FTDI (the company practically synonymous with TTL serial adapters) uses 3.5mm tip-ring-sleeve connectors for this. In other words, a common headphone jack.
I added one to my single-board computer enclosure, following FTDI's wiring. Now I can easily connect whenever I need to use the serial console, and a standard 3.5mm audio extension cable will let me reach across the room without moving my main computer. Replacement parts, if I ever need them, are cheap and easy to find.
That really sucks for hotplugging since TRS connectors sliding in basically make random connections before seating properly.
Granted, you shouldn't hotplug TTL serial, but everyone™ does it anyway. (In some situations you're even forced to, to avoid reverse powering something.)
My scenario is connecting terminal emulators to getty or u-boot, which I think is a common one. A blip of line noise when plugging in would be barely considered an annoyance, easily cleared by pressing Backspace.
But yes, if someone happens to be using their serial line for some kind of sensitive signaling, then I would agree that choosing a more isolated connector (or just avoiding hotplug) would make sense.
Thing is, for a serial TTL connector to become ubiquitous, it needs to cover at least something like 99% of scenarios. Or maybe 95%. 3.5mm TRS ain't that, and thus just increases general diversity in connectors.
In my experience noise on hot-plugging the serial adapter is a complete nonissue. I'm sure it matters for you, sometimes. But for this kind of embedded work, I'd say it's important far, far less than 1% of the time.
> I assume that all connections use 3.3V signal levels. Should I ever come across 5V signals, I would probably use a larger Julet M8 for differentiation. OpenWrt writes that some routers already ship with incompatible 1.8V or 2.5V logic. Should the time arrive that I need to work with those, I would probably use this little fella called Micro Q ....
At that point please, please, please just switch to the 4-wire configuration exposing V_IO (aka V_target aka ...). So many problems go away when you have that available. It's nice to have at 3.3V as well, but a lot of OTS 3.3V adapters aren't configured to make use of it.
on Aliexpress, Julet M6 pigtail pair is $2.99, while USB-TTL breakout (with female USB-C connector) is $1.40 - even for nice CP2102 with unique serial number.
If you have a portable device which needs serial connector, don't bother with weird connector and bespoke adapters which will get lost. Go for USB-C directly. As an extra bonus, USB-C is significantly more electrically rugged than raw UART input.
(this does not apply if you have some sort of multi-port serial port server, but those are pretty rare, and most likely use real RS232 voltages anyway)
If you have money to burn, get some LEMO push-pull connectors. They'll run you about $80 for a pair of each gender, but when you use them, you understand why trendy audio gear and instrumentation/optical OEMs use them. Occasionally when I make a project I treat myself and use them. Cheaper options are Mini XLR and Hirose HR10. Molex make a lot of wire-wire, but they're typically not rated for more than 25-50 cycles.
Those magnet couplings are nice. EDAC make a non-Aliexpress version, about $25 a pair.
And yes, they're magnetically polarized according to the datasheet. To answer the sibling comment, they're through-hole soldered but frustratingly are 2.7mm not 2.54 so you'd be squeezing a little into an existing header (ignoring clearance on the side anyway). I'm not sure why the 7-pin is obsolete, they make up to 5-way active.
I would tell anyone who is doing a new traditional serial connector/cable to add the following -
1. Automatic DCE-DTE detection and an interface which will rewire itself as needed to be the correct way, or you automatically know DCE vs DTE by connector gender.
2. Automatic Voltage Detection - 232 levels, TTL 5v, TTL 3v - and interfaces that are isolated enough to deal with the wrong voltage (clamping diodes or whatever), or different cable sizes for each.
3. Automatic type detection - TTL/RS-232, RS-422, RS-485, or different connector types by each.
Ideally I'd do this on a 8p8c or 10p10c connector, because of ease of making cables, with various resistance values across pins 1-8, or 1-10 to tell you what kind of interface it was.
That sounds completely useless. Who needs a standard that is not actually standardizing anything?
If you are willing to design a board with with 8p8c connector, identification resistor, and protection parts, then you might as well put RS-232/RS-422 transceiver. Those are very cheap nowadays, might even be cheaper than the connector itself. And you standardize on crossover cables, there will be no need DCE-DTE detection.
(Not that anyone would need this today. If you are connecting to existing serial port, there is USB, and USB-serial adapters are very cheap. And if you are designing a serial-based communication system, say for the robot, then CAN or RS485 are much better choices)
At that point it's not a debug connector anymore. Note there's a pseudo-standard for V.24/RS232 on RJ45 already, and nobody uses it for debug connectors since (a) you'd need a RS232 transceiver and (b) RJ45 connectors are honking huge.
The point is to shave off the last cent, which is why you get a possibly-unpopulated 1×4 or 1×3 2.54mm header. Bonus points if the manufacturer designed series resistors into the board (let's say 0402 or even 0201) and left those out too to save the last 0.01 cent.
What you are describing is going to be nightmare to work with - i.e. when you will have automatic detection of levels and it will decide to push RS232 into 3V3 MCU then you will have dead, maybe one of the kind prototype or dead expensive production device
The only things with a chance at succeeding in this space are putting USB-serial directly on the board + USB-C, or alternatively bluetooth classic RFCOMM profile. (The latter is a very long shot.)
Apart from that, this doesn't even touch upon the various voltage levels for logic-level serial ports, or the question of whether to Vref or not to Vref. (Or RTS/CTS.)
Ideally you wouldn't need to expose a TTL serial debug port to begin with. Maybe on a prototype you would want this but I'd rather just have a single connector that can expose everything (jtag or swd). Bonus points if the interface chip is on the board so it's just a USB port
FUSB307B is amazing for this, it has a hardware DBG output pin indicating that a USB-C debug accessory mode connection has been made. Perfect for connecting a mux and putting SWD & serial on some USB-C pins.
Too bad there's no standard for debug accessory connections. Also, at that point (putting a USB-C TCPC on your board) you might as well do full usbserial…
NB: there's no orientation detection in debug accessory mode.
Well the first problem you will hit I'd that very likely you will need to protect or isolate those lines from ESD. This will raise price of device and it will get denied just in these grounds.
ESD diode costs are nothing compared to the connector headers or even dealing with the mechanical complications of exposing Just One More Thing. Seriously, they're cheap.
I always wonder why Molex makes only black and white connectors. This makes it needlessly difficult to use different connectors for different purposes. Same with phoenix contact but there it's mostly green terminal blocks. Is it so difficult to just offer some more colors?
They’ll make them in any color you want if you order enough.
Adding more colors would multiply their logistic and inventory requirements, which are already hard with so many different connectors multiplied by number of pin options. Multiplying that even further by number of colors that most people would never buy isn’t a win.
I cannot count how many PCBs I did with various quick connect ideas to have a fast way to debug..
- Chop a PCI connector and have edge fingers on the PCB
- Skedd connectors
- Micro usb with a toggle switch or solder blob to switch between SWD/UART or USB
- Low profile usb-c and have D+/- as normal, and RX/TX over the accessory pins (like audio)
- Pogo pin clips
- GH1.25 connectors
- Tag-connect meh
- If thickness of pcb allows, your PCB can plug directly into a USB-A port (Thicc pcb) or if its too thin, it can plug into a male usb-c connector from a charger cable(might bend some pins though)
etc. etc.
So just like the author, anything but Dupont connectors ;)
I added one to my single-board computer enclosure, following FTDI's wiring. Now I can easily connect whenever I need to use the serial console, and a standard 3.5mm audio extension cable will let me reach across the room without moving my main computer. Replacement parts, if I ever need them, are cheap and easy to find.
Here's the pinout:
https://www.ftdichip.com/Support/Documents/DataSheets/Cables...
Granted, you shouldn't hotplug TTL serial, but everyone™ does it anyway. (In some situations you're even forced to, to avoid reverse powering something.)
But yes, if someone happens to be using their serial line for some kind of sensitive signaling, then I would agree that choosing a more isolated connector (or just avoiding hotplug) would make sense.
At that point please, please, please just switch to the 4-wire configuration exposing V_IO (aka V_target aka ...). So many problems go away when you have that available. It's nice to have at 3.3V as well, but a lot of OTS 3.3V adapters aren't configured to make use of it.
If you have a portable device which needs serial connector, don't bother with weird connector and bespoke adapters which will get lost. Go for USB-C directly. As an extra bonus, USB-C is significantly more electrically rugged than raw UART input.
(this does not apply if you have some sort of multi-port serial port server, but those are pretty rare, and most likely use real RS232 voltages anyway)
Those magnet couplings are nice. EDAC make a non-Aliexpress version, about $25 a pair.
</jk>
Less of a joke though: those aren't polarized, how do you not accidentally 180° them? Are they magnetically polarized or what?
[ed.: I didn't initially see it, they're mechanically polarized, one "short end" is flat, the other rounded.]
And yes, they're magnetically polarized according to the datasheet. To answer the sibling comment, they're through-hole soldered but frustratingly are 2.7mm not 2.54 so you'd be squeezing a little into an existing header (ignoring clearance on the side anyway). I'm not sure why the 7-pin is obsolete, they make up to 5-way active.
Or Adafruit? https://www.adafruit.com/product/5468
1. Automatic DCE-DTE detection and an interface which will rewire itself as needed to be the correct way, or you automatically know DCE vs DTE by connector gender.
2. Automatic Voltage Detection - 232 levels, TTL 5v, TTL 3v - and interfaces that are isolated enough to deal with the wrong voltage (clamping diodes or whatever), or different cable sizes for each.
3. Automatic type detection - TTL/RS-232, RS-422, RS-485, or different connector types by each.
Ideally I'd do this on a 8p8c or 10p10c connector, because of ease of making cables, with various resistance values across pins 1-8, or 1-10 to tell you what kind of interface it was.
If you are willing to design a board with with 8p8c connector, identification resistor, and protection parts, then you might as well put RS-232/RS-422 transceiver. Those are very cheap nowadays, might even be cheaper than the connector itself. And you standardize on crossover cables, there will be no need DCE-DTE detection.
(Not that anyone would need this today. If you are connecting to existing serial port, there is USB, and USB-serial adapters are very cheap. And if you are designing a serial-based communication system, say for the robot, then CAN or RS485 are much better choices)
The point is to shave off the last cent, which is why you get a possibly-unpopulated 1×4 or 1×3 2.54mm header. Bonus points if the manufacturer designed series resistors into the board (let's say 0402 or even 0201) and left those out too to save the last 0.01 cent.
Apart from that, this doesn't even touch upon the various voltage levels for logic-level serial ports, or the question of whether to Vref or not to Vref. (Or RTS/CTS.)
Too bad there's no standard for debug accessory connections. Also, at that point (putting a USB-C TCPC on your board) you might as well do full usbserial…
NB: there's no orientation detection in debug accessory mode.
Adding more colors would multiply their logistic and inventory requirements, which are already hard with so many different connectors multiplied by number of pin options. Multiplying that even further by number of colors that most people would never buy isn’t a win.
The alignment mechanisms on TagConnect/EdgeConnect negate any space advantage.
- Chop a PCI connector and have edge fingers on the PCB
- Skedd connectors
- Micro usb with a toggle switch or solder blob to switch between SWD/UART or USB
- Low profile usb-c and have D+/- as normal, and RX/TX over the accessory pins (like audio)
- Pogo pin clips
- GH1.25 connectors
- Tag-connect meh
- If thickness of pcb allows, your PCB can plug directly into a USB-A port (Thicc pcb) or if its too thin, it can plug into a male usb-c connector from a charger cable(might bend some pins though)
etc. etc.
So just like the author, anything but Dupont connectors ;)