Hardware device for driving antique Teletype machines.
For use with 60mA current loop Teletypes at 45 baud.
Version 3.3 works, successfully driving a Model 15 Teletype. It's also been built by Steve Garrison and used to drive a Model 28 Teletype.
This is a board to allow connecting antique Teletype machines to a computer through a USB port. It's for Teletype Model 14 and 15 machines, which use 60mA current loops. This board needs no external power supply other than the USB port. This approach uses only about 1 watt of power, drawing 200mA from the USB port.
The Teletype needs a current of 60mA and an initial voltage of 120V to operate the selector magnet. The 120V is needed to overcome the huge inductance of the selector magnet; once the magnet is charged, the required voltage is only a few volts. So the circuit charges up some capacitors and dumps them into the selector magnet at the start of each 1 bit (called "MARK" in the Teletype world). After a few milliseconds, a low-voltage sustain supply takes over to maintain the 60mA current.
All this is powered entirely from a USB port. So there's a custom switching power supply.
At the USB end is a CP2012 breakout board U3. The Silicon Labs CP2102 is a USB to serial converter, one of the few which can be reprogrammed for 45 baud operation. We reprogram it to map a request for 600 baud to 45 baud. We also reprogram it to request 400mA from the USB port.
There are two power supplies for the output. One is a custom switching power supply which, during SPACE, charges up a pair of 1uF ceramic capacitors C1 and C2 to 120V. There's a board position for a C3 capacitor as well, in parallel with C1 and C2, but this has not been necessary. These are through-hole components because very large SMT capacitors have capactance which declines with voltage, which is unacceptable here. They're ceramics for low internal resistance.
These caps provide the initial power to pull in the selector magnet, pushing 60mA through the huge (measured) at 5.5H inductance of the magnet coils. The other is a 15V supply U8 which provides sustaining current at 60mA once C2 and C11 have discharged. Both power supplies feed, through diode D6, a solid-state relay U4. The relay is controlled by the transmit data line TxD from U3.
The switching power supply is controlled by an LT3750 capacitor charging controller. This is the right tool for the job, but with pin spacing of 0.5mm, is hard to solder. The LT3750 uses comparators and flip-flops to control charging.
The switcher is an isolated boost supply, consisting of U1, T1, C1, C2, and some passives. The controller signal GATE turns the FET in Q1 on and off. Turning Q1 off produces an inductive kick in T1, which has a 1:10 turns ratio. This can produce over 120VDC, which is used to charge C1 and C2.
Charging occurs only during SPACE. During MARK, TxD goes high, inverter U9 inverts that signal, and so, on a MARK to SPACE transition, the CHARGE input of U1 sees an edge which starts the charging process.
When the charging process starts, a flip-flop inside U1 turns on, and this turns on the GATE signal to Q1. This allows current to flow through T1. The current ramps up linearly as the inductor charges. Resistor R3, which is only 0.091 ohm, is used as a current sense resistor to tell U1 when to shut off GATE, stopping current though the transformer primary. The value of R3 controls the maximum current the charger will draw.
Then the transformer, disconnected from power, dumps its magnetic field into the secondary, charging C1 and C2 through D1. As this happens, the voltage across the transformer primary drops, and this is sensed via the RDCM pin of U1. When that voltage difference is near zero, a comparator in U1 turns the flip-flop back on, turning on GATE and starting the cycle again. That's the oscillator in this.
There's also circuitry in U1 which senses the peak voltage level to which the primary of U1 rises during each charge cycle via RVout. This is used to stop charging once the desired voltage has been reached. It takes about 14ms to charge C1 and C2 to 120V, and then the charging stops. The value of R4 controls this goal.
The sustain supply U8 is always on, but there is no load on it during SPACE, because U4 is turned off. So the two power supplies take turns drawing power from the USB port, which keeps the peak current down.
To allow this circuit to support various selector magnet resistances, there's a linear current regulator at the end, consisting of Q2 and R5. Q2 is a depletion-mode MOSFET, an unusual component useful for current regulation over a wide voltage range. This limits the output current to 60mA. When driving a 220 ohm selector, this circuit isn't doing much current limiting, but when driving a 55 ohm selector, it is. It also provides short circuit protection, limiting short circuit current to 60mA. (A previous version of the board had a removable jumper for adjusting the output current, but that was too 1980s. Now it's fully automatic.)
On the keyboard side, U6, a very small 5V to 24V DC-DC converter, produces enough voltage for Teletype keyboard contacts, which may have oil or dirt on them. A 5V logic level is known to be too weak for this. Another opto-isolator, U5, isolates the keyboard from the logic level circuitry. There's a BREAK button, SW2, so that a BREAK can be sent even if no keyboard is present.
A standard 5V solid state relay, such as a Crydom CSW2410-10 can be plugged into jack J3 for Teletype motor control. This will turn on when the USB serial port raises Request to Send and power is on. The middle green light of D13 will also turn on. The "Baudotrss" software package supports this function.
This board is powered entirely from a USB port. There are strict rules about drawing power from a USB port, and most modern laptop computers have a USB port controller wihich strictly enforces them. If a device draws too much current, even for a millisecond, the port turns off power, and usually won't turn back on again until the laptop is turned off. Devices must negotiate with the power source for how much power they want, and the power source can say no. This happens when the device is plugged into the USB port, or when the laptop turns the port on.
USB devices are guaranteed 100mA, but if you want more than that, you have to ask. We ask for 400mA, below the 500mA maximum for USB versions 1 and 2, by programming a register in the CP2102 USB interface. We also set the speed for the serial port to 45 baud when 600 baud is requested. Programming is done with Silicon Labs' Simplicity Studio program.
If the host device says yes to the request for 400mA, the CP2012 turns on /SUSPEND after successful completion of the power handshake. Turning off the device in software (as when the host computer goes to sleep) will turn off /SUSPEND.
U2 is a AP2553W6 power control IC intended for USB ports. When /SUSPEND goes high, it turns on and lets power into the rest of the board. It also has a built-in power limit, set to 400mA by resistor R16, which limits the inrush current as C1 charges at power up. If turn-on is successful, and SW1 is on, the VPWR line comes up and everything on the board gets power, including the top LED of D13.
With some operating systems, this won't happen until the USB port is opened by software. If the computer goes to sleep, everything will turn off. Most USB hubs and some "smart" USB extension cables will reject a request for 500mA, and the light won't come on. If the light will come on when the USB cable is directly plugged into a computer, but won't come on when plugged in through a long cable or hub, that's the reason.
R1 and C3 are a filter for the inductive spikes from the Teletype selector magnet. The back to back 120V Zeners D10 and D11 protect against the inductive kickback spikes from the selector magnet.
On the input side, C8, C12, and L3 keep the inductive kickback from T1 from getting back into the USB power supply.
The AP2553W6 provides overload protection to the USB port. It first acts as a current limiter, and if the overload continues, it heats up, detects the overtemp condition, and cuts power. The CPC1510G solid state relays also provide current limiting, in case the printer output is shorted. A dead short across the output will not damage anything.
On the Teletype side, everything is isolated from ground and from the USB side.
- W1 - low-voltage ground
- W2 - low-voltage power (4.5-5V to W1 GND when powered up)
- W3 - Low side of transformer primary. Watch the switcher action here.
- W4 - High side of switcher current sense resistor
- W5 - high-voltage ground
- W6 - high-voltage high side
J1 is the 1/4" Teletype printer jack, supplying 120VDC at 60mA for a Teletype Model 14 or 15. Only one printer can be driven; there is not enoug power to drive a whole chain of machines.
J2 is the 1/4" Teletype keyboard jack, supplying power for a Model 14/15 keyboard. If nothing is plugged in, the jack shorts and the BREAK button still works.
P2 is a connection for a current meter. A 100mA meter is suggested. If a meter is not used, this plug must be jumpered.
J3 is a smalll jack for a solid state relay (5V in) to turn on the Teletype motor. Use of this is optional.
SW1 is the main power switch.
SW2 is a BREAK button, interrupting the keyboard circuit. This works even if no keyboard is present.
D13 is a 3-high set of green LEDs.
Top - Power ON Middle - Motor ON Bottom - Data output
This unit appears to most computers as a serial port. No special driver should be required. The port should be opened and set to 600 baud to request 45 baud output. (The CP2102 must be programmed for this, because many systems can't request 45 baud. 600 baud is used because that speed was never used for any historical devices.) When power is on, Data Set Ready will turn on. When power is on and the motor is running, Clear to Send will turn on. Power management works; if the computer goes to sleep or suspends, the board will turn off.
The USB to serial converter part on the board must be configured to output 45 baud. See the README file in the "firmware" directory for instructions. The driving software must request 600 baud to get 45 baud, because 45 baud is no longer a standard baud rate.
Standard Gerber and drill files are in "board/fab/fab.zip". These are known to work with Seeed Studio in Shentzen, and should work with other board shops.
The board is 75mm x 120mm, and will fit in a Hammond 1455K1202 box. Patterns for cutting the end plates for the box are in the directory "case". The Corel Draw files will work with Epilog laser cutters. We use 1/16" black acrylic.
The bill of materials is in "board/ttydriver01.xml", in KiCAD format. If this file is run through the Python program in "tools/src/kicadbomtovendor.py", CSV files are generated. From the command line, when in the directory "board", try
python3 ../tools/src/kicadbomtovendor ttydriver01.xml --split=VENDOR
This will generate CSV files for Digikey, Robotshop, and Coilcraft. The Digikey CSV file can be read by most spreadsheet programs, and can be fed into the Digikey BOM system, which will look up and price all the parts. Various parts may be out of stock, but every part on that list has been shipped to us from DigiKey at least once. Substitute out of stock resistors and caps with equivalent parts if needed.
Mount the CP2102 board to the main board without sockets using pin headers. If you mount it with a socket, the board won't fit in the box. The CP2102 board comes with angled headers soldered on. Those have to be removed. Solder straight 0.10 inch headers soldered into all the connector holes on the CP2102 board. The CP2102 board comes with the needed header pins.
The manufacturer isn't critical. None of the resistors are doing anything exotic. R3, R5, and R16 are 1% for a reason; they're all for current sensing. The others aren't critical and 5% resistors will work.
This board has been assembled by using solder paste for the surface mount parts, placing the parts, then reflowing in a reflow oven. It's a straightforward job by modern SMT standards. This will require SMT skills and tools. U1 has 0.5mm pin spacing, which is tight.
Check for solder bridges, especially between pins of U1. The pads of U1 are laid out so that solder bridges tend to occur near the ends of the pins, where you can get at them with copper braid and a pointed soldering iron tip.
No software required yet. See that the computer recognizes it as a USB device.
W1 is ground, W2 is Vcc. Should see 4.8V between those two with the power switch on. Bottom green light should turn on. This may not happen unless some program has the USB port open, which turns it on.
If you haven't reprogrammed the SiLabs baord for 45 baud, do so now.
Send data from Baudotrss or Heavy Metal at 45 baud. Watch the lights on the serial board blink. Top green light should blink.
W5 is HV ground. W6 is the high side of the capacitors. Expect to see 120V there with no load on the output. You have to send some data to get a voltage there, because the switcher only charges the cap after a MARK to SPACE transition. R14 will discharge the caps in a few seconds, so after power-off it doesn't retain high voltage. So you need data to keep high voltage.
With no data being sent but with something driving the port, hook up a current meter to the printer jack. Should register 60mA, even into a dead short. The current limiter can handle this. Q2 will heat up, but it's rated for 150C and it only heats up to 80C or so driving a dead short for 20 minutes.
If all those tests work, you're probably good to run a Teletype. Enjoy!