Another Bush internet.tv mod thread

[BubbleWrap]

nSIOCS1 and nSIOCS2 also have no track or via on the component side unless it is under the chip. We may need to lift off one of these SoCs and put it on your A7000 to find out what things are connected to!

That's quite severe.

Did you make a trace-finder brush like @mikeselectricstuff showed? Trace back from the CS on the modem eeprom.

The modem EEPROM and SRAM appear to be driven entirely by the 1670T itself, they aren't connected directly to the ARM.

Pins 44 and 45 of the 1670T are connected to pin 200 (nEASCS) of the ARM7500.
I don't have a datasheet for the 1670T, but this is the only signal I've found so far that looks like a CS.

[IanS]

Do we have a dump of the EPROM/Flash? (Does it look like a podule rom, I suspect it isn't)
It may be that the modem is connected with a serial connection to the smsc IO controller.

[thecellartroll]

Did you make a trace-finder brush like @mikeselectricstuff showed? Trace back from the CS on the modem eeprom.

Not yet. I'll need to make one from a fluffed out mains cable or something.

[philpem]

I had a google around on my lunch break and found a kernel patch to run Linux on the IBX: http://www.kyllikki.org/software/armlinux/

Looks like the IR keyboard is sat on COM2.

Wild guess, maybe the modem is on COM1. Picking apart the NCOS ROM will probably give that away (look for whatever the dialler is configured to talk to).

Re continuity testers - I think Mike Harrison (of Mike's Electric Stuff and Whitewing Logic fame) did a short video on DIYing one. It was some kind of steel wire brush, I can't remember what for but possibly something for a Dremel tool or abrasive pencil. They apparently work quite well for this kind of "brush" testing.

The brush-type continuity testers used to be a standard item in electronics catalogues, but I haven't seen them on sale for some years. Maybe they were a Wavetek/Meterman or Vann-Draper product?

Edit - oh, Mike's video was mentioned back on page 3! viewtopic.php?p=374771#p374771

Edit 2 - The RISC OS NC ROMs from Myelin's post (viewtopic.php?p=266923#p266923) seem to unpack okay using the RISC OS 3.20 tools (viewtopic.php?t=14110). You'll need to increase the Wimpslot to about 8.5MB in the !ExtractMod Obeyfile, and possibly change the filename of the ROM image in ExtractMod BASIC program. The output from running the Obey file is a directory full of modules.

DiallerB seems to be responsible for running dialler scripts.
ISPCapture seems to be at least partly responsible for ISP configuration.
LED seems to handle the LEDs, via various interfaces.

There's a "Modem" module in the NCOS ROM. No clue if that's the modem driver or something else. Perhaps the modem isn't a podule at all? It seems to respond to one command,

Code: Select all

*ModemInfo

.

There's also a "SoakTest" module which has a tantalisingly-named "parallel bootstrap" (possibly a way of loading code over the parallel port?) and a "TestStart" module which claims to detect a parallel-port selftest dongle.

[IanJeffray]

Perhaps the modem isn't a podule at all?

I can say that on the NC at least, it definitely is, as at Oregan we created custom ROMs for it - but it really was just a standard Archimedes podule back then, so couldn't have used the serial port interface on the 7500, so maybe simplified for the STB. Good shout.

[thecellartroll]

There's a "Modem" module in the NCOS ROM. No clue if that's the modem driver or something else. Perhaps the modem isn't a podule at all? It seems to respond to one command,

Code: Select all

*ModemInfo

.

Interesting that it is in the NC ROM! That module disappeared when I pulled the little PROM on the bottom left of the board, which I assumed was for the modem.

[BubbleWrap]

Do we have a dump of the EPROM/Flash? (Does it look like a podule rom, I suspect it isn't)
It may be that the modem is connected with a serial connection to the smsc IO controller.

I've attached what my EEPROM programmer read out of it.

The 1670T appears to be connected to La[2..4] and BD[0..15]. There's no serial connection I can find.

[IanS]

So it looks like it's effectively an ISA card.

RISC OS won't know anything about that, afaik.

[thecellartroll]

RISC OS won't know anything about that, afaik.

I wonder how to stop RISC OS from seeing the podule interface as fully occupied.

[philpem]

Interesting that it is in the NC ROM! That module disappeared when I pulled the little PROM on the bottom left of the board, which I assumed was for the modem.

Well, to be clear -- it's in the "-11" version Myelin posted!

At a guess - that might be because it didn't load because it couldn't detect the modem... which makes sense if the modem ROM had been removed.

I've attached what my EEPROM programmer read out of it.

Yep, that's modem code. Full of AT commands,

The 1670T appears to be connected to La[2..4] and BD[0..15]. There's no serial connection I can find.

That and this string kinda gives it away... the modem implements a serial port of its own, over the PC ISA bus:

Code: Select all

Steve Kerner Ltd. Ver:02-00-01 Last compiled:19:12:59 12/11/1999
Venus 33.6 ISA I052099V

It probably has a chip-select pin which is linked up to an LA or CS-decode. I doubt it's a Podule. I mean, why implement it as a Podule if you don't have to?

There's a product sheet for the chip here: https://www.digchip.com/datasheets/down ... umber=1673

Looks like Steve Kerner Ltd / Kerner Telecommunications was a consultancy, specialising in modem design: http://web.archive.org/web/200502141135 ... erner.com/



The phantom Podules are probably from a data bus pull resistor being missing. One of the Acorn service manuals (possibly the RISC PC one or the Podule spec) talks about this. The minimum Podule has a single-byte ID, but a certain bit must be set low for the Podule to be detected as being present, and only when it's selected (otherwise you get phantoms in other slots, or worse).

[thecellartroll]

The phantom Podules are probably from a data bus pull resistor being missing. One of the Acorn service manuals (possibly the RISC PC one or the Podule spec) talks about this. The minimum Podule has a single-byte ID, but a certain bit must be set low for the Podule to be detected as being present, and only when it's selected (otherwise you get phantoms in other slots, or worse).

Like the unpopulated LK that I found, which if made would have pulled one pin low via a resistor

[thecellartroll]

Not that anyone will be surprised, but adding the diodes to the PS2 socket does not change behaviour at all:
DEL power-on does not work properly, though it does do something.

The system will not restart from a Ctrl-Break with a mouse plugged in.

If anyone else wants to add the diodes, do it before you solder the PS2 socket in.

[thecellartroll]

Similarly making LK8 does nothing at all

[IanJeffray]

Similarly making LK8 does nothing at all

What does LK8 do? How did you know?

[thecellartroll]

What does LK8 do? How did you know?

No idea what it is meant to do. Just an experiment to see if the resistor it is connected to, which pulls one pin to ground as Philpem described. Twas a (very) long shot that wasted about 30 seconds of my life

[IanJeffray]

What does LK8 do? How did you know?

No idea what it is meant to do. Just an experiment to see if the resistor it is connected to, which pulls one pin to ground as Philpem described. Twas a (very) long shot that wasted about 30 seconds of my life

Oh. Hm. Be careful. You checked it had "some function" (a resistor) but note that Acorn in general have many LK positions on boards which are just power and ground probe points - you can short things out by just shorting links.

What's the current status with 3.7? Still not managed to get a clean CMOS reset happening? But keyboard is ok? Mouse is ok? Video is...?
So we could do with a custom 3.7 ROM with, ideally, !Zip in it, to allow easier further faffing? Zip's not small, but I suspect there's plenty of modules we can throw out to make room.

[thecellartroll]

Oh. Hm. Be careful. You checked it had "some function" (a resistor) but note that Acorn in general have many LK positions on boards which are just power and ground probe points - you can short things out by just shorting links.

Noted; it wasn't really a sensible plan.

What's the current status with 3.7? Still not managed to get a clean CMOS reset happening? But keyboard is ok? Mouse is ok? Video is...?
So we could do with a custom 3.7 ROM with, ideally, !Zip in it, to allow easier further faffing? Zip's not small, but I suspect there's plenty of modules we can throw out to make room.

Keyboard and mouse still don't play nicely together. It is not possible to restart the machine with a crtl-break if there is a mouse plugged in, or if the keyboard is plugged into the "mouse" socket.

No clean CMOS RAM reset happening. Not surprising as I haven't changed anything there.

Video is ATM still connected via a BBC-B RGB cable. I've direct soldered a wire to the un-routed HSYNC pin on the SoC and my scope says there is a nice square wave there, so you can guess my next plan.

[thecellartroll]

Much oddness. Connected to the hideous AKF18 I get the normal pink then blue solid screen followed by shortly by the RISC OS 8mb and Keyboard Not Found scrolling vertically.

Obviously the sync mode isn't right. Thing is, the keyboard doesn't work at all any more so I can't change it!

Switch back to RGB on SCART and the keyboard is back. Beyond me

[thecellartroll]

https://youtu.be/ZhLcTEFBh90

[thecellartroll]

Even more curious...

A really nice picture on my Acer monitor and the keyboard works too. Most screenmodes unavailable though as it will only work with Monitor Type set to 0. Setting it to 4, as it should be for SVGA I think, results in a blank screen. This is a 15khz compatible monitor.

[BeebMaster]

What part of RISC OS gives the icon bar? On my NC the Desktop is just a grey screen with mouse pointer and I don't know how I can get the icon bar to show up, if at all. There's no F12 or CTRL-F12 function either. I can start tasks in a single-task window within the "Desktop" using *DESKTOP BASIC or *WIMPTASK GOS etc. but that's about all I can make it do.

[thecellartroll]

What part of RISC OS gives the icon bar? On my NC the Desktop is just a grey screen with mouse pointer and I don't know how I can get the icon bar to show up, if at all. There's no F12 or CTRL-F12 function either. I can start tasks in a single-task window within the "Desktop" using *DESKTOP BASIC or *WIMPTASK GOS etc. but that's about all I can make it do.

There is a module called "Iconbar" so I'm guessing that.

[BeebMaster]

Interesting, it's unplugged in my RISC PC (3.70) so it didn't show up in *MODULES but the icon bar works as normal!

I can't see it in RISC OS NC 1.11 though, which makes sense as there is no icon bar.

[IanJeffray]

There is a module called "Iconbar" so I'm guessing that.

No, that's the Toolbox module that provides iconbar functionality for applications.
The WindowManager module is what provides the iconbar itself. There are various levels of pockling of WIMP for the NC. I just load the latest Acorn WindowManager 3.98 module softload.

[thecellartroll]

An update on VGA out.

Monitortypes 0 and 8 are still the only ones that work.

Investigations show that *configure monitortype 1-4 actually turn video output off altogether! *configure sync 0 or 1 alternates whether the sync is held high or low, but in any of those monitor configurations there doesn't appear to be any actual signal on HSYNC, VSYNC, R, G or B. Weird. Does RISC OS actually require the ID0 line to be connected for VGA monitors to work?

[steve3000]

An update on VGA out.

Monitortypes 0 and 8 are still the only ones that work.

Investigations show that *configure monitortype 1-4 actually turn video output off altogether! *configure sync 0 or 1 alternates whether the sync is held high or low, but in any of those monitor configurations there doesn't appear to be any actual signal on HSYNC, VSYNC, R, G or B. Weird. Does RISC OS actually require the ID0 line to be connected for VGA monitors to work?

How is the VIDC20 clock circuitry different to the PLL used in RPC/A7000? What you describe sounds like no clock is selected when you switch to VGA so the VIDC20 just stops....

[baz4096]

In this thread you mention this link:

http://www.kyllikki.org/hardware/bushstb/mod-sound.html

Components need to be
X2 Crystal Oscillator 11.2896 MHz
(X3 This is the HCLK Xtal for VGA monitor timing and not for sound)
IC1 Some sort of 74HCxx TTL - not sure of type
IC2 Philips TDA1545A DAC (Crickewood Electronics sell an equivilent)
IC3 Line Driver TDA1308 or similar
C6, C13 Ouput Capacitors 100uF 16 v ?
PL1A 3-pin header for sound output (line/headphone levels)
Plus some other resistors and capacitors...

This implies there’s an empty space for a VGA crystal on the IBX pcb?

[baz4096]

In this thread you mention this link:

http://www.kyllikki.org/hardware/bushstb/mod-sound.html

Components need to be
X2 Crystal Oscillator 11.2896 MHz
(X3 This is the HCLK Xtal for VGA monitor timing and not for sound)
IC1 Some sort of 74HCxx TTL - not sure of type
IC2 Philips TDA1545A DAC (Crickewood Electronics sell an equivilent)
IC3 Line Driver TDA1308 or similar
C6, C13 Ouput Capacitors 100uF 16 v ?
PL1A 3-pin header for sound output (line/headphone levels)
Plus some other resistors and capacitors...

This implies there’s an empty space for a VGA crystal on the IBX pcb?

[thecellartroll]

Oh yes! I think you guys may have cracked it

[thecellartroll]

This implies there’s an empty space for a VGA crystal on the IBX pcb?

There is a space for the XTAL. From the RPC schematic it looks like it should be 24Mhz.