hand solder vs reflow oven

[ge]

Has anyone seen a performance difference between soldering parts by hand vs epoxy & reflow oven?

I have a board with 2 good SMA connectors and 5 relays. When run thru the reflow oven the insertion loss increases by 2.5dB at 3GHz.
The loss does not appear to be associate with the connectors, relays or thickness of the solder or paste. I've ruled out the connectors and relays by replacing them. I've ruled out solder thickness by adding/wicking off material. The results are repeatable.

Odd epoxy?

thanks,
GE

 

[KlausST]

Hi,

Humidity in epoxy may change e-r.

There willbe less humidity after reflow process..
Other chemical mutations may also affect e-r. This is especially true with two components like epoxy. E-r will change during chemical process and it will have big thermal drift.

Klaus

 

[FvM]

There willbe less humidity after reflow process.

And restores to the usual equilibrium value within a few days maximum. Is it a regular two-layer FR4 board? In case of a (bad quality) multilayer board, delamination may take place.

2.5 dB sounds severe though. You should be able to locate the change with a network analyzer. Are yoou sure that it isn't a relays problem?

 

[ge]

Top and bottom layer of board is 20mil Ro4350. RF is grounded coplanar with an 8mil gap.

Yes, 2.5dB is very severe. I can isolate it to this board. There is only SMA connectors and Axicom HF3 relays. I replaced the relays in a complete path, no change. I tested the old relays on a protoboard, no problem.
I sent the reflow oven temp profile to Axicom. Axicom have no problem with it. We are 40 degrees under suggested temperature profile with similar duration.

The board house is sectioning one of my boards Friday / Monday.

 

[FvM]

Do you say it's a hybrid (Rogers + FR4) stackup? Then delamination is more likely to happen, although series 4000 can be bonded with FR4.

Is the RF part of the stackup "all-Rogers" (RF groundplane is bottom side of the Rogers substrate)?

 

[biff44]

the relays are a big thermal mass. Maybe they left it in the reflow oven too long so that the pads got soldered well, but the insides of the relays also reflowed and got screwed up?

 

[vfone]

"ge" mentioned in both above posts that the relays were replaced and got no change.

Right now I am working to fix a project having similar temperature problems. A small dimensions multilayer PCB board is going very hot (~92 deg Celsius) due to high DC current of some components (and few on-board DC-DC convertors).
Initially I believed that the increase of the system insertion loss was caused by the degradation of performances of the on-board SAW filters, but later I found that the PCB was the cause, specifically delamination.

 

[ge]

Thanks all. I still have not found the problem. The board is at the board house being dissected.

All RF is on layer 1. Could vias break due to mishandling or reflow oven?

I mispoke the stackup:
Layer1/layer2 RO4350B 1Oz/1Oz
1xRO4450F prepreg
0.006" RO4350 filler 0Oz/0Oz
1xRO4450F prepreg
Layer3/layer4 RO4350B 1Oz/1Oz

 

[biff44]

vias can definitely break. Think of how they are made, there is a sharp ridge at top and bottom, and the metal halfway down the hole might be thinner. If there is a microcrack, it might open up only at some temperatures, and be fine at other temperatures.