SMPS to meet peak current demands of cellular radio

G'day,

I am currently designing a power supply (SMPS) for a Telit cellular module. Usually I tend to over engineer something like this and be done with it, but this particular design needs to be both small, lightweight, and power friendly. So it needs to be pretty close to spec! I'm wondering how to realize a SMPS able to deliver peak current spikes. To explain this better, the cellular module datasheet specifies when running in GSM mode average current consumption will be 176mA however, given the nature of GSM, peak currents as large as 2A can be expected at the transmit rate of 216Hz. (duty cycle approx 8% i'm guessing??). I should also mention, the supply will have to be rated to a least 600mA continues to meet the WCDMA HSDPA power requirement.

What does one need to account for when designing a SMPS capable of delivering the high current peak demands?

Hope someone can shed some light!

Contuinuous Current = 600ma

Peak current = 2 amp (for how many ms ?)

Size allowed = ??? dimensions

Weight allowed = ???

max ripple voltage allowed?

operating frequency?? to avoid interference with the cellular module?

From my expirience.
Average current for TELIT was about 450(mA). So, you have to design power supply taking into account 600(mA) for HSDPA. I would recommend to set target to 800(mA) to be sure.
In order to ensure 2A spike, use big capacitors -C = (1/dV)*432*1e-6(F), so for dV = 0.1 about 4320(uF), for 0.2(V) about 2200(uF).
Do not use one big capacitor, better 3-4 smaller - in order to have smaller ESR. Do not forget that capacitors have limited pulsed current value.
Jurijs

There are two basic approaches. Either make a low bandwidth control loop with a large output capacitor, or a high bandwidth converter with a low output capacitor.

The first option allows the output voltage to droop, but as long as the capacitor is large the droop can be limited to whatever you need. Also the first option takes longer to recover from droop. The second option will allow the output voltage to be tightly regulated and respond quickly to transient loads. The amplitude of droop/overshoot may be higher than the first option, but recovery time is much faster. Also keep in mind for this approach to work, you likely need a very low impedance source for the converter, since it will also demand large transient currents.

You should specify the allowable droop or overshoot that is tolerable on the output voltage, first of all.

Thanks kot_b and mtweig very helpful,

As long as I can replenish the charge in the storage caps before the next current pulse I should be right.... meaning the max current of the switcher will be determined by the amount of time I have to recharge the caps.. (I.e. off time with regards to duty cycle). If for example I have a switcher with Imax = 800ma, would I need a controller that had some form of current limiting (current folding)? As each time there was a pulse current demand, the storage caps would provide this over by which duration there would be some voltage drop (within spec of course). The switcher would be looking at very low impedance (i.e. half empty caps) meaning it would go 'full tilt' to charge them up again. Would this not be similar to shorting the output momentarily, causing additional power consumption from inductor saturation? Hence a requirement for a current limited supply in such case?

Side note... Vin for the Telit is between 3.2V and 4.2V (designed for Li Poly battery) so I have some pretty wide limits.