TEF1002-02 EMC

[SteveC]

Hi,
Sundance is developing a new FMC 1G SPS ADC part FM582 using the TEF1002-02 carrier.
I have observed a significant amount of RF interence in the range 250 to 450MHz, which I have traced to the TEF1002.
The TEF1002 uses an Analog Devices LTM4638 monolithic DCDC to generate 5V from the external 12V supply.
This device is clearly the source of the noise, because a 1cm diameter wire probe placed near the LTM4638 shows direct correlation of the magnetic field.
I have attached a scope plot and a photo. The plot shows the following:
Ch1 = Current probe, type Singer 94111-1, around external 12V supply to TEF1002.
Ch2 = 1 turn wire diameter 1cm located 1cm from the TEF1002 main 5V DCDC
Ch4 = P6247 differential 1GHz probe across the input of IC19 on FM582
The photo shows the TEF1002, current probe, P6247 differential probe attached to FM582.

My question is about the LTM4638 connections:
The LTM4638 data sheet has a section titled “SW Pins and snubbering circuit” on page 16.
This describes how to stop EMC caused by switching in this device by adding a snubber.
Is there a snubber already fitted ?
If not please could you describe to me the location of the "SW" pin net, and the 0V net, so that I can add a snubber circuit.

Thank you
Best reagrds
Steve Carpenter

[Martin R.]

Hi Steve,
up to now we have not heard about a problem like this. Online available schematic page 9 shows the circuit used:
https://shop.trenz-electronic.de/trenzdownloads/Trenz_Electronic/Modules_and_Module_Carriers/4x5/4x5_Carriers/TEF1002/REV02/Documents/SCH-TEF1002-02-default.PDF
SW pin B7 is floating and due to the BGA package not easily accessible. Ball B7 is at the edge of the component (see attachment) therefore it may be possible to heat up and stick a tiny wire to the ball but all at own risk.
Other things to try which are influencing EMI is using other working Mode, but we saw that continuos mode is not working well for 12V to 5V (part gets easily hot).
Further options would be (to simulate the circuit and) try other compensation (there are options, see schematic and compare assembly diagram in download area) or switching frequency.
If you need the LTSpice schematic we used please write an request to support@trenz-electronic.de

I do not have a Revision 02 board available, so i can not check further in hardware. We noted that for a new revision (which will be needed in near future due to EOL status of Enpirion parts) of this board, having SW pin accessible at least via a test point would be good.

PS.: If more details needed, layout (Altium project) for our carriers is also online in download area.

Best regards
Martin

[SteveC]

Hi Martin,

Thanks for your reply.
I have found a good work around, which is to attach an external 5.20V supply across C56 and C57.
This reduces the EMC to zero because the LTM4638 does not supply any current.

I think it would be very useful to bring out the SW (B7 ball) pin to a series resistor and capacitor to 0V, so that a snubber can easily be added.
It would also be useful to bring out the RUN (B4 ball) pin to a pull up resistor, with optional pull down resistor so the LTM4638 can be disabled.

Thanks
Steve Carpenter

[Martin R.]

Good to hear that you found a solution for you.
We will take your recommendation for revision 03 into account.
Thank you!
Martin