ISOW7821: ISOW7821 Ripple and radiation problems

The supply pins require very good decoupling (see section 12), and especially the input supply requires large storage (see section 11).

Please show the schematic and the board layout.

I'd guess that you will need to add more capacitors/inductors.

Hi Varun,

We have already placed a ferrite bead in my circuit. The above picture is the result with ferrite bead and decoupling cap (100uF, 10uF, 100nF) 

The ferrite we use is Wurth Elektronik PN: 742 792 097.

Layer 1  

Layer 2 

Layer 3 

Layer 4 

Hi Vincent,

Thanks for the details on Ferrite Bead and Decaps. The Ripple Waveform is on VISO or VDD?

Can you please share the part number of the 10uF and 100uF decaps used.

Also please share the layout pictures for these decap and ferrite bead placement. The Layout pictures you tried sharing above are not visible.

Hi Varun,

The ripple and the EMI are from VDD.

Please see if below picture are okay.

L1  L2  L3  L4 

100uF:

10uF: 

Here is the cap used.

100uF: Murata, GRM31CR61A107M

10uF: TDK, C3216X5R1E106KT

Hi Vincent,

Thanks for the Inputs. Please Find my inputs below:

1. After looking into all the details, It doesn't look like a Decap or Ferrite Bead issue, rather it has to do with the loading of ISOW7821 based on the switching waveform shared.  
2. Looks like that the ISOW7821 is getting overloaded and not able to be in regulation, hence the sudden DC Voltage Drop and at the same time we have voltage spikes which means device is out of regulation:
   1. 
3. Please check the Transient Current requirement of the load that VISO is connected to. This should not exceed the Max Load Current in the respective mode

Hi Varun,

Sorry for the late response.

My test set up was shown below.

As I was only using a resistor as loading at the isolated side, I don`t think the IC is overloaded.

Input current from the power supply is 33mA and isolated current at the load is 10mA.

I have not connected any other I/O pins (all floating). And I have observed an even larger noise ripple at the power supply side VCC periodically.

Would you provide us some idea on this? 

Thanks,

Vincent

Hi Vincent,

Thanks for the details. So you don't have any switching load and this waveform is at VCC.

Couple of questions and inputs:

1. Scope shots: The Second image shows 71mV of Vp-p ripple, where as previous shared image has 146mV. Did Something change ?

2. Input Supply: What is the current limit on the input side supply used? If it is too low then and not able to deliver the current demand when power converter switches ON, then also you we can see these high ripples.

3. PCB Layout: Most Important, I reviewed the device layout in more detail, and found one fundamental issue. The GNDs of the Device on either side are not connected strongly. The ISOW78XX requires strong ground connection for most optimistic performance and decaps to be effective.

Hi Varun,

Thanks for your quick response.

For the input supply, I was using a DC power supply which output current limit was set to 500mA. From the above picture the loading is only 33mA at input supply. Therefore I think it may not be an issue.

I have also alert the Grounding on the layout and I have made an other small PCB to verify this. Therefore you may see the different in the voltage ripple. The 71mV Vp-p is from my new PCB which has a much stronger Ground connection.

There is the layout for my new PCB,  the test set up was shown above (10mA resistive load)

Top

L2 

L3 and bottom (Only GND plane)

Hi Vincent,

Thanks for the Update on the Boards. Looks good to me and great to see the ripple numbers improving by 50%.

So the Ground is taken care. Good!

Now coming to decaps:

1.  The 100uF Cap used in your board does not have high frequency filtering capability. SRF<1MHz. 

Vincent Wong said:

100uF: Murata, GRM31CR61A107M

2. The 10uF TDK Cap is discontinued to can't comment on it's SRF but make sure the SRF lies right of the noise frequency of failure,

What is the Noise Frequency in Spectrum which shows failure? Did you check for emissions with new board, were they under limits now?

Hi Varun,

For the VCC, yes the overall performance is improved but there is still spikes appears periodically. Not sure if those will cause the EMI problem.

For your reference, the waveform on the new PCB are shown below. (Green - primary side VCC, Yellow, - Secondary side, Viso)

For the decoupling capacitors.

1) For larger cap values, it is not easy to find a high SRF one. Would you suggest the reference component for me? Maybe from TI's EVK?

2) For the 10uF TDK cap, the SRF is similar. Below graph is what I found from their WEB.

From the spectrum, the fundamental noise frequency is around 60MHz-70MHz.

Please see if you had any comment on this.

Thanks,

Vincent

Hi Vincent,

Thank you for sharing all the useful information so far.

Regarding the capacitor recommendations, I don't have anything specific to recommend. But I now understand that the output ripple issue has been addressed but there are radiated emissions that are being observed in the spectrum analyzer.

Since ISOW7821 integrates a high frequency DC/DC converter, the radiations are going to be picked-up easily when monitored using a probe and spectrum analyzer in lab. This is not going to be the same as the radiated emissions results obtained from a certified CISPR 32 anechoic chamber. We recommend that customer test the product in a certified EMC chamber to observe actual test results.

To further address radiated emissions results, customer may have follow the CMC suggestions as described in the E2E post below.

https://e2e.ti.com/support/isolation-group/isolation/f/isolation-forum/806938/isow7841evm-isow7841evm

I understand that customer was also using CMC or FB already in their system, can you please share complete schematic (along with component values/part numbers), PCB layout and emissions test setup so that I can review and make suggestions to improve results?

Regards,
Koteshwar Rao

Hi Koteshwar,

We will be soon sending our samples to the certification lab. As we don`t have chamber so the near field coupling method is just for our reference.

Here i attached the schematic for your reference. The component choose are not exactly the same on my test board.

1) The EMI filter was replaced by 0R resistor

2) The CMC we are using is Wurth 744232102. I saw your pervious post and I will try to get those CMC suggested to test again.

3) Although the suggested CMC may help, it seems that the EMC issue is not completely eliminated,  is there any alternative isolated DCDC that would not have such a high frequency switching from TI ? 

Thanks,

Vincent