1
\$\begingroup\$

I am trying to convert a 3.3V 50 Ohm Impedance TTL Pulse-per-second into other formats (GPIB, USB, RS-232, etc.). I've found some format adapters/convertors like the MAX232, but I am either misunderstanding something or missing something.

So I know the TTL line has a 50 Ohm impedance, and from looking at the RS-232 standards, the impedance is 3k-7k (?). I imagine the MAX 232 doesn't convert/impedance match. So, do I have to do my own impedance matching from 50 --> MAX 232 --> Whatever the RS-232 is (any idea how I can find out from the spec since if it is indeed 3k-7k, thats pretty much pure reflection if im not mistaken).

\$\endgroup\$
1
  • \$\begingroup\$ Reflections don't appear inside ICs, they appear in transmission lines. You should describe those. \$\endgroup\$ Sep 7, 2020 at 9:04

1 Answer 1

0
\$\begingroup\$

If you have a long transmission line at the 3.3V signal, you only have to worry about terminating that properly, however if the output has 50 ohm source termination, whatever reflection happens at the end will get terminated at the source, so it may not be a problem. Then, the RS232 driver outputs have slew rate limiting at the driver output and it requires no termination, as serial port lines do not use impedance matching, they use voltage as the signal so output impedance is low and input impedance is high.

\$\endgroup\$
12
  • \$\begingroup\$ What defines long? Does 1m/~3ft count for long? I'm confused regarding the 50 ohm source termination. From this datasheet serialcomm.com/USB_adapters/USB_converters/… there seems to be no mention of source termination. Most converters for anything dont seem to list anything about termination or impedance. I'm sure im missing something? Also, doesn't having a slew rate essentially destroy the whole purpose of the PPS since slew delays/ruins the integrity of the signal? \$\endgroup\$
    – bchang32
    Aug 31, 2020 at 17:44
  • \$\begingroup\$ Long is actually a relative term regarding to bandwidth of the signal, or slew rate. If rise time is slow enough in comparison to the length and speed it travels along the wire, it is not considered a transmission line so transmission line effects like termination become unimportant. Perhaps RS232 is not a right physical interface for your signal if it does not allow for skew or delay. \$\endgroup\$
    – Justme
    Aug 31, 2020 at 18:06
  • \$\begingroup\$ Is there a formula for this to figure out if its considered a transmission line? If my line is 6 ft long and rise time is typically 1.5 ns (so 1.5E-9 < ~2m/3E8 == 1.5E-9 < 6.667E-9). My options of interface are USB2.0, USB3.0, GPIB, RS-232, and ethernet. From my research it seems like USB, Ethernet, and GPIB have a large latency so seems like 232 is the best option? (also to convert 3.3TTL to USB, it seems like I need to convert to Rs-232 anyway?) \$\endgroup\$
    – bchang32
    Aug 31, 2020 at 18:23
  • \$\begingroup\$ Common rule of thumb is that if the the cable length is greater than 1/10 of the wavelength, it can be considered as a transmission line. Typically RS232 is used at below 200kbps, so it has a max slew rate of 30V/us, and tranceivers can easily have propagation delay of 500ns. Also USB RS232 serial ports can be quite limited, as they are typically devices that use 1ms polling rate at the USB interface. And USB TTL serial ports exist, they just don't have a separate TTL to RS232 tranceiver, so no, there is no need to convert to RS232 anyway. \$\endgroup\$
    – Justme
    Aug 31, 2020 at 18:57
  • \$\begingroup\$ Oh wow, this is a lot to process and to digest. Thank you! Do you have any advice, recommendations, or suggestions on the best route to ingest a 1 PPS with the least amount of latency? My PPS is from an NTP server so I also have to convert from a BNC female port to something suitable too (surprisingly annoying finding BNC to ... converters/adapters). \$\endgroup\$
    – bchang32
    Aug 31, 2020 at 19:39

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