calculating pcb trace width for low duty cycle

Thread Starter

Gibson486

Joined Jul 20, 2012
355
I have PWM pulses at about 4A. However, the pulse is only about 15 micro seconds long, so it is a really low duty cycle. When i calculate the trace width on the PCB, do I still need to calculate it as if it were 4 amps?
 

kubeek

Joined Sep 20, 2005
5,794
... the pulse is only about 15 micro seconds long, so it is a really low duty cycle
This doesn´t say anything about the duty cycle, it could be on for 15μs and off for 1μs. You need to state the repetition rate or length of the other half of the pulse.
 

crutschow

Joined Mar 14, 2008
34,285
Not necessarily. What is the duty cycle? You said the pulse if 15μs but you didn't state the pulse period or frequency.

At low duty-cycle the concern would be the voltage drop due to the 4A of current, not trace heating from the current (which is the concern for DC current).
 

Thread Starter

Gibson486

Joined Jul 20, 2012
355
Duty cycle is pretty low... like 5%. I need to measure the actually frequency, though. I think it is 90 Hz, but i may be wrong. I need to check my notes at home.
 

JDT

Joined Feb 12, 2009
657
The limiting factor is temperature. This is where the duty cycle is important. Gives the track time to cool down before the next pulse.

But there are other things: Voltage drop, not duty-cycle dependant. Inductance, important with fast pulses.

Best to have the track as thick and short as possible. Even if it has to narrow at one section, keep the rest as wide as possible.
 

Thread Starter

Gibson486

Joined Jul 20, 2012
355
Well, I was off on the frequency... it is actually approx 750-800 hz.

Duty cycle is high for about 15us low for about 1.25ms (1250 us).

On the supply side, it says current is about 2mA being drawn. not sure if that can be an accurate way of measuring the avg current over time.
 

Thread Starter

Gibson486

Joined Jul 20, 2012
355
The limiting factor is temperature. This is where the duty cycle is important. Gives the track time to cool down before the next pulse.

But there are other things: Voltage drop, not duty-cycle dependant. Inductance, important with fast pulses.

Best to have the track as thick and short as possible. Even if it has to narrow at one section, keep the rest as wide as possible.
I am really constrained space wise. Could I have really thick traces coming from the connectors, but shrink them the remainder of the way? Or do i have no choice but to make the traces wide enough and have more than a 2 layer board?
 

Thread Starter

Gibson486

Joined Jul 20, 2012
355
Thanks, so when I put this in the trace width calculator, can I amply adjust current to my duty cycle times pulsed current to get avg current? Does that suffice in terms of current calculation?
 

kubeek

Joined Sep 20, 2005
5,794
Like others said, the thermal capacity of the trace and pcb will average the current in terms of carrying capability, but the question is if your circuit can tolerate the higher resistance of a thin trace and thus the voltage drop at the hight current.
 

JMac3108

Joined Aug 16, 2010
348
Thanks, so when I put this in the trace width calculator, can I amply adjust current to my duty cycle times pulsed current to get avg current? Does that suffice in terms of current calculation?
As others have said ...

(1) Sizing the traces for average current will ensure that the traces do not get too hot.

(2) However, in a case like yours where the average current is low, but the peak current is high, traces sized simply for average current may be too small. The result will be a lot of voltage drop acsoss them when the peak current is being drawn. This may or may not be a problem in your circuit.

As a starting place I would attempt to size the traces for my peak current. If other resitrictions prevented this I would reduce them, but not as small as the average current calculation would indicate.

.... Kubeek said exactly what I said but more concise! ...
 
Last edited:

vrainom

Joined Sep 8, 2011
126
If you absolutely must make a thin trace then fill the trace with a little bit of solder to make it more robust. Or even solder some copper wire to the top of the trace.
 

MrChips

Joined Oct 2, 2009
30,720
Notice that I did not size the trace for average current. I used the average current to set my absolute bottom limit. Now we can look at trace widths:

500mA 5mil
1A 12mil
2A 30mil
4A 80mil

Looking at these figures, I would not go lower than 20 mil anyway which is good for 1.4A, assuming 1oz Cu.
 
Top