Can type 2 compensation cure boost smps problems?

[eem2am]

type 2 compensation

Hi,

I'm working for a company in Chicago and doing continuous mode boost smps at 17V in and 27V out to drive LEDs at 350mA for signage. (10W)

Anyway:

The following circuit uses components C1, C2 and R5 to give a Type 2 compensaton to the boost:-
.....

Here is the LED current at start up for the above circuit......

...please notice how there is serious overshoot of the current up to 0.9A at switch_on...this is due to the resonant behaviour of the boost's L's and C's at the application of V(in).
Please also notice that later on, the LED current reaches 350mA but first there is overshoot of the LED current up to about 400mA.

These are serious effects, LEDs do not have the surge current ratings of normal diodes and these LEDs will die much quicker than well-treated LEDs.

--------------------------------------------------------------
To correct for the above effects, a 10uF capacitor C9 (below) is added across the LEDs.
(supplementary diodes D11 and D13 ensure that C9 does not effect the circuit after turn_on ...D13 gives a discharge path for C9 when the boost is switched off)

C9 basically gives a soft start to the boost.....here is the augmented circuit......

and here is the LED current waveform with this added C9......

...Please notice how the resonant surge current in the LEDs, and the start-up overshoot LED current are both now TOTALLY CURED.
-------------------------------------------------------

Can readers confirm that that the augmented circuit with C9 provides Type 3 compensation?

Can readers also confirm that the first circuit, with inly C1, C2 and R5 as compensation components (no C9), would NEVER be able to cure either the switch_on resonant surge current OR the start-up overshoot LED current. -No matter how the values of C1, C2 and R5 were modified ?

 

[FvM]

compesation problems

To my opinion, there is too much auxilary stuff involved in the circuit. At first sight, it seems to be far from a lean
final production design. Can you post the *.asc file and additional libraries to allow some tests?

 

[eem2am]

problems on compensation

here is the .asc file

https://www.quickfilepost.com/download.do?get=5b76fad35bdaf86d38fa47f36e4bc160

 

[FvM]

type 2 compensation circuit

Because the initial inrush current hasn't to do with active boost operation, it can't be influenced by the loop compensation,
of course. It can be removed e.g. by ramping the input voltage, which happens anyway in most real applications. Also
selecting a more realistic (higher series resistance) boost inductor can considerably reduce it, even without a
preload diode D6.

The overshoot in boost operation start can be removed by adjusting the loop compensation, I used e.g. C1=6.8n,
C2=100p, R5=7.5k.

L3/C8 won't be needed,if the load is mounted near to the converter, also L2 necessity can be questioned, because
L1 is achieving an almost smooth input current.

 

[eem2am]

what are the problems of compensation

Thankyou FvM,

your C1, C2 and R5 values do indeed get rid of the start-up overshoot.

However, as you correctly state, they do not get rid of the switch_on resonant current surge peak in the LEDs.

You spoke of ramping the input voltage. -unfortunately i do not have the luxury of that....my Vin is just a sudden step voltage of 17V.

and I do not have any room for additional components to ramp the input voltage.

Also, i cannot select an L1 with higher series resistance as it will heat up too much in the small space where this SMPS is placed.

I am surprised to hear you say that EMC inductors L2 and L3 and C8 could be omitted.

With L2 and L3 and C8 all removed i got 160mA of input ripple current and 3.5mA of output ripple current.
-with current "strict" EMC laws, there is surely no way that i will get away with this?

This is a 10W converter.

By the way, may i ask do you have any thoughts on the need for common mode chokes and general common mode noise from this circuit?

Added after 34 minutes:

FvM,

I can appreciate that your values for C1, C2 and R5 stop overshoot when there is no C9.

However, these compensation components that you have selected have a lower gain and phase marginn then my values for C1, C2 and R5.

(-And in any case, C9 completely cures the overshoot problem anyway...)

So what i am saying is , that your values of C1, C2 and R5 make the converter less robust than my values for these components as shown in the above diagrms.

-so your values for C1, C2 and R5 would not be as robust against component changes with age or temperature....and yours my mean the converter breaking into oscillation at some point in its life.

May i ask do you agree?

 

[max0412]

soft start boost smps

I would first start out with just a bare bones cct. There will be patristic RCL in your actual layout caps have esr,source impeadance etc. Your inductor is grossly over sized if it actually only has 15mohm DCR. A 0.5A inductor shielded SMD at that inductance is usually about 0.7 to over 1 ohm DCR. This would help to limit the turn on surge if not place a small R before your start up diode. Then add components as they are deemed necessary ie your input PI filter or maybe a Diode for decoupling and a simple RC filter may be good enough.

A 0.5A inductor is much smaller then the several amp inductor you are useing if size is critical which it usually is.

This is my result with a 0.8ohm dcr inductor (COILTRONICS UP0.4C-470-R) and a 4.7ohm resistor.

 

[FvM]

common mode choke led boost

I think the compensation is sufficient for a simple LED switcher. Your initial discussion was about overshoot, not any additional requirements.

The EMC filter design depends completely on the application. I said, the output filter can be omitted with the load near to the switcher, I think that's obvious. In this case, also the output capacitor can be possibly reduced, which would be the best way to get rid of the inrush current problem.

To decide about the necessity of an input filter and suitable means to fight the inrush current problem, the real application input source must be specified. If you connect a simulation zero ohm source to the switcher, the ripple current flows exclusively into the source.

I didn't want to go to deep into the circuit dimensioning details, however it seems questionable, that an economic design would use an 8 or 10 A inductor for less than 1 A, but a 120 mohms source resistor at the same time.

You said, that you don't have place for additional components, but you added a 10u capacitor and e.g. D6 to fight the inrush current. It seems to be meaningful to think about more effective solutions. But the prerequisitions, particular the real source impedance and power-on conditions must be known.

 

[max0412]

boost smps inductor

With some further thought I would also omit your start-up diode D6 as FVM already suggested. The combination of your source impedance and cap esr as well as the DCR of a more practical inductor would likely solve all your problems. If your inrush persists in hardware I doubt it would though then maybe add a small R at your input.

 

[eem2am]

designing of smps input surge protection circuit

hello,

thankyou max0412 and FvM,

I am using MSS1048-473 Coilcraft 47uH inductor and it has 130mR resistance. (not a 10A inductor)

-So it really is low DC resistance even though its only a 2.2A inductor which it needs to be since this must work down to 8V where Iin peak = 1.5A.

I cannot put in series damping resistances as it will heat up my small circuit too much and the pcb area is small so not enough planar copper to dissipate heat....i must keep it well cool as its also enclosed.

Also max0412 circuit above did not use input EMC inductor, so you have obviously not suffered the same high level of resonant LC effects that i will suffer as in my top diagrams, also max0412 has not output inductor.

I will need all these components to pass EMC test

So i cannot surely omit my C9 and diodes D6, D11 and D13. i need them because surely you cannot even for short time put 400mA plus through a 350mA LED. ?

LEDs are delicate to current surge? i thought so?

The source for these signs will be a car battery. So it is very very low impedance indeed......so it will not help to dampen my resonances. Car battery is a few 10's of milliohms source resistance.

So now do you think that i really do need my surge and overshoot protection ?

 

[FvM]

compensating diode ckt. solution problem

I assumed 8 - 10 A because of the 15 mohm inductor Rs in the simulation. A 2.2 A inductor seems appropriate.

I understand that in an automotive application, you have to assume the source hard switched to the circuit,
so there's no supply ramping. The load placement is still unclear, however. If it's near the switcher, I would
reduce C1 in a first attempt to fight the inrush problem. If this isn't suitable or doesn't help, a softstart
circuit with a series resistor and a FET switch to short it involves most likely the least effort.

When using schottky diodes in preload circuits, you should consider, that they are rather sensitive to current
spikes. The 4.7 ohm resistor can protect the diode, but it also reduces the effectivity of preload, passes
a larger current amount to the boost inductor.

As another point, the circuit should have an additional overvoltage protection. LTC1871 maximum voltage
rating of 36V ist sufficient for automotive without a protection device. e.g. a transil diode.

 

[eem2am]

no load overshoot boost smps

Thankyou FvM,

In the actual circuit i have my LTC1871 supplied through a 10R and then there is a 33V Zener to protect it (at least i hope so, since i did not go to the extent of a transil diode)

The load will be a series chain of LEDs mounted a few cms away from the SMPS PCB. -The SMPS output wires to the LEDs is just say 10cm as it has to curl round and about.

FvM

If it's near the switcher, I would
reduce C1 in a first attempt to fight the inrush problem
[*QUOTE]

...My apologies, i am not sure what you mean by this, are you actually saying that you would adjust one of the compensation capacitors to counteract the inrush?

Trying to limit inrush with a FET and a resistor ends up needing far more components than i thought............................

....since this thing had to be operational from 8 to 17V i had to end up using the LTC1871's INTVcc output to turn the FET on and switch out the inrush resistor.

There is also an input which powers the converter down, which also is seen here, using an optocoupler.

I couldnt get the FET inrush circuit using any less components than this, and it was too much , so had to bin it.................and in the end used my C9, D6, D11 and D13 as in my secong diagram down.

Regarding the Scottky sensitivity to current spikes, if in the datasheet the "IFSM" figure is upwards of 12A, then i presume it is OK, would you say this is correct?

Also, are we in agreement that a surge of 400mA through some 350mA LEDs, even if only occurring for one millisecond per day (at switch on) would severely reduce the life expectancy of the LEDs?

Added after 3 minutes:

oh sorry, i forgot to say that in this simulation, the inrush FET is not much good, as its Drain-source capacitance passes virtually the entire inrush current even when the FET should be off !

here is the .asc file of the inrush FET version...

https://uploading.com/files/9ETW4EU... NFET _OVERSHOOT REDUCTION _080609.asc.html

 

[max0412]

overshoot smps

I assumed 8 - 10 A because of the 15 mohm inductor Rs in the simulation.

This is how I assumed a large inductor was used.You had only 15mohm DCR for your inductor in the sim.

If this isn't suitable or doesn't help, a softstart circuit with a series resistor and a FET switch to short it involves most likely the least effort.

Along those lines there are IC's that control a high side P-FET gate dv for inrush limiting. I don't recall a particular part number at the moment I think TI makes one though. There is a wide selection of pfets in the 55 to 60v range.

The 4.7 ohm resistor can protect the diode, but it also reduces the effectivity of preload, passes a larger current amount to the boost inductor.

Yes I thought of that after my first post and it's why I added my second omiting it.

Also max0412 circuit above did not use input EMC inductor, so you have obviously not suffered the same high level of resonant LC effects that i will suffer as in my top diagrams, also max0412 has not output inductor.

I will need all these components to pass EMC test

Only you know what your acceptable harmonic levels are for EMC (specific test).

I suggested pretty much build a prototype bare bones then determine what needs correcting. The cct is low parts count and you could make a PCB in short order. Then you would know what real problems need to be addressed instead of worrying about problems that may exist only on your PC.

Edit

I found one of the PFET slew rate control drivers.Its not for your voltage range but there are similar ones, you just have to look for them at your supplier. This should give you some key words to use in your search. Here is the pdf "slew rate controller".

https://www.fairchildsemi.com/ds/FD/FDG901D.pdf

EDIT AGAIN

oh sorry, i forgot to say that in this simulation, the inrush FET is not much good, as its Drain-source capacitance passes virtually the entire inrush current even when the FET should be off !

What? Are you sure it's not conducting through the body diode? Spice is a little quirky try it in real life.

 

[FvM]

smps boost no load

I actually meant reducing C3 rather than C1, sorry for causing confusion.

A NMOS FET in the negative supply must have it's S and D pin exchanged to work correctly, because the most negative node is on the left.

 

[eem2am]

boost loop compensation

Hello,

FvM : I used your values of compensation components C1, R5, C9 (respectively 6.8nF, 7.5K, 100pF ) in this slightly improved but similar circuit................................................

....this time the input supply voltage is stepped from 8V (Vin min) to 17V (Vin max) with a period of 4ms.

Unfortunately, with these values of compensation components, the LED current suffers severely large overshoot to some 560mA and these 350mA LEDs would be damaged and have greatly reduced life-time..............

showing input volts (blue) and LED current (green)

......However, with the 10uF capacitor (C2) and diodes D15 & D16 included , the LED overshoot current is considerably less, though still not acceptable..........

...circuit...

...and the waveforms

(input volts blue, LED current in green)

So i am now wondering, what components can be used to get rid of this LED current overshoot which occurs in the continuous mode boost when its input voltage goes from Vin_min to Vin_max ?


(Please remember that the solution must also eradicate the switch_on surge current as depicted in the first post)
----------------------------
here is the .asc file

https://uploading.com/files/A2KY10JY/LTC1871 LED DRIVER _OVERSHOOT REDUCTION _080609.asc.html

Added after 4 minutes:

...i do beg your pardon.....

...here is the Led current (green) and Input voltage (blue) for the circuit that does NOT use the 10uF cap C2, or the diodes D15 and D16

...You can see the terrible LED current overshoot which will kill the 350mA LEDs.

 

[FvM]

led smps boost overshoot

You should also display the input voltage to the right of L2 to understand, what's the problem in input transisent response.
It's not primarly a compensation issue in my opinion. This kind of input filter is simply unsuitable.

 

[eem2am]

compensation types in smps

hello,

the following schem does a good job of reducing the overshoot from vin_min to vin_max..

And wavefroms for LED current (green)and Input volts (blue)

....overshoot vastly reduced.....any improvements?

i cannot forego the input filter i have used.....it is necessary to stop noise and EMC.

Added after 58 seconds:

beg your parsdon...here is the waveform LED current in green ,, input volts in blue...