The Definitive Collection of V.90/V.92 Dial-Up Modem Sounds

Nice work.

I had no idea the old dialup modem was so complicated.

There is a generation that has no idea of what dial-up means, so it's nice that history is captured for the day they want to explore.

Mark

You think that was fun, I found the various AT commands more intriguing with so many variations. I kept a few sheets of script notes around for help.

Clem

Yep been and played with a few of them.

Luckily we were mainly dialling out in a relatively dumb mode, but we did need to start locking DTE rates when modems climbed out of the 2400 baud age. ....

Mark

I mainly was changing modems DIL to pick optimal exchanges since some of their choices conflicted with our Robotics X2 technology and failed to connect.

You may have a grudge against the vendors and their semi-arbitrary choices of AT commands for the same "features", but this came about because the company that first bought in the smart-modem (Hayes) essentially lost market leadership and control of the command set, and others were probably afraid of litigation due to copying the other vendor's commands (e.g. the +++ escape sequence resulted in the semi-compatible +++AT "time independent" escape sequence, aka TIES). The later modems, not hamstrung by the expense of ROM/EEPROM/Flash memory, started replying nicely if you enquired it about its value (AT+MS?) and also responded as to the values it would accept (AT+MS=?), so you could poke around even without the AT command reference.

at+ms=?

+MS: (0,1,2,3,9,10,11,56, 64,69),(0,1),(300-33600),(300-56000),(0,1),(0,1)

OK

On the whole, at least dealing with later modems, the important commands are easier to remember because they became less bit-mapped S-register values, and more "actual" values with onboard help. Modulation-select commands were common to "help" stabilize modems by limiting their connect rate, and since most modems were Rockwell/Conexant based, the first port of call was AT+MS, whereas the Lucent/Agere had AT-V90 and some others had no options at all.

Modems also had nice diagnostic output, at least some of them, but again, different chipsets had different tendencies. USRobotics/Texas Instruments based modems were ATI6 and ATI11 with ATY11 line frequency-attenuation readout capability that no other modems had, whereas Lucent/Agere stuck with ATI11, Broadcom was ATI6 & ATI11. Oddball Rockwell/Conexant had older modems responding to AT&V1 (as did Motorola), and later modems with AT#UG. Ambient/Intel chose ATI20.

e.g. ATI6, ATI11 from TI chipset

ati6

 

U.S. Robotics 56K Professional Message  Link Diagnostics...

 

Chars sent                    3      Chars Received               82

Chars lost                    0

Octets sent                   3      Octets Received              82

Blocks sent                   3      Blocks Received               4

Blocks resent                 0

 

Retrains Requested            0      Retrains Granted              0

Line Reversals                0      Blers                         1

Link Timeouts                 0      Link Naks                     0

 

Data Compression       V42BIS 512/16

Equalization           Long

Fallback               Enabled

Protocol               LAPM

Speed                  42666/28800

V.90 Peak Speed        42666

Current Call           00:00:11

 

Online

 

OK

ati11

 

U.S. Robotics 56K Professional Message  Link Diagnostics...

 

 

Modulation                  V.90

Carrier Freq     (Hz)       None/1920

Symbol Rate                 8000/3200

Trellis Code                None/64S-4D

Nonlinear Encoding          None/ON

Precoding                   None/ON

Shaping                     ON/ON

Preemphasis      (-dB)      0/4

Recv/Xmit Level  (-dBm)     8/16

Near Echo Loss   (dB)       2

Far Echo Loss    (dB)       0

Carrier Offset   (Hz)       NONE

Round Trip Delay (msec)     585

Timing Offset    (ppm)      18244

SNR              (dB)       41.9

Speed Shifts Up/Down        0/0

Status :                    ua,5,12N,12.2,-8,0N,0,41.2,23.7

OK

e.g. ATI11 from LT/Agere

ati11

 

 

    Description                         Status

    ---------------                     ------------

    Last Connection                     NA             

    Initial Transmit Carrier Rate       NA             

    Initial Receive  Carrier Rate       NA             

    Final   Transmit Carrier Rate       NA             

    Final   Receive  Carrier Rate       NA             

    Protocol Negotiation Result         NA             

    Data Compression Result             NA             

    Estimated Noise Level               NA             

    Receive  Signal Power Level  (-dBm) NA             

    Transmit Signal Power Level  (-dBm) NA             

    Round Trip Delay             (msec) NA             

Press any key to continue; ESC to quit.

 

    Description                         Status

    ---------------                     ------------

    Near Echo Level              (-dBm) NA             

    Far  Echo Level              (-dBm) NA             

    Transmit Frame Count                0              

    Transmit Frame Error Count          0              

    Receive  Frame Count                0              

    Receive  Frame Error Count          0              

    Retrain by Local  Modem             0              

    Retrain by Remote Modem             0              

    Call Termination Cause              0              

    Robbed-Bit Signaling                NA             

    Digital Loss                   (dB) NA             

    Remote Server ID                    NA             

 

OK

e.g. AT#UG from Conexant HSFi

at#ug

<VERSION=10>

<CALL SETUP RESULT=Detected Data Answer Tone>

<MULTIMEDIA MODE=Data Only>

<DTE-DCE MODE=Async data>

<V.8 CM octet string="">

<V.8 JM octet string="C16513948">

<RX SIG POWER DB=16, TX SIG POWER DB=18, S/N RATIO DB=40>

<NORMALIZED MEAN SQUARED ERROR=48>

<NEAR ECHO LOSS=6>

<FAR ECHO LOSS=200>

<FAR ECHO DELAY=177>

<ROUND TRIP DELAY=177>

<V.34 INFO bit map =0>

<TX MOD=V.90, TX SYMBOL RATE=3200, TX CARRIER FREQ=1920>

<RX MOD=V.90, RX SYMBOL RATE=8000, RX CARRIER FREQ=0>

<TX FIRST DATA RATE=28800, RX FIRST DATA RATE=46667>

<CARRIER LOSS=0, RATE RENEG=0>

<RETRAINS REQ=0, RETRAINS DET=0>

<TX FINAL RATE=28800, RX FINAL RATE=46667>

<PROTOCOL NEGOTIATION=LAPM>

<EC FRAME SIZE=80>

<EC LINK TIMEOUTS=0>

<EC LINK NAKS=0>

<COMPRESSION NEGOTIATION=V42BIS>

<COMPRESSION DICTIONARY SIZE=800>

<TX FLOW CONTROL=2>

<RX FLOW CONTROL=2>

<TX CHARS SENT FROM DTE=9>

<RX CHARS SENT TO DTE=3F>

<TX CHARS LOST=0>

<RX CHARS LOST=0>

<TX FRAME COUNT=8>

<RX FRAME COUNT=4>

<TX FRAME ERROR COUNT=3>

<RX FRAME ERROR COUNT=0>

<TERMINATION CAUSE=No previous call>

<CALL WAITING EVENT COUNT=0>

<LSD Vdd=56806200, Power Bits=362CB38>

<V8BIS FLEX VERSION=0, V8BIS DSP VERSION=0>

<RATE RENEG REQ=0, RATE RENEG DET=0>

<HIGH PASS FILTER ENABLED=0>

<ROBBED BIT PATTERN=EB>

<DIGITAL PAD DETECTED=0, DIGITAL LOSS ESTIMATE=0>

<Cnt1=0, Det1=0, Res1=0>

<EQM=30, TRAINING EQM SUM=30>

<NEGOTIATION TIME=22854>

 

OK

Believe it or not, Microsoft tried to have a hand in this diagnostic business, and tried to standardize a command to dump modem diagnostics. The command is AT#UD for "unimodem diagnostics", but the problem was (like in SMART) they never standardized anything aside from the dump format, so you end up with cryptic Key + Value pairs which you can only decode if the driver contains the strings which define what each key meant - and some modem manufacturers never got that far. Others really goofed up, and lost all the stats once the call was hung up, so that's very useful!

e.g. AT#UD output from the SL2800:

at#ud

OK

DIAG <2A4D3263 00=FF 01=F 02=F 03=F>

DIAG <2A4D3263 04=000000000000000000000000>

DIAG <2A4D3263 05=000000000000000000000000>

DIAG <2A4D3263 16=FFFF 17=FF 18=ffffffff>

DIAG <2A4D3263 20=ff 21=ff 22=ffffffff>

DIAG <2A4D3263 23=ffffffff 24=ffffffff>

DIAG <2A4D3263 25=ffffffff 26=ffffffff>

DIAG <2A4D3263 27=ffffffff 30=FF 31=FF>

DIAG <2A4D3263 32=FF 33=FF 34=FFFF>

DIAG <2A4D3263 35=FFFF 40=FF 41=FFF>

DIAG <2A4D3263 42=FF 43=FF 44=FF 45=FFFF>

DIAG <2A4D3263 50=F 51=F 52=ffffffff>

DIAG <2A4D3263 53=ffffffff 54=ffffffff>

DIAG <2A4D3263 55=ffffffff 56=ffffffff>

DIAG <2A4D3263 57=ffffffff 58=0 59=0>

DIAG <2A4D3263 60=01 61=FF 128=ffffffff>

On the whole, I very much liked AT commands for their simple in-band management and the fact that the S-registers reminded me of just how simple you can make your gear. Changing the S-registers are literally changing the numbers by which the modem code would use to perform its routines, and the resulting changes sometimes helped users get better compatibility of their modems (e.g. Apple users with an adapter cable use DTR signal differently to PC users, or whether the modem reported DCE to DCE or DTE to DCE speeds and what return codes it would supply), or even push the speeds (I got 54667bps connect just *once* in my lifetime, but even doing that for a fraction of a minute was an achievement).

My biggest regret at this stage was not having the opportunity to document 3com/USR's x2 modulation. In Australia, I don't remember x2 being very popular, and the number of high quality x2 handshake recordings are nought (at least, to my knowledge). The number of x2 modem banks are probably very low as well, as only 3com answer equipment has a chance of supporting it, and no ISP readily advertises their gear nor their capabilities aside from the ITU standards. I never got a chance to carefully examine K56Flex either for that matter, or record it, but I do clearly remember K56Flex having 2k connection increments (44, 46, 48, 50, 52k being common speeds to see) whereas V.90 changed that to 1.333k so we got 44, 45.3, 46.6, 48, 49.3, 50.6, 52k, so an early way of telling if your modem was V.90 was to see whether it had 3's and 6/7's in the connect speed line (although if you connected at the common rates, you couldn't tell).

Fond memories ... very fond memories. I was on dial-up a lot longer than I should have been, partly because of the house-hopping and renting, but it was mostly a reliable partner.

- Gough

You may have a grudge against the vendors and their semi-arbitrary choices of AT commands for the same "features", but this came about because the company that first bought in the smart-modem (Hayes) essentially lost market leadership and control of the command set, and others were probably afraid of litigation due to copying the other vendor's commands (e.g. the +++ escape sequence resulted in the semi-compatible +++AT "time independent" escape sequence, aka TIES). The later modems, not hamstrung by the expense of ROM/EEPROM/Flash memory, started replying nicely if you enquired it about its value (AT+MS?) and also responded as to the values it would accept (AT+MS=?), so you could poke around even without the AT command reference.

at+ms=?

+MS: (0,1,2,3,9,10,11,56, 64,69),(0,1),(300-33600),(300-56000),(0,1),(0,1)

OK

On the whole, at least dealing with later modems, the important commands are easier to remember because they became less bit-mapped S-register values, and more "actual" values with onboard help. Modulation-select commands were common to "help" stabilize modems by limiting their connect rate, and since most modems were Rockwell/Conexant based, the first port of call was AT+MS, whereas the Lucent/Agere had AT-V90 and some others had no options at all.

Modems also had nice diagnostic output, at least some of them, but again, different chipsets had different tendencies. USRobotics/Texas Instruments based modems were ATI6 and ATI11 with ATY11 line frequency-attenuation readout capability that no other modems had, whereas Lucent/Agere stuck with ATI11, Broadcom was ATI6 & ATI11. Oddball Rockwell/Conexant had older modems responding to AT&V1 (as did Motorola), and later modems with AT#UG. Ambient/Intel chose ATI20.

e.g. ATI6, ATI11 from TI chipset

ati6

 

U.S. Robotics 56K Professional Message  Link Diagnostics...

 

Chars sent                    3      Chars Received               82

Chars lost                    0

Octets sent                   3      Octets Received              82

Blocks sent                   3      Blocks Received               4

Blocks resent                 0

 

Retrains Requested            0      Retrains Granted              0

Line Reversals                0      Blers                         1

Link Timeouts                 0      Link Naks                     0

 

Data Compression       V42BIS 512/16

Equalization           Long

Fallback               Enabled

Protocol               LAPM

Speed                  42666/28800

V.90 Peak Speed        42666

Current Call           00:00:11

 

Online

 

OK

ati11

 

U.S. Robotics 56K Professional Message  Link Diagnostics...

 

 

Modulation                  V.90

Carrier Freq     (Hz)       None/1920

Symbol Rate                 8000/3200

Trellis Code                None/64S-4D

Nonlinear Encoding          None/ON

Precoding                   None/ON

Shaping                     ON/ON

Preemphasis      (-dB)      0/4

Recv/Xmit Level  (-dBm)     8/16

Near Echo Loss   (dB)       2

Far Echo Loss    (dB)       0

Carrier Offset   (Hz)       NONE

Round Trip Delay (msec)     585

Timing Offset    (ppm)      18244

SNR              (dB)       41.9

Speed Shifts Up/Down        0/0

Status :                    ua,5,12N,12.2,-8,0N,0,41.2,23.7

OK

e.g. ATI11 from LT/Agere

ati11

 

 

    Description                         Status

    ---------------                     ------------

    Last Connection                     NA             

    Initial Transmit Carrier Rate       NA             

    Initial Receive  Carrier Rate       NA             

    Final   Transmit Carrier Rate       NA             

    Final   Receive  Carrier Rate       NA             

    Protocol Negotiation Result         NA             

    Data Compression Result             NA             

    Estimated Noise Level               NA             

    Receive  Signal Power Level  (-dBm) NA             

    Transmit Signal Power Level  (-dBm) NA             

    Round Trip Delay             (msec) NA             

Press any key to continue; ESC to quit.

 

    Description                         Status

    ---------------                     ------------

    Near Echo Level              (-dBm) NA             

    Far  Echo Level              (-dBm) NA             

    Transmit Frame Count                0              

    Transmit Frame Error Count          0              

    Receive  Frame Count                0              

    Receive  Frame Error Count          0              

    Retrain by Local  Modem             0              

    Retrain by Remote Modem             0              

    Call Termination Cause              0              

    Robbed-Bit Signaling                NA             

    Digital Loss                   (dB) NA             

    Remote Server ID                    NA             

 

OK

e.g. AT#UG from Conexant HSFi

at#ug

<VERSION=10>

<CALL SETUP RESULT=Detected Data Answer Tone>

<MULTIMEDIA MODE=Data Only>

<DTE-DCE MODE=Async data>

<V.8 CM octet string="">

<V.8 JM octet string="C16513948">

<RX SIG POWER DB=16, TX SIG POWER DB=18, S/N RATIO DB=40>

<NORMALIZED MEAN SQUARED ERROR=48>

<NEAR ECHO LOSS=6>

<FAR ECHO LOSS=200>

<FAR ECHO DELAY=177>

<ROUND TRIP DELAY=177>

<V.34 INFO bit map =0>

<TX MOD=V.90, TX SYMBOL RATE=3200, TX CARRIER FREQ=1920>

<RX MOD=V.90, RX SYMBOL RATE=8000, RX CARRIER FREQ=0>

<TX FIRST DATA RATE=28800, RX FIRST DATA RATE=46667>

<CARRIER LOSS=0, RATE RENEG=0>

<RETRAINS REQ=0, RETRAINS DET=0>

<TX FINAL RATE=28800, RX FINAL RATE=46667>

<PROTOCOL NEGOTIATION=LAPM>

<EC FRAME SIZE=80>

<EC LINK TIMEOUTS=0>

<EC LINK NAKS=0>

<COMPRESSION NEGOTIATION=V42BIS>

<COMPRESSION DICTIONARY SIZE=800>

<TX FLOW CONTROL=2>

<RX FLOW CONTROL=2>

<TX CHARS SENT FROM DTE=9>

<RX CHARS SENT TO DTE=3F>

<TX CHARS LOST=0>

<RX CHARS LOST=0>

<TX FRAME COUNT=8>

<RX FRAME COUNT=4>

<TX FRAME ERROR COUNT=3>

<RX FRAME ERROR COUNT=0>

<TERMINATION CAUSE=No previous call>

<CALL WAITING EVENT COUNT=0>

<LSD Vdd=56806200, Power Bits=362CB38>

<V8BIS FLEX VERSION=0, V8BIS DSP VERSION=0>

<RATE RENEG REQ=0, RATE RENEG DET=0>

<HIGH PASS FILTER ENABLED=0>

<ROBBED BIT PATTERN=EB>

<DIGITAL PAD DETECTED=0, DIGITAL LOSS ESTIMATE=0>

<Cnt1=0, Det1=0, Res1=0>

<EQM=30, TRAINING EQM SUM=30>

<NEGOTIATION TIME=22854>

 

OK

Believe it or not, Microsoft tried to have a hand in this diagnostic business, and tried to standardize a command to dump modem diagnostics. The command is AT#UD for "unimodem diagnostics", but the problem was (like in SMART) they never standardized anything aside from the dump format, so you end up with cryptic Key + Value pairs which you can only decode if the driver contains the strings which define what each key meant - and some modem manufacturers never got that far. Others really goofed up, and lost all the stats once the call was hung up, so that's very useful!

e.g. AT#UD output from the SL2800:

at#ud

OK

DIAG <2A4D3263 00=FF 01=F 02=F 03=F>

DIAG <2A4D3263 04=000000000000000000000000>

DIAG <2A4D3263 05=000000000000000000000000>

DIAG <2A4D3263 16=FFFF 17=FF 18=ffffffff>

DIAG <2A4D3263 20=ff 21=ff 22=ffffffff>

DIAG <2A4D3263 23=ffffffff 24=ffffffff>

DIAG <2A4D3263 25=ffffffff 26=ffffffff>

DIAG <2A4D3263 27=ffffffff 30=FF 31=FF>

DIAG <2A4D3263 32=FF 33=FF 34=FFFF>

DIAG <2A4D3263 35=FFFF 40=FF 41=FFF>

DIAG <2A4D3263 42=FF 43=FF 44=FF 45=FFFF>

DIAG <2A4D3263 50=F 51=F 52=ffffffff>

DIAG <2A4D3263 53=ffffffff 54=ffffffff>

DIAG <2A4D3263 55=ffffffff 56=ffffffff>

DIAG <2A4D3263 57=ffffffff 58=0 59=0>

DIAG <2A4D3263 60=01 61=FF 128=ffffffff>

On the whole, I very much liked AT commands for their simple in-band management and the fact that the S-registers reminded me of just how simple you can make your gear. Changing the S-registers are literally changing the numbers by which the modem code would use to perform its routines, and the resulting changes sometimes helped users get better compatibility of their modems (e.g. Apple users with an adapter cable use DTR signal differently to PC users, or whether the modem reported DCE to DCE or DTE to DCE speeds and what return codes it would supply), or even push the speeds (I got 54667bps connect just *once* in my lifetime, but even doing that for a fraction of a minute was an achievement).

My biggest regret at this stage was not having the opportunity to document 3com/USR's x2 modulation. In Australia, I don't remember x2 being very popular, and the number of high quality x2 handshake recordings are nought (at least, to my knowledge). The number of x2 modem banks are probably very low as well, as only 3com answer equipment has a chance of supporting it, and no ISP readily advertises their gear nor their capabilities aside from the ITU standards. I never got a chance to carefully examine K56Flex either for that matter, or record it, but I do clearly remember K56Flex having 2k connection increments (44, 46, 48, 50, 52k being common speeds to see) whereas V.90 changed that to 1.333k so we got 44, 45.3, 46.6, 48, 49.3, 50.6, 52k, so an early way of telling if your modem was V.90 was to see whether it had 3's and 6/7's in the connect speed line (although if you connected at the common rates, you couldn't tell).

Fond memories ... very fond memories. I was on dial-up a lot longer than I should have been, partly because of the house-hopping and renting, but it was mostly a reliable partner.

- Gough