Solution Finder Get Tech Support

UHF Emissions

Posted by: smps077 on

New to this forum and probably asking a question already asked, but....

I have used the PI tool to design several low voltage (+/-15V) supplies. I have followed the component selection and layout guidelines as close as I can get.

Each of the supplies is very quiet at the switching freq and harmonics. No problems there.

However, all of them emit at various peaks between 50MHz and 300MHz. These emissions are non-trivial and prevent these supplies from passing certification.

I assume that I am not the first to encounter this, so just wondering of the PI team or fellow users can help me track this down to solve it.

Thanks.

 

Comments

Submitted by PI-NANO on 04/15/2016

Hi,

What is the device that you are using in your design?

Could you please send us the schematic, PCB layout, PIxls design, Transformer consturction document etc?

Regards,

PI-NANO

Submitted by smps077 on 04/15/2016

Thanks. Where should I send the info?

Submitted by smps077 on 04/19/2016

Bump. Please tell me where to send the information you requested.

Submitted by PI-NANO on 04/20/2016

Hi, 

You can attach the files when you are replying in this post. If you scroll down you will see a section called Files. There you can add the new files.

Most of the peole share their designs over the forum, however if you are not comfortable sharing the files over forum, Please post your email Id here so that I can get in contact with you.

Regards,

PI-NANO

Submitted by smps077 on 04/21/2016

Thanks. First, here's the uds file from the PI Expert tool. This should be most of what you've asked for. The transformer was made by rtss fast proto service from this file.

Attachment Size
smps077 uds file.uds 335 KB
Submitted by PI-NANO on 04/21/2016

In reply to by smps077

Hi,

PIExpert gives a first pass design, however you need to optimize the power supply for efficiency, EMI, PIV's etc. I saw that snubber on the secondary rectifier diodes are very small. Please take a look at secondary rectifier waveforms and increse the amount of snubbing if required. Generally high frequency noise generated due to sharp dv/dt and di/dt. 

Also try changing the diodeds D3 and D4 to Schottky diodes and see if it makes any difference. You can also try using slow diode (for example 1n4007) in place of D1. 

If you still need help, please let me where you are located so that I can give you our sales location details. You can get in contact with the local FAE.

Regards,

PI-NANO

Submitted by smps077 on 04/21/2016

Thanks. Snubbers are what the tool created. I can look at the waveforms and they will tell me something, but won't necessarily tell me if the emissions pass cert only if they relatively go higher or lower. Can you recommend what changes to the snubbers? And at what point is it too much snubbing? I'll look for some Schottkys also.

I don't need to optimize the efficiency any further bc it's efficient enough for my purposes. :)

Also, would it be better to connect RTN to the Source of the primary Mosfet instead of to the top of the clamp?

I may need help, but will try this first. 

Submitted by PI-NANO on 05/11/2016

Hi,

As I said PIExpert gives a first pass design, however you need to optimize the power supply to achieve certain performance. Look at the secondary rectifier diode waveforms and if there is excessive ringing on the wavefrom increase the snubber capacitor value gradually and reduce the snubber resistor value also to get optimum results. Snubber capacitor value do not increse more than 2.2nF.

If you are talking about C6 connection, yes you can try that option as well.

Regards,

PI-NANO

Submitted by smps077 on 05/30/2016

A follow up for PI-NANO and anyone else who might have interest in this particular problem.

None of the recommended steps made any difference.

1. Increasing snubbing on the rectifiers, zero detectable changes.

2. Using a slower D1 made things worse.

3. Schottkys made no change or maybe slightly worse.

After doing some investigating I snubbed the DS of the chip with 330pf. Bingo, almost 10db less emissions on every single peak at the UHF where it was radiating. Higher capacitance didn't improve things and lower capacitance yielded higher emissions as I reduced the value.

Seems to me that this is something you guys probably know about, Right? I wonder why the PI Expert tool doesn't offer this as an optional modification? Or does it and I just missed it?

Submitted by smps077 on 06/01/2016

Again, for anyone interest, here's one more follow up.

The PI Expert tool calls the common connection (ground) on the secondary side RTN (Return). This RTN line receives junk from the primary side through the Y cap. Since the RTN is effectively the ground of the product on the secondary side, this introduces crud into the ground planes of everything.

I placed an inductor between the RTN and system ground and this reduced the noise coming into the system ground from the smps and, hence, the radiated emissions from all the pours in the PS. The drop in emission level here was significant, larger than the effect of the DS snubber cap.

I'm assuming this is also a trick known to smps designers who are better than I am. I should have realized this early on, but it just took a lot of experimentation to get here. :)

Submitted by PI-NANO on 06/02/2016

Hi,

If the radiation is coming from the primary drain switch and RCDZ clamp circuit, try D1 to a slow diode and you can also try to increase the R2. Using a small capacitor like 10pF to 22pF is OK on the primary drain. However as you increase the capacitor value the switching losses in the device will go up and you may run into thermal issue.

The function of the Y-cap is to bypass some of the noise currents from entering the earth ground.

Regarding placement of inductor You mean to say that you have placed inductor between primary gnd and secondary gnd or between secondary gnd and system ground, which is earth connection?

 Regards,

PI-NANO

 

Submitted by smps077 on 06/03/2016

Thanks.

As noted above I tried a slower D1. No change.

To get the emissions down to a reasonable level I had to use 330p on the DS. I ran the unit for several days straight with load. Using my hand infrared thermometer there was no detectable rise in temp from normal. I moved around the chip as best I could to look for hot spots. In the absence of an infrared camera, I would say that keeping the mosfet in its linear region for a little longer is not causing much extra power dissipation.

The Y cap is bleeding primary side noise to the RTN (secondary ground). When the RTN is connected directly to my system ground the entire ground plane is flooded with UHF and radiates off the board and out the the power cord ground, the connector grounds, and box itself. It is an array of peaks from 50MHz to 350MHz, many of which exceed CE maximums for their respective bands. Don't you guys know about this?

If you look at the schematic supplied by your own PI Expert tool you will see the RTN line. I have put an inductor between this RTN line and the actual system ground. The safety ground from the power inlet is connected directly to the system ground and box and the entire PS floats on the other side of the inductor.

This was the only way, among the things I tried, to stop the emissions. :)

Thanks for the help.