I'm trying to characterize the performance of a power supply module, one of the criteria being power supply noise rejection, in particular at different frequencies.

For testing 10Hz to 100kHz, I was planning on using a power opamp to inject a function generator onto the input power supply (12V, <2.5A). The bandwidth of the opamp stops you injecting frequencies any faster.

However, the power board is being used to power an ASIC which operates in the GHz range, and is sensitive to noise 1GHz to 40GHz. Realistically, I think most significant sources of noise and going to be 5GHz and below (Wifi, GSM/LTE-4G etc), so 5-10GHz test and measure is a reasonable target. The entire system will be running in a lab environment, which is supposed to be no mobile phones/wifi etc, but its rarely enforced and it's unlikely our customers will enforce it either.

The power supply board is a fairly typical cascade of LDOs hanging off switching regs, and we've never had issues with it, so I am not anticipating any problems with it. However, it has never been fully benchmarked before.

Can anybody think of a way to inject sine waves at high frequency onto a 12V, power supply? I have access to function generators, oscilloscopes and spectrum analyzers at 20/40GHz.

Perhaps some kind RF amplifier with a bias set to 12V?


Basically, you take an RF generator and couple it to the 12v line via a capacitor. The output impedance of the power supply is very high at RF (let's say, 10 MHz and above), so the RF will ride on the supply voltage just fine. The resulting RF level will be dependent on parasitics and physical layout though, so be warned.

|improve this answer|||||
  • \$\begingroup\$ Surely this will add the peaks of the waveform, but not the troughs since it can't pull the 12V line down? \$\endgroup\$ – Oliver Apr 22 '14 at 14:33
  • \$\begingroup\$ Nevermind, I'm being an idiot. That solution should work, but it completely depends on the output impedance of the power supply. I've been playing with it in spice, and the amount of current coming from the function generator is dependent on the impedance of the power supply (naturally). I'm wondering how high the impedance is, and whether I can damage the function generator by sinking too much current in it. \$\endgroup\$ – Oliver Apr 22 '14 at 15:14
  • \$\begingroup\$ Perhaps I can just stick a PI filter on the 12V supply, then couple on the RF with a cap. Something like this \$\endgroup\$ – Oliver Apr 22 '14 at 15:31
  • \$\begingroup\$ In principle, yes. However, you need to take a close look at the specs for your 12v supply. It probably (if it's a good one) will show you the response to a load transient, and this will give you some idea of its impedance at higher frequencies. Since you're doing this in a commercial environment, you actually need to characterize your power supply by test. Like I say, at the frequencies you're talking about, geometry and parasitics become critical. I recommend you hire somebody with experience in the field. Earn while you learn is an iffy proposition in these circumstances. \$\endgroup\$ – WhatRoughBeast Apr 22 '14 at 18:49

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.