Under frustration finding a decent high power DC-DC converter for my use (mainly my RC helicopters) I decided to utilize my spare time at uni to make my own.

There were certain characteristics I was after in my converter:
1. Zero Dropout voltage
2. Low noise
3. Input from 5 – 30 volts
4. Output a rock solid 6volts with NO droop/drop under heavy loads
5. Have the capability to output other voltages ranging up to Vin
6. Handle 10 amps continuous
7. Have a surge capability of 20 amps
8. Extremely efficient operation
9. Very small and lightweight.

So, I searched far and wide (5minutes on the net) for a controller IC that had the capability to deliver the needs. The LTC1625 from LT was chosen. Anyways, I ordered some samples from Linear Technology, and as usual, the samples arrived within a few days, total of 12.

Datasheet here.


Pictures of chip.


Quick background info: There are two types of controllers, voltage mode (which have one control loop that just measures the output voltage to regulate), and current mode controllers (which have two control loops, one for the output voltage and another for the inductor current). Although voltage mode control offer high efficiency and a simple design, it is difficult to compensate feedback under high current outputs (and droop occurs), they have poor rejection of input voltage transients and they do not inherently limit output current under fault conditions which leads to fried components.

Normally current mode controllers would need a current sense resistor to measure the inductor current, which results in an efficiency loss, thus making voltage mode controller more preferable in high efficiency applications. However, the LTC1625 eliminates the current sense resistor by measuring current using the MOSFETs on-resistance, thus, making it just as efficient as the voltage control topology, with all the current mode benefits. Current-mode chips have rock-solid output regulation even under extremely high loading conditions. Also, the use of synchronous rectification eliminates the low side diode typical buck converters have, and this also adds to the efficiency.

Also, a quick note, Linear says the LTC1625 is the industries highest efficiency current mode controller.

Anyways, the data sheet for the LTC1625 contains a reference design example for delivering a 3.3volt/7amp output, with a few modifications and a change of components, I had the circuit designed (10 minutes later ) to deliver between 1.2 – Vin volts at 10amps continuous.



The switching frequency of the LTC is 150 kHz, which is pretty high, so I needed to keep an eye on the gate charge of the FET’s to minimize the switching losses and also keep in mind the on-resistance because I wanted high current capability. So, I chose the Infineon IPS040N03 which has the industries best FOM (gate charge x on-resistance product), handles a theoretical 90amps **cough** and is in a neat small SMD TO252 package.



Ok, so now I know my component values, and I’ve decided to go full SMD. I’ve chosen a good FET, caps are standard low-ESR. Now, I need to choose an inductor. Most of the noise generated from DC-DC converters radiates from the inductor, so I went for a shielded inductor (most expensive part on the bugger at $4 from farnell ) which takes up to 15 amps with only 20% saturation.

Finally, I had sourced all my parts (took me a damn 5 hours at least, some came from overseas because sourcing electronic parts from AUS is shocking).

So I started the PCB layout……..



…………and finished after 2 hours. I poured grounded copper around all free nets to form an EMI shield, and I kept the SW (that’s NetCb_2 in the layout) net as small as possible, because that’s a high current, high swing line where most of the EMI is radiated from.

Finished layout in 3D




Guys at work fixed me up with PCBs for free (actually cost me a few beers at the bar 2 weeks ago)



And after the solder paste/hot air soldering…






The 12 gauge wire is just for bench testing purposes, realistically you'll use 16 - 18 gauge.
I’ve been testing it extensively in the past few days, and the results are amazing.

Stock, it puts out a solid 6 volts from anything from 6-30 volts. With an optional trimpot, it can output anything you like up to 99% of Vin. Also, an optional on/off switch can be added otherwise it'll always stay on.

This has yet to be tested on any of my fleet (initially designed as a regulator for my helicopters ), i will do further testing (EMI, heat, short circiut etc..) before it goes in.

Specs so far are

INPUT: 4 – 36 volts
OUTPUT: 6 volts (or anything from 1.2volts to Vin with added trimpot)
CURRENT: 10amps (stays very cool after a few mins)
BURST: 20amps
RIPPLE: only 45mV tested with scope @ 8amp load.
DROOP: -0.01volts. (voltage actually rises 0.01volts @ 10amp load @ 6v)

Ill post a full report on the results in the next few days.

COST:

Very cheap. Total parts cost was around $12, but that's without the PCB cost and controller factored in. Realistically, itll cost around $20 to make . The FETs were hard to find, came from mouser in the US along with some other SMD caps, the controllers were sent directly from LT. The rest came from Farnell in AUS.

Also due to the extremely low dropout operation (only 1% of Vin), this can be used to deliver a regulated 5-6v output from existing 2cell LiPO / 5cell NiMH setups. However, i initially designed it to use straight from my 6s Lipoly trex packs at 22.2volts.

Ok, some quick bench test results with pictures:

Here i have the BEC (battery eliminator circuit) running 3 50watt downlights @ 6 volts. The input is the home made 18 volt unregulated power supply you can in the background. There was NO voltage drop at all as i plugged the lights in one by one. The total drain current of the three downlights was 7.89 amps at 6 volts.

I left the power running for 10 mins, and measured the temp. The core temp rose to 52 degrees after ten minutes. Ambient temperature was 27 degrees.



During the test, i took some quick results with my scope.

Here we can see the output ripple at 48 mV with the 7.9 amp load. The internal frequency of the controller is set to 150 KHz. The ripple can be decreased by adding larger output capacitors, but i think 48mV would be ok.



A quick FFT analysis shows the main frequency components at the output occur in 150KHz multiples and dying out after around 500KHz, with some other background noise. I tested all the way up to 100MHz but my scope wont go any further. I need to wait to get back into uni next week to test it up in the GHz range with the spectrum analyzer (just to make sure it wont affect our 2.4GHz systems) but i think it'll be fine.



The lightest load Ive tested this on is running a charger drawing 600mA, all seemed clean. At no load, it goes into burst mode operation and runs at 400Hz just to keep the output caps charged at the correct voltage.

hardware rev b

A ripple of only 49millivolts is very good considering its only a 10 micro Henry inductor. Very nice indeed.

For higher current operation, Linear recommends the use of the LTC1775 which is exactly the same as the LTC1625 but it has 300mV Maximum Current Sense Voltage.

This means you can pass more current through your MOSFETS without going into over current shutdown.

I had some samples of the LTC1775 sent out to me as well. They are 100% pin compatible with the LTC1625 and make a direct replacement.

LTC1775 data sheet attached here:

Ive had a few question about this unit so i think ill answer them publicly so everyone can see...

Ive been looking at the data sheet for the ltc1625...and your schematic... I see you have pin 4 floating.... so it wont know whether to run in continuous mode or burst mode?? depending on how it floats??

FCB actually rises to INTVcc voltage when left open. The pin pull-up current on the FCB pin is around 5uA. This makes the controller go into burst mode operation at low current outputs. It could of been tied to ground, or directly to INTVcc, but it was left open on purpose to keep my options open incase i wanted to switch modes (i can just short it to the surrounding ground plane).

Burst mode is more efficient at lower output currents (im talking just few mA), but produces a larger voltage ripple, and not recommended for highly sensitive devices.