Why high ripple present in my first SMPS project

[jagspaul]

I have made one 12 volt 2 amp smps using uc3844 pwm controller with MOSFET switching frequency 100 Khz. When I measure the output with digital and analog multimeter the output is stable for both meters. But when I connected the output to oscilloscope (2volt/dev and 2us/dev) high ripple voltage is showing on the CRT screen.
Please find the CRT image file attach here with.

Somebody please come forward to solve this problem and explain why this ripple has come. (I have used TL431 in control loop.)

regards
jagspaul

 

[FvM]

Without real information about the circuit and the PCB design, let me just say, it looks quite common. SMPS are generating strong interferences, it's not easy to handle them. In most cases an effective filtering is required at the in- and outputs of the switcher. You also have to care, that the high frequent communtating currents are kept inside the power stage and don't spread all over the PCB

 

[jagspaul]

Thanks for reply.
The Max ripple amplitude is 6volt p-p 100khz. Is it acceptable ?
Tomorrow I will send the schematic and PCB layout file.

 

[eem2am]

is that off-load?

it looks like it.......it looks like it is bursting on and off....as if it were on no load or light load.

ripple i s at switch freq.....so it is probably mostly due to your ESR of your output caps.

 

[dick_freebird]

That's not "ripple", that's ringing.

Potential causes include excess switch node capacitance
and inductor self-capacitance, poor input supply
decoupling (quantity or Q), and possibly even things
like ground bounce induced supply reswitching (a
direct coupled ramp=comparator to output path can
easily be perturbed by 100mV of ground noise).

The sources may be distinguishable by looking at the
frequency of those bursts; parasitic LC effects are
liable to be 100MHz+ while through-chip paths will
likely be 10MHz or below (maybe faster in low voltage
technologies, but classic 40V PWMs are pretty slow).

You should look at the switch node for signs of
reswitching / chatter, more likely on the H-L edge.
Look at the supply pin for gross droop or ringing.
Look at ground plane vs some exposed internal
signal if you can find one, see if there's an evident
bounce inside the bond wires or core:board.

I've had to redesign eval boards for ground bounce,
tight tight tight ground routing is the key.

 

[jagspaul]

is that off-load?

it looks like it.......it looks like it is bursting on and off....as if it were on no load or light load.

ripple i s at switch freq.....so it is probably mostly due to your ESR of your output caps.

Yes this is off load, I checked it with full load also and found more or less same picture with high p-p value.

Added after 27 minutes:

That's not "ripple", that's ringing.

Potential causes include excess switch node capacitance
and inductor self-capacitance, poor input supply
decoupling (quantity or Q), and possibly even things
like ground bounce induced supply reswitching (a
direct coupled ramp=comparator to output path can
easily be perturbed by 100mV of ground noise).

The sources may be distinguishable by looking at the
frequency of those bursts; parasitic LC effects are
liable to be 100MHz+ while through-chip paths will
likely be 10MHz or below (maybe faster in low voltage
technologies, but classic 40V PWMs are pretty slow).

You should look at the switch node for signs of
reswitching / chatter, more likely on the H-L edge.
Look at the supply pin for gross droop or ringing.
Look at ground plane vs some exposed internal
signal if you can find one, see if there's an evident
bounce inside the bond wires or core:board.

I've had to redesign eval boards for ground bounce,
tight tight tight ground routing is the key.

I have no differential probe, only have a single ended 1X/10X probe. So I can not connect it with high voltage section (non Isolated). I found same result any where in the secondary (isolated) the probe is connected with ground reference. I think you can guess more when you would get the schematic. Please let me send it.

If possible please send a schematic with your suggestion as a reference design, where this type of symptom has been removed properly.

 

[eem2am]

also,

is it flyback?

maybe you get bad ringing from output diode.

maybe you need to snub it.


did you use RCD snubber across primary?

It is weird, cuzz your output cap should not allow its voltage to ring like that.....unless you have very very bad layout

 

[jagspaul]

also,

is it flyback?

maybe you get bad ringing from output diode.

maybe you need to snub it.


did you use RCD snubber across primary?

It is weird, cuzz your output cap should not allow its voltage to ring like that.....unless you have very very bad layout

Please download the schematic attach here with and post your suggestion.

 

[jagspaul]

I have planed, first develop different types of prototype SMPS to gather the skill on this field as I am new in this field. I know lot of integrated MOSFET SMPS controller is available in the market but I am interested only with discrete MOSFET and SMPS controller chip and soft ferrite core by COSMO FERRITES LIMITED. COSMO FERRITES is Indian company and in kolkata market it is readily available.

I have gone through lot of application notes, finally I have made the circuit with reference TI application note.

I am sending data sheet of the COSMO FERRITES. Please have a look on it.
Details about the ferrite transformer which I have used in my project as given below.
Core Part Number = EE2507 (of COSMO FERRITES)
Material Type = CF129

I have test with two different Transformers having different trans ratio and wire size.
But regarding ringing found same result with both transformer. I have changed the airgap of the core but found no effect on ringing.
Transformer #1
Pri – 26 + 26 Trans, 28SWG 1 stand
Bias(18V) – 13 Trans, 34 SWG 1 stand
Sec(12V 2A) – 8 Trans, 21 SWG 1 stand

Transformer #2
Pri – 32 + 32 Trans, 31SWG 1 stand
Bias(18V) – 18 Trans, 34 SWG 1 stand
Sec(12V 2A) – 12 Trans, 28 SWG 2 stand

Air gap is 0.5mm in all leg, so effective Air gap is 0.5 x 2 = 1 mm

Please note that center leg gaped core is not readily available in kolkata market, so I have to design only with all leg gap core.

Regards
jagadish

 

[svhb]

You must be very carefull by measuring anything on an smps design. Start with keeping the ground lead of the oscillocope probe as close as possible to the probe (minimising the loop). Most of the times you can see a difference. Also try to measure with the bare probe tip and the ground lead wrapped as a spiral around the probe.

With some probes for oscilloscopes, there are some additional pats shipped (screwdriver, colored rings, ... and a piece you can use to fit over the probe tip to connect the probe to a BNC connector (it connects the ground of the probe to the ground of the BNC)) Do never throw the BNC thing away : it you solder a little pin to it, you can use the pin and the tip to measure with a very small ground loop.

I had somewhere a picture of this, but can not find it now. You can look on page 5 and following of document : https://focus.ti.com/lit/an/slpa005/slpa005.pdf to have an idea of what I'm talking about.

Stefaan

 

[jagspaul]

Hi everybody,

I think the issue may be due to bad PCB layout. I request you once again to please send me a PCB layout which quite matches with my design (if required little bit changes, I will do this).

One question, the ringing tendency is less in case of flyback continuous mode rather than discontinuous mode. Please tell me whether it is correct or not.

If it is correct then if I could change the design to continuous mode the ringing problem will be solved.

Please let me know. I am waiting for your answer.

Regards
Jagadish

 

[amrith]

hi jagadish can you plz send me the transformer specifications of your schematic,
i'm on R&D of SMPS using different MOSFETS and pwm controllers.

 

[kanishuw]

One question, the ringing tendency is less in case of flyback continuous mode rather than discontinuous mode. Please tell me whether it is correct or not.

If it is correct then if I could change the design to continuous mode the ringing problem will be solved.

Please let me know. I am waiting for your answer.

Regards
Jagadish

This is exactly correct. Indeed it is the reason why you get a ringing effect in your output. When you have no load connected at the output the mean output current (Io) is very low. As you may already know always there is a ripple in the output current. If the mean current is too low, due to this ripple there comes a situation where the current crosses the zero line which we identify as the DISCONTINUOUS mode. In this situation you get a high impedance at the cathode of the Schottky diode, which causes oscillation in the LC parts. This oscillation then causes the ringing effect.

To get rid of this issue, the first thing you should have tried is connecting a high load (low resistance) at the output which will increase the Io hence the mean value of Io. This will make the system operate in continuous mode.

Hope this helps. Regards !

Kanihska

 

[BADFLOPPY]

In secondary filter capacitor, Only Electrrolytic Capacitor is used. Please use with parallel an non polar capacitor with each and check Moreover you can check with 1n5822 diode without snubber and another 1n5822 as free wheeling at output choke. As I am also from Kolkata, possibly, I can help you.

Thanks