Basic circuit issues magically being fixed by the oscilloscope

I have a very basic circuit where a PIC is putting a 2kHz square wave to a FET that is driving an 8 ohm speaker. Everything is going well except for one issue. The speaker randomly does not work. Troubleshooting has narrowed it down to two situations.

1) It works. This happens when my circuit’s ground is connected to ground on my oscilloscope. Whether or not the oscope is even on. And it has to be the ground.

2) It doesn’t work. This happens in every other situation. Also, the circuit is running at 5VDC, but when it isn’t working, the voltage drops to 3.7VDC (circuit is being powered by a PICKit 2).

For some reason my benchtop multimeter ground doesn’t make the speaker work. My understanding is that the ground impedance for the multimeter would be extremely high, but so is the ground impedance for an oscope, correct?

When I substitute a high impedance piezoelectric buzzer, it oscillates at the appropriate frequency and everything is good. In summation, I’m assuming that it is a ground issue related to the relatively high amount of current the speaker draws.

I know I’m missing something basic, but I’m concerned that this issue will be propagated into my final design, which inherently needs to be battery powered. Any thoughts would be greatly appreciated.

Your oscope's ground is typically connected to earth ground, unless you have a special battery-powered style oscilloscope such as a Tektronix TPS series.

The fact that your voltage is dropping to 3.7V seems telling. Are you powering your entire circuit from the PICKit2? This board is USB powered and has a 100mA max current for Vdd Target.

How much current is your overall circuit drawing?

Is the output from your 5V supply reading 3.7V directly from positive to negative when the speaker is hooked up? If so then I would check to make sure the supply isn't being overloaded while attempting to drive the speaker.

If the supply is still at 5V directly from positive to negative however, and you are measuring the 3.7V at another physical point in your circuit (across the speaker?), then I would try to find where the other 1.3V has gone missing with some multimeter probing around connectors, wires, and common return paths. Even a couple ohms of miscellaneous resistance inline with an 8 ohm speaker will divide off a volt from your supply after all.

Joe is correct, most oscilloscope's ground clips are hooked directly to earth ground through the power outlet. This is actually a LOW impedance path in every wall powered scope I've seen, and if the problem is some miscellaneous resistances as explained above, then this would make sense why the speaker works while the oscilloscope ground clip is attached. With the clip on, you may be shorting out any miscellaneous resistances in your return path to your system common (now referenced off of earth ground), and magically it works. Remove this short and reference to earth ground, and those nasty miscellaneous resistances come back to undercut your voltage again.

I also thought that 3.7V was an interesting number, not coincidental, especially since Vdd dropped, not just over the speaker. The entire circuit is being powered by the PICKit2. My memory was that the max current was 500mA, but the datasheet agrees with you. Important lesson #1, datasheets have better memories than I do. Current was going to be 5V/8ohms = ~625mA with a 50% duty cycle, I figured that would be enough cushion with a 500mA max.

Unfortunately, my power supply shorted out the other day (which is why I was using the PICKit2 Vdd) and I don't have a new fuse yet (Important lesson #2, don't physically try and move your power supply when plugged in. I also still need to take it apart and see what moved to cause the short).

And it is important to note that my understanding of the oscope ground was wrong. Important lesson #3. Thank you both for your help and speedy replies!

Hello Joshua,

The first thing we need to understand is the ground pathway. If I were to make a guess about your system without seeing the schematic, it would be that your ground path is not solid (Low Impedance). When you plug in your scope, it will connect you to earth ground. Your computer's USB port is also connected to earth ground. Thus your pathway for ground return is probably through the ground plug of the oscilloscope. When disconnected your circuit ground is probably floating at 1.3V above ground. This could be because of a forward voltage drop across a diode or two, or possibly a Vce sat across a transistor or something similar.

You may also be experiencing ground bounce in your circuit. This is a condition in which large currents across an impedance in the ground system cause a voltage drop. This raises the relative ground floor giving you the voltage you see. Your scope would be much lower in impedance and very little bounce would occur.

You can make a very inexpensive 5V source from 4AA cells in series with the addition of a diode or two in series. this would eliminate the ground impedance in the circuit, be it active or passive.It would also provide a stiff supply to provide current for the switching in the circuit. I say this because of the possibility of current starving the switching system. This is the reason for bypass and decoupling caps. The reason behind them is different, and there are many app notes to explain bot theory and application.

Hope this helps.

Richard