce losses in high side mosfet diode.   I had thought the .12v schottky diode I found was lowest power loss, but finally found some non-TI mosfets that actually have lower power loss.  must add a schottky diode in series with M2 mosfet, not a problem.

I gave up on webench, due to cannot add new mosfets, and serious errors in Rsense power loss.  Also, if you change frequency, webench makes you start over--it deletes all my custom parts.  I did 3 webenches, using 3 different TI boost chip circuits:

1.  2 said .11W loss (each?  unclear), but that is not correct, has to be about .45W.

2.  3rd said 0 loss.  really wrong, and nothing I could try made a difference.

there is no webench method to use inductor DCR Rsense, stuck with having to add a discrete sense resistor.  that greatly changes power loss and efficiency.  work around is do webench, then make separate docs showing this change.  that is not easy.

the TPS43061_boost calctool rB.xls does some useful calculations that will make my circuit design easier.  but I mention this tool because it mostly has inputs and not enough outputs.  has few power loss results.  the use of "odd" exponential input values made it difficult, most are not on even E3, E6 etc..  example, would have made it better to input Rdson in mohms, have your equations accept mohms and multiply by 1000 internally.  Rdson is displayed as 650.00E-5, that is neither mohms or ohms notation.  spreadsheet is locked, and that is not good.

so I created my own spreadsheet, and it is awesome and accurate, with all humility.  Massive search on internet, read over 330 pdfs about boost circuits, produced correct detailed equations.  did you know many TI pdfs have different equations for same formula?  better to have consistent correct formulas!  TI would benefit from having one boost pdf for calculations, then have each chip reference it.  I literally read every boost chip TI has to offer.  that took quite a while.

my design efficiency for 19.5-24v in, 5A, 32v out, 160W, is 99%.  I paid extra attention to mosfet Coss, Rdson, diode drop, Qd, rise/fall times, Vgs threshold, Qgd gate-to-drain.

Am using the inductor for current sense.  scary, because pdfs "says" it is not accurate enough.  inductance change vs. temperature can be large.  I chose an inductor series that has almost flat variance vs. temperature and frequency.  the power loss through the 2 extra resistors is significant--downside.  I adjusted my frequency, and also chose an inductor with DCR resistance that closely matches the calculated sense resistor, and has   My spreadsheet can model this.  I could not find any pdfs that followed through enough to make a decision, so plowed ahead.

bottom line, staying with TI tps43061.  thank you TI.

Hi Vern
1 If you use TPS43061, it doesn’t have output short protection because of the body diode of MOS.
I recommend you to find a controller with true disconnection.
2 I find you pay more attention to gate drive voltage.
You think if the gate drive voltage is high, the loss will be increase.
But it depends on the application.
Driving MOS with a higher voltage, the Rdson will be decrease, so the conduction loss will be decrease. And the gate drive loss will be increase, it is a balance.
You could take this paper as a reference. www.ti.com/.../slua341.pdf
3 you could add a schottky diode in parallel to reduce losses in high side mosfet diode. It is right.
Please let me know if you have any questions.
Thanks
Jing

Good. do you have other question? if not please click the green button.

thank you Jing, I mentioned many serious issues with webench. Do you think TI will fix soon?

Hi Vern

  Could you summarize the Webench problem clearly? and this will help the webench team to fix them.

copy/paste my details here again, they are clear enough.

I gave up on webench, due to cannot add new mosfets, and serious errors in Rsense power loss. Also, if you change frequency, webench makes you start over--it deletes all my custom parts. I did 3 webenches, using 3 different TI boost chip circuits:

1. 2 said .11W loss (each? unclear), but that is not correct, has to be about .45W.

2. 3rd said 0 loss. really wrong, and nothing I could try made a difference.

there is no webench method to use inductor DCR Rsense, stuck with having to add a discrete sense resistor. that greatly changes power loss and efficiency. work around is do webench, then make separate docs showing this change. that is not easy.