I read an interesting article in QST by Art Heft K8CIT on the design of a pressure sensor based paddle (QST, Feb 2019, pp 42,43). This appealed to me as I’ve been interested in a no-moving-part paddle for SOTA activations for both compactness and hopefully reliability in a range of operating conditions (rain, cold, humidity etc). I have previously tried capacitive touch-based paddles, but found they don’t work too well in damp/humid conditions and are also quite susceptible to RF interference in a portable situation. The pressure sensor based paddle should be more robust as the sensors are sealed and the threshold voltages for switching are much higher than capacitive sensors and should therefore be more robust to RF interference.

Art’s design is based on using a simple resistive divider to feed an op-amp based comparator. This required a 9V battery to power the op-amps and thus makes a relatively large paddle.

A second source of inspiration was a simplified design, also inspired by the QST article, by Tim Keene K5DEZ which he reported on the SOTA Reflector as part of the thread here. Tim’s circuit was much simpler and he reported good results. However, like the original circuit, it relied on a battery to power it, something I was keen to avoid.

I decided to have a go at my own version using power scavenged from the key line bias which is common to all rigs. In order to detect a paddle closure, all circuits I’ve looked at use a pull-up resistor between the key line and a supply of typically 3.3V or 5V (at least in modern solid-state rigs). Key closure drops the voltage on the line from the pull-up value to close to zero which is then detected by the keying circuit (usually a microcontroller input line these days).

Looking at a few rigs, the pull up resistor value is typically between 10k and 50k which means there is between 16uA and 125uA available to power circuitry depending on the supply voltage and pull-up resistor value. I figured that if I could scavenge this power to charge a large capacitor, I could then power the paddle from this source and avoid the battery altogether.

To make the circuit work with a full range of rigs, the power scavenging needs to be done in a way that will work with voltages from 3V up and provide consistent performance independent of supply voltage. In my circuit, this is achieved by regulating the voltage to 2.5V using a precision reference. This may seem overkill, but there are relatively few devices that will regulate voltage at extremely low current levels (e.g 80uA).

The capacitor sizing is determined by the fact that it is the only source of power when both paddles are activated simultaneously (iambic sending.) In this case, both lines are grounded and hence no current flows to the capacitor. When only one paddle is activated, the capacitor is replenished by the other paddle line. The period over which the voltage on the transistor gate is held high enough to maintain it active is determined by the charge level on the capacitor and the total resistance of the divider network. The latter is determined by the resistor (R1/R2) and the pressure applied to the pressure sensor – hence quite variable.

The R1/R2 sets the sensitivity of the paddle with higher values leading to higher sensitivity. In practice, I’ve found that setting it to anything between 100k and 470k seems to work fine.

One of the most complex elements of this design is scavenging enough current to keep the capacitor charged without dropping the key line voltage below the “high” threshold of the key line sense port. This is typically a minimum of 75% of Vcc. This sets the maximum current that is practically drawable to about 82uA for a 10k pull up resistor to 3.3V. This current limit could be achieved using a resistor, but this would have the negative effect of slowing the capacitor charge significantly. A better (albeit more complex) solution is to use a current source.

I’ve implemented the current source using a well known JFET current limiter in series with a Schotkky diode to prevent back-flow of current when the key line is active. There is some significant variability in the limited current due to a wide tolerance in the JFET parameters, so there may be a few cases where the value of the resistor will need adjustment for reliable operation.

In this design, the current source is set to around 82uA which covers the worst case of 10k pull up and 3.3V supply. Theoretically, with a 50k pullup, there simply isn’t enough current available to charge the capacitor faster than the sensor depletes it. However, in practice, I’ve found that the key seems to work fine with such rigs (tested on KD1JV MTR 2 and KD1JV Soda Pop). For rigs with higher than 10k pull ups, it may be necessary to modify them to add 10k pull up resistors on each paddle line to the appropriate supply line (the one feeding the microcontroller that terminates the paddle lines) to achieve reliable operation. This is very unlikely to cause any problems with the rig.

The MOSFETs chosen for this application were selected based on the low threshold voltage (about 0.5V at 330uA drain current). The low threshold voltage means that the circuit will work down to the low 2.5V regulated supply voltage.

The PCB is designed to allow a couple of mounting options. There are 4 screw holes in flanges that can be used to attach the board to some sort of package or support such as a piece of wood or plastic to provide a grip or mounting. Alternatively, the side flanges can be cut off and the remaining PCB encased in heat shrink tubing to provide a minimalist paddle. Note that there are now also PCB gerbers for a paddle with or without mounting flanges (see below).

The PCB is designed to have a 3 core cable attached directly and held in place by a small cable tie through the provided holes.

The whole paddle and cable weighs in at just 26g, so is probably about as light as you can get for a portable CW paddle.

I have provided build data below if anyone wants to try it out.

Appendix – Build data

PCB gerber files:

There are three versions of the PCB available. There are versions with and without the mounting flanges and also a longer version without mounting flanges for those with larger hands.

Here are the links to the gerber files:

• V1.2 – Standard with mounting flanges
• V1.2A – Standard length with no mounting flanges
• V1.2B – 100mm long with no mounting flanges

For PCB manufacture I use AllPCB.com which will produce 5 of these boards for US$5 plus shipping – they are located in China and provide fast turn around and excellent quality in my experience. Note that when ordering, the board sizes are:

• V1.2: 34 x 63mm
• V1.2A: 22 x 63mm
• V1.2B: 22 x 100mm

Bill of materials:

The following table contains all the components together with stock numbers for Digikey.

Part	Qty	Value	Case	Desription	Manufacturer	Manufacturer part no.	Digikey number
Q1,Q2	2		SOT23	N-Channel MOSFET, 300 mA, 20 V,	Rohm	RYC002N05T316	RYC002N05T316CT-ND
U1	1		SOT23	2.5V voltage reference	TI	LM4040CEM3-2.5/NOPB	LM4040CEM3-2.5/NOPBCT-ND
C1,C2	2	100u	1206	Ceramic X5R capacitor	Murata	GRM31CR60J107ME39K	490-7217-1-ND
C3,C4	2	10n	805	Ceramic NPO capacitors	Murata	GCM2195C1H103JA16D	490-4783-1-ND
R1,R2	2	100k	805	1% resistor	TE Connectivity	CRG0805F100K	A106052CT-ND
Q3,Q4	2		SOT23	JFET N-CH 40V 0.225W SOT23	ON Semi	MMBF4117	MMBF4117CT-ND
D1,D2	2		SOD323	Diode Schottky 30V 200mA Surface Mount USC	Toshiba	CUS520,H3F	CUS520H3FCT-ND
R3,R4	2	5.6k	805	1% resistor	Vishay	CRCW08055K60FKEA	541-5.60KCCT-ND
S1,S2	2			Force sensor	DFRobot	RP-C18.3-ST	1738-SEN0294-ND

Pressure sensors:

The sensors used in this project are now available from Digikey, but may also be found on Amazon, Aliexpress and Ebay. I suggest you Google: RP-C18.3-ST

Construction hints:

• All components are located on one side of the board except for the dash pressure sensor
• Here is a link to the component overlay that will help with component placement – Overlay
• Suggest fitting the surface mount components first using solder paste, a hot plate and hot air gun – the component order is not important
• Fit the sensors last – they usually have adhesive already on the sensor covered by a protective paper. Remove the protective paper and align them carefully to each side of the board with the contacts centred on the corresponding pads and solder.
• Here’s a closeup of the assembled board (this is version 1.1 without trimpots or sensors) courtesy Jerry AA1OF which may help with parts placement and orientation: