LM3478MM Sepic problem

Hello experts

Still rather new to the world of SEPIC, I have a problem with a design.

I have made a circuit based on the datasheet for the LM3478. It is intended
to run with Vin from 10V to 32V and deliver 24V out with a maximum of 4Amps.

My testsetup runs at 32V input. The circuit is as can be seen here:
www.impc.dk/scm/lm3478sepic.pdf

A colleague of mine recently built something quite similar but with the
LTC1871 controller instead. But I'd rather use the national part, since we
already has this part in stock.

When I run with no load, output is 24V and everything is nice. But when I
load the darn thing, even only by one amp, everything gets so hot and
eventually the sense resistors crack.

I plan to change R79 so that I run at 200kHz, because according to my
calculations 100kHz is a little to slow for my 22uH/13A inductors.

This heat issue has me puzzled somewhat but before i dive into measuring, I
wonder if anyone would have a comment on the schematic. Especially regarding
the values for R103/C16 and R145 which I have been struggling to
determine.But also if I need anything or need to change anything. Not much
is to be found on google regarding SEPIC with this chip, unfortunately.

I am eager to learn, but still a beginner in this field, so be gentle ;-)

Thanking you in advance for any and all help.

Best regards
Henrik

 

Dear Joerg,

Thank you for your reply.

About C73/C74, I always thought that using electrolytics was more safe. I
though I read that somewhere. But I could easily be wrong and cofusing two
unrelated subjects. But the psysical size of ceramics should be quite large
to handle the job, shouldn't they? Do they handle currents better than
electrolytics?

I'll check L4, I use these
http://www.chilisin.com.tw/en_new/product/images/pdf/14-tdh.pdf
the TDH2420 series, last columns in the table. Will my 1000uF suffice for
this?

My sense resistors should be able to handle 12Amps. But something made them
break with a one amp load tonight.

I'll check if my MOSFET does avalanche. I think I have a few 100V lying
around if need be.

Time to pull out the oscilloscope and get this beast running ;-)

Hehehe... actually we have been discussing that electrolyte schematics
symbol ourselves. There is indeed a rather large difference between european
and american symbols, some use a solid box to designate negative and an
unfilled box to designate positive. Some use a straight line for positive
and a curve for negative and the some use the symbol we use. I admit it is
rather unclever and I would like much if there was just one standard for
this. No need to complicate matters with a "european" and an "american"
standard. In this day and age, with everything bein so global and "one
village"-like, we should just settle on one system that satisfies all and
avoid the confusion.

Best regards
Henrik

 

I have not had experience with a SEPIC, but I have worked with several
single inductor boost converters. Looking at a similar circuit in the
LTSpice library, I see that the capacitors are 20 times smaller for a 1 amp
output, so I think you might be able to use something like 100 uF for
C73-74 and 200 uF for the output filters. Also, all the electrolytics must
be lew ESR high frequency types, and should be bypassed with about 200 nF
ceramics. If the capacitors get hot, they have too high ESR.

Using the scope you should be able to see if saturation occurs, which will
cause very high power in the sense resistors. Use a 10 amp fuse or circuit
breaker in the supply, and monitor the input current. Slowly increase the
load, and watch for a sharp increase in current. Check the duty cycle at
this point and see if it makes sense.

I also recommend using LTSpice to simulate your circuit. They have
something similar to the LM3478. Try LT1241, LT1247, or LT1317 as a start.

Good luck.

Paul

 

Keep in mind that both SEPIC and Cuk converters have their switches
seeing the sum of the input and output currents and voltages. Thus,
at 24V-to-24V and 4A loading, the stresses on the switches are easily
above 8A and 48V.

At 32V input, MBR2545 would see at least 56V. It will certainly
avalanche this Schottky diode which is only rated at 45V.

At least, look at the voltages at the gate and at the source of the
power MOSFET.

 

Ceramics have lower ESR than electrolytics. And they aren't polarised.

About the only factor that electrolytics have in their favour is that
large values are available and affordable. Looking at Farnell's site, a
[email protected] ceramic costs £0.70, versus £0.02p for an electrolytic. And
that's the largest ceramic they list. So if you actually wanted 1000uF of
ceramics, that's £7.

But as Joerg says, 1000uF seems excessive, although admittedly I only have
the vaguest clue about power conversion circuits.

4A through 22uH = (1/2)*L*I^2 = 0.5*22e-6*4^2 = 176uJ.
24V on 1000uF = (1/2)*C*V^2 = 0.5*1e-3*24^2 = 288mJ.

That's a >1600:1 energy storage ratio.

Doing a back-of-the-envelope simulation in LTSpice (with a fixed duty
cycle and parasitic values pulled out of ... thin air) doesn't show a
great deal of difference between 1uF and 1000uF for the pass caps.

LTSpice model appended in case anyone wants to point out flaws in my
reasoning.

Version 4
SHEET 1 1192 680
WIRE 224 16 64 16
WIRE 320 16 224 16
WIRE 464 16 400 16
WIRE 528 16 464 16
WIRE 656 16 592 16
WIRE 768 16 656 16
WIRE 928 16 832 16
WIRE 1072 16 928 16
WIRE 1136 16 1072 16
WIRE 64 80 64 16
WIRE 224 80 224 16
WIRE 464 80 464 16
WIRE 656 80 656 16
WIRE 928 80 928 16
WIRE 1136 80 1136 16
WIRE 416 96 384 96
WIRE 384 112 384 96
WIRE 336 144 320 144
WIRE 416 144 336 144
WIRE 320 176 320 144
WIRE 64 288 64 160
WIRE 224 288 224 144
WIRE 224 288 64 288
WIRE 320 288 320 256
WIRE 320 288 224 288
WIRE 464 288 464 160
WIRE 464 288 320 288
WIRE 656 288 656 160
WIRE 656 288 464 288
WIRE 928 288 928 144
WIRE 928 288 656 288
WIRE 1136 288 1136 160
WIRE 1136 288 928 288
WIRE 64 320 64 288
FLAG 64 320 0
FLAG 384 112 0
FLAG 1072 16 out
FLAG 464 16 P1
FLAG 656 16 P2
FLAG 336 144 clk
SYMBOL voltage 64 64 R0
WINDOW 123 0 0 Left 0
WINDOW 39 24 132 Left 0
SYMATTR SpiceLine Rser=50m
SYMATTR InstName V1
SYMATTR Value 24V
SYMBOL cap 208 80 R0
SYMATTR InstName Cin
SYMATTR Value 1000µF
SYMATTR SpiceLine Rser=50m
SYMBOL ind 304 32 R270
WINDOW 0 32 56 VTop 0
WINDOW 3 5 56 VBottom 0
SYMATTR InstName L1
SYMATTR Value 22µH
SYMATTR SpiceLine Rser=50m
SYMBOL cap 528 32 R270
WINDOW 0 32 32 VTop 0
WINDOW 3 0 32 VBottom 0
SYMATTR InstName C1
SYMATTR Value {C}
SYMATTR SpiceLine Rser=50m
SYMBOL ind 640 64 R0
SYMATTR InstName L2
SYMATTR Value 22µH
SYMATTR SpiceLine Rser=50m
SYMBOL cap 912 80 R0
SYMATTR InstName C3
SYMATTR Value 3000µF
SYMATTR SpiceLine Rser=50m
SYMBOL res 1120 64 R0
SYMATTR InstName R1
SYMATTR Value 6R
SYMBOL sw 464 64 R0
SYMATTR InstName S1
SYMBOL voltage 320 160 R0
WINDOW 3 24 104 Invisible 0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR Value PULSE(-1 1 5us 50ns 50ns 2.6us 5us)
SYMATTR InstName V2
SYMBOL schottky 768 32 R270
WINDOW 0 32 32 VTop 0
WINDOW 3 0 32 VBottom 0
SYMATTR InstName D1
SYMATTR Value 1N5817
TEXT 320 328 Left 0 !.model SW SW(Ron=50m,Roff=1Meg)
TEXT 32 352 Left 0 !.tran 0 10ms 9.95ms
TEXT 784 328 Left 0 !.ic V(out)=24V
TEXT 704 384 Left 0 !.step dec param C 1uF 1000uF 2

 

What is the point of R97?
It does not do anything significant , that I can see anyway. Maybe I
need to get new glasses :0)
Unfortunately this does not point to your problem though :0(

Cheers
Rob

 

24 * 10/190.47 = 1.260041

What do you think the nominal reference voltage of the LM3478 is?

Some scope traces would help. Maybe something is way off causing
saturation of those drum inductors.

 

I personally grew up with zig-zag lines, but I can cope with either.

When you're reading a 2nd generation photocopy of a schematic (or a
crappy, scanned PDF), you're likely to prefer 4K7 to 4.7K.

As for polarised capacitors, yes, it's pretty weird to see the solid bar
being the positive terminal.

 

Ok, changed my diodes to 100V types and lowered the input voltage to
20Volts, but still no cigar :-( I can still burn my fingers on the Mosfet
and still fry my Rsense when I load with one amp.

I've taken a scope picture of my DRIVE pin, and it looks really not like I
would expect.

www.impc.dk/scm/gate1_noload.tif is the DRIVE pin on the LM3478 when viewed
with 10uS timebase. I run RFA as 180K.

www.impc.dk/scm/gate2_noload.tif is the DRIVE pin on the LM3478 when zoomed
in a little more. I would have expected this to be much nicer waveforms.

What could be happening here? This is a no load situation. There is no
heating in the no load situation and output is 24V as I would expect.

Thanking you all in advance for any help


Best regards
Henrik

 

What does the MOSFET current look like?

 

Well, I've taken af few more scope shots:

First the sourcepin, there is neglible distance between the source pin and
the Rsense.
www.impc.dk/scm/source1_noload.tif
www.impc.dk/scm/source2_noload.tif

The drain pin:
www.impc.dk/scm/drain1_noload.tif

Any help or suggestions is gladly appreciated.


Best regards
Henrik

 

Hello Joerg,

I've taken a few screenshots:

First the sourcepin, there is neglible distance between the source pin and
the Rsense.
www.impc.dk/scm/source1_noload.tif
www.impc.dk/scm/source2_noload.tif

The drain pin:
www.impc.dk/scm/drain1_noload.tif


Amy help is gladly appreciated.

Best regards
Henrik

 

"Joerg" <> skrev i en meddelelse

Well, output is at 24V as expected, so nothing is shortened. But as you
describe later, it may be the MOSFET that does this?

I use Rfa = 180Kohms. The datasheet for the inductors are here:
www.impc.dk/scm/tdh-series.pdf

The partnumber for the inductor is TDH2420T-220K-N

I have 100V mos-fets coming in this late afternoon. The courier is on his
way ;-) I thought I had a few lying around, but none that was sufficient.

Could this be the entire cause of what I am seeing here?


I really do value your comments. This is a great learning experience for me.

Best regards
Henrik

 

It's a 4 layer PCB but i am having trouble generating a PDF of the
individual layers. Maybe our CAD specialist will have more luck that I. The
inductors are close to the LM3478 (a few mm) but the traces are short, all
components are situated by only a few mm from the LM3478 and there is lots
of copperplane around. Only exception is the RFA signal which unfortunately
has a longer trace due to the V2 attachement.

Choking, is this somehow related to the MOS-FET being, not up for the task?
Or is there another reason for this?

13 amps max at 190mV, I'll switch to 15mR or maybe even 20mR. That should be
fine for the inductors I use, right?

According to my calculations L4 has an average current of 8,7Amps when I
have 24V/3.5A out with a Vin of 10Volts. And this should yield a peak Switch
current of 12.7Amps, which is what Rsense is seeing. Am I right?

Yes, I am aware that the inductors could be better, but the ones that i have
used were what we had at the time of design. I have learned a lot since
then, but for now I am stuck.

Trying to generate PDF's unless you are able to read Cadstar or gerber
files?

Yes, I am now painfully aware that the road to the flat part of the learning
curve is through burning FET's and fingers ;-) But as long as the light in
the horizon is still visible, I don't mind the steep climb. However, diving
into this headfirst, has significantly renewed my respect for all you
hardcore guru's of this field ;-)

 

Those screen shots indicate some rather extreme ringing at a higher
frequency than the applied PWM drive. I have seen noise like that on my own
100 kHz switcher. I have tried to measure the MOSFET current with a scope
clipped directly across a flat metal strip sense resistor, and there is a
lot of high frequency noise. But I also see it when I clip the ground lead
of the probe to the tip, which indicated that it is picking up radiated or
induced RF. I have read that you must make a special probe tip adaptor to
get a good signal.

I had some bad experiences with current mode switching supply controllers,
particularly the LT1247 and the equivalent UC1843A. When I tried to get
more than about 20-30 watts, it became unstable and I experienced
overheating of the inductor(s) and the MOSFET. Of course, as things heat
up, even more power is consumed in I2R losses, and a runaway condition can
result. I did find that it was important to provide a very good gate drive,
especially when I used a MOSFET with low RdsOn and high current and voltage
ratings. The gate capacitance and charge caused a lot of switching losses,
so I used a 9A gate driver UCC27321 with some improvement.

Now I am using a PIC16HV616 with PWM output for the switching boost
converter, and it usually seems stable, but sometimes it tries to regulate
at a higher current than it should. But at least everything seems to be
more predictable and is easier to measure on a scope. I am only measuring
the output current and adjusting the PWM in a software loop, rather than
attempting cycle-by-cycle regulation.

The previous current mode converter had such variable waveforms that I
could not get a good picture of the operation. The duty cycle was all over
the place, and I saw bursts of PWM signals alternating with quiescent
periods.

My circuits are unique in that I am regulating an output current into a
non-linear load (a string of high power LEDs), so I could not run the
circuit at no load (which would be a short circuit). But I only had major
problems when I exceeded about 20 watts, and it really gave problems at
about 50 watts. Yet it is only a 1" x 2.5" PCB, and I was trying to get
better than 85% efficiency. I could easily obtain 75%. It sounds like you
have even more severe problems.

Paul

 

Yes, one plane is a full groundplane. I haven't had the luck to build PDF's,
CAD is not my area of work, but here are the gerberfiles. The PSU are
located in the left side. Note there are two of them, but they are similar.

Well, I got my 100V mosfets and new 100V diodes today, but everything still
burns, when loaded :-( I've taken new scope shots

Here the drain, which has changed, but still looks weird.
www.impc.dk/scm/drain1_noload.tif

Here the source, which is so "grassy"
www.impc.dk/scm/source1_noload.tif

and the source grass up close:
www.impc.dk/scm/source2_noload.tif

The gate, which has oscillations, I think
www.impc.dk/scm/gate1_noload.tif

The gate, up close:
www.impc.dk/scm/gate2_noload.tif

I almost don't know which was more painful, the actual fingerburn, or the
hardhitting fact that the bigger mosfet did not just fix my problem ;-(

I've tried changing R145 to alter the slope, but so far to no avail.
www.impc.dk/scm/lm3478sepic.pdf

Yes, I know that you are right, but having no more blisterfree fingers, I
grasp at every straw ;-)
www.impc.dk/scm/gerber.zip

No, I can definetely see that this is not at all easy and that my approach
from the beginning was probably wrong, but I keep the hope alive ;-) I
really like this design to run in the end. And learning along the way should
make this easier next time ;-)

Best regards
Henrik

 

I think you have other problems but maybe try a resistor at the gate.I
don't think that's resonant oscillations (I've never seen them like
that before) I could be wrong though but a small value resistor in the
gate wouldn't hurt.

You might also want to put a Schottky on the current sense pin to
clamp reverse voltages. If the chip doesn't have leading edge blanking
you also may want to place an RC filter on current feedback, sometimes
this is still recommended with an IC that has LEB internally.

Also maybe swap out the controller if you have a spare that would at
least eliminate it as the problem.

All those data sheet application schematics are usually bare bones so
they can say look how few components you need. Sometimes they work
sometimes (likely) you will need to modify it.

I'm by no means an expert but that's been my experience from some of
my previous experiences in SMPS's.

Good luck

 

Not sure about that grounding path for the LM3478..




Best regards,
Spehro Pefhany

 

I was going to say that the scope signals were noisy, until I saw the
drain waveform. Still, even when things are normal, you'll need to
watch scope probe ground lead length. Please apply a minimum load when
scoping signals - this may avoid other problems that aren't currently
of interest (ie heat build-up and death are the real concern here).

You're going to have to clean up the current sensing signal, as the
160mV threshold of the controller won't tolerate a lot of noise. It
appears to be reacting to drive noise, as is. R145 won't add much in
the way of slope compensation if it is below 1K.

You might also move C11 to the other side of R145 (and reduce it to
<100pF if R145 is increased to 1K). The capacitor in the present
position will be hard-pressed to absorb gate drive charge, as
intended, at a 10 milliohm level. The app notes for this part seem to
address much lower power levels, sensing 50 milliohm parts without
slope compensation.

Current-snub the mosfet drain and bead/voltage-snub the schottky
rectifiers.

Four paralleled sensors will work better than a series/parallel
combination. These better be film parts.

Your scoped waveforms are crazy, showing the device driving the mosfet
at some MHz, within the intended lower frequency duty cycle of 200KHz.
Make sure R79 is really >=100K.

Higher frequency driven oscillation shouldn't be possible if current
sensing is latched, and is unlikely if the integrated driver's
impedance is in excess of 15R, as per spec.

May need a larger decoupling cap on Vin.

Nat Semi seem to be pretty optimistic in the functional diagram for
this device, showing internal analog level shifting on the current
sensing signal between pins 4 and 5. I'll believe it when I see it.

RL

 

On Fri, 27 Mar 2009 13:57:18 +0100, "Henrik [7182]"

Try not to save iterated waveform plots with the same name as previous
plots. It's going to confuse things.

Giving the name date and time stamps is the simplest ie gate03280947
is gate plot at 9:47AM on the 28th.

RL