Power MOSFET Chasis Wiring

[acoustixman]

Hi, all!

Chassis Wiring Question:

I am working on installing a pair of my present circuit design into an existing chassis (a toasted SAE). The existing location of the heatsinks and most convenient [by wide margin] position of my existing PCBs in the enclosure requires several inches of wiring be led from the PCB to the MOSFETs for each lead (gate,drain,source). The circuit is a complementary source follower (single FET pair in use in this instance).

What I am wondering is how these wires are best managed without causing oscillation. My first guess is a twisted triplet, since currents are balanced and extreme ultraudio-frequency feedback (over 120kHz) is taken from the previous [VA buffer] stage which should compensate the phase margin for lousy parasitics on this wire (I think?). The logic of this feedback (as explained by W.M.Leach . Obviously wire length should be preferentially kept to minimum, but the whole idea is to prevent costly mechanical rework and I am wondering if there are any folks who have run into this and didn't have oscillation problems. The only problem I can foresee is of producing an inductively coupled positive feedback loop at high frequency between the drain lead and gate lead, but the foul influence from the drain lead should be carried in opposite kind [and thus cancelled] by the source lead, right?!

Any thoughts from the expert pool? Thanks in advance!

Happy and safe listening to all!

[Leolabs]

What's the length of the wires???

[richie00boy]

Twist the drain and source wires, but keep the gate wire separate. I think you'll also need a gate resistor (try 100 ohms for starters) directly on the gate pin.

[acoustixman]

Originally posted by Leolabs
What's the length of the wires???

Originally posted by richie00boy
Twist the drain and source wires, but keep the gate wire separate. I think you'll also need a gate resistor (try 100 ohms for starters) directly on the gate pin.

In my existing work, I used about 4 inches of wire each (FETs are TO-3 on nearby heatsink), and twisted as you would suggest, Richie, but use gate stoppers at 470 ohms before the wire (on PCB). No problems.

In the retrofit I am using more like 8 inches of wire each in twisted triplet (will change back to original plan of twist S-D and not G, I think) and same 470 ohms on board. Seems to ring but requires excitation. Maybe I should relocate the resistor to the gate pins as you also suggest... I suppose that should make stray energy in the gate wire find a preferential escape back to the control circuit.

I wouldn't think 8 inches is a killer; I'm sure I've seen longer. I do aim for shortness of wiring, but if this can work that would save a lot of rework on an otherwise pretty good chassis...

[jgedde]

I design motor controllers using totem pole MOSFET stages; very similar to audio amplifiers.

The problem is two-fold:
The gate is very high impedance and has fairly high capacitance. Therefore, the small inductance introduced by the wire can form an LC circuit. Furthermore, it is prone to pickup.

Try the following:
Split the required gate resistor into two parts, each half of the total required. One goes at the source, the other at the gate itself.

You can also try making a twisted pair out of the source and gate wiring.

The best solution, of course is to keep that wiring as short as possible, even if it means a change in your design.

[jgedde]

Oops. I think I may have made my reply confusing. When I meant one resistor at the gate, the other at the "source" I meant your gate driving circuit, not the source on the FET.

John