Fooling Around with RT4812

Description

So I've been interested in finding a *Quality* Li-ion boost DC-DC circuit these past couple months. Unfortunately almost everything out there is moving to BGA or MFQ-type non-leaded packaged. To say that sucks is an understatement. I believe medium to high power ICs should almost always have a leaded option. Not everything is made for cell phones... So my search landed me with the RT4812. It's a perfect package type for me (read: robust) coming in a nice SOT-23-8L. I made an Eagle footprint for it too, which I'm happy to share. These pins can be soldered without reflowing and that makes it ideal for my project (and for prototyping). The specs are too good to be true. >91% efficiency, 2A output, all without an external diode. The datasheet is decent EXCEPT for the layout section...what a mess! One great advantage to this chip is the inductor is not tied straight to the output. That keeps the output from draining VBATT_IN when the IC is powered down

Details

Fortunately I came across this great Hackaday page by Patrick!

https://hackaday.io/project/9461-lifepo4weredpi

He was nice enough to answer questions and help me with my design. My previous 2 layouts were failures, shutting down after only <4.5W output.

I noticed the recommended inductors for this thing are >4A! for only 2A output current I thought this was strange.

Anyway, I designed this new proto board to run further testing. I'll report back!

So far I learned:

*CFF is not necessary. It is used for large transient load regulation. The recommended 10pf completely screwed up the output. A more modest 1pf seemed to work better (both 50v NP0 rated)

*Actual efficiency is closer 80-85%

*ILIM should be tied to EN if unused

*Ground loop layout is extremely critical

Questions I have so far:

Q: GND and PGND being separate makes me wonder if GND should be tied only to signal ground, i.e. the FB resistor? But the datasheet layout shows them joined so...that's what I did

Q: How will it funtion with a 3A-4A Inductor, rated 2.2uH?

Q: What is the actual sleep mode current draw?

Q: What is the EN pin current draw (wondering if a flip flop might actually be causing an issue driving this) though that seems unlikely

Files

IMG_4951.JPG

JPEG Image - 2.41 MB - 12/07/2017 at 22:18

Preview

IMG_4946.JPG

JPEG Image - 2.28 MB - 12/07/2017 at 22:18

Preview

IMG_4942.JPG

JPEG Image - 2.03 MB - 12/07/2017 at 22:17

Preview

IMG_4945.JPG

JPEG Image - 1.98 MB - 12/07/2017 at 22:17

Preview

IMG_4940.JPG

JPEG Image - 2.19 MB - 12/07/2017 at 22:17

Preview

Project Logs

Collapse

Success!!

segasonicfan • 12/07/2017 at 22:14 • 0 comments

Finally got the circuit working as intended.

The difficulty was primarily due to layout considerations (improper ground pours + ground loops) and having a bulk electrolytic on output. With 3x 10uf ceramics it is working great!

I used a 3A 2.2uH inductor which is not exactly to spec so I'm getting ~1.5A output, but that's more than I need anyway. I think the reason the datasheet calls for such an enormous inductor (>6A Isat!) is due to low impedance windings. This is needed to achieve the ultra-high efficiency (I think). The datahseet recommends an incredible 9.7mohm with an Isat off 11A! This is shocking for a DC-DC that only pushes 2A / 5W of power...

The substitute I found (their recommended one is obsolete) is here:

https://www.mouser.com/ProductDetail/Bourns/SRN6045TA-1R5Y/?qs=sGAEpiMZZMsg%252by3WlYCkU3tyM9HcjVri8VL%2fIY%252bgGEQ%3d

This has slightly higher resistance but I think will do the job well. Right now with my successful layout I am getting 70-85% efficiency. Let's see if this part bumps that up higher...

Also, I am going to expose the ground copper beneath the ship for better thermal dissipation. Not that it needs it, it didn't break a sweat running for hours at 1.45A. It's good practice though.

Discussions

segasonicfan wrote 12/08/2017 at 03:06

woops, thanks for catching that typo. Cool, what are your takeaways from that chip? Did you get over 90% efficiency at any point?

Are you sure? yes | no

Jarrett wrote 12/07/2017 at 23:33

*Richtek

I'm using the RT6150 in some stuff, similar chip from the same manufacturer :)

Are you sure? yes | no

Fooling Around with RT4812

Description

Details