2
\$\begingroup\$

My design involves 2 micros that operate at 3.3v. I have output line drivers that operate at 5v. One of the micros is a wifi device so I want to stay away of switching power supplies.

5v or 12v can be applied to the POWER IN to the board (Vin). I have a P-Fet providing reverse battery protection (90 milliohms ).

I need to get to 5.0v ... from there I have a 3.3v LDO reg supplying the micros.

My question is the best way to implement the circuit between the Rev-Bat P-Fet (either 5v or 12) ... and get to 5v.

I would normally use a boost-buck regulator but not with the radio being so near.

I need about 400 milliamps ... so I'm designing for 1A.

I can get several 1 or 2 amp 5v linear reg (under $3) that have an enable control. The main micro can come up and keep current low until it decides how to configure the vreg control.

So - I can have the 5v linear on by default (assuming 12v is applied) and measure the input voltage. If I determine Vin is actually 5v -- I can turn off the 5V vreg and enable a bypass switch to route the Vin of 5v to the input of the 3.3 LDO.

So ... after all of that --- what is the best circuit design for the bypass that can be controlled by the micro? More than likely another P-Fet -- I'm just not sure of the actual circuit design.

Another question ... I've been out of hardware design for 45 years -- now software. At 400ma and 7v being dropped across the 5v Vreg when Vin is 12v -- that almost 3W the regulator has to dissipate. How hot is that? How hot with a finned heat sink?

\$\endgroup\$
  • 3
    \$\begingroup\$ I wouldn't be too worried about the dc-dc interfering with the wifi unit too much to be honest, as long as it's got a small loop area for the high switching currents and you have lots of ceramic capacitors, I can't see much of a problem. The frequencies that the wifi radio operates are so far above what the dc-dc would spit out (even the harmonics) that it's not likely to cause interference (if it's a half decent wifi module it should have it's own internal power regulator/filter anyway). The wifi module will also have error correction and narrowband filters to further shield it from noise. \$\endgroup\$ – Sam May 2 '16 at 1:46
  • \$\begingroup\$ To reiterate @Tom's comment: if your buck converter interferes with Wi-Fi, you are doing it wrong. Use a buck converter to convert the 5-12Vin to 3V3. \$\endgroup\$ – uint128_t May 2 '16 at 2:10
  • \$\begingroup\$ I can't go directly to 3.3v -- I have to have 5v for my line drivers. I've never designed a switching reg before. I know there are a lot of app notes on the subject. I was hoping to stay in technology I know and am comfortable with. \$\endgroup\$ – JHinkle May 2 '16 at 2:20
  • \$\begingroup\$ Well, have the buck go to 5V, and use a 3V3 linear reg for the low power micros - they won't need much power. Do you even need 5V line drivers? (I'm assuming the high power device is something at 5V, use the buck to feed the high power device and a linear reg for the low power stuff) \$\endgroup\$ – Sam May 2 '16 at 2:22
  • \$\begingroup\$ There are some pretty plug 'n' play dc-dc converter chips out there, just add an inductor and some ceramic caps and away you go. (some even have the inductor built in as well, but they cost more) \$\endgroup\$ – Sam May 2 '16 at 2:24
1
\$\begingroup\$

A solution to keep the circuit simple would be to choose a buck regulator that is able to handle 100% duty cycle. I found two of them with a bit of searching : LTC3864 and NCV8852. There are probably more.

When the input is 12V, it will efficiently drop the voltage down, without heatsinking required. When the input is 5V, it will maintain the PFET on, thus having only the sense resistor, the PFET, and the inductor that will drop some voltage from the DC jack to the powered circuit. You can manage to have a total resistance for this (Rsense+PFET+inductor) in the range of 150mOhm, which would drop 60mV with a 400mA current. This is certainly acceptable.

LTC3864 can be simulated with LTSpice, by the way.

And you shouldn't worry much about the Wifi and switching regulator interference. Keep them as far as possible on the board, and follow closely the recommendations for the layout of the regulator (there are design references for both chips I mentioned), and you'll be OK.

\$\endgroup\$
  • \$\begingroup\$ Thanks. I had stopped considering just a buck when all the documents I looked at stated Vout min was .8Vin. I primarily use Mouser and Digikey for my parts selection since my volume is usually less than 100. What is the key specification that would identify Coincident Vin tracking? 100% PWM? Is there a way to filter for that or is it just look at all of them until you find one. I was hoping the cost to implement the buck would be less - right now it appears to be about 12% of the parts cost. \$\endgroup\$ – JHinkle May 2 '16 at 13:50
  • \$\begingroup\$ Keeping it far from the radio is a relative statement -- far for me will be maybe an inch if I can manage that. Again - thanks for the answer. \$\endgroup\$ – JHinkle May 2 '16 at 13:50
  • \$\begingroup\$ There is no "key specification" as such, in that case. When looking up such unusual parts, I just use google. "100% duty cycle buck" are the search terms I used. You could also try "P channel buck" as the controllers that have such capability typically use a P channel mosfet whereas the more usual ones typically use a N channel MOSFET. \$\endgroup\$ – dim May 2 '16 at 17:35
  • \$\begingroup\$ I'm sure an inch is fine. Actually, unless you use an unshielded inductor that you stack above the wifi module, I'm sure it will work. \$\endgroup\$ – dim May 2 '16 at 17:39
  • \$\begingroup\$ Otherwise, as uint128_t said, you could use a SEPIC (step up+step down) converter. That's somewhat overkill, but there will certainly more parts available and that may actually end up cheaper. \$\endgroup\$ – dim May 2 '16 at 18:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.