C*NET Audiocodes MP112

[ThomasCH]

Hallo zusammen
Ist das CNET ein Begriff? Ein Netz wie iTelex, aber für Telefone weltweit www.ckts.info . Ich bin da zwar schon länger dabei, nun konnte ich aber mit der grossen Hilfe eines Telefonsammlers meine Geräte wieder richtig nutzen. Zuerst hatte ich den CISCO SPA 112 für zwei Linien. Leider kann der aber weder Gebührenimpulse erzeugen noch Impulswahl verarbeiten. Mit dem Audiocodes MP112 ist das alles wieder möglich. Das Münztelefon kassiert wieder, wie zu alten Zeiten..... Der MP112 ist aber eine echte Knacknuss um ihn richtig zu Konfigurieren. Ohne Hilfe wäre ich nie zum Ziel gekommen. Unglaublich was man da alles definieren kann... und nur wenn alles richtig ist, dann gehts auch.
Für mehr Infos bzw Kontaktdaten einfach bei mir melden. Das CNET ist doch eine hübsche Ergänzung zum iTelex.....

Viel Grüsse
Thomas

[M1ECY]

There is talk from the UK C*Net group of connecting Telex devices to the network - I do not know how serious this possibility is, but before long here in the UK, this will be the only way to use old technology telephones.

I joined the UK group a few years ago, but as yet have not made any progress towards building a system

[DF3OE]

The idea of using a VoiP net for data communications (even 50 Baud Telex) is bullshit (sorry). We have enough proof that is does not work.
To use direct IP connections as i-Telex uses is ingenious. It has been the best idea of Fred ever.
And the success and reliablity of the i-Telex network speaks for itself.

[Fernschreiber]

I can't harmonise to hennings statement completely. Of course an VoIP audiopath is not the first choice for transmitting data. But if modems with approval are used, it works quite well, but is sensiive to biterrors under certain circumstances. I made lots of tests with the old analog telex-system (when I got a second one) and those old 50 baud modems of that time. Works all well in local or not-local VoIP-mode. The fact that the modern i-Telex analog card does not work so well may have it based in having no approval. Just little deviation in levelling the tones, pre/ deemphasis will lead to overrun the codecs (sum of power) , bad adjustment of hybrid (echoes) or stress the echo-compensation which is optimised for for a fixed amplitude range. Between Henning and me is still an open issue regarding setting up an analog system (Einkanal-Modem) like in the old days 1930's when telex startet on existing telephonelines. I set up a one-channel-modem by plans and with equipment of the old days. I'm still waiting for Hennings system as a counterpart to send and receive data to/from. I leveled everything including some safety-space according to the todays level-plan of VoIP- systems. After some testing it was clear, that only g.711 codecs can can do that job (ISDN-Quality) and level has to be kept as low as possible to keep echoes away or get them under the limit of recognition. In that szenario the only bad guy was my wireless-LAN access. Moving direct to my analog interface on the router, there were 99% success. I have a terminal-station somewhere in the Telekom Telefone network (fritzbox) that works as an analoge answerunit . But instead having an voicefile to listen to, there is an audiofile I created with my moden and stored it there. I can even send data to be stored on that unit and can read back. I test that from time to time and it still works fine.
But to be honest: using VoIP connection with nonvoice sound today is still full of risk. The terminals will handle down to the lowest common codec if not even forced by the provider for keeping traffic low. On international calls there is no guarante for a seemless datastream (length of path can change rapidly), jitter is a great enemy. Even different properties like my-law or a-law used in g.711 codecs (european/us version) don't make it easyer.
But for a hobby-network it is still good as silver(not gold) to make connections between terminals. It gets you back to the very ols days when guy's tried to connect continents with radio or simple cabels through the ocean.
For me as a specialist in transmission-technique it woud be much fun to find someone with an old modem of that time (1,5 or 1,7 Khz) somewhere, set up an direkt VoIP channel whithout telco provider between and try to connect two teletypewriters that way. Hopefully it's not only a dream.
So generally telling VoIP is "bullshit" for data-transmission is not on my line. With some basic knowhow it is useable quiet well. But that old knowhow is going to vanish unluckily day by day but the physics of transmission (copper or fiber or wireless, its all analog on layer0) do not.
Regards

Willi

[Casandro]

Well Data over VoIP works fine, if you have a decent Internet connection and you neither use terrible codecs nor terribly bad ATAs. Cisco and Grandstream are the worst offenders when it comes to quality. I don't know about older devices from Audiocodes, but at least the ones with ISDN ports seem to work.

At least in Germany we are in a lucky situation as the most common router series (Fritz!Box) actually is a very decent ATA/IAD.

The main problem (for voice and data) is that VoIP has no sample clock, so both ends of a conversation will drift in time. Bad ATAs will just drop and add voice frames to hide that problem. Professional devices have a lock input so you can sync them to a frequency standard. Decent devices will have a decently accurate clock, but measure the average rate the packets from the other side come and adjust their clock accordingly.

Then there is Grandstream. Somehow they manage not not only have a severely wrong clock, but also somehow manages to move every voice frame in time by a random amount, with a means that jumps in steps every minute or so. Do not expect a device like a Grandstream HT802 to be useful for anything but silence.

[Fernschreiber]

Hello,
my experience as a professional and in private use showed , that using those ATA's in local networks to realise e.g. centralised voicerecording on analog PABX systems was always accepted, the quality was good enough for that purpose. Wide area systems are indeed a greater challenge. But even the worst ATA's available work just fine in pure audiomode when in use by people. Dropimg packets or wrong values do not interfear the call.
Human way of listening is extremly tolerant regarding errors. In addition lots of drops happen in the time of silence, conversation is mostly halfduplex.
But on the other hand even low amplitude echoes are recognised as interference when time exeeds a certain value. That's a new problem caused by bad hybrids and high level signals delayed by distance and routing-servers, it is no realtime anymore (there it was called sidetone masking). GSM-Codecs are (especially halfrate codecs) much more worse but calls can be done. Halfrate codecs were never used much, people conplained too much. so for humans we just need a certain recognition of sillables. The words we believe to hear are produced by our brain. This works excellent whith people we know and still good enough with those we never talked to. That's the big difference when VoIP is used for data. Loss of pakets can not be equalised and the carrier with its information (phase/amplitude) is missing causing biterrors. Errorcorrection can only be used by reducing speed because bandwidth is small. Same effects can be found in video. The illusion of seeing (moving) pictures is and was eversince brainmade, the errors can even be higher than in audio.
In the IP-world we have to use UDP for so called "realtime" services, because TCP with it's theeway Handshake is too slow even for the todays bitrates available. We wouldn't miss a paket, but delayspread is extremely and to sort in a longjorney paket ( or requested again) into an existing stored array takes too much time done by the application-layer. To drop is the best and easiest way.
Back to the telexspeed. 50 Baud is quite small and the used bandwidth is around 80 Hz. In the old days up to 24 of those channels were put into one telephonechannel with 3.1KHz bandwidth. This channel was realised by carrierfrequency or 64Kbit/s PCM. Using a G.711 channel (64kbit/s) via UDP in the IP-world means around 100-120kbit/s. This bandwidth for just one 50baud telexapplication is far oversized. A few dropped packets are not the problem.
Also a small jitter is tolerated , we have a symboltime of 20ms. The only real disturbance is the point when caused by reflected signals an echo can grow and then travels between the terminals until both echocancel-units find a way (interrupt path) to stop it. Non of the old days modems can cope with that effect. A very accurate levelplan and defined impedance is essential in the analog world even today keeping quality (phone or data) good enough.

Regards
Willi

[M1ECY]

The idea of using a VoiP net for data communications (even 50 Baud Telex) is bullshit (sorry). We have enough proof that is does not work.
To use direct IP connections as i-Telex uses is ingenious. It has been the best idea of Fred ever.
And the success and reliablity of the i-Telex network speaks for itself.

Quite agree Henning,

The C*Net users in the UK that have suggested this is possible, to my knowledge have not actually managed a connection - even if this is possible, I can only imagine under v.21 protocol - at least this is at audio.

I think the reasoning behind the interest in using c*Net for Telex is based on the already existing equipment at the collector's home, and less based on technical feasibility.

Cheers
Sean