I have been interested in digital modes for some time. I purchased a SignaLink USB in the early days, even when I was an F-call, anticipating my full call would be issued before it arrived. This turned out to be the case. It would be a bummer if I could not use it! Note: in VK, F-calls may not transmit anything aside from SSB, AM, FM, PM. and hand generated (inc using an electronic keyer) CW.
My early days were focused on Easypal, a DRM based SSTV mode. I also did a little bit of playing around with RTTY, PSK and other narrow band based modes. Eventually SOTA started in VK, and I began to think about the possibility of doing digital modes on summit, but not taking a computer and the Signalink.
I use a Samsung Galaxy S2, and I began to think about what would be needed to use this for PSK and RTTY. Wolphi has some andriod apps on Google play for these modes, and also of interest, sells an interface. He also has put some designs up on the web here and here.
Version 1 of these interfaces uses a transformer to boost up tx audio from the phone device, so a transistor can be switched to pull down the PTT on the radio. Version 2 removes the transformer and instead relies on a voltage that is present on the microphone pin. There is also a bug in the schematic of v2, as the rx audio does not go to pin 1, it needs to go to pin 4, plus add a capacitor.
Separate to this, VK3XJM developed an interface between his FT-817 and his ipad. His initial try was broadly similar to the v1 Wolphi interface, and I then made him aware of v2. His next revision is similar to the Wolphi v2.
My design effort
Well, homebrewer extrodinare, I wondered if I would buy one of these interfaces, or build my own. Answer: design and build my own!
The first issue to consider is that line devices are supposed to present high impedance to the audio out device. While the mobile(cell) phones will think that they are driving headphones, these still will typically be 200ohms plus. The idea to use a transformer to boost up voltage is going against the principle of trying to minimise the load. So, I will not use the transformer approach. Both VK3XJM and Wolphi have found that this approach is barely viable anyway.
So the next step: time to find out more about this microphone voltage. It has been highlighted that this varies quite a bit between devices, and this can cause some of the above interfaces to not work so well. What is this voltage? It turns out that most devices will put this voltage on as part of the “Plug-in Power Supply” system. This is to provide power for Electret Condenser microphones. These devices have a permanently statically charged film over a metal plate. This forms a variable capacitor, which changes based on captured sound. The tiny signal is placed on the gate of a common source JFET, with the signal taken from the drain. “Plug-in Power Supply” provides the voltage on the drain to drive the JFET. Typically the voltage is 3V with a 6.8Kohm resister in series in the supplying device. This is not supposed to be a high current power source!
Unfortunately, there is no one standard for the voltage, plus the series resistance of this source. On the net, there are stories of voltages anywhere from 1.5V to 3V. Hmmmmmm, this variability could be a problem.
I took a hybrid circuit based on a combination of the VK3XJM and the Wolphi v2 design and put it together in gschem. After a bit of stuffing around to find libraries for various devices, I used gnetlist to use the gschem saved file to generate a spice netlist for spice simulation.
The first thing I found is that I was sucking just way too much current from the voltage on the pin! If the supply voltage was high, the circuit still worked, but it was marginal. Things needed to be improved. I also found on birdwatching web sites (these guys are using these electric condenser microphones) that many devices will not supply much more than 300uA, and my circuit was pulling far more than that. Time to put this circuit on a diet!
The biggest pull of current was the first transistor Q1. It was time to get the Ic down, to 200uA tops! Using some design principles in Ludeman Introduction to Electronic Devices and Circuits, I ended up selecting a 20k collector resistor. This then flowed through to a 1.1k emitter resister, although my choice here was a little arbitrary. I was still trying to keep Ve low, when Ludeman recommends Re to be 1/3rd of Rc. From there, the Ib can be determined, and Vbe found using a formula. This then sets what the two resistors used as a voltage divider should be. My resisters used were far higher than the initial circuit. Ludeman also provides formulas for the input, output and emitter bypass capacitors, based on what low pass is required. I chose 100Hz. Its quite low, but it means that there should be no problems doing PSK at 500hz AF. I was prepared to fall back to 200Hz if required, but the really high impedances made capacitor choices easier. Note that Ludeman points out that the emitter bypass capacitor is not just set by Re, but there is also a AC path out of the base of the transistor. This path can be quite significant in the capacitor sizing.
The ngspice simulators showed the Q1 switching nicely on TX input, while presenting very low load to the signal. Next up was dealing with the PTT switching transistor Q2. The FT-817ND has a 11k load from a 5V source on the PTT pin based on the circuit diagram. Both the initial circuits basically rectify the AC output of Q1, filter it with a cap and feed this to the base of Q2. I first used Schottkies like VK3XJM. Hey the lower drop across the diode should help? I could not get the ngspice model working properly, so I went to ordinary 1N3891 diodes. I got these working nicely. I also bumped up a resister on the base to 20k, but also Q2 effectively provides a high resistance to the tiny weany current that is going through the base. This also allowed me to slice the filter cap right down to 100nF. This is the only cap that actually has to be charged in this circuit, and the early high current versions of this circuit required much higher values of this cap to work. It also meant that this cap was presenting too much load to the “Plug-in Power Supply” system.
So what I have now will work quite happily on anything from 1.2V to 3.5V, and I have not bothered testing outside these ranges ’cause it should cover just about everything. Once the voltage is below about 1.5V, it takes around 80ms to pull down the PTT on a 500Hz AF source, but it will pull it down. I think it would even work with a 1V supply! At 3V, it basically pulls it down in 2ms, which is on the first signal waveform! I am happy with these results, because this circuit should be very robust.
Current requirements: At 3.5V, it pulls around 80uA average, 140uA peak from the supply. Well within the 300uA for a mono electric! When the supply is 1.2V, it pulls 70uA peak and 20uA average. This circuit has certainly come out trim after its diet!
Where there any consequences in using such a tiny filter cap at 100nF. Well, yes, but they don’t matter. At 3.5V supply, the ripple is 200mV, but the signal at the base is plenty high enough, and the thing is so current limited that it does not matter than Vbe is 0.9V to 1.1V!
When the voltage is 1.2V supply, the ripple is hardly anything. So 100nF is perfectly fine for the filter cap on the base of Q2.
I have a screen shot of the circuit below. Click for full size.
I plan to build one of these on variboard. If people were interested, I could design either a through hole or a surface mount circuit PCB for it and make a kit available.
Regards, Wayne VK3WAM
EDIT: I replaced the ordinary diodes with Schottkies as their lower voltage drop across the diode helps the robustness of the circuit. The updated circuit diagram, with light background, shown below, can be clicked on.
EDIT2: Updating of the schematic as microphone audio in (to the phone) is taken from pin 4 as discussed in the text. I had kept the old one up, but Gerald DL3KGS noticed the difference.