The Tech Blog

Was discussing a project with a client when they mentioned needing a “5-Watt” speaker for their product. I asked how they arrived at that number, and they said they’d used one previously and liked its loudness. I want to clarify a common mistake folks make. More Watts Doesn’t Always Mean More Volume. In fact, Watts might not even be the right way to choose a speaker. Let’s discuss this.

Speaker loudness is actually measured by Sound Pressure Level (SPL), expressed in decibels (dB), not watts. Watts only indicate how much electrical power a speaker can handle. SPL tells you how loud a speaker will get at a given distance and power input. Specifically, speaker sensitivity measures loudness in decibels at 1 meter away using just 1 watt of input. This is your real guide for loudness, and you will find this value in good speaker datasheets.

To get just a 3dB increase in volume, you must double the power. So, if one speaker has a sensitivity rating of 90dB and another is rated at 80dB, the first speaker is significantly louder at the same wattage. For instance, a 50-watt speaker with a 90dB rating will outperform a 100-watt speaker rated at 80 dB. Why? Because the higher sensitivity speaker converts electrical power into sound more efficiently.

SPL(dB) = Sensitivity(dB) + 10 log(Power)
  For the 90dB speaker at 50W : 90 + 10 log(50) = 90 + 16.99 ≈ 107dB
  For the 80dB speaker at 100W: 80 + 10 log(100) = 80 + 20 = 100dB

Remember, sound decreases by approximately 6dB each time the distance doubles, so you use this a tip to estimate levels at longer distance from the speaker. A speaker producing 90dB at 1m drops to about 84dB at 2m.

BTW fun fact, humans perceive a 10 dB increase as roughly twice as loud, but this actually requires about ten times more power.

So next time you’re selecting speakers, first check the sensitivity rating (dB/W/m) and not Watts alone.

I was doing some high speed PCB designing and thought AC coupling capacitors that are placed on high speed differential lines might be a good topic to discuss today. These are series capacitors you will see placed in USB 3/4 lines, PCI Express, HDMI etc.

Like any capacitor, at DC, it’s an open switch, at high frequency, it’s a dead short. Why do you need it? It blocks any DC voltage. Every differential link has two voltages: the desired differential swing (like +/-400mV) and an unwanted but expected common-mode voltage that both wires sit on, measured with respect to ground. The series capacitors block the DC part, letting each side of receiver/transmitter to coexist without any issues.

If there is any small ground offset(in mVs) between two sides, it would push a continuous long term DC current and waste power. Coupling caps block them. It also allows for hot plugging. Meaning if one side is powered, the other side is off, without any series capacitors, live driver will push DC current straight into pins that are still at 0 V, which is bad. By inserting series caps, you ensure no DC flows until both ends are fully powered and switching.

How to choose one? Most specs put cap values between 75nF and 220nF. With the 100Ω termination, that puts the high-pass freq in the tens of kilohertz, well under the spec to pass gigabits of data. I usually prefer using 0201 type capacitors, purely because they are small(low inductance) and they fit the differential pair lines width without majorly spacing it out. Placement matters. I place one cap in each leg, side by side, same orientation, pads inline, so the pair runs straight through. Keep stubs under half a millimetre and return to 100 Ω within a millimetre. Most spec/chips will tell you what you need to do.

Please take care of them while routing. They look tiny but if you get their placement or value wrong and your gigabit link will fail. It’s hard to debug them unless you have some high-end oscilloscopes. So treat them well.