솔루션 찾기

Large ripple on output voltage and voltage dropping at higher load

Posted by: saiknaram on Wed, 10/27/2021

Hello PI Engineers, always great reading these help topics. I need your help again. 

I am using an LNK304DG-TL chip with a 680 uH output inductor to have 100 mA output load. 

I am seeing 456 mV of ripple on the output. I am measuring with a very short ground loop on my oscilloscope probe. 

I have tried using a 0.1 uF and 0.0022 uF capacitors parallel to the output and have gotten it to reduce to 308 mV. 

Please find the design attached. We want to get it down to 100 mV. Would really appreciate your help. 

Also, when loading the output at 70 mA and over the voltage drops to 10.75V, when we expect 11.70V. 

Thank you,

Sai

Files

첨부 파일 파일 크기
LNK304design-highripple.PNG 231.82 KB
LNK304baseddesign-highripplePCB.PNG 301.57 KB
Loadregulation.PNG 17.25 KB

댓글

Submitted by saiknaram on Tue, 11/02/2021

I tried out a 1mH inductor - RLB0812-102KL and that seemed to drop the ripple to 175 mV from the 456 mV it was showing before. The larger inductance and the lower DCR helped in reducing the ripple. To reduce the ripple further, we might have to go up to RLB9012 series. The tradeoff here is that with larger inductor diameters, the lower the DCR and we would be losing board space at that time. 

 

Submitted by PI-Elex on Tue, 11/02/2021

Hi saiknaram,

Thank you for using PI Product.

Upon checking the datasheet, the ESR of the output capacitor affects the output voltage ripple. In addition, it should not exceed the rated ripple voltage divided by the typical current limit of the chosen LinkSwitch-TN. Can you check the output capacitor (C207) used and try changing it with a capacitor with different ESR? While doing so, try it with and without the added capacitors (0.1 uF and 0.0022 uF) to see the effect.

Also, the feedback (FB) pin should be set to 1.65V. Can you check it? I suggest changing the feedback resistor (R206) to 12.5kΩ and see if it can resolve the voltage drop issue at higher load (above 70 mA). While you're at it, can you also check if the Bypass (BP) pin is maintained at 5.8V?

Lastly, may I know if you used PI Expert in designing this circuit?

Thanks and best regards,

Submitted by saiknaram on Tue, 11/02/2021

C207 has a ripple current rating of 120 mA. It didn't have the dissipation factor and impedance calculation. My FB voltage is set to 1.74V while the max is 1.76. I will try with a different set of resistors. How does changing the feedback voltage decrease the resistance? Wouldn't that relate to the output voltage directly? 

I will check the BP pin. I used PI online Excel sheet.

Thank you,

Sai

Submitted by PI-Elex on Tue, 11/02/2021

Noted on the inductor. Can you check the suggested DCR on the PI online Excel sheet? If it suggested 2 Ohms, see if you can try RLB0812-681KL as an alternative. 

When using the 1mH inductor (RLB0812-102KL), aside from the decreased ripple voltage, is there an effect on the higher load (above 70mA) voltage drop problem ?

Thanks and best regards,

Submitted by saiknaram on Tue, 11/02/2021

Using a 1 mH inductor with a DCR of 3.9 Ohms, better ripple and no dropping of voltage from 0-150 mA load. I am using a 16V 100uF for C207. Is cap derating an issue here?

Submitted by saiknaram on Tue, 11/02/2021

I tried a 100uF 25V capacitor and 12k ohms on the FB resistor. Ripple shot up to 300 mV. I think I will keep the design at 11.8kohms FB resistor, 100 uF 16V and change the inductor to 1mH 0812-102KL. That should bring the ripple to 172-175 mV

Submitted by PI-Elex on Thu, 11/04/2021

Hi Sai,

Fully noted on the changes you made. In the event that you change your PCB design to use the larger inductor, I would recommend optimizing the source trace as suggested on page 5 of the datasheet.

With that, I'd like to know if your circuit is already working properly or if you need further help.

Regards,

Submitted by saiknaram on Tue, 11/09/2021

Hello PI-Elex,

A new problem appeared for us. Steadily increasing the electronic load from 0 to 150 mA maintains relatively constant 11.7 to 12V output. However, when the load is instantaneously increased from OFF position to 45-60 mA, we see 3-4 V of ripple on the 12V. 

I've taken plots of the waveform under load with the output load increasing steadily from 0 to 150 mA. Ripple looks good and the output stays around 12V. There are the abnormal operation plots that show the blue waveform which is 12V which has 3-4V ripple. If we look at the red waveform, this is the inductor current, and the yellow waveform, which is the SW node, they seem to shutoff and turn on again. The duration for the waveforms to be ON is ~10 ms and OFF is ~6 ms. Why does this happen? It cannot be that we are hitting the over temperature limit at 142C and going back down to 75C in that short span of time. 

The real world affect of this is that we are hearing audible noise from the inductor that sounds like "air escaping tires". Please take a look at the design files above for the schematic and layout. 

Just for reference, I am still using an LNK304 and output inductance of 1 mH - RLB0812-102KL.

Your response is greatly appreciated. 

Regards,

Sai

Submitted by saiknaram on Fri, 11/12/2021

I saw another forum post https://www.power.com/community/forum/ac-dc-conversion-forum/2020/auto-restart-under-normal-conditions-top256 and understood that the overload protection might be causing the auto-restart condition. So I reduced the 3.3k output resistor to 2.4k ohms. This removed the large ripple like behavior I saw.

However, when leaving the electronic load on at 40-50 mA and turning on the switch causes the power supply to go into auto-restart mode again. Keeping the electronic load off and turning it on after the power supply is on causes no auto-restart behavior. 

Do you think this is because of the large VF of the fast recovery free wheeling diode? I was wondering if it would make a difference as the topology is different between this Buck and the design in the link above. 

Submitted by saiknaram on Thu, 11/18/2021

Hello PI-Elex or someone else, can you please help?

Submitted by PI-Elex on Fri, 11/19/2021

Hi Sai,

Can you check if the following will be of any help?

  • While D202 is rated at 35ns, can you use diodes with smaller reverse recovery time (Trr)? The operating mode (MDCM or CCM) will affect this.
  • Try using another diode so that the Vf of D203 matches that of D202. See also if D202 with lower Vf helps.

In addition:

  • Can you also measure drain current if it's still below the rated peak current of LNK304?
  • With regards to the inductor audible noise, does this only happen during that high ripple condition (40-60mA load)? Based on the current ripple waveform, a peak current of 250mA is measured. The inductor might be saturating since it's only rated to 120mA. Can you check if the inductor is heating up?
  • What happens to the output ripple when the load is instantaneously dropped from full-load to that 40-60mA load range? 
  • Aside from the inductor and output resistor (R207), what other changes were made?

Regards,

Submitted by saiknaram on Fri, 11/19/2021

We have two problems at hand:

1) Ripple caused when a load of 40 mA is applied to the 12V rail. This causes that large ripple seen. Using a 220 uF cap on the output and not the 2.4 k resistor instead of 3.3k resistor this is suppressed. However, when this load is inductive such as triggering a relay, 12V rail for the power board above sees almost 5V of ripple. However, switching back to the original BOM with 100 uF and 3.3 k with the 1 mH inductor allows minimal ripple. 

2) The other problem in both cases is EMI seen when switching on the relay (press a button to turn on a light). What are your ideas for EMI suppression?

Now, I have inductors with larger current limit in today along with the low VF diodes to try out. These diodes at VF=0.93V and recovery time of 20 ns. The ripple can probably drop with the lower losses but EMI suppression is probably not guaranteed. 

I will have to measure the drain current of the chip. It is confusing why the inductor current is at 250 mA when the load is only at 50 mA+10 mA for the control chip. 

Submitted by PI-Elex on Fri, 11/19/2021

Hi Sai,

Here are additional things that you can checkout:

  • I tried checking with PIExpert and the suggested output inductance is 820uH, assuming 85-265Vac input. Can you share your PIExpert result? Also, the inductor current rating should at least be 30% higher than the max current limit of the device. For LNK304, it's 345mA. Try using one that's about 450mA.
  • I also rechecked the application example in the datasheet (page 4) and noticed that it doesn't have a resistor similar to your R205. Try returning the replaced components to the original ones but without the R205 and see the effect if the initial and current ripple concerns still exists.
  • For the inductor audible noise, checkout conformally coated chokes, such as the Kemet SBC Series.
  • Capacitor values above 100 µF are not recommended as they can prevent the output voltage from reaching regulation during the 50 ms period prior to auto-restart.
  • Since you're testing it at variable load configurations, what would be the usual/expected load requirements?
  • The PCB Layout needs some modification to further optimize the performance.
    • The loop formed between the inductor (L201), freewheeling diode (D202), and output capacitor (C207) should be kept as small as possible.
    • The loop between the input capacitor C203, IC DRAIN pin, SOURCE pin, freewheeling diode (D202) cathode and anode should be as small as possible.
    • The EMI performance of half-wave rectified designs is improved by adding a second diode in the lower return rail. Checkout D4 on datasheet page 5.
    • Source trace shouldn't overlap with FB and BP pins.
    • C204 should be located physically close to the SOURCE and BYPASS pins. Relocate the traces so that no trace is passing through the center of the IC and make sure that enough clearance is provided, especially around the drain pin and the control components.
    • C203 should closer to drain pin.
    • The output trace is too close to the AC input trace. Observe the proper power flow from input to output as suggested in page 4-5 of the datasheet.

Furthermore, I would also suggest checking out a similar forum here, and see the provided application note in it.

Note that the italicized insights above are taken from either the datasheet or the application note.

 

Hope these can be of further help.

Regards,

Submitted by saiknaram on Fri, 11/19/2021

Hello PI-Elex, Thank you for your response.

Our load which is the relay and ATTINY 1607 (on 12V regulator) should never exceed 60 mA. That's the ATTINY, LEDs, buttons, relay and so on. So I don't understand why we need an inductor with such a high current rating. 

R205 was listed to be used on LNK320X series designs but since those chips were not available, I decided to go with LN304. Is it a problem to include it? I will try without the resistor. 

I will use KEMET SBC series. I have one for 1mH 220 mA. 

I removed the 220 uF and now using the 100 uF capacitor. 

I checked the loop with L201, D202, C207 is small. Should that be made smaller?

Loop for C203, Drain are on one side of the chip and D202 is on the other side as it is a free wheeling diode. If I move it close to C203 and Drain, I will be moving D202 away from Source (SW node). Is this necessary?

You mean add a second diode in the neutral line? Why is this needed? Neutral usually should not have high voltage there right?

C204 is very close to Source and Bypass, it is below the chip. Is this what you mean? The Bypass_BP/M trace going through the center of the IC is a problem? I see that it is going next to the Source trace. Is this a problem with coupling? Again, same thing with having testpad on top layer?

C203 is part of the PI filter - is the concern that we should have a closer coupling to Drain?

I realize that the AC input trace is close to output but at this point the voltage is rectified and 0.72 mm away. How much clearance do you suggest?

Thank you,

Sai

Submitted by saiknaram on Fri, 11/19/2021

Also, I used RF201L4SDDTE25 for D202 and SE20PJHM3/84A for D203 and I see a higher ripple than with US1GWF-7 and S1J-E3_5AT, respectively. 2-3V ripple vs 520 mV for the original case. Please help with this too.

Thank you very much for your help. 

Have a great day,
Sai

Submitted by saiknaram on Mon, 11/22/2021

Here is the excel sheet for the design

첨부 파일 파일 크기
LNK304-spreadsheetdesign.pdf 157.34 KB
Submitted by PI-Elex on Wed, 11/24/2021

Hi Sai,

How is the performance of your design now? What is the effect of removing R205 with all the original components in placed?

Regards,

Submitted by saiknaram on Fri, 11/26/2021

There was noise being generated from the armature of the relay closing which affected a button GPIO. This was addressed using a debouncing algorithm in firmware. 

Now, I still hear inductor noise. I was looking in the application note AN-70 and found there was no recommendation for the output inductor. So by changing out LNK3206/LN304 for a LNK3202, I was able to keep the rest of the components intact. However, I found that I can only go as high as 40 mA (relay) and 35 mA (separate load). The inductor is rated for 390 mA. https://www.mouser.com/ProductDetail/652-RLB9012-182KL

The noise on the inductor is due to bunched pulses or pulse skipping right? So by using double or triple heat shrink on the inductor, can the noise be reduced? The conformally coated inductors are quite expensive. 

We also want to use this power supply for up to 360 mA with the LNK3206 but the inductor is noisy when LEDs are ON and pulsing. Also, the current output goes as high as 210 mA but not even close to 300 mA even when the inductor is the same as above. Do you think its a problem with using a lower current rating? 

The application note AN-70 writes that RFB1010-122L is a good inductor to use but only guaranteed for CCM at 225 mA? How can I use up to 360 mA?

Also, what did you think about the design excel sheet that I attached? Is using 2 2.2uF caps good instead of 2 4.7 uF caps?

Thank you again for your help. I am incorporating your layout recommendations for the next board spin. Have a great weekend. 

Regards,

Sai