Gluing JFETs to a copper ground plane

[johnferrier]

I’m thinking of making a JFET headphone amplifier (with 16 JFETs). This week's plan is to use a thermally conductive epoxy to glue the FETs down to a common ground plane. Then I will use point to point wiring to connect the parts. I’m doing this for three reasons: one, it adds some amount of heatsinking to plastic JFETs; two, it eliminates the PCB material (FR4 or Teflon); and three, P2P allows me to use three dimensions (as JCarr has wrote) to wire the circuitry.

Now, I was concerned about creating small value capacitors to ground through the plastic encapsulation epoxy. However, after doing some internet searches, it seems like any lead to ground capacitance will be on the order of < 1pF. I base this on calculating a capacitance for the largest area of the top of the FET and a dielectric constant of 4. (Diameter 5.1mm, distance of 1mm, and again DC of 4.) This calculation gave me about 0.7 pF (this is quite worse case). Also, I found information on IC packages that describe a high typical inter-electrode capacitance of 0.2 pF.

This means that the capacitive reactance will be about 8M ohms at 100khz. The only part of my circuit that has that high of impedance is the input. If I look at the specs of the JFETs they have an input capacitance of 10-30pF (one to two orders of magnitude higher). And I may isolate the input FETs from my common ground plane.

In a way, what I'm doing would not be much different than a printed circuit board that uses a solid ground plane. With such a PCBA, all the leads and traces have small capacitance to ground (likely higher capacitance).

There probably is a reason that parts aren't glued to a ground plane. Certainly, for RF applications this wouldn't work. Anyone have experience with components glued to a common ground? I plan to use a thermally conductive epoxy for use in electronic applications. I'll use a thin glue line for the best thermal conduction and make sure the components are free of oils that may degrade the bond. I expect that heatsinking would only add to the lifetime of components that will now become difficult to replace. Also, I'll do the matching and perhaps breadboard the circuit before using the epoxy...


JF

[peranders]

You don't have to worry at all about the capacitance towards the groundplane. My current feedback headphone amp has transistors really tight towards the groundplane and the bandwidth is > 10 MHz, according to simulations approx. 25 MHz!

About your calculations: You can't include the plastic package as "area". Which area has the die inside the plastic? 0.7 pF is way too much.

I say, go ahead with your ideas and you can use any glue for that matter.

[johnferrier]

Thanks Per-Anders!

So you are using a ground plane on the back side of your PCB?


: )
JF