2
\$\begingroup\$

I'm interfacing to the "Dynamixel TTL" bus, which is a simple TTL serial bus with multiple talkers. Thus, while I'm not talking, I need to tri-state the output, and be prepared to read input. There is a single coordinating master, so collision detection is not necessary.

The bus is specified at 5V, and my microcontroller is 3.3V, but it's 5V tolerant so as long as I can drive 5V and 3V counts as "high" for the inputs of the buffer, I'm fine.

I've been using 74HCT125 chips for interfacing, using two of the elements. This is annoying, because I need to invert one of the "enables" in the buffer to get output-vs-input to be decided by a single pin.

There are a number of RS-485 drivers that have the right "shape -- OE, !RE, driver, and receiver. They can even translate voltages. But they output differential signals (A/B.) I just need a single-ended signal.

I'm considering using a RS-485 buffer chip, and just tying the "B" wire to a resistor divider to decide the high/low detection point of the "A" wire, but I'd hope there's something more elegant -- something not requiring more external resistors, for example. It should also be an active surface mount part, not something I can only find in '70s basement clean-out sales :-)

The 74AUP1T45 from Nexperia is almost right, except it only goes to 3.6V. (I don't control the bus specification; I need to drive at 5V and be tolerant to at least 5.5V.)

Parametric searches at Mouser and Digi-Key don't show something obviously better than the re-wired 485 driver option, so I'm turning to this bigger brain: What's a great part for these requirements?

SN74ABT245 has the right topology and function, but it has way too many pins because its an 8-channel device, and it's also TTL and thus has low output impedance (only guarantees 3V output with 5V VCC.)

enter image description here

\$\endgroup\$
1
\$\begingroup\$

If the only thing that the 74AUP1T45 does not do right is the voltage, then you can use the same gate in a different family that supports 5 V, such as LVC. The (SN)74LVC1T45 is available from many manufacturers.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks for the suggestion! I'd previously discounted the LVC family because it specifies input-high as 0.7 * VCC, which at 5V translates to 3.5V. I'd prefer something that recognizes high at 3.0V even at the 5V node. (In this particular chip, there are two supplies, so it'll go to 2.0V at 3.3V VCC, which is good for that case.) \$\endgroup\$ – Jon Watte Jul 4 '17 at 21:19
  • \$\begingroup\$ Also, the 74xx145 will not isolate the "RX" from the "TX" on the MCU side, so I'd re-receive every byte that I send. That's not a huge problem, but it does increase interrupt load slightly, so the Toshiba part actually works better! I tie in/out on the bus side, but keep them separate on the MCU side, and the MCU is 5V tolerant. \$\endgroup\$ – Jon Watte Jul 4 '17 at 21:57
0
\$\begingroup\$

I finally found a part that works!

TC7WT241FU

It's the TC7WT241FU. It has very tolerant Vinhi of 2V even with 5V Vcc. It has two enables, one of which is the inverse of the other. It outputs at least 4.18V (typical 4.5V) for high, at 4.5V Vcc. It switches in less than 40 ns, which is a bonus.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.