TRF7960EVM PCB Gerber Files

Eddie,

Thanks very much for providing those files it will help immensely for me moving forward with this. I have read most of that app note and will use the information learned and this coil to implement my design. Thanks again.

Brennan

Eddie,

I am doing some layout and was referencing those gerber files that you posted in this thread. One thing that I started to wonder about was the trace width used in the RF section of the EVM and how its characteristic impedance was controlled or calculated. In the gerber files it specifies a 2 layer board using 2 oz copper with total copper-to-copper thickness of 62 mils. If I assume 2oz copper results in traces that are 2.8 mils thick, and the traces in the RF section seem to be 25 mils with a 12 mil pullback of ground pour, and assume that the transmission line is a colplanar waveguide, then I get a resulting characteristic impedance closer to ~60 ohms. In some places the ground pour pullback is up to 20 mils which results in an impedance of ~75 ohms, and in places where there is no ground plane under the transmission line and no ground pour near it (C22 and C21), it would seem the impedance is undefined. As you may recall I am only using this IC and coil to supply power and not do any actual communication, but I am wonder how one should go about constructing an appropriate transmission line given the erratic nature of the tx line in the EVM. If I just go about it ignoring the layout of the EVM and try to make it 50 ohms throughout, would the matching then be different and result in lowered performance if I used the components specified in the EVM BOM? Thanks for the help.

Brennan

Brennan -

This is magnetic (H or B) field system, not electric field system. You are looking to make a coil which you can then resonante with passive elements here and the characteristic complex impedance of the coil you are making should be in the neighborhood of 1+j85 (1uH) to 1+j170(2uH). in this case it is not good enough just to get on the 50 Ohm circle (real), you must also take care to also hit 0 ohms complex.  

You then impedance match that back to 50 Ohms, taking care to make the BW appropriate for what you are doing (maybe almost nothing there in your case, so high Q is OK)

 The problem with going high Q is that the ones of pF (and even fractions of one pF) really matter in the tuning solution so then check each one as you build them.

See below for two tuning solutions using same 2uH inductance value tuned to 50 Ohms, but with two different BW resistor settings.

Josh,

Thanks for the insightful answer, and as you know I am only going to be using the system for power transfer and not data communication, so I am modeling my final project very closely on the EVM board. I guess the bottom line question would then be, assuming I am using a coil and circuit identical to the one in the EVM, should I just use traces of the same widths during my layout? Also, assuming I am using the same coil and circuit as the EVM, how much of an effect would using soldermask over the coil have on its impedance and performance? I only ask because the EVM has the coil exposed while in my application it would be better if I could put soldermask over the coil.

Brennan

correct -

and i would add that you can use same coil or make it two turns on top, two on bottom (for better winding efficiency), then if you have the test tools to measure the resulting inductor at 13.56MHz and the test tools to measure the resonant circuit, then you can experiment with different passive values (BW control resistor) and tuning caps to achieve best energy field out for your application. Remember that the higher Q tuned ckts are easier to detune, too. So you may find some happy medium between field strength and ease of repeatability in a production setting.  

Great, at first I will likely start from something very close to the EVM and then if possible try something like what you describe, as I do have access to a network analyzer that I can use at 13.56 MHz. Just to clarify, would having soldermask on top of the coil have a noticeable effect on performance at all?

no - at this frequency - the skin depth is not very deep, so even most of the trace that we have there is not actually used anyway.