FreeFi-232 v2

[danielj]

As I think I mentioned elsewhere (I'm losing track of where I say anything now!) I started playing around with a proper design for a wifi modem, mainly because I found the closedness of the existing ones to be rather meh (especially as no one was building them), together with the fact that they were apparently not releasing sources which were based on GPL'd code.

Anyhow, after watching a "discussion" evolve on the VCFED forum around all the modem control lines being available, I figured it might be nice to make something complete including DTR/DSR/DCD and RING - all of which should be easy to implement via software. I'd really welcome some feedback on the schematic.

It's fairly straight forward - it implements the MAX3237 application note and the Arduino IDE ESP8266 programming circuit. Is it worth including a jumper to turn on hardware flow control and swapping the dtr/rts routing pads for jumpers? I'm prepared to make the PCB a bit larger than my original iteration, so jumpers instead of pads are feasible. Is a micro-usb a sensible power source, alongside a pair of power pins?

Anything else?

d.

[iainjh]

Cool idea daniel. Power pins are a must as some of us are reckless enough to want to tap power from inside the host computer rather than a usb psu

About how much current do these draw anyway?

[danielj]

About 250mA max I'd guess (the esp8266 is up to 170mA).

I think I might ditch the pads, a real modem would just give the pins it gives and it'd be up to the user to make the appropriate cable if their end requires something odd. I note that the arc is a bit peculiar.

d.

[danielj]

OK - another iteration - I've removed the pads to rejiggle the connections. This is a modem replacement, ergo it should be wired as such. If a computer needs special jiggery pokery in its connections, then those need to be made in the cable. I've included a jumper to select hardware flow control from the off.

[flibble]

By the time of the A5000/A3010/A3020/A4000/A4/A7000/RPC the serial port was normal enough to work with regular modems.

[danielj]

So what's the deal with the bizarre wiring diagrams in here?

http://chrisacorns.computinghistory.org ... AN/234.pdf

RI on the computer should be linked to RI on the modem, but instead it's being used as an output from the archimedes. Rather odd, no?

Or is it just that you need to use an odd cable?

d.

[flibble]

So what's the deal with the bizarre wiring diagrams in here?

Not sure, but a difference between earlier 'Archimedes' and later ones is RS423 vs RS232. But I can't find any diagrams that explain the difference between those two.

RI on the computer should be linked to RI on the modem, but instead it's being used as an output from the archimedes. Rather odd, no?

Or is it just that you need to use an odd cable?

I've not heard of anyone using a non standard modem cable

[danielj]

In which case a 9 to 25 on this should work fine then, I hope! Ultimately all of the additional lines will be implemented in software (DCD/DSR/DTR/RING), so it'll just be a question of turning them off or reconfiguring if they're misbehaving

[flibble]

Found this bit in the docs.

"Risc PCs and A5000s use an 82710 or 82711 peripheral controller, which provides the driver hardware for the serial port; as this IC is used in many IBM PC compatibles, it is possible to software-configure the computer to use the serial port in exactly the same way as an IBM PC compatible does. However, the system is pre-configured to behave as an Acorn machine, so Archimedes cables which require DCD and have flow control via DSR will work correctly without any reconfiguration. Some software communications packages have an option to reconfigure the serial
port on A5000s and Risc PC so that IBM-type cables can be used."

This seem to gel with my my half remembered memory that the later machines were more (or could be configured to be more) 'normal'

[danielj]

I love a healthy dollop of non-standard

d.

[danielj]

So, I tried making it smaller and then got angry at it, so I filled the empty space with a silk screen instead!

Anyway, I think I shall get some of these boards ordered and see if they can be made to work

d.

[1024MAK]

Nice

Ideally a modem should have it's interface compatible with the RS-232 (EIA 232, TIA 232) standards and with the V.24 and V.28 standards (it being DCE). As then, if the DTE has a standard RS-232 port, it can be plugged in directly or via a standard 'straight through' cable.

I very much prefer normal barrel power connectors (OD 5.5mm, ID 2.1mm) rather than mini or micro USB connectors. The micro USB connectors are not that durable IMHO.

If you have spare space on the PCB, maybe include some status LEDs (with resistors) to monitor the RS-232 data and control lines. SMD types would be fine.

Mark

[Elminster]

If enough real estate on board could have space for either barrel or micro usb?

[danielj]

There's not much space around the edges and it's not the easiest board to route unfortunately. The power pins can be used to wire on a barrel connector if so desired. Anyway, it's all a bit moot as the boards are currently in production . I'm not going to be selling these or building them, so people really are free to take the design and update it! As yet it's unproven though, so I'd hold fire until I've at least tested it works. Assuming it does I might have a go at rejiggling the layout.

d.

[Elminster]

As of 45 mins ago I don’t like barrel connectors anyway, having thought I picked up a 5v psu, when it in fact 12v. The item in question is now not working. Ho hum. Time to workout what I blew up.

[daveejhitchins]

As of 45 mins ago I don’t like barrel connectors anyway, having thought I picked up a 5v psu, when it in fact 12v. The item in question is now not working. Ho hum. Time to workout what I blew up.

I'm surprised these type of connectors haven't been colour coded - Can't believe the Chinese have missed such an opportunity!

Dave H

[1024MAK]

As of 45 mins ago I don’t like barrel connectors anyway, having thought I picked up a 5v psu, when it in fact 12v. The item in question is now not working. Ho hum. Time to workout what I blew up.

I can't believe you didn’t test with a multimeter for polarity and voltage first...

Mark

[Elminster]

As of 45 mins ago I don’t like barrel connectors anyway, having thought I picked up a 5v psu, when it in fact 12v. The item in question is now not working. Ho hum. Time to workout what I blew up.

I can't believe you didn’t test with a multimeter for polarity and voltage first...

Mark

Why would I? Polarity and voltage are correct. Or they would have been if I had picked up my 5v PSU, and not my 12v.

[1024MAK]

I learned the lesson many years ago at work. We had some telecommunications equipment (effectively a specialist auto-dialler for loop-disconnect systems). There were two versions, 12V and 50V. Both used 5 pin '180o DIN sockets. Guess who accidentally plugged the DIN to DIN connection lead into the 50V test socket on the 'Test Desk' with a 12V unit on the other end
There was a loud crack sound, and then total silence (these units tended to click their relays as if dialling out when first powered up).

I unscrewed the lid, only to find half of a 4000 series logic chip floating about. Some of others did not look very happy either. In the end, I had to replace every logic chip on the board

Whoops!

Luckily we had spare chips in stock, and the supervisor and manager never found out

So from then on, I have nearly always tested before connecting, unless I have already labelled the lead, or double checked that the lead connects to the PSU I think it should connect to...

Mark

[Elminster]

Moment of madness. The board I plug it into will take 12v with optional power transistor, caps and a jumper in a different place. Just I don’t have those.

[Elminster]

Added to the hardware list, looks like a first for Daniel.

[danielj]

We don't know if it works yet!

(Thanks! )

[Elminster]

Of course it will work. No pressure.

[danielj]

So - I've done the first test build of this Hooray!

It sort of not quite works from a programming front....

GPIO0 (iDTR on the schematic) needs to be at 3.3V for normal operation and pulled low for programming mode to be activated. For some reason it's sitting at around 2V which is nicely right on the edge of anything sensible and causing all sorts of glitching. I've got it connected to 3.3V via a 10k pullup (as per the recommended configuration for auto-programming with the arduino IDE), but it's stubbornly sitting there at the wrong voltage - this is when not connected to anything at all... Any ideas? the same pin on the esp12-E on my nodemcu is sitting at 3.3v (via 12k pullup), so it's clearly something wrong here. I can't see or detect any shorts

d.

[1024MAK]

Must be a current path to GND somewhere (via a resistance of around 15k ohms)...

Mark

[cmorley]

Internal configurable pulldown?

[danielj]

There is one, but it shouldn't be active on boot... Going to do a bit more testing today

[danielj]

Fixed - basically the flash was borked, it was booting badly and leaving that pin having a fight with the pull up. A bit of reflashing with esptool.py and it's back up and running properly

d.

[danielj]

Or not. Got to the bottom of it now though.

Right - I have a few version 1 PCBs, they will need a bit of fettling to work (a couple of tracks need cutting, a resistor needs adding, another resistor isn't needed, some chunkier wire is needed to beef up the power, and a couple of GPIOs need re-routing with kynar).

Is anyone interested in one of these PCBs for postage?

Parts needed: 100nF 0805 caps, 10k 0805 resistors, ESP-12E/F module, 25-pin female d-sub, some pin headers, an LM1117-3.3v reg, 2x10uF tant caps (case style B), kynar, a MAX3237. You need a micro-USB connector + little diode if you want to use one of those for power.

For programming you need a USB-Serial that does 3.3V, and has DTR and RTS outputs. The cheapy FTDI ones seem to work.

I'll post detailed modifications here if anyone wants one of these ones, and obviously contributions to the firmware, and testing of it are very welcome!

d.

[Elminster]

There are better people placed for doing this sort of stuff, but if you struggle to shift them, then I can always take one.