Why would dynamic current consumption increase with frequency?

[KD494]

So I have an analog path of a few stages of op amps. I tested it over a wide range of frequencies (100kHz-20MHz) and got more or less the results I expected in terms of gain, bandwidth, noise, and quiescent current. However I noticed that once I reach 5MHz or so the current from my supplies is suddenly jumping by a factor of two or so. The output voltage pk-pk had not changed, the load is an ADC that can handle these frequencies and should still be high impedance and the chain is not visibly going unstable. What other reasons could there be for the increases power supply draw that is so dependent on frequency of excitation?

 

[chuckey]

Its the input capacitance of the transistors within the IC causes more charge/discharge current as the frequency increases.
Frank

 

[KD494]

I was thinking about capacitors since I also have an RC filter right at the input to the ADC but shouldn't the average current through the capacitors be zero since they store and then release energy rather than dissipating it as heat like a resistor? I'm measuring current draw from my supply with a fluke multimeter so I thought this wouldn't show up since it gives an average over many cycles.

EDIT: Maybe what I said doesn't make sense because I suppose the average current through a resistor can be zero and power will still be dissipated. What I was getting at was that since a capacitor ideally doesn't burn energy away as heat I didn't expect the charging and discharging to increase the current draw on my supply.

 

[Warpspeed]

It still requires energy to charge and discharge a capacitor, even though all the stored energy in the capacitor is fully recovered.

Think of it like compressing and relaxing a powerful spring with your arm muscles.
The spring returns all its stored potential energy, but you will still quickly raise a sweat.

 

[crutschow]

The power isn't dissipated in the capacitor that's true, it's dissipated in the resistor or transistor that's charging/discharging the capacitor.

 

[chuckey]

In the 1960s there was in the UK a very popular 10W audio amplifier (Henry's amp), at 1 KHZ it would take .5A at 10 KHZ the current was 1.5A. There is a similar effect with micro processor ICs as enthusiasts modified their clock speed (over clocking) the dissipation would rise a lot, so added cooling was needed.
Frank

 

[c_mitra]

Dissipation is always a function of frequency and always increases with frequency, although it may not appear much at lower frequency and hence invisible.

I like the Warpspeed example in post #5 and I shall copy that!

Capacitors store, charge and discharge energy without loss. During charging and discharging, there will always be some loss depending on the associated series resistance (in a real capacitor).

At 1Hz, there will be 1 unit of dissipation; at 10Hz, the same dissipation will be ten times more, because it is being charged and discharged 10 times in the same time (1s).

Usually this contribution is small (there are other dissipations) at low frequency but becomes prominent as the frequency becomes higher.

In general, all transitions have a term that is dissipative in nature

 

[FvM]

Dissipation is always a function of frequency and always increases with frequency

Only if the amplifier supply current is signal dependent and the amplifier is driving a frequency dependent load, e.g. the said capacitor (or has respective internal capacitances). That's the case for class AB and B amplifiers, but not pure class A.

 

[KD494]

although it may not appear much at lower frequency and hence invisible.

That makes a lot of sense actually, I was wondering why my quiescent current seemed to stay stable until the MHz range where it really started to add up.

Thanks for all the input!

 

[c_mitra]

It is the same basic reason that the dissipation within a CPU (that has thousands of transistors that act as switch) increases with the operating frequency. A switch has dissipation only during transitions.