[SOLVED] what is the highest switching frequency for the dc/dc converter from 12V to 1V@15A?

[max.wangxin.sh]

Hi Guys, happy new year! I am applying for the job position in https://www.coolstartechnology.com and have discussed the issue with Dr. Shuming Xu, CTO in the company. He claimed the latest answer is 4meg HZ and 94% efficiency. Frankly it is beyond my experience. In my opinion 500K HZ is the highest switching frequency for this spec. would you please share your experience/opinion? Do you know this company or CTO? Any response is appreciated greatly and many thanks for your attention and sharing.

 

[BradtheRad]

Think carefully whether he might apply some 'trick solution' to obtain his figures. For instance, by interleaving eight converters, each running at 500 kHz, he has 4 MHz, in a manner of speaking.

Another possibility is that the inductor starts acting more as a choke at very fast frequencies, rather than for energy storage. It suggests you might make a different approach to the design. My notion is merely tentative and it may not apply in a 15A converter.

 

[mtwieg]

That sort of spec is achievable with GaN transistors, if you use very good magnetics. See here.

Also you may want to be careful dropping the name of your company and boss in a public forum...

 

[badea]

in high frequencies like 4Mhz, capcitors and inductors loose their specification

as you see, there is a inter winding capacitance, this capacitor in high frequencies will provide an easy way for current instead of inductor, and inductors losses it's specification, this problem wil faced you again in capacitors, Xc for capacitors in very high frequencies will be raised

regards

 

[dick_freebird]

We got to about the stated efficiency and frequency
on a family of 5V-1V buck ICs using RF CMOS SOI and
using only Coilcraft kit inductors & caps we bought
from Digi-Key, with no "de-embedding of passive
losses" marketing trickery. Just "wallplug". So I think
this can be done with superior switch devices. The
RF CMOS was not very sensitive to fSW, in the
efficiency.

But startup CTOs (in fact startup CxOs of any
function) are by their nature slippery folks and you
should parse carefully anything that's said, and try
to keep track of what's left unsaid. "Forward looking
statements, not to be relied upon". Repeat it to
yourself.

I'd recommend that since the pitch is about aggressive
DC-DC design, you come in with the sensible questions
about how cute theory becomes manufacturable fact.
Be concerned, and slightly knowledgeable, and maybe
they'll figure you can help.

Getting a 12:1 stepdown at 4MHz means you have
about a 20nS on time, and need nanosecond-range
risetimes from the output stage. That's gonna be
some ugly EMI and tough to do current mode control
(expect ringing tails will not be quenched by the time
you're trying to do your current compare). Maybe a
non-fixed-frequency-PWM operating mode could do
better here. But I'd expect mucho ugliness for a more
traditional topology.

 

[max.wangxin.sh]

Dear all, many thanks for your sharing and reminding. As dick_freebird mentioned it would be RF CMOS SOI for the power converter. It seems pre-research tech. It maybe be the trend in the future and maybe not. Anyway your answer enlarged my views. thanks a lot.

 

[mtwieg]

For silicon devices, even SOI CMOS, I don't think 94% at 4MHz is plausible unless it ignores inductor losses.

 

[max.wangxin.sh]

Hello mtwieg, many thanks for your comments. It is the reason i started the thread. maybe dick_freebird has more. As i told i do NOT have the experience.

 

[schmitt trigger]

The devil is in the details.

It is one thing to develop a single prototype, another to assemble a mass-produced product with real components.

I am not saying it can't be done.
But many technical claims actually are "vaporware", unless there are actual customers employing the product. Reliably.

 

[dick_freebird]

The stepdown ratio is more the killer, than the frequency,
in my experience (but I got to use a pretty good high
frequency technology, and my stepdown ratio was 5:1
not 12:1).

In my interaction with customers, not -one- had any
interest in going past 1MHz, due to attributes of the
available passives. We put a fair bit of effort into
making it work at 5MHz and nobody cared.

Heroes' reward is often posthumous.

 

[max.wangxin.sh]

I will visit the company this month and I hope to test their demo board by myself. maybe there is a miracle.