Retronetworking

h1. T-DisplayTel

This is probably the only ISDN Phone with built-in [[BTX]] Terminal that was ever made.

h2. Pictures

FIXME

h2. Hardware

The hardware consists of 3 circuit boards:

* Mainboard

* Display Board

* Keyboard

h3. Mainboard

h4. Main CPU

* 1x NEC uPD70236AGD-12 (V53A) 16bit CPU (I101) attachment:UPD70236_NECElectronics.pdf

* 1x Samsung KM29C010-15 128k x 8 CMOS Flash Memory (I281) attachment:KM29C010.pdf

* 1x TI TMS27C240-12 256k x 16 EPROM (I271) atachment:TMS27C240.pdf

* 2x Samsung KM658128ALG-8  28k x 8 PSRAM (I231, I241) attachment:KM658128.pdf

* 24.576 MHz XTAL

* TI TL770SACP Supply Voltage Supervisor

h4. Keyboard Interface

* 1x OKI M80C51F-368 8-bit 8051 controller with 4k mask ROM (LOEWE 349-025514) (I611) attachment:M80C51F.pdf

* 11.0592 MHz XTAL

* connects to keyboard via 2x8 header (W651)

h4. Display Interface

* 1x OKI M6355 Display Controller (I361)

* connects to Display via 14pin FPC (W371)

* 1x Samsung KM62256BLG-10 32k x 8 CMOS SRAM (I381) attachment:KM62256.pdf

* 1x Motorola 34063AP1 KDI DC/DC connverter controller (for display supply voltage)

h4. ISDN Interface

* 1x Siemens SAB 82526 HSCX-1 V2.1

* 1x Siemens PSB 2186N V1.1 ISAC-S TE

h3. Keyboard

h3. Display board

h2. Making the BTX / Datex-J terminal work

@laforge has been attempting to find out what needs to be done to get the BTX/terminal part of T-DisplayTel to work.

On the D-channel signaling side, it is establshing a normal _UDI (unrestricted digital information)_, no specific HLC/LLC.

The more interesting question is what it speaks on the B-channel.

h3. Accepting the call, dumping raw B channel

When accepting a call (from the simulated BTX server side), the T-DisplayTel unexpectedly doesn't send HDLC but some 0xff-padding based signaling:

<pre>

00000000  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

00000040  ff ff ff ff ff ff ff ff  ff ff ff ff fb 05 7c 59  |..............|Y|

00000050  df ed f7 ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

00000060  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

00001f20  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff fb  |................|

00001f30  05 7c 59 df ed f7 ff ff  ff ff ff ff ff ff ff ff  |.|Y.............|

00001f40  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

00007ce0  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff fb  |................|

00007cf0  05 7c 59 df ed f7 ff ff  ff ff ff ff ff ff ff ff  |.|Y.............|

00007d00  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

0000daa0  ff ff ff ff ff ff ff ff  ff ff ff ff fb 05 7c 59  |..............|Y|

0000dab0  df ed f7 ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

0000dac0  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

00013860  ff ff ff ff ff ff ff ff  7e 01 5f d6 77 fb fd ff  |........~._.w...|

00013870  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

00019620  ff ff ff ff ff ff fb 05  7c 59 df ed f7 ff ff ff  |........|Y......|

00019630  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

0001f3e0  ff ff ff ff 7e 01 5f d6  77 fb fd ff ff ff ff ff  |....~._.w.......|

0001f3f0  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

000251a0  ff fb 05 7c 59 df ed f7  ff ff ff ff ff ff ff ff  |...|Y...........|

000251b0  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

*

0002b310  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ab ab  |................|

0002b320  ab ab ab ab ab ab ab ab  ab ab ab ab ab ab ab ab  |................|

</pre>

So we see two messages, with a lot of 'ff' padding in between:

<pre>

fb 05 7c 59 df ed f7

7e 01 5f d6 77 fb fd

</pre>

If we re-order the bits of each byte and then HDLC decode (e.g. with https://gitea.osmocom.org/cellular-infrastructure/osmo-e1-recorder/src/branch/master/src/hdlc-test.c) we get

<pre>

fffb057c59dfedf7ff => F 01 3f eb df F

ff7e015fd677fbfdff => F 01 3f eb df F

</pre>

so it's actually a repeat transmission of the same frame.

h3. establishing a call against an ISDN TA

Establishing call towards ISDN TA ([[US_Robotics_Sportster_ISDN_TA_Ext]]) and playing with the B-channel protocol on the TA, I can get a "proper" connection established when putting the TA into *T.70-NL* mode (@ATB22@).  No success with HDLC, X.75 or V.120.

When the T.70-NL call is established, the DisplayTel cleaars its screen, and any characters sent by the TA are displayed on the screen. Great.

However, no characters are transmitted from the DisplayTel in the opposite direction.  In fact, it doesn't even ever leave the 'ff' mode, i.e. no HDLC flag octets are sent by it.  No keypress on the DisplayTel generates any non-ff data to be transmitted on the B-channel.

I'm attaching raw B-channel traces for both directions (attachment:displaytel-32-rx.raw and attachment:displaytel-32-tx.raw)

Interstingly, in this situation, the uplink from the Displaytel only contains a single, non-ff message:

<pre>

df a0 3e 9a fb b7 ef

</pre>

Decoded (bit order inversal + HDLC decode):

<pre>

F 01 3f eb df F 

</pre>

The same behavior can be observed when replacing the hardware ISDN TA with a software implementation: isdn4linux/ttyI0 in BTX mode (AT&X1) on Fritz!Card PCI.