From an early age I’ve had an interest in historical electrical communications systems such as the electrical telegraph, local battery telephone systems, Morse code over radio and mechanical teleprinters.
That’s why I was delighted to find Les Kerr’s MorseKOB system (https://sites.google.com/site/morsekob/home ) while surfing the internet. Les’ MorseKOB is a system that allows an individual to connect (through a simple electronic interface) a telegraph key and sounder to a Windows computer, and then connect the Windows computer to a multi-circuit telegraph conference bridge via the internet. Some of the “wires” (channels) of the conference bridge can be used to communicate via telegraph instruments, with other individuals anywhere in the world who have internet service, either one-on-one or in party line fashion. Several other “wires” are broadcast channels that continuously send a variety of content, such as current news stories, historical news stories, weather information, and even OS (reported On the Sheet) train movement reports.
I’m moderately proficient with international Morse code, but have little ability with American Morse. I hope that, when time permits, that I can gain some conversational ability with American Morse. But until that happens, I enjoy having a MorseKOB enabled sounder make telegraphic music when I’m in my workshop.
With that as background, I’d like to describe my own implementation of a MorseKOB interface. I did two different versions: one makes use of a USB to RS-232 Serial interface, the other makes use of a USB to TTL Serial interface. The voltages with RS-232 are bipolar and inverted, while the TTL version is unipolar and not inverted. Both designs are very similar to Les’ original design; I just used a few different components, primarily a MOSFET to drive the sounder rather than a Bipolar Junction Transistor. I’m not sure that the extra protection devices I included are actually needed, but when absolute minimum cost is not a goal I sometimes adopt a “belt and suspenders” approach to circuit design.
I’ll first discuss the RS-232 version. Figure 1 (at the bottom of this page) shows the schematic of my version of an RS-232 to MorseKOB interface. You will see that it’s very similar to Les’ original design. Some of the changes I made were to use an LM358 as the op-amp and a MOSFET as the switch. I also added an dim LED to serve as an RX Mark state indicator. Additionally, I’ve included a voltage spike snubber across the sounder coil. This is essentially the same snubber that I use on selector magnets in teletype machines. I need to do more testing with these snubber values to see if further optimization can be achieved for the different sounder windings and their associated inductance. With that as background, let’s get into RS-232 interface circuit.
One key component not shown on the schematic is a USB to RS-232 serial converter itself. This is a device that connects to a USB port on a computer and presents an RS-232 terminal interface to a modem (the modem in this case being the MorseKOB Interface). These devices usually look like a cable with a USB-A connector on one end and a DB-9 male connector on the other. These devices are available from many vendors and cost between $7 and $15 dollars. The DB-9 male connector on the converter cable connects to P1, shown at location D1 on the schematic.
In conventional serial port operation, the serial data from the computer to the modem would appear on pin 3, the TXD lead. For the MorseKOB system the serial data from the computer to the sounder appear on pin 7, the RTS pin. This RTS signal is bipolar, swinging from approximately -6 volts for sounder off to +6 volts for sounder on. The RTS signal is passed to the gate of Q1, the MOSFET switch, through resistor R9. Because of the extremely high impedance of the Gate pin to Source pin on Q1, there is no substantive voltage drop across R9. R8 ensures that the gate is grounded if the RTS signal should ever go high impedance. An “RX Mark state indicator” is formed by LED DS1, R7 and D4. DS1 is a high efficiency LED so that the current needed for the LED is low (around 1mA) and I want it to be dim, not bright. D4 is included because some LEDs will tolerate only a small amount of reverse voltage. Although Q1 physically appears to be a “beefy” device and is constructed to easily mount on a heat sink, in the MorseKOB application Q1 acts as an on-off switch, dissipating virtually no power, so a heat sink is not needed or desired.
When the gate of Q1 is more positive than about +2,5* volts, Q1 will turn on, conducting current from Drain to Source when the telegraph key is closed, thus causing the sounder to pull its armature down onto its anvil. When the gate of Q1 is less than +1 volt (keep in mind that “less than +1 volt” includes negative voltages) Q1 is in the off state and no current flows from Drain to Source, thus interrupting current flow through the sounder. Note that virtually all of the current that flows through the key and sounder also flows through Q1 and out the Drain pin of Q1, passing through R10 (1 ohm) to ground. R10 has no operational effect on the circuit, other than it provides a means to easily measure the current that is passing through the sounder. For every millivolt across R1 there is a millampere of current passing through the sounder/key loop. So, if the voltage across R1 is 150 mV then the current through the sounder is 150mA
The purpose of op-amp U1A and associated components is to inform the computer, via the DSR signal, the state of the telegraph key, without regard to whether Q1 is conducting current or not. Let’s look more closely at how this works.
An interesting aspect of the RS-232 MorseKOB interface is that power for the op-amp is derived from RS-232 signals themselves, in particular the DTR signal for a positive voltage source and the TXD signal for a negative voltage source.. When running the MorseKOB software both the TXD and DTR signal leads stay at there respective voltages and do not change. A reference voltage of approximately +0.7 volts for the negative op-amp input (pin 2) is developed by R3 and D1. The positive input op-amp input (pin 3) monitors the voltage level at the Drain of Q1, raised by approximately 1.4 volts due to the voltage drop across D5 and D6 when current flows through them.
When the key is open, regardless of the state of Q1, the voltage across R5 is zero (ignoring op-amp input bias currents which are 10s of nano-amps) and thus the voltage applied to the positive input of the op-amp is zero. In this condition the output of the op-amp (DSR lead) goes to (near) the negative supply voltage, -6 VDC, telling the computer that the key is open.
When the key is closed and Q1 is off (no sounder current flowing), the voltage across R5 is nearly equal to the loop supply voltage, somewhere between +5 and +130 volts. The voltage across R5 is also applied to the clamping circuit formed by R6, D2 and D3, to keep the voltage applied to the positive op-amp input between DTR+0.7 volts on the high side and -0.7 volts on the low side. Additionally, the current flowing into the op-amp due to excessive voltages is limited by R6. Regardless of the specific voltage of the loop supply, when the key is closed and Q1 is off, the voltage on the positive op-amp input is greater than the voltage on the negative op-amp input (+0.7 VDC) and in this condition the output of the op-amp (signal lead DSR) is near the positive supply voltage, +6VDC, telling the computer that the key is closed.
Now consider the case of key closed and Q1 fully on, conducting current. Also imagine that D5 and D6 were shunted by a shorting strap, so that they were no longer operational. In this case with the key closed and Q1 on, causing sounder loop current to flow, the voltage across Q1 would be nearly zero and thus the voltage across R5 would be nearly zero (some voltage would be across R5 due to the voltage drop across R10) and the positive op-amp input would be near zero, which is the same condition as when the key is open. So the computer would not know the correct state of the telegraph key when Q1 was conducting.
Now consider the shunt removed from D5 and D6, so that they are operational. With D5 and D6 returned to the circuit, and with the key closed and Q1 conducting, the voltage across R5 is now at least +1.4 volts due to the voltage drop across D5 and D6. That +1.4 volt input, is greater than the +0.7 volts on the negative op-amp input, causing the output of the op-amp (signal lead DSR) to be near the positive supply voltage, +6VDC, telling the computer that the key is closed. With D5 and D6 in circuit as shown on the schematic, the DSR signal now reflects the state of the key regardless of whether Q1 is on or off.
Figure 1 – RS=-232 Version Click here to see this as a PDF
The TTL version of the interface is similar to that of the RS-232. The major changes are that the DSR and RTS signals need to be inverted with respect to the signal levels for the RS-232 version, as provided by Q2 and Q5. Also the power source is +5V as provided by the USB to TTL converter and the return is the USB ground signal. Referring to the schematic below, you will see a number of generic 2N3904 NPN transistors. In quantity the price of these transistors is less than one cent each, so I often use them without regard to cost. Q5, R10 and R11 take the place of the op-amp with a switching transition of approximately 1.3 volts at room temperature. An op-amp would provide tighter switching voltage, but I wanted to try a circuit without an op-amp. Q3 and Q4 form a darlington driver for the LED. I used this configuration so that there was as little loading as possible on the collector of Q2, so that when the voltage to the MOSFET Q1 was intended to be high, it would be as close to +5 as possible. This is more of a concern when IRF type MOSFETS are used instead of the IRL MOSFETs. The IRF type MOSFETS would like a gate voltage of at least 4.5 volts for the on state, while the IRL MOSFETS are happy with 2.5 volts for the on state.
A FEW ADDITIONAL THOUGHTS AND COMMENTS
*PARTS: Good choices for the MOSFET are IRL620, IRL640, IRF620, IRF640. The IRL parts have a gate threshold voltage of about 2.5 volts while the IRF parts have a gate threshold voltage of about 4.5 volts. If you are buying parts rather than pulling from your junk box, I’d suggest you get the IRL640 as that would be the most flexible (lower gate threshold voltage, higher drain source voltage). For the LED indicator that I added, one would normally run about 10 mA to the LED to ensure it was well lit. In this case, I want a dim indicator, not a lighthouse. So, I used a 10K ohm resistor and my LED is a high efficiency type. Adjust the value of the LED current limiting resistor to give you the brightness (or dimness) you want in the visual indicator. I have lots of parts, because I often buy in lots of 10 or 25. If you need parts feel free to contact me to see what I can offer for free. If I have spares I’ll be happy to give you some so you can make your own interface. I may ask you to cover postage.
Earth ground is your friend. There’s a basic rule in electrical design that says “no floating metal”. All metal masses in the system need to be committed to a known voltage relative to earth ground (there are exception cases, but they are infrequent). The known voltage can be time varying or AC, but it must have a galvanic connection to earth ground (even if it’s through a 1 Meg ohm resistor). In order to be a safety ground, rather than just a static dissipating ground, the path to ground must be one of low resistance. Like most telegraph systems of old, the exposed metal of the telegraph key is not at earth ground potential. In most cases the voltage on the key is relatively low, but if you are using a loop supply with an open circuit voltage of more than 48 volts you need to keep in mind that the key is “hot”. When your computer is connected to the AC mains via a 3 wire power cord, the circuit ground on the USB cable will likely be at earth ground potential, and that provides the earth ground reference for your MorseKOB system. If you are in doubt about earth grounding via the USB cable, disconnect your computer from AC power and check with an ohm meter for continuity between the MorseKOB ground and the ground pin on the computer power cord. When using a laptop computer from internal battery, without a line cord from laptop to AC mains, it’s likely that your MorseKOB circuit common is not at earth ground potential. In this case it would be good practice to have a ground wire from your MorseKOB circuit common to some source of earth ground. A ground wire from the MorseKOB circuit common that ends in a banana plug which is inserted into the ground pin of an AC outlet will solve that problem. One could also use a connection to a cold water pipe (assuming it’s metal pipe all the way to where it hits dirt) or even a ground rod. All this may sound excessive, but it’s really just good practice.
Do you really need a fuse in your loop supply? I think the answer is yes. Fuses are cheap and they provide protection. Telegraph sounders are not easily replaced in the 21st century yet fuses are readily available. So just do it. Use a fuse. I don’t expect the fuse in the MorseKOB interface to ever need replacing, unless I do something stupid, which does happen. Even so, I don’t use a fuse holder. Instead I just solder the fuse into the circuit. To make that easier, I cut off the leads of a 1/4 watt resistor and then reuse the leads by soldering them to the ends of the fuse. Parts catalogs show leaded low voltage fuses, but the minimum order quantity is often 1000 pieces. Minimum order for low voltage fuses intended for a fuse holder is 1, so I buy those and add my own leads to the fuses that I use.
The loop supply for a MorseKOB interface should be configured as shown in the schematic: A series circuit composed of a DC voltage source with negative side grounded or connected to circuit common, the positive side to a fuse, then to the loop current limiting resistor, and then the other side of the resistor is the “output” of the loop supply. The output of the loop supply would then be connected as a series circuit through the sounder and key, to the MorseKOB interface which returns current back to ground or circuit common. The fuse should be rated for about 1.5 to 2.0 times the expected typical loop current. If you connect the output of the loop supply (composed to voltage source, fuse and resistor) to ground for 30 minutes the fuse should not blow and your loop current limiting resistor should not fail. If your loop current limiting resistor gets too hot touch (but is still within the resistor wattage spec) you may need to provide for some ventilation holes in your loop supply enclosure.
The inductance of your sounder magnet coils goes up as the resistance of the coils in your sounder goes up (longer wire, higher resistance, more turns yields more inductance). As the inductance goes up, you will likely need a higher loop voltage and larger current limiting resistor, in order to have the sounder produce “snappy” action. When the inductance gets higher (think 120 and 400 ohm sounders) the only way to lower the L/R time constant is to increase the series resistance. Then, in order to get the necessary current, the voltage source needs to increase. I don’t have any 400 ohm sounders but I do have a 120 ohm sounder. I measure about 500 mH of inductance with this sounder. The 120 ohm sounder only needs about 6 volts to operate, but with 6 volts there’s no current limiting resistor, other than the resistance of the sounder itself, which produces a long L/R time constant, and the sounder sounds sluggish on the downward movement of the armature. The solution for these large ohm value sounders is to have a larger driving voltage and a large current limiting resistor. My 120 ohm sounder sounds OK at 24 volts but much better with 48 volts and an even larger current limiting resistor. I’ve found that an easy way to get 48 VDC is to get two 24 volt wall-wart power converters and connect the outputs in series. In the case my 4 ohm sounder, I measure only 5 mH of inductance. I drive the 4 ohm sounder with 9 volts and a current limiting resistor to yield about 300 mA of current. With these values the 4 ohm (local) sounder produces nice, snappy action, thanks to the relatively low inductance of the magnets.
Many electronics hobbyists (sometimes myself included) fall into the trap of thinking that the voltage drop across a silicon diode is always exactly 0.7 volts. It’s not; it’s approximately 0.7 volts, but the actual value will vary depending on the amount of current passing through the diode and the temperature of the diode. According to its data sheet, a common 1N4148 diode at 25C (77F) will have a voltage drop of about 0.7 volts at 5 mA; 0.5 volts at 100 uA; and 1.0 volts at about 180 mA. Temperature is an additional factor: higher temperature yields lower voltage drop, and vice versa. It’s OK to use 0.7 volts as an approximate value, but always keep in mind that it is an approximate value.
The electrical component of a telegraph sounder is an electro-magnet, which can have a significant value of inductance. As the old adage goes “the current through an inductor can not change instantaneously”. So when the key of a sounder loop goes open, the supply of current to the sounder is cut off, but then consider that “the current through an inductor can not change instantaneously”. What that means is that when the key goes open (or Q1 goes open), the inductor, at least for a few microseconds or milliseconds, as the magnetic flux collapses, will try to force that same amount of current into that open circuit. So if there was 100 mA sounder current flowing, and then the key goes open, the sounder “sees” an infinite resistance (because the circuit is now open), and tries to shove 100mA of current into that resistance (for a few microseconds or milliseconds). The resistance of the open circuit is not actually infinite, as there are always leakage paths, but the open circuit resistance IS BIG. Let’s assume it’s 10,000 ohms (although it’s probably much larger). Ohms law applies, so the sounder wants to send 100mA into 100,000 ohms, at least for a few microseconds. So 0.1 amperes x 10,000 ohms equals is 1,000V. That voltage spike creates two problems. One is that the high voltage spike can damage transistors (semiconductors) that are in the control circuit and the other is that voltage spikes will produce radio frequency interference (RFI). The RFI may not be of much concern to you, unless you are operating near an HF radio receiver, but the voltage spike can damage Q1 in the interface circuit. Which brings us to “snubbers”. Snubbers are circuit elements that limit the amplitude of the voltage spike produced by an inductor when the driving current through the inductor is interrupted. WU apparently used a 22 ohm resistor in series with a 100nF (0.1 microfarad) capacitor as a snubber across the sounder (and perhaps relay) coils. I use a series RC network as well, but I also add an additional 1nF (0.001 microfarad) capacitor in parallel with the coils for my snubbers. Without the 1nF capacitor, when the key (or Q1) goes open, the voltage spike will be limited in amplitude by the resistor in the snubber, but the voltage will jump up to that limited amplitude very very fast, almost a step function, which will produce a click on a radio receiver. The 1nF capacitor is effective for only a few microseconds after the key (or Q1) goes open, but it will slow the rise of the voltage that appears across the snubber resistor and will greatly reduce the key clicks as heard on a radio receiver. So, to summarize, the 1nF snubber cap is to kill key clicks as heard on a radio receiver and the the 100nF cap in series with a resistor are to limit the peak voltage produced by the sounder coil when current through the sounder coils is interrupted – I used both at the same time. My snubber resistor of 470 ohms was picked because it’s what I use for Teletype magnets. That may not be the best value for telegraph sounders. I need to do more testing to see what the best value may be. Also, the best snubber resistor value for a 4 ohm sounder may not be the best value for a 120 ohm sounder; but something is probably better than nothing. I mount my snubbers on the bottom of the sounder, where they are less likely to be visible. Keep in mind that semiconductor components don’t always die instantly from a single large voltage spike. In some cases it can take 100s or 1000s of spikes before ill effect is evident. However dead components are dead components, regardless whether they die with first voltage spike or the 10,000th voltage spike.
 If you use an RS-232 serial card installed internally in your computer, rather than a USB to RS-232 cable, you may find that the signal voltages are +12VDC and -12VDC rather than +6VDC and -6VDC as found on the USB to RS-232 cable. The MorseKOB interface should work with either system voltage.
 In this document I use the convention of “electrical current” flowing from a positive electrode to a negative electrode. Other readers may be more familiar with the convention of “electron flow” from a negative electrode to a positive electrode. Both perspectives are correct.