prototype SEPIC DC-DC Converter module

[djQUAN]

I was planning to make this module a long time ago. just had barely enough time last weekend to complete it. ;) only powered it up but not yet checked specs if they were met.

http://i119.photobucket.com/albums/o128/djquan128/Image004-1.jpg

I designed it to take in 10-16V input, regulated 14.4V out (or any that you wish, adjustable with a trimmer) at about 2amps. very few components and no SMD. :D

would be perfect for an amp32/amp3 running on three or four Li-Ion cells in series.

initial specs I got is efficiency at around 80% and idle current at no load is only ~10mA. still need to tweak compensation network as I get overshoot when load is suddenly removed.

[djQUAN]

I want it to do more than 5A but LT1170 chip can do only 5A peak so output current is limited. this is just a prototype though so I could get a grasp on SEPIC converter topology. after this we go bigger. ;)

[krilli]

I rather do like it!

[djQUAN]

I found the LT1170 almost thrown away at work so I got it. :D

It is originally used for boost topology but since SEPIC also use negative side switch, I used it there. seems to work fine.

although at higher currents I would prefer push pull so it would be easier for the output capacitors. converters like SEPIC/boost/buck put too much peak current through output caps. I had to get Panasonic FM's from Jan for this one to handle the ripple currents involved.

[V-bro]

Nice! This is way more advanced than an LM1084 circuit. I'm sure it stiffens up the voltage nicely to improve the amplifier performance!

Today I had an email that my boards are shipped, five pieces of DC up converter boards, the silkscreen says thanks to mr. Quan! :)

It is about the size of a Truepath board and can deliver at least 600W of clean power in any desired voltage, single or dual!

Here's the latest silkscreen:

[V-bro]

You can always parallel to double the power rating...

[djQUAN]

Although it's not enough for a BIG amp:
I think it may be very worthwhile to check out the LT1170/1270!
- very compact designs
- input 3-30Vdc
- output upto 60Vdc
- 5 to 10A
- pos to neg or...pos to symmetrical
- can be used for lot of amps for car use.
- AMP-1x, -4x, -5, -9x, -10x
Can make a cheap 12-24Vdc/5-10A converter, like proposed?

if someone could send me a couple I'll try and build a design. ;) else, I'm stuck with TL494 since those aren't available over here :P

[V-bro]

Ok, my boards arrived, I've got six pieces! :)
543

[V-bro]

If anyone wants a proto board I sell it at cost price...
Power board: 29,35 euro
PWM board: 14,26 euro

[V-bro]

IT WORKS!! there are one or two small cosmetic changes to be made, but the circuitry is up and running without a flaw! Now I am still waiting on better toroidal cores, last time I ordered ones that were too high perm... Then I want to find the ultimate FET for this purpose, right now we still ise the STP75N06, we tried some 40V fets with lower Rds on and incredible current capability, but the waveform was turning sinusoidal, so there must be too much parasitic capacitance...

When this is all solved we can do some brutal power testing!

[V-bro]

Too bad it does not work with STW35.... or STW50....


It might work, but it would be far from ideal, they are 200V... There will always be much better choices...

[djQUAN]

use IRF3205 or IRF1404 IIRC.

what is the value of the gate resistors used and what is your gate driver?

[V-bro]

Thanks!

The Rg is 22R and the SG3525 is the driver... Should I have added resistors? Funny thing is that neither commercial designs use a separate driver, just straight from the SG3525 ouputs...

With IRF1404 I had some problems when I paralleled them, single ones do reasonably well, but the waveform gets messed up from more than that. It starts to look a bit like a sine wave... Seems like they suffer from too much input capacitance...

[djQUAN]

when it turns into a sine wave does the input idle current change? have you tried adjusting deadtime?

[V-bro]

Yes, increasing dead time wasn't the solution. Idle current, well, I dont remember exactly, the wave looked so messed up that I didn't look much further to investigate the performance...

[djQUAN]

I haven't experienced that yet but I'm also guessing the high perm ferrite has something to do with it.

[V-bro]

This wasn't with the high perm ferrite core, but with a few stacked common mode filter cores (shaped like E-I, but then one solid piece)

[V-bro]

Here's how my first test build looks by the way...

http://www.zelfbouwaudio.nl/forum/download/file.php?id=27070&mode=view
http://www.zelfbouwaudio.nl/forum/download/file.php?id=27071&mode=view
http://www.zelfbouwaudio.nl/forum/download/file.php?id=27072&mode=view
http://www.zelfbouwaudio.nl/forum/download/file.php?id=27073&mode=view

[djQUAN]

a short little update. two little problems with the prototype.

first was the LT chip was going full tilt once turned on. cause was that the package I used from Eagle has the pins for control and feedback reversed for the LT1170. cut the tracks and it regulated just fine.

second problem was that there was an overshoot when the supply was driving full load and then the load goes to zero. cause was that I took feedback after the output filter. it was causing a delay on the output causing problems with the feedback compensation. I was hoping I could counter the drop across the output inductor but it caused more problems than benefits so I cut another track and took feedback before the output filter.

SMPS is now powering amp32 with constant 13.5V from an 11.1V Li Ion battery pack with no problems.

[V-bro]

Hey welcome back!! Good one! .

So what operating voltage range does it have now and at what current?

[djQUAN]

I designed it with around 9-16VDC input and around 14.4V output 2amps. the 2 amps will probably be a lot less as the SMPS IC heats up (but only warm to the touch) with only amp32 and 8 ohm drivers.

[djQUAN]

Two years later and that project went on collecting dust due to low output current capability.

I designed a new PCB using a different controller (the old TL494) and high side current sensing using MAX4172. Tested up to 2A current was easy but was afraid to go higher as the mosfet does not have a heatsink. It should do atleast 5A cont with proper cooling. Board is pretty compact too, about the same size as a bar of soap. It should be able to drive AMP9 or similar single rail amps.

The PCB can be configured for SEPIC or boost converter topology. It also makes a nice CC/CV power supply for driving high power LEDs or charging large Li-Ion/LiFePO4 battery packs.

Pics to come once I gain access to a camera (dad took it on a trip)

[V-bro]

Oh man, I know how time just flies by without stuff getting finished, and sadly I have had to leave people waiting for ages for some projects.... It's been due to all sorts of circumstances, this also being the reason I haven't been around here all that much...

At work I did have some opportunities to go on with the DC-DC converter project as I sometimes have to train students and we've moved up with the converter with a new modulator board that features a sense line towards the output and will enable the duty cycle kind of voltage regulation and slight output voltage adjustment! The board has yet to be ordered and should fit the original design main board.

[djQUAN]

Took me a while to figure this out. The converter works fine in CV mode but goes crazy in CC mode. the current limiting feedback coupled with the power stage transfer function makes a nice oscillator!

I took some reading and also ended up in this thread from diyaudio (see post#10)
http://www.diyaudio.com/forums/power-supplies/127033-difficulties-sepic-converter.html

I also posted a schematic and PCB layout image on that thread (in the next page)

realizing a SEPIC is so troublesome to stabilize it is only suited for supplies with very slow transient response. I ended up slowing the current feedback so much that if the load is connected when the supply is on, there is a very large surge of current before current limiting kicks in. The solution was to simply connect the load before powering up the circuit as it has soft start. My first application was a battery charger for a 4cell 15Ah LiFePO4 battery which needs about 14.60V at 3A. The slow transient response is no problem for that. Seems to be working fine with an efficiency of about 85% and charging the battery right now.

I think this would be the most power output I would be designing a SEPIC for. Too much hassle with the control loop! The PCB can also be configured for boost converter application. I haven't tested it for that yet but I think I can squeeze more power in that topology using this board. Probably useful for my amp9 board that is gathering dust. ;)

[djQUAN]

had some spare time at work and set it up on their powersupply and electronic load. worst case 11V input and 14.4V output max current. max power I was able to get was about 40.7W (as shown on the e-load) before current limiting kicks in. Let it sat at that power level for a few hours and came back it still kicking. :) disregarding cable loss, the total efficiency of the converter was 85%. Output was rock solid at 14.4V (I previously set it to 14.6V but lowered it for the cell balancer circuits to release) from no load to full load and 10V to 18V input.

Now all I have to do is make a case for this. I'll post pics once it's done.

[djQUAN]

Pics of my completed converter module is posted in DIYaudio.

go here: http://www.diyaudio.com/forums/power-supplies/127033-difficulties-sepic-converter-2.html#post3244999

[V-bro]

Looks very professional mate! :)

[djQUAN]

Thanks! seems to work pretty well. Now planning to make a boost version to power an amp9 basic. :D