Help with PC board design

[trident9]

Greetings

This is my first attempt at designing a PC board, your help in correcting errors will be much appreciated. The input signal is 25kHz PWM from an open drain output. Maximum power dissipation in the pass transistor will be 2 watts for 1 second or less at start-up, continuous dissipation will be 1 watt or less. The gate resistor is purposely large to limit the output slew. The board will be 1 oz copper on both sides.
1) Will the heat sink be adequate for 30 degrees C ambient in still air?
2) Will RFI from input transistor switching be an issue?

Thanks in advance for you help.
_______
trident

[Boncuk]

Hi Trident9,

first off, I guess the PCB can neatly be done single sided with Q3 mirrored onto the solder side.

If you can afford to have more space try to enlarge the heatsink plane.

Further I see "FG" on two connectors, bot not elsewhere in the schematic. What is it good for?

Boncuk

[MrDEB]

nice layout but no need for 2x board
just 1 or 2 jumpers

[Boncuk]

Q3 is located on top of the cooling pad.

Zoom it.

[Boncuk]

Originally Posted by MrDEB

nice layout but no need for 2x board
just 1 or 2 jumpers

How do the traces reach the SMD transistor (Q3) on a single sided board?

[MrDEB]

first off, I guess the PCB can neatly be done single sided with Q3 mirrored onto the solder side.

If you can afford to have more space try to enlarge the heatsink plane.

I missed Q3 but mirriring is solution

[trident9]

Can someone point me to info for designing thermal pads as to the area required? How many and what size via are appropriate?

Any opinions as to how much radiated noise Q1 and Q2 may generate, would a ground plane under U1 be an effective shield?

Thermal pad for Q3 is double sided to minimize the board size, currently there is a little more than 0.65 sq inch on both sides of the board. FG is a signal from the load that needs to go back to the controller.
_______
trident

[mneary]

In still air, or enclosed? Will the surrounding air stay at 30c or less after some time?

[Boncuk]

Hi trident9,

I have reworked your circuit with a large cooling surface for Q3 including 14 thermal drills.

Board size is 57.4675X39.6875mm (2.2750X1.5625").

It's fully single sided and uses two fills, ground fill on the left hand side and drain Q3 fill on the right hand side.

I'd be interested to know how much drain current you use for Q3.

With the given dimensions of the cooling pad the temperature will be about 45deg/C at max current.

Boncuk

[mvs sarma]

Originally Posted by Boncuk

Hi trident9,

I have reworked your circuit with a large cooling surface for Q3 including 14 thermal drills.

Board size is 57.4675X39.6875mm (2.2750X1.5625").

It's fully single sided and uses two fills, ground fill on the left hand side and drain Q3 fill on the right hand side.

I'd be interested to know how much drain current you use for Q3.

With the given dimensions of the cooling pad the temperature will be about 45deg/C at max current.

Boncuk

Nice and cute board Boncuk!! Which artwork do you use?

[Boncuk]

Hi Sarma,

there is no artwork program used. It's simply Eagle schematic and board design software.

I used the OP's original PCB layout and did a few changes to avoid wire jumps. (also moved the SMD transistor to the solder side.)

Regards

Hans

[MrDEB]

NICE rework BONCUK
did you ever figure out what "FG" was??

[Boncuk]

Thank you, MrDEB,

I didn't find out what it is good for (missing connections), but here is what the OP planned it to use for:

FG is a signal from the load that needs to go back to the controller.

Boncuk

[trident9]

Greetings mneary
The board will be vertical in still free air @ 30 degrees ambient, the ambient temp will be constant. From the Fairchild data-sheet: thermal resistance from junction to case is 12 degrees C per watt. Max junction temp is 150 degrees C. Referencing the transient response curve, transient resistance would be 0.2 for a single 1 second pulse. If I am understanding the math correctly a thermal pad that can dissipate 1 watt continuous would be able to dissipate a single 5 watt 1 second pulse. I would like to error on the safe side using 125 degrees C as the max. junction temp. and 35 degrees C as the ambient. What I do not have is a value that relates area to thermal resistance for a double sided 1 oz copper thermal pad to plug into the equation.

Greetings Boncuk
I much appreciate you posting the alternate layout. I am approaching this project as a learning opportunity. If someone does all the work for me that opportunity will be lost. I am using EAGLE 5.6.0 with custom package outlines, If you like, I can post the files. I will not be etching the board myself, so going double side is not an issue. Max. current would be at 1/2 Vss, ~330mA @ 6 volts.

How did you calculate the thermal rise for Q3? also your layout does not address several items I believe are important: 1) A low impedance ground path from J1 to J2 for load power, 2)The gate resistor for Q3 should be as close to the transistor as possible, 3) no trace for the FG signal.
_______
trident

[Boncuk]

Originally Posted by trident9

How did you calculate the thermal rise for Q3?

I very seldom calculate. Just used the data sheet and the suggestions for the thermal pad according to size and copper weight and estimated the resulting temperature. According to my experience my estimates are close to calculated ones, saving headaches.

also your layout does not address several items I believe are important: 1) A low impedance ground path from J1 to J2 for load power, 2)The gate resistor for Q3 should be as close to the transistor as possible, 3) no trace for the FG signal.

1) A low ground impedance is given by the ground fill. 300mA is just laughably low.

2) I moved the gate resistor towards the gate, but I think this has absolutely no effect on the function. If you are afraid that noise will be the problem it could be inserted in the trace between the OpAmp and the resistor.

3) If you had labelled the traces instead using flags I'd known they must be connected in the schematic and on the PCB. They are connected now using a wire jump.

Boncuk