Using a FET as a low-side switch to modulate an RF carrier; is it detuning the RF transmissions?

Thread Starter

procnias

Joined Jun 2, 2016
9
Hello all,
I am a wildlife biologist with a strong interest in the use and development of radiotelemetry devices for use in tracking small animals. A few months ago I wrote a post about getting help with analyzing a circuit of one of the world's smallest VHF radiotags (see attached circuit diagram and this post) for a species of bird that I work with in the Neotropics; a design published by Naef-Daenzer and others (2005).

Over the last few months I have moved from this simple RC-pulsed design to a similar design, but where the RF pulses are controlled by an MCU. Although I have already gotten a circuit working using an ATtiny20 MCU, the transmissions have completely detuned (i.e. the harmonic filtering that led to a strong emphasis of the 3rd harmonic [144 MHz] in the original analog design no longer results in any appreciable filtering of the fundamental, or any other harmonics).

The changes I've made to the design are fairly modest. WRT the attached circuit I have removed R1 (replaced it with a short), replaced R3 with a N-channel MOSFET to serve as a low-side switch, and attached the MCU to VCC, GND, and one I/O to the gate of the FET. Although the circuit works fairly well otherwise, the detuning of the RF carrier has made the transmitter useless in its current form.

I have been consulting with three very bright electronics students at my university, and one of them suggested that the parasitic capacitance and/or the on-state resistance of the FET (~250 milli Ohm) could interfere with the harmonic filtering of the RF oscillations. I have tested the circuit's harmonic output using a manual switch (replacing the FET) to control output, and the harmonics are filtered as intended (I believe that result lends further evidence for the idea that the problem is causally linked to the FET).

First, does this sound like a reasonable diagnosis of the tuning problem? If so, does anyone out there have any suggestions for 1) a location where the FET could be moved where it would not interfere with the RF carrier (possibly using a high-side switch instead)? or 2) passives that could be inserted to isolate the parasitic effects of the FET from the RF output of the BJT (e.g. reinserting R3 so that it lies between the BJT and the FET)?

Any advice or knowledge you could offer for any of these questions would be greatly appreciated!

Many thanks,
Julian
 

Attachments

GopherT

Joined Nov 23, 2012
8,012
Hello all,
I am a wildlife biologist with a strong interest in the use and development of radiotelemetry devices for use in tracking small animals. A few months ago I wrote a post about getting help with analyzing a circuit of one of the world's smallest VHF radiotags (see attached circuit diagram and this post) for a species of bird that I work with in the Neotropics; a design published by Naef-Daenzer and others (2005).

Over the last few months I have moved from this simple RC-pulsed design to a similar design, but where the RF pulses are controlled by an MCU. Although I have already gotten a circuit working using an ATtiny20 MCU, the transmissions have completely detuned (i.e. the harmonic filtering that led to a strong emphasis of the 3rd harmonic [144 MHz] in the original analog design no longer results in any appreciable filtering of the fundamental, or any other harmonics).

The changes I've made to the design are fairly modest. WRT the attached circuit I have removed R1 (replaced it with a short), replaced R3 with a N-channel MOSFET to serve as a low-side switch, and attached the MCU to VCC, GND, and one I/O to the gate of the FET. Although the circuit works fairly well otherwise, the detuning of the RF carrier has made the transmitter useless in its current form.

I have been consulting with three very bright electronics students at my university, and one of them suggested that the parasitic capacitance and/or the on-state resistance of the FET (~250 milli Ohm) could interfere with the harmonic filtering of the RF oscillations. I have tested the circuit's harmonic output using a manual switch (replacing the FET) to control output, and the harmonics are filtered as intended (I believe that result lends further evidence for the idea that the problem is causally linked to the FET).

First, does this sound like a reasonable diagnosis of the tuning problem? If so, does anyone out there have any suggestions for 1) a location where the FET could be moved where it would not interfere with the RF carrier (possibly using a high-side switch instead)? or 2) passives that could be inserted to isolate the parasitic effects of the FET from the RF output of the BJT (e.g. reinserting R3 so that it lies between the BJT and the FET)?

Any advice or knowledge you could offer for any of these questions would be greatly appreciated!

Many thanks,
Julian
With out the schematic of the modified circuit, it is difficult to see, exactly, what changes have been made. Additionally, we need some model numbers of the MOSFET and any other parts you've added.

As a first observation, what MOSFET did you use because most will not turn on with such low voltages shown in your original schematic.
 

Thread Starter

procnias

Joined Jun 2, 2016
9
Hi GopherT,

Thanks for the offer to take a look. I've attached the modified circuit and the FET and MCU datasheets. All R, C, and L values are the same as in the original diagram. Any input you have would be appreciated, but I'm especially interested in the general answer to whether it's theoretically possible that a FET, functioning as a low-side switch, might interfere with RF filtering through parasitic effects.

Regarding the threshold voltage for the FET, it's 0.7 V and is being driven to ~ +3 V (Vcc = 3V) by the MCU.

Thanks again,
Julian
 

Attachments

GopherT

Joined Nov 23, 2012
8,012
There are just too many things going on here to diagnose. Some things you can look at...

is the microcontroller drawing too much current from the battery and causing the voltage that supplies the oscillator to vary and cause a frequency shift.

Are your pulses of the FET so short that the oscillator doesn't have time to come back on (achieve an oscillating state). Normally, the oscillator is left powered on in a stable state and another node is switched (Between oscillator and antenna).

Remember that a real battery is just an ideal battery with a resistor in series (internal resistance). The internal resistance increases as the battery reaches end of life.

What voltage battery are you using (1.5, 3.0, other?). Low voltage supply means high on-state resistance for MOSFET.
 

Thread Starter

procnias

Joined Jun 2, 2016
9
Thanks again GopherT,
In reference to your very reasonable suggestions:

1. I had considered that the supply current to the MCU might cause a voltage drop in Vcc for the rest of the circuit, but the issue remains when I isolate the power supply for the transmitter and for the MCU. Also, the transmitter produces a nice pure 3rd-harmonic RF carrier when I run the MCU with the same supply, but remove the FET and replace it with a hand-operated switch (suggesting that the MCU current draw isn't the problem).

2. The start-up time for the crystal is a real issue I've had to deal with, but once the crystal starts (I allow it to stabilize for a few ms) I can rapidly modulate the signal with on-off keying and still maintain a stable crystal frequency. Nonetheless I have tested the possibility that the rapid modulation is affecting filtering by programming the FET to turn on the oscillator without any modulation (i.e. long stable RF carrier tone). The "detuning" issue remains in this case as well.

3. Using extremely small batteries, which I am for my work with hummingbirds, results in a large voltage drop with each pulse unless a large capacitor is put in parallel with the battery. I have tested the circuit with both a regulated 3V supply, and with the secondary coin cell (3V, 1mAh) I have used for the original (analog) version of the tag. Both have the same problem.

I have not yet measured (empirically) the on-state resistance of the FET, but if it is higher than the range recommended for R3 in the original design (0 - 100 Ohm; which seems highly unlikely) I can imagine that this could influence tuning. Certainly worth a measure.

None of your suggestions point to the possibility that capacitive parasitics of the FET could influence the tuning of the circuit. In your opinion, does this seem less likely to you than other issues?

Thanks,
Julian
 

GopherT

Joined Nov 23, 2012
8,012
@procnias
Those were the first ideas I had without seeing the thing in action.
Note that the load on the microcontroller outputs can significantly drop the voltage on the output pins. What is connected to the jumper pins? Is that just for programming the microcontroller or data dumps, or are they connected to something during radio operation?
 

GopherT

Joined Nov 23, 2012
8,012
@procnias
Also, just because 0-100 ohms is allowed in the design doesn't mean it won't change the oscillation frequency.

My last thoughts (maybe someone else here can help with these)
- oscillators with inductors can have a kick as voltage is cut off, can that cause the body diode of the MOSFET is conduct and somehow shift the oscillation frequency.

- Is the output capacitance of the MOSFET possibly changing the frequency of the radio? Your C3 and C4 are only 15pf and 3.3pf respectively. Now you add a MOSFET below T1 with an output capacitance (Coss) of about 20 to 30pF at 2 volts. I think that can have an impact.

upload_2016-10-24_14-48-38.png
 

Attachments

Thread Starter

procnias

Joined Jun 2, 2016
9
Those jumper pins are essentially left floating (unless otherwise indicated in the diagram). The jumpers are for programming only, and are cut off the tag to minimize size once it's programmed. Any unused pins (after the MCU is programmed) are programmed to be inputs with internal pull-ups activated (not positive about this, but I've done whatever is recommended by Atmel for unused pins and to minimize power consumption; one unused pin is set as an output low so that it can be grounded).

Regarding the second two suggestions - the interaction between the RF inductors and the body diode, and the significant output capacitance of the FET - any ideas about how to get around this behavior assuming it's happening? A high-side switch with a P-FET in place of R1 in the original circuit (L1 and C2 interact to serve as an RF choke, so this part of the circuit should be less likely to interfere with the RF filtering due to funky FET parasitics)?

Thanks,
Julian
 

GopherT

Joined Nov 23, 2012
8,012
Those jumper pins are essentially left floating (unless otherwise indicated in the diagram). The jumpers are for programming only, and are cut off the tag to minimize size once it's programmed. Any unused pins (after the MCU is programmed) are programmed to be inputs with internal pull-ups activated (not positive about this, but I've done whatever is recommended by Atmel for unused pins and to minimize power consumption; one unused pin is set as an output low so that it can be grounded).

Regarding the second two suggestions - the interaction between the RF inductors and the body diode, and the significant output capacitance of the FET - any ideas about how to get around this behavior assuming it's happening? A high-side switch with a P-FET in place of R1 in the original circuit (L1 and C2 interact to serve as an RF choke, so this part of the circuit should be less likely to interfere with the RF filtering due to funky FET parasitics)?

Thanks,
Julian
what happened to the PDTA114 in your revised schematic?
 

Thread Starter

procnias

Joined Jun 2, 2016
9
The PDTA was a device that helped increase the charge time of C1 in the the RC-controlled (R1 and C1) duty cycle. This made it possible to have longer gaps between transmissions without having to increase the duration of each transmission. Because the duty cycle is now controlled by the FET that functionality is no longer needed.

Importantly though, the tag works as it should (in terms of the frequency and harmonics) in the revised schematic (i.e. without PDTA114, R1, and R3, and with the MCU installed) when the FET is replaced with a manual switch.
 
Top