Is this ~150W 12V dc-dc supply realistic?

[MrPerfectionest]

A guy on mp3car.com is offering up a very compact 12V supply and I am
wondering if the claims he is making are legit. I had contemplated a
similar supply myself but my knowledge was limited to boost / buck
configurations and the high cost of low esr caps to manage output
ripple kept me from taking the idea to fruition. Did I severely
underestimate the power of a full/half bridge or forward converter
setup? He is being mum on what the topology is but from the looks of
it, it appears full bridge to me. What he has posted about it:

I just got done designing a new DC-DC power supply that takes in 6-20

volts and puts out 12V on the output. With the output between 10-16
volts the output load can be 10 amps sustained, 13 amps peak. It is
optimized for car use and will survive cranking no problem with a 10
amp load.

It measures 1.9" by 3.5" by 1.2" and has 4 standard 0.25 inch

automotive faston connectors.

There is a power on LED and a On/Off Jumper.

It can be ordered as a fixed output supply, or as a supply with an

output range adjustable by a small pot.

Sounds too good to be true? It's real. I even surprised myself when I

measured my first sample and I measured 97.5% efficiency with a 85-watt
load!

The supply is 95+ percent efficient over the 8-15 volt range.

http://www.mpegbox.net/DSX12V/DSX12V_2.jpg
http://www.mpegbox.net/DSX12V/DSX12V_top.jpg

Thanks

 

[Fritz Schlunder]

A guy on mp3car.com is offering up a very compact 12V supply and I am
wondering if the claims he is making are legit. I had contemplated a
similar supply myself but my knowledge was limited to boost / buck
configurations and the high cost of low esr caps to manage output
ripple kept me from taking the idea to fruition. Did I severely
underestimate the power of a full/half bridge or forward converter
setup? He is being mum on what the topology is but from the looks of
it, it appears full bridge to me. What he has posted about it:

I don't think this is a half/full bridge or forward converter. All three of
those types of converters typically employ at least two magnetic devices
(one transformer and at least one inductor). This device only appears to
have one magnetic device (which appears to only have 2 wires on it, so only
a simple inductor, although that isn't 100% guaranteed given the low
resolution of the images). Additionally a full bridge takes more than four
semiconductor power devices (four MOSFETs + typically two output
rectifiers), but the board only appears to have four on it.

Exactly what topology this is isn't fully clear to me given the resolution
of the images.

However, given the modern design approach that appears to employ synchronous
rectifiers, the output power and size specifications are quite possibly
realistic. The efficiency claim seems awfully high, but on the other hand
not totally unrealistic using synchronous rectifiers and modern MOSFETs.

It looks legit to me.

 

[James Meyer]

I don't think this is a half/full bridge or forward converter. All three of
those types of converters typically employ at least two magnetic devices
(one transformer and at least one inductor). This device only appears to
have one magnetic device (which appears to only have 2 wires on it, so only
a simple inductor, although that isn't 100% guaranteed given the low
resolution of the images).

There is a solder pad visible between the "inductor" and the caps. My
guess is that it's for a center-tap on the inductor. I seem to recall a CUK or
SEPIC topology that uses a center tapped inductor for a buck/boost converter.

My guess is that the converter is legit too.

Jim

 

[Ted Edwards]

A guy on mp3car.com is offering up a very compact 12V supply and I am
wondering if the claims he is making are legit. I had contemplated a
similar supply myself but my knowledge was limited to boost / buck
...
volts and puts out 12V on the output.

If you are interested in buoilding your own, see
http://www.maxim-ic.com/appnotes.cfm/appnote_number/1051/ln/en

Ted

 

[MrPerfectionest]

I am not so sure Jim. I didn't mention but when I was trying to
accomplish this myself I looked into the SEPIC configuration,
particularly with the LM3488 chip from national and to be able to
manage the ripple to the ATX spec (<120mV) was going to require like
4-5 large 2200uF low esr caps at low end of the voltage range. I
wanted to stick hard to the spec over the full voltage range but with
Vins near 12V (as most cars will be during non cranking times) then its
much easier.

Now thinking about this some, you are probably right about it being
SEPIC. I inquired to him about what the output ripple was and he
quoted me 60mV with a 10A load. Looking at some .jpgs he posted of his
scope shots, a lot of his test conditions were around the upper 9V
range or 13V range which isn't quite starting to stress SEPIC yet.

I guess the two SMD devices there in the middle of the board that are
side by side are the CS capacitors.

It all makes sense now I think.

 

[Ian Stirling]

My guess is that the converter is legit too.

I'd guess that the efficiency is probably a bit off.
It's not hard to get measurement errors, and a 97% efficiency is moderately
hard to obtain.

 

[[email protected]]

I think it is a bridge converter with two of the fets being used in
reverse current.

 

[[email protected]]

Based on the Linear Technology LTC3780. It operates as boost,
buck-boost, or buck depending on Vout - Vin.

 

[Winfield Hill]

[email protected] wrote...

Based on the Linear Technology LTC3780. It operates as boost,
buck-boost, or buck depending on Vout - Vin.

They show a 60W example on the front page, although they don't give
any critical part values. One surprising thing, I couldn't find a
FET gate-current drive-capability spec anywhere in LTC's datasheet.
This is an important parameter for a converter running at 400kHz.
They do mention you'll need to use logic-level FETs, due to the low
internal Vcc = 6V drive voltage. The VCC regulator is spec'd with
a 20mA load (5mA per FET), and the absolute maximum rating is 40mA,
which implies rather modest average FET gate-drive currents. Their
worked-out 36W example uses Vishay Si7884 MOSFETs, which feature a
low 10-milli-ohm Ron, have a fairly-high 2000pF gate capacitance,
and a 22nC gate charge at 6V. Note, 22nC * 4 * 400kHz = 35mA, oops!
Hmm, wonder why they didn't tie PLLFLTR low and run at 200kHz.

Apparently the LTC3780 is capable of driving moderately-serious,
but not too serious, power FETs.

It may be possible to scale up the 36W design example, but be aware
that higher operating currents mean higher dI/dt, which means larger
V = L dI/dt voltage spikes from fast FET switching and high-current
slew rates, which means blown out stuff, if you don't know exactly
what you're doing. Suggestion: Build the 36W and 60W versions first.