1
\$\begingroup\$

I tend to be very paranoid when it comes to making PCB's to the point where I try to use 50+ mil tracks for ground wires when routing the board. After, I fill the remaining space on the board with a ground plane.

I have come across a situation where I tried routing with fewer wire jumpers and even though routing is 100% successful, the ground pin from a DIP IC leading to some components consists of a 20mil wide track with a length between 5mm and 1cm before the track turns into a large ground plane (at least 200mil wide).

Since wires have resistance, I just want to make sure that such traces leading to ground planes are acceptable. If I could, I'd connect the DIP IC ground to a middle of a > 200mil wide ground plane.

So will my circuit be negatively affected by resistance/inductance if the ground of a DIP IC is connected to a large > 200mil ground plane through a 20mil track of 1cm length?

If no, then what's the maximum length I can use at 20 mils before noticing any bad effects?

My circuit will be done on a single-sided FR4 PCB with 1oz copper thickness, and My other PCB tracks and spacings are minimum 10.5 mils.

The power going through the circuit is 5VDC and the speed of operation is no more than 24Mhz.

I will add that the IC's used in my circuit are the 74HC types and cmos types.

\$\endgroup\$
  • 3
    \$\begingroup\$ If you're designing a high-speed board I most certainly wouldn't recommend a single-sided board. Generally for high-speed you should have at least a two-layer board, and even better a four layer with two signal layers and two ground planes. This way you can drop a via (or even multiple vias) straight through to a plane and avoid having any discontinuities in your ground or power connections. It also prevents the possibility of a ground plane having to neck down or being cut off between components, increasing loop area and making the signal flow less efficient \$\endgroup\$ – DerStrom8 Sep 8 '17 at 19:09
  • \$\begingroup\$ How many ICs do you have on board? \$\endgroup\$ – Ale..chenski Sep 8 '17 at 19:11
  • 2
    \$\begingroup\$ I second @DerStrom8 comment, and would add blindly filling with copper is not always the right solution either. You have to be careful not to leave islands and loops and little antennas all over the place. \$\endgroup\$ – Trevor_G Sep 8 '17 at 19:20
  • 3
    \$\begingroup\$ You should give equal consideration to power distribution. However, with 74HC and other CMOS parts, the supply and ground currents will be very low, so voltage drops overa couple of cm of 20 mil tracks will be insignificant. It is still important to have power supply bypass capacitors as close as practical to the IC power and ground pins. \$\endgroup\$ – Peter Bennett Sep 8 '17 at 19:22
  • 1
    \$\begingroup\$ The resistance is largely irrelevant. For low speed boards the track length (mostly) doesn't matter. For high speed boards, you need to consider the inductance. \$\endgroup\$ – Neil_UK Sep 8 '17 at 19:33
2
\$\begingroup\$

For 2OZ copper, 1cm of 20mil track is about 5mOhm. Assuming your mcu sinks a max current of 100ma, you'll get a vdrop of about 500uV. This is pretty tiny. The inductance will likely not be a big deal either. With microcontrollers, the most important thing is how quick the current draw changes. You can mitigate this by putting a small decoupling cap as close to the IC supply pins as possible. This is what inductance affects, so you want to make the loop as small as possible (i.e. the length is more important than the width).

\$\endgroup\$
  • \$\begingroup\$ I just made a recent board in eagle with large ground plane and an unusual power track arrangement where the track is 1.7mm wide and a couple inches long and one section of that track next to the microcontroller is expanded to about 7mm wide by about an inch long. Would this expansion be an OK substitute for a decoupling capacitor? Most of the VCC track is next to the ground plane with about a 0.4mm spacing. \$\endgroup\$ – user152879 Sep 9 '17 at 23:26
  • \$\begingroup\$ I'm not sure I get what you mean, can you post a picture? \$\endgroup\$ – BeB00 Sep 10 '17 at 20:15
0
\$\begingroup\$

It is not the clock speed that matters for power and ground inductance but rise time and ringing due to impedance mismatches. f_equiv.=0.35/t

Then compute prop delay. ad have adequate caps for distributed decoupling or use low L ground + power planes.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy