RFM69/RFM12B data line ratings

[captcha]

In one of my customized PCB designs I've hit a little snag where switching on the (RFM12B) radio module using a fet causes a big dip in voltage because the MCP1703 can't keep up (very poor PSRR and transient response characteristics). This results in a reset of the mcu every time I turn on the radio. I've looked at many LDO alternatives but can't seem to find a part that fits my absurd list of requirements (cheap, SOT23, low dropout voltage, high PSRR, low Iq, Imax 250mA).

So I've decided to ditch the LDO between the battery and the mcu and instead feed the ATMega328 and friends (RTC, DS18B20) directly off the 1S lipo (4.2V max). The idea is to still use a fet to switch on the radio, but now I'm using the fet to switch on the LDO first, which in turn only powers on the radio.

My question relates to the radio's control signal / data line voltage levels. In the past when the board ran on 3.3V life was easy, but now that the mcu runs on (up to) 4.2V and the specs for the RFM12B and RFM69(H)W say that their maximum input voltage is around 3.8V I'm investigating the best way to add the right level of protection for the data lines.

The radio's Vcc is nicely fed with 3.3V from the switched-on LDO, but the signal lines that come out of the mcu are at higher voltage levels, hence my concern to add some protection.

I could use some fets as bi-directional logic level converters.




Or simple two-resistor voltage-divider.



The data lines I'm trying to protect are:

- IRQ
- SEL
- SDI
- SDO
- SCK

However, does anyone know whether these radios need that kind of protection on ALL lines?

PCB real estate is at a premium and if I can safely leave out some parts it would make the overall design a lot neater.

[Felix]

Would buffering with a ceramic 100uF help?
Why the RFM12b?

[captcha]

Caps? Just inline? (sorry, analog electronics is not my forte..)

I bought a couple of RFM12B's back in 2012 and don't mind using them for something useful. I have already bought some RFM69CW's as a drop-in replacement but for this particular project I really don't need the range (unless you tell me that the data lines of RFM69's are capable of higher voltages).

[TomWS]

@captcha, I suspect the dip is due to the transient response of the Moteino VR to a large capacitive load on the RFM12.  Since you're switching with a P Channel FET, this is relatively easy to address.  Put a resistor between the GPIO pin and the gate of the FET and add a capacitor between the drain and the gate pin of the FET.

The result is that the switching time is linearly controlled by the current out of the gate node determined by the Vgsth and the resistor from the GPIO.  The capacitor completes the timing equation as it is equal and opposite to the current.  Net is you have TOTAL control of the switching time, ie exact control of the rise time of VDD on the RFM12 circuits and any transient load they may present.

Reasonable starting values are: R=20K, C=0.1uF.

Tom

[captcha]

Wow, I was actually hoping for getting into an X-Y question/dialog, where by I'm trying to solve X but it's actually caused by Y which I'm taking for granted.

I was clearly under the impression that the transient response of the LDO was something that couldn't be smoothed out. Now you've given me some real pointers I can try out without having to resort to drastic measures as per my original plan of attack.

The measured voltage dip was around 8us in duration and dipped the voltage from 3.3V down to about 2.5V. These chips can run on 1.8V but not at 8MHz which I'm using.

I'll turn on the soldering iron and report back when I have some results.