[GreenKeys] Diddle, Real Metal, & Indeterminate Line Length
Bob Camp
ham at cq.nu
Thu Feb 24 18:35:03 EST 2005
Hi
Diddle diddle diddle ....
There is one minor problem with diddle. It has to do with the pesky
stop bit problem. These days we more or less stick with 1 or 2 stop
bits in ASCII code. Rarely if ever will you see an 8 bit guy talk about
1.5 stop bits. Oddly enough a lot of mechanical gear was set up for a
stop bit length of 1.42.
If you are trying to run a numerical flywheel you need to know the stop
bit length. If you don't then the history buffer pretty much becomes
useless at the end of each character. Either you need a very fancy
adaptive approach or your flywheel will only work will on computer
generated signals.
You need to be careful in any case to pick a diddle character that is
easy to synchronize to. A character with only a start bit and
everything else the same sense as the stop bit is easy to synchronize
to no matter when you pick it up. This pretty much drives you to pick
LTRS as your diddle character.
If you are going to inter-operate between computer generated systems
and mechanical systems then you probably want to transmit a stop pule
that is two 2.0 units wide. You definitely want to receive with a UART
set to a stop pulse 1.0 units wide. On receive the UART will simply
ignore the rest of the stop pulse.
I have no idea what difference any of this makes unless somebody is
getting ready to code up a new RTTY program. It does have at least a
minor impact on a system of audio tone TTY over the Internet.
Take Care!
Bob Camp
KB8TQ
On Feb 24, 2005, at 10:36 AM, Jerry wrote:
> Some good comments and explantions regarding diddles from Brian Beezley
> that are timely....
>
> Jer -n6jp-
>
>
> The Importance of RTTY Diddle
>
>
> When I got back on RTTY in 1995 after having been away
>
> since the 1970s, the first thing I noticed was the widespread
>
> use of diddle. Diddle is the transmission of a do-nothing, idle
>
> character (usually LTRS) when there's no text to send. An RTTY
>
> signal without diddle is just a steady mark tone.
>
>
>
> Diddle originates in your terminal software or modem.
>
> Usually a parameter is provided to enable or disable the
>
> feature. The purpose of this note is to motivate those of you
>
> with diddle disabled to turn it on. Transmission of an idle
>
> character may seem curiously irrelevant, but using diddle has
>
> many benefits, some profound.
>
>
>
> To begin with, diddle allows someone tuning across your
>
> signal to immediately identify it as RTTY. Without diddle, the
>
> steady mark tone your modem emits when it runs out of text is
>
> indistinguishable from an unmodulated carrier. You might just
>
> as well be someone tuning up, a computer birdie, Radio Moscow
>
> between programs, and so on. Diddle uniquely identifies your
>
> signal as RTTY.
>
>
>
> An operator encountering a diddled signal can
>
> immediately determine its baud rate and frequency shift
>
> (experienced operators can do this by ear). Diddle thus allows
>
> an operator to quickly set his modem to receive your transmitted
>
> signal. Without diddle, the operator must wait for text before
>
> he can set modem parameters and tune you in. If you pause long
>
> or type irregularly, a frustrated operator may pass your signal
>
> by and look for one easier to decode.
>
>
>
> The advantages of diddle mentioned so far are receiving
>
> conveniences. If you're patient, once you've set the right
>
> modem parameters you'll copy the same text with diddle or
>
> without as long as the signal is strong. But diddle has another
>
> advantage: It can help recover text when signals are weak.
>
>
>
> One way it does this is to correct the receive case
>
> whenever your modem incorrectly decodes a FIGS character due to
>
> noise. As soon as it decodes a LTRS diddle character, your
>
> receive case will again be synchronized with that of the
>
> transmitter. This can help prevent printing strings of numbers
>
> that should be letters.
>
>
>
> Another benefit of diddle is more subtle. Many modems
>
> use automatic threshold correction (ATC). This feature
>
> automatically adjusts the decision threshold that determines
>
> whether a received bit is a mark or space. When the mark and
>
> space signals have equal amplitude, the best detection threshold
>
> for the mark-minus-space signal is zero. But when selective
>
> fading, IF-filter ripple, or audio rolloff cause the two signals
>
> to be received at different amplitudes, a threshold of zero is
>
> no longer optimal. ATC continually adjusts the threshold based
>
> on an estimate of the mark and space amplitudes. But it's
>
> impossible to do this when there's no space signal to sample!
>
> This is the case during a non-diddled idle. Thus an optimal
>
> decision threshold is unavailable for the first character sent
>
> after a pure-mark idle. (Digital ATC can compensate for this
>
> limitation by peeking into the future and extrapolating a
>
> threshold back into the past. Analog circuits don't have this
>
> luxury.) Using diddle ensures that ATC systems have the signal
>
> they need to work effectively.
>
>
>
> Here's an even stronger reason to use diddle: It allows
>
> a receiver greatly enhanced immunity to loss of synchronization
>
> due to noise. RTTY uses asynchronous transmission. The five
>
> data bits of the Baudot code are preceded by a start bit, which
>
> is always a space. Your decoder waits for a start bit, collects
>
> the data bits, and at some point in the stop bit becomes ready
>
> for a new character. But what if while you're sending no
>
> character, a noise burst on the space frequency overrides the
>
> steady mark signal and masquerades as a start bit? Your decoder
>
> must then commit itself to decoding an entire character. But if
>
> you happen to begin sending during the middle of the decoding
>
> cycle, the receiver loses sync. It may take a dozen characters,
>
> each garbled, before the decoder locates your start bits and
>
> resumes decoding correctly.
>
>
>
> Leaving a steady mark signal between characters is just
>
> waiting for an accident to happen. If you're lucky, a noise
>
> burst or signal fade will be decoded as a single bogus character
>
> and the modem will resume waiting for a start bit. But if
>
> you're unlucky, you'll lose sync and print garbled text until
>
> the transmitter and receiver resync. Using diddle greatly
>
> minimizes the chance of this happening. Diddle commits the
>
> receiver to decoding an idle character whenever there's nothing
>
> to send. This forces the receiver to maintain sync with the
>
> transmitter. A noise burst can then cause a false trigger only
>
> by occurring in the very brief interval between the receipt of
>
> the stop bit and the beginning of the start bit.
>
>
>
> The width of this open window is a modem design
>
> parameter. It involves a trade-off between rejecting false
>
> start bits and missing valid start bits of signals whose timing
>
> is fast. The window for detection of a new start bit might
>
> begin 22 ms into a 45-baud stop bit. The window ends when the
>
> diddling transmitter emits another start bit, typically 9 ms
>
> later. Thus, out of a nominal character length of 163 ms, the
>
> receiver can be thrown out of sync only during an interval 9 ms
>
> long. Diddle thus improves noise immunity from 0% to 94.5%.
>
> That is, the receiver will falsely trigger on sufficiently
>
> strong noise only 5.5% of the time rather than 100%!
>
>
>
> While these arguments for diddle are compelling, I think
>
> this is the most powerful of all: Modems with advanced
>
> synchronization algorithms can lock to asynchronous character
>
> streams and essentially receive RTTY synchronously. The
>
> critical asynchronous-protocol dependency on a start bit can be
>
> virtually eliminated. A modem can do this by implementing what
>
> amounts to a numerical flywheel whose rotation is synchronized
>
> in phase and frequency with the transmitted characters. Timing
>
> marks engraved on the flywheel, rather than noisy start pulses,
>
> determine when characters begin. With a flywheel of enough
>
> mass, a modem can maintain sync through deep fades in which the
>
> signal completely disappears. A numerical flywheel permits you
>
> to recover text that would otherwise be lost.
>
>
>
> Use of a numerical flywheel also improves decoding by
>
> greatly reducing timing jitter. When optimal channel filters
>
> are used to receive RTTY, the exact point at which the signal is
>
> sampled to determine whether it's a mark or space becomes
>
> critical. Optimal filters do not output rectangular or rounded
>
> pulses. Instead, they generate triangular waveforms. The
>
> optimal sampling point is at the peak of the triangle, but this
>
> location can't be determined by waveform inspection when the
>
> signal is noisy. Instead, sample timing must be derived from
>
> the timing of the start pulse. But when that's noisy, data
>
> samples will be mistimed and a mark may be mistaken as a space.
>
> A well-implemented numerical flywheel can virtually eliminate
>
> decoding errors due to timing jitter.
>
>
>
> But here's the rub: Without diddle, a numerical
>
> flywheel has no way to maintain lock when you stop typing or
>
> type irregularly. After a prolonged mark tone or intermittent
>
> character emission, the flywheel will have wandered off and
>
> become useless.
>
>
>
> RTTY modems with robust numerical flywheels have been
>
> available since 1995. Your signal can take advantage of their
>
> advanced synchronization capabilities only by using diddle. For
>
> this reason and for all of the others mentioned above, do
>
> yourself a favor and always enable diddle.
>
>
>
> Brian Beezley, K6STI
>
>
>
> ----- Original Message -----
> From: "George B. Hutchison" <w7tty at readysetsurf.com>
> To: "GreenKeys" <greenkeys at mailman.qth.net>
> Sent: Wednesday, February 23, 2005 9:00 AM
> Subject: [GreenKeys] Diddle, Real Metal, & Indeterminate Line Length
>
>
> : GreenKeyers - - -
> :
> : I am no math expert in these matters, but I do bel;ieve
> communications
> : theory states that a circuit tends to be more reliable if some manner
> of
> : signal transmission is continuously present on the circuit.
> :
> : Diddle accomplishes this. In the days of slideback detectors (Bob
> : Weitbrecht, W6NRM,) and ATC/DTC contention (Irv, W6FFC) diddle did a
> : nice job of keeping the circuitry's determination of mark and space
> : right around the zero crossing point. It irritated the hell outa me
> when
> : Larry, WA6JYJ (now W7JYJ) first started using it, but, it made things
> : better.
> :
> : The downside of running diddle is that if one is running a reperf,
> you
> : can chug out miles of tape with nothing but LTRS on it. Brian
> Beezley's
> : RITTY program has an RS-232 output that has a "Diddle Filter" on it,
> in
> : that after two or three LTRS characters it shuts off output and goes
> : idle until a character other than LTRS comes across the circuit.
> Saves
> : lotsa tape.
> :
> : RITTY is still available from Brian, he just does not publicize it at
> : all. If you are interested, e-mail me and I will make arrangements
> for
> : you to contact him. RITTY is a DOS-based program that uses some very
> : sophisticated DSP algorithms. A dirt-bag DX-66 computer with RITTY is
> a
> : push for an ST-8000 or Universal 8000 when it comes to digging down
> into
> : the mud. Cost is 100 dollars, and each copy is personalized with your
> : name and call.
> :
> : Indeterminate line length is caused by the RTTY software geeks
> relying
> : on word wrap to determine the end of line point, rather than calling
> a
> : specific point (usually 68 to 72 characters) as the machines are
> : designed to run at.
> :
> : Auto Cr-Lf and non-overline is nice, but, for those who like to see
> the
> : really intricate overstrike TTY art, both will tend to turn a piece
> of
> : artwork into a piece of crap.
> :
> : Nowadays, when most on-air RTTY communications consist of CQs and
> Signal
> : Strength reports, maybe a defined EOL point is superfluous, because
> the
> : lines never get that long. I prefer DEFINED EOL-CR points. It's just
> : more professional, and the machines LIKE THEM. BLACK SPOTS on the
> right
> : hand side are very hard to read.
> :
> : If your software has a means of setting the EOL-CR, it is best to use
> : it.
> :
> : Regarding a RATTS net, or autostart net, I am all for it. I am close
> to
> : getting back on the air, and it would be nice to have something like
> : that going.
> :
> : The RATTS net here in the Northwest is one of the longest surviving
> nets
> : of its type. They seldom handle any traffic, but by golly, the means
> and
> : the good-will is there.
> :
> : Two or three years ago I participated in an Emergency Communications
> : Exercise wherein the Red Cross was to be the central focus of the
> : operation. The Red Cross did not know sawdust from Arbuckle's Coffee,
> in
> : that they did not seem to know they were a part of it.
> :
> : I turned off my transmitter in disgust. The RATTS guys tried.
> :
> : W7TTY will be back on the air soon.
> :
> : Stay Tuned, film at eleven...
> :
> : 73,
> :
> : George - W7TTY
> :
> : _______________________________________________
> : GreenKeys mailing list
> : GreenKeys at mailman.qth.net
> : http://mailman.qth.net/mailman/listinfo/greenkeys
>
>
> _______________________________________________
> GreenKeys mailing list
> GreenKeys at mailman.qth.net
> http://mailman.qth.net/mailman/listinfo/greenkeys
>
More information about the GreenKeys
mailing list