Related Topics:
 • Homebrewing
 • Soundmodem
 

 Build a Thru-modem

Using the TCM3105 Chip...

If you are not sure what a thru-modem is, here is as close as I can come to a definition. It takes a bit stream of 1's and 0's on the RS232 line and converts them to tones, the mark being 1200Hz and the space being 2200Hz (Bell 202 Standard). This digital stream emanates directly from the output port on the serial COMs card and passes through the thru-modem at 1200 baud, or whatever the device has been designed to "translate," emerges as an analogue stream properly tonally encoded, and enters the radio input to be transmitted on the air as FM signals.

This is also known as Frequency Shift Keying (FSK) or AFSK, Audio FSK. There is no computer on board and no software of any kind, no eeprom, etc. Sometimes it is referred to as a "dumb" modem, but I prefer the term thru-modem.

---

Quick Headings Reference List Hardware Construction A Few Construction Hints Property Disclaimer Parts Availability Construction Procedures photo: Container Front photo: Full Circuit Board photo: Close up of Board Populating the Board The I/O and PTT Interface Going "Live" Faults Normal Operation Theory of Operation The TCM3105 The 4011 The 4093 The 78L05 Filter Capacitors

The active components for the circuit are:

• the TCM3105, a modem chip
• the 4011, a nand gate chip
• the 78L05, a voltage regulator

A word of caution about the modem chip. They don't make them anymore. However, that doesn't mean you can't get one, you just have to hunt around at the supply houses and ask, or look for defunct TNC's that use the 3105 and re-cycle it into this project! (NOTE: Since this is an amateur radio site, I cannot promote any particular product line. However, a site that may prove useful as a beginning point to obtain the TCM3105 chip is www.tcm3105.com .)

A Few Construction Hints:

An Experimenter's Plug Board

If you are a beginner at circuit board soldering, or this is your first project, I recommend asking for help. This is at least an intermediate level project. I estimate the time required to be between 10-20 hours of board time, not including the gathering of needed parts. Also, I strongly suggest that the circuit be built on a "project experimenter" plug board , also known as a prototype board, first. Then I recommend labeling the feed lines and any others that warrant. Make sure the circuit works before transferring it to the circuit board! (An obvious, but often overlooked design strategy.)

Also, there is no pre-printed circuit board that I know of currently that you can order that matches the schematic shown here, so you have to "translate" it to whatever board you choose and this can be challenging! With a bit of practice, I found that I was able to do this without too much trouble, but it was difficult at first, so have patience. And there is no particular board, or container that I recommend. You are free to rummage your "junk" box or be as "creative" as you like :)

I also suggest using chip sockets, and using 2 nine pin connectors, a male and a female for data and radio input/output, respectively. (However, the radio port only uses three or optionally four wires, so you might want to use a different kind of connection.)

An Intellectual Property Caveat and Disclaimer:

The circuit design that is shown on this page is of my own design. It was taken originally from an article that appeared in QST and I have modified it based on experiments with various chips and components. I have not tried to intentionally clone any existing design. If this is the case, then it is by accident that that has happened... You are free to do as you wish with this design, to share it or to improve it where possible in the spirit of Amateur Radio.

Why I like this circuit:

• A Minimum Number of Parts, chips consolidate transistors
• Very Reliable, crystal controlled, highly stable transmission signal
• No External Power Source Needed, RS-232 signals double as current supply

Parts Availability:

Most of these parts are readily available from either your "junk" box, or parts supply houses. The more difficult parts to obtain, as mentioned, are the TCM3105, and its companion crystal. (You might try entering these data values from your favorite search engine on the net and see what comes back.)

Electronic Parts Description Table & Schematic Part Label Part Value Part Description   1. U1   4011   CMOS Nand Gate, high impedance "buffers"   2. U2   TCM3105   AFSK Voice Band Modem, main chip   3. U3   78L05   Low Power Voltage Regulator   4. D1-6   1N914   PIN Diodes (low current)   5. Y1   4.4336 MHz   Crystal for TCM3105 Modem Clock   6. R1   56 k   1/4 watt resistor, part of voltage divider   7. R2   68 k   1/4 watt resistor, part of voltage divider   8. R3   100 k   low power variable resistor, audio output   9. R4   330-1000 Ohm   1/2 or 1/4 watt Push-To-Talk resistor 10. C1-2   20 mmf   Oscillator Capacitors, (use good quality) 11. C3, C7   0.01 mf   Decoupling Capacitors 12. C4, C5   0.1 mf   Audio Coupling Capacitors 13. C6   0.33mf   Input Filter Capacitor 14. C8   100 mf   Output Power Filter Capacitor

NOTE: All of the pins for the above IC chips begin with the number one (1) in the upper left corner and increment in a counterclockwise direction until either reaching 14 for the 4011 (U1), or 16 for the TCM3105 (U2).

Pinout Diagram for the TCM3105

RS-232-C Pin Number to Signal Assignment Table DB25 Pin (alternate) DB9 Pin (male) RS232 Signal 2 3 TXD 4 7 RTS 5 8 CTS 7 5 GND 20 4 DTR DB9 Pin Number to Radio (KAM Compatible) DB25 Pin (alternate) DB9 Pin (female) Radio Ops - 1 AFSK out - 3 PTT - 5 Audio in - 6 GND

Construction Procedures:

I begin most of my projects by assigning a ground (negative) bus and a supply (positive) bus on the printed circuit board. I often tie the positive lands with a red wire and the grounds with a black wire. Once, I have these power "zones" marked out, I solder the sockets onto the board so that they are close to the feed (+) and drain (-) points. I also strongly suggest that you use sockets for the chips. One is a 14 pin, the other is a 16 pin. You can use a transistor socket for the power supply chip if you like. (I soldered mine directly, but high up from the surface of the board. Be sure to use a heat sink for this operation!)

Then I usually solder all the needed joints for a given chip, doing one chip at a time. You might begin with the 4011, then the TCM3105, and finally the regulator device. Leave the sockets empty and use a volt meter set to resistance to check the continuity of different sections of the circuit. Take your time and do frequent checks against the schematic to verify the connections. Lightly buff the printed board with steel wool before you do any soldering, and don't use too much heat, just enough to "run and puddle" the solder joint! Also, use the best quality board you can obtain so that the lands won't lift off if you accidently overheat a joint.

To adjust the audio output, you must preset R3, the 100k Ohm variable resistor. I set this to about 25000 Ohms coming from the chip output to the slider output, or about 75% of full audio power. I have found that this is a reliable starting point. (If you can't seem to connect to anyone, you can try changing this setting, but I have always been able to connect at this point on the variable resistor.) Using your Ohm meter, make this setting before you place the chips in their sockets.

The first active component that I place on the board is the regulator chip. As I mentioned, I soldered it in place using a heat sink. If you don't have a heat sink, you can use needle nose pliers. Take an elastic band and wrap it around the handgrips so that the pliers are forced shut. Then place the regulator in between the pinchers of the pliers. Situate the chip and solder in place with only enough heat to do the job. Even with the heat sink, don't keep the gun on the board for more than a few seconds at a time. If you like, you may test the supply voltage by using a 9 volt battery and placing your volt meter across the "rails." You should read between 5 and 5.5 volts. (Apply the battery voltage across the regulator's input pin and ground.)

Perhaps the trickiest part of this entire project is attaching the leads for the two nine pin connectors You are free to choose any type of connector that you wish. I just happened to select a male nine pin (from the computer) and a female nine pin (to the radio), since most of my TNC's use a similar convention. This is why I recommended that you label all input/output wiring for the computer side and all input/output for the radio side. You may use any labeling scheme you like. This greatly simplifies the hook up of these wires to the board! I used numbers on the computer side and brief labels like "PTT" or "TX" on the radio side. You could also use RS232 signal names like CTS or DTR for the computer side connector.

This close up shows the 9-pin connectors and their locations on the front of the container. The male on the left receives the RS232 line from the computer, and the female on the right receives a 9-pin connector from the radio. Note: You need to consult your radio user's manual in order to build the radio side connection correctly. The three required lines are GND, RX, and TX. You may wire the female connector using any pins that you like as long as they match the joining male plug from the radio. (Check on how you want to handle the external PTT line, it may be optional.) Note: You may not need the external PTT line since that is handled internally by bringing the TX line to ground via R4 behind the output capacitor C3. (I had followed a connector diagram for interfacing to a handi-talki which did require a PTT line and a 2200 Ohm resistor bridging the TX and PTT lines inside the 9-pin connector to the radio.) So the PTT line is shown as a possible external line in the schematic; but, it is optional. Here is a photograph of the completed circuit board. The first thing you might notice is the large amount of space left on the board. Whenever I finish a project, it seems there is always one more part I want to add; so for this one, I left room for 2 more chips, just in case! Please note that I left the labels on the wires from the board to the 9-pin connectors! You can see the pin numbered labels on the left and the radio labels on the right. The nine pin on the left (computer) is a male and the one on the right (radio) is a female. Here is a close up of the circuit board. The regulator chip, in the upper left, looks like a transistor. The 4011 is the chip on the left and the TCM3105 is the chip on the right. Note: The labeled wires that feed the connectors to the outside of the box are quite evident. (You may remove the labels later if you like.) Note: The large black wire on the lower right is a "strap" joining two ground busses on the other side of the board.

Populating the Board:

At this point you should have all the passive components and the regulator chip on the board. Carefully insert the remaining two active components, the 4011 and the TCM3105, into their sockets. Sometimes this can be tricky! Don't jam them in or you will end up with a bunch of bent pins on the chips! Take your needle nose pliers again and gently move a rack of pins toward the upright position, 90 degrees from the base of the chip itself. This will straighten them and then try the insertion again. I make a point of this because one time I applied so much pressure to a chip that it flipped over and became "inserted" into my thumb. An experience I will not soon forget! :)

As a final word of caution, you can't desk check the circuit board too much. Every minute you spend checking the board against the schematic can represent hours of trouble shooting that you won't have to do. It's worth the extra time. And, let some time go by before your last check. It is truly amazing what your mind will accept as OK when in fact it is not. Have a friend with fresh eyes critique your board...

The I/O and PTT Interface:

As mentioned above, there are many ways that you might want to handle the PTT design. Numerous options abound! You could come right out of the modem container with audio cable (600 Ohm) lines, an RX and a TX, not using a separate PTT line. Or, you could use a PTT line through some connector arrangement as I have done with a 9-pin.

I chose the latter option since I had many connectors already made up that had been used on handi-talkies, HTX-202's, in the past, and I wanted to ensure compatibility. Specifically, I had set up the radio side to work with a KAM, my first commercial TNC, still in reliable service today after over a decade! The table above entitled "DB9 Pin Numbers to Radio" suggests one possible set up for the modem 9-pin output based on the KAM pinout. You are free to choose any design that is compatible with your equipment! As noted, this is a hardware area that is wide open for experimentation!

A Thru-modem cable for a Handi-Talkie: Nine Pin to 1/8 and 3/32 inch plugs

My thru-modem cables have embedded in the 9-pin (male) connector body an audio capacitor and a pull-down resistor which isolate the DC and AC signal components to the radio input. One side of the resistor is located between the capacitor and the mic input center pin, the other is attached to the PTT pin on the connector. When this DC line is pulled low, the mic is keyed and opened for data transmission. Using the circuit shown above, you can directly connect a mic input line to TXA since its internal PTT network already has a pull-down resistor R4 and audio capacitor C5 built in. (Or, you can use a specially-made cable like mine, as shown here, which effectively places the two resistors in parallel and the two capacitors in series, offering the maximum AC/DC signal isolation. Both approaches work fine.)

Since radios vary widely in their cabling/connector hardware, and may continue to change in the future, such variability certainly presents challenges to the builder! That is why I observed that this is a "highly" experimental area. (For example, you could even build a vox-like circuit which detects data on the RS232 line DTR, thus eliminating the need for an RTS TX signal, one less wire in the cable... using 4 wires instead of 5, perhaps suggesting the application of different cable types and connectors! Ah, decisions, decisions...)

Going "Live:"

OK, you are ready for a real live test. Don't put the cover on yet. Run your test with it off so that you can measure voltages if you need to.

NOTE : Remember, this device can be used with any OS or any platform! It is just a "translator," converting between tones and RS232 square waves. The HDLC protocol, needed to manage packets, must be resident within the software running on your computer. It can be any software running on any hardware platform! However, if you are running a Linux system, and a *NOS network application, and you need some help with setting up the software to complete the hardware tests, please see: Setting up the Software...

Hooking it all up together, plug in your computer RS232 serial line to the modem, and plug in your radio connections from the modem; watch your radio as you make this connection. Sometimes, it will be "stuck" in transmit. If this is the case, then leave it detached until you have all the software in place. Now start your software application, plugging in the radio connection line. If all is going well, the radio will be in listening mode and tracing packets if enabled. Attempt a contact and watch or listen to the radio for clues. If it works, congratulations! If not, then you need to find the problem and fix it.

Faults:

The two most likely causes of faults are:

• miss-wired parts
• bad solder joints

This covers about 95% of the mistakes made in home brew projects! After removing the chips, go back and "continuity test" the circuit, comparing to the schematic for errors. Take a look at all the solder joints. Using a magnifying glass, see if any lands have been bridged. You should be able to run a small screw driver through all the lands without encountering any solder globs. Make sure all the joints have a smooth shiny appearance, and are not crinkly, or just partially soldered, or just plain weird looking as compared to the rest. If any need a re-heat, then carefully apply a small bit of new solder over the old until it is completely melted or joined together.

If this still does not fix the problem, then you may have damaged a part in assembling the board, or you may have received a bad part. If you have spares, try replacing them. Double check the joints around the crystal part of the circuit for the TCM3105. This can be a weak spot. (Don't give up, you will find it and fix it!)

Normal Operation:

With all your software installed and all the connections made between the computer and the modem and the radio and the modem, you should begin to see your station log filling up. If all is going well, this modem will look and behave just like a regular TNC, except it requires no independent power source and often is more sensitive than a TNC. For best receiving, the signal strength should be an S5 or greater, but I have copied stations as low as an S2, something my TNC's cannot do. Hope you have enjoyed building this project and find it useful. For about $25-30 (today's prices), you can't find a better bargain for packet radio fun!

The TCM3105:

Paraphrasing from the TI spec sheet, this chip is an asynchronous Frequency Shift Keying voiceband modem that uses gate CMOS technology to implement a switched capacitor architecture. It contains a group delay equalizer, automatic gain control, carrier detect level adjustment, and bias distortion adjustment.

It should be noted too that this chip was never designed to be used in a radio context. Therefore it responds to a raised energy level that it assumes is carrying intelligence. It can't really tell the difference between noise and signal! Sometimes this chip is gated via a PLL chip which can tell if the "disturbance" is uniform and within a known bandwidth. Generally, though, I see very few "bogus" signals making it through onto my monitor.

The diagram below presents in detail the major functional components and how they relate to each other. One obvious, but sometimes overlooked, consideration is that the transmission path and the receive path are two distinct and independent entities! Another way of saying this is that the device can be operated full duplex! And conversely, this implies that you could theoretically use just one of the sub-units at a time, or more commonly as this article has proposed, both working alternately in half duplex. For example, you could build a unit using just the TCM3105's transmission capabilities relying on its superior crystal controlled, oscillator stabilization, an outstanding feature of this chip. You then could build a separate receiver unit which might be a PLL. Recall that the TCM3105 was designed for landline telephony, requiring loud decibel signals on receive, something that radio reception does not always guarantee! A PLL unit might detect signals that don't even show up on the radio's signal strength meter! (Now whether this is a "good thing" is up to the station operator's judgment; but it might be doable and have some very interesting experimental "side-effects," such as hearing stations you can't reach!)

Functional Block Diagram for the TCM3105
Illustration Courtesy of Texas Instruments Corporation

To display a thumbnail image of the pinout diagram for the TCM3105, press: . Adjust the pop-up window dimensions to suit your monitor and browser aspect accommodation. Then reposition the window off to one side so that you may compare the pin locations on the chip with the descriptions on the block diagram. (NOTE: You need JavaScript enabled to do this...)

If you are an experimenter, and you wouldn't be here if you weren't, you may replace the comparator threshold voltage divider, feeding pin 7, with a 100K variable resistor and approximately replicate the values of 56K and 68K, setting the pot near the mid position. The recommended values from the data sheet are a minimum of 2.3v, a maximum of 3.1v, and a typical value of 2.7 volts. The frequencies, 1200Hz and 2200Hz, capacitively sum their own associated voltages, generated by each frequency and an internal regulator. It's the job of the comparator to translate these voltages into a rail-to-rail square wave signal that becomes the RS-232 data stream for the computer input. All this "threshold adjust" really does, via pin RXB above, is to regulate the duration of the on and off parts of the square wave, the periodic symmetry of the wave, as it passes through and out the final comparator via pin RXD .

If you are a perfectionist, you might hook up your scope to the output stream and tweak the comparator level until you see that perfect square wave! I recently installed a potentiometer instead of the voltage divider circuit as noted in the above schematic. One more knob to twiddle! Did I see a dramatic improvement in anything? Not really. But, I learned just a bit more about how this part of the chip works. Got me thinking about how to do some interesting loopback tests to verify voltage levels. If you like to "tinker" and are interested in how these fascinating devices really work, here is another opportunity to explore the world of digital communication! And, the fixed resistor, voltage divider works just fine too! :)

That said, generally there is little room for design experimentation using this chip in the circuit described above. You must follow the recommended pin settings if it is to work at all at the baud rate that you want. It has been well documented and is easy to implement... However, this doesn't imply that there is no room for experimentation. I accepted the conventional design as found in the "literature." You may see something that can be improved on! (That's what science and technology are all about.)

The 4011:

This chip is a NAND gate CMOS low power logic chip, and is being used in this circuit as a low power, high impedance switching device. It manages the RS232 signals coming from the computer. In a previous circuit, it replaces about four transistors with a host of supporting parts. I saw this application as a way to reduce overall part count, increase reliability and hold down the cost.

The 4011 can sink about 10ma of current which in most cases is enough to trigger a PTT line. The TX line from the radio has a DC component on it so that when that radio line is pulled low, it triggers the transmitter to send. In the circuit above, the DC is blocked by capacitor C5 and the "pull down" resistor R4 is what "creates" a ground when the 4011 gets a high signal from the RTS (computer) line and inverts it. (By tying its two inputs together, we are making an inverter out of each gate.)

An alternate chip could be the 4511. I built a circuit several years ago using this chip instead of the 4011 because I was worried about the amount of current the 4011 could sink, i.e., would it trigger the PTT line? This chip will sink about 45ma and is set up as an inverter. Either one will work fine.

The 4093:

This chip may be considered as a substitute for the 4011. It is a NAND CMOS low power device identical to the 4011 except that it is a Schmitt triggered device. This means that it "squares up" waves very well. If it even hints of a square wave, this chip will force it into a square wave! If you are operating in a noisy local environment, this could improve signal clarity. I have replaced one of my 4011's with this chip. Did I see a dramatic improvement? No, I didn't. It won't pull weak signals out of the "mud" as a PLL would, but it adds just a little extra assurance in cleaning up the line of any low level noise.

Actually, the modem which uses the 4093 has never displayed an "AX25: bad header!" message on a received packet, while the 4011 might present this message about every 1000 or so RX's. And why this is so is still not clear to me. Whether this is a noise spike, a collision, or a subtle variation in the circuit layout, is difficult to tell. But the 4093 seems to avoid all of these potential problems... It's worth the extra few cents.

The 78L05:

This little device is truly amazing! When you consider that the power sources are the RS232 lines themselves and are constantly changing, it is a wonder that the output voltage can be so well stabilized. I tried many other voltage regulators and, despite its size, this one performed the best. You might also note that I added a diode to the ground line on this device. The effect is to raise the output voltage to about 5.5 volts. This seemed to be a very inexpensive way to get just a bit more voltage across the "rails."

Filter Capacitors:

Another interesting note is the large number of different capacitors used in this circuit. There are three categories: input filter capacitors, output filter capacitors, and decoupling capacitors. Since three signal lines are used as current sources RTS, DTR, and TXD, capacitor C6 helps to smooth out the transient nature of the source. There are two decoupling capacitors that ease the "spikes" that may occur across devices: C3 on the TCM3105 and C7 on the 78L05. And there is the large output filter C8 which helps to "store" the final voltage and supply reserve against a surge of current drain.