CW Contest Sending Speed

The speed of the CW we send in a contest is strongly correlated to our CW skill level. It's true even though we rarely use the paddles, by relying on computer software to do the sending for us. Is this sensible or is another strategy more effective? In a contest, effective means a higher score, and we score higher by having more QSOs, and that typically involves completing each QSO as quickly as possible. High speed CW depends on the skill and tactics of both stations in each QSO.

This is an important question since in a contest it is routine for highly skilled CW contesters to work less skilled operators. Indeed, many operators in this era of no-code licenses only learn the code in order to contest. There is often a wide discrepancy of ability that everyone needs to understand and deal with.

Since many CW contesters are old timers who learned the code decades ago, the difficulty newcomers face learning the code, and especially at the high speeds common in a contest, is not always appreciated. It takes time and practice, sometimes quite a lot. Few of us became proficient overnight! We forget that and become impatient when we are forced to QRS or are delayed by slow sending and repeats.

Hams with a natural talent for CW are fairly rare. I am not a member of that select group. Even today I sometimes struggle to send and receive at high contest speeds, and more when I take to the paddles. Years ago, memory keyers helped and today computer software takes over much of the effort. I built my first contest keyer in my teen years soon after I gravitated to contests. What a pleasure that was for a clumsy CW hack like myself!

Contests are not about having conversations. It is very possible to score well with modest CW skill. The computer does almost all the sending and copy is easier since the format is predictable and short. Although the sending speed you ought to select should be intuitive after spending an hour or two in a contest, I am always surprised by how many make choices that impair their results. 

Many casual operators do not care, and that's okay. Those that do want to increase score potential should reflect on the points I raise in this article.

For those who can't copy at contest speeds

A prevalent "mistake" by casual CW contesters is to send no faster than they can copy. When I ask why, they tell me that's what they're comfortable with. If that's 20 wpm, well, that's what they set the computer to send. This is odd because a large majority of contesters can copy much faster. Both operators are delayed and it isn't necessary. Go ahead and crank the speed up to 30 or 35 wpm when you call someone. They'll copy you just fine.

Consider S & P (search and pounce): You already know the call of the station being called, and that's probably the hardest information to copy. The exchange may be fixed (e.g. CQ WW) and if you use the call history feature of your logging software, the exchange may be pre-filled, or you already have it from their previous QSO. You need only verify that it is correct and make changes where necessary. 

For variable exchanges such as serial numbers, well, that you have to copy. But, again, you may already know what to expect based on their previously sent number. In any case, numerals are long characters (5 elements) and there are rarely more than 4 of them in a serial number. That isn't too challenging. Cut numbers are another matter which I won't get into (e.g. A for 1, E for 5, T for 0, etc.).

One thing you can be certain of is that the running station will not QRS to your sending speed. Not only does it throw off their rhythm and require extra work, they won't see the point since you obviously copied their call sign before calling them. Also, you probably already copied all or most of the exchange. Sending slowly does not help anyone in this most common circumstance. Be more aggressive; that is, be a contester and get out of the slow lane.

I know CW contesters who are not very competent at high speeds. They may use a CW reader application on their computers as an operating aid. They'll glance at the display to confirm what they copied or to fill in the gaps. A glance at the super check partial (SCP) window of known contester call signs is also useful as a sanity check: if it isn't in there you may want to listen again or request a repeat. 

There's no shame is using these aids as you climb up the lower slope of the learning curve. Check the contest rules since your entry class may be affected by the use of code readers.

Competition in the pile up

Another danger of sending slowly is that you will lose to the faster competition when other stations are calling. I deal with this quite often. Let's say that I am running at a good clip, sending ~35 wpm, with 2, 3 or more stations calling at a time. That's contest heaven: lots of callers but not so many that I can't pull one full call sign out of the bunch. 

I don't always pick the loudest or the one who starts sending first. Often it is the fastest that goes into the log. The reason is that I can copy it easily in a second or two. Time is of the essence. When I am operating SO2R it is critical that call sign reception take no longer than my transmission on the other radio or I will find myself having to copy on both radios at the same time, and that isn't easy. I pick the fastest since it is to my advantage.

Slow sending and long call signs upset my rhythm. It slows everything down and puts more burden on my short term memory, where your call resides until I type it in. When I pick the slower station, the subsequent delay discourages other callers since they don't want to wait the extra 10 seconds. They click on another spot and disappear. I not only lower my immediate rate and I might never hear that other station again. In my experience, slow stations don't immediately QSY when I answer a faster caller.

If you're not a contester, this reluctance to wait a few seconds might seem ridiculous. It is the reality we deal with and it affects scores. Send faster and you will become more competitive. If you have a weak signal or there is QRM or QRN the station you're calling will request a repeat. That may be a good time to QRS a little, but only then.

Help the other station by being fast. Unnecessary repetition of a call sign or exchange halves the effective speed. If you're already at 20 wpm, that takes you down to 10 wpm. That's breeds impatience and you may not get through at all. 

When there's QSB, QRQ is recommended because when it takes several seconds to send your call or exchange there is a strong possibility that one or more characters will be missed during a signal dip.

Running? QRQ could be a problem

Your speed does matter if you are running to attract callers. Many will QRS a few wpm. This is about more than courtesy: they want to move on quickly and if you ask for a repeat because they're sending too fast for you to copy the QSO takes more time. 

Of course there are those that never QRS, and that's a problem for both parties. Valuable time is wasted on repeats. I hear this all the time, and it's only funny because I am not one of the parties in the QSO. Operators who persist in calling slower stations at high speed often fail to rank high in the contest standings. That's not a surprise! The smart operator will QRS to get you in the log quickly so they can move on to the next contact.

Let's say you have to ask for a repeat of the QRQ stations' call sign or exchange. The effective speed of their sending is less than half of their actual sending speed. After all, the information is being communicated twice (or more) plus the time for the request to be made and the multiple turnovers in transmission and reception. 

Unrelenting QRQ slows the QSO rate for everyone. The slow runner is perfectly justified to ignore the persistently crazy fast caller.

Effective speed

The common method for calculating sending speed is by measuring the duration of character elements (dots and dashes). Character and word spaces are included in the calculation, using assumptions for character mix and average word length. A machine can do it easily, and they do.

CW Skimmer software measures speed quite well. Turn on skimmer spots on your favourite DX spotting network and soon your band map will be filled with skimmer spots. Watch the data in the Telnet window or (easier) hover your mouse over the call in the band map. The screen shot shows how it's done in N1MM Logger+.

That's only half the story. We are not machines and context matters. What we measure as CW speed (wpm or words per minute) is actually bit length: the time to send a dit. However, the true information rate is that of characters and words, and there are important variations.

Let me give an example of two call signs I worked in a recent contest: EE5X and OM0M. Despite both being 4 characters long there is an obvious difference between them. Using the common specification for CW keying, the number of dit lengths in the first call sign is 31 versus 53 in the second. That is, one of those calls takes almost twice as long to send at the same speed.

For many contesters, the one that is faster to send may seem attractive since it saves time. But therein lies a problem. The operator copying the call must work at twice the mental speed since the 4 characters come in half the time of the other call. That is the cause of many copying errors, and errors spawn repeats and busted calls and exchanges.

The Morse character set is deliberately optimized this way, similar to Huffman coding, by assigning short codes to letters that appear more frequently in typical English language text. Being less common, numerals and punctuation are given longer codes. Thus E is the shortest letter, at one bit, and J and Q are the longest, at 13 bits. It is no coincidence that E is worth 1 point and J and Q are worth 10 points in English version of the game Scrabble.

To bring us back to the point of all this, when you send at contest speeds, the message content matters. For call signs with short characters it can be very worthwhile to send more slowly. Conversely, call signs with long characters can be sent more quickly. It is quite easy to do so in most contest logging applications by altering the speed before and after call signs and to insert padding between characters.

Dealing with dits (again)

Following from the previous discussion, it is no surprise that strings of dits cause problems despite our best intentions. Ears and brains must work more quickly than for other element sequences. If your call sign has one or more of these strings, it is worthwhile giving some thought to how fast you send it.

This is a subject I've mentioned before, and more than just once. As much as I like my call sign it is a problem on CW. The VE3 prefix may seem to be a problem with its strings of dits, yet it rarely is. The prefix is so common that almost everyone recognizes the sound of it and can copy it correctly, no matter how fast it is sent. A few mistake it for VE2. For me the trouble lies with the second V in my call. 

My solution (first link in the previous paragraph) is to program a function key that sends the suffix slowly and with extra spacing. When the inevitable occurs I can correct the other operator with minimum effort and time. Unfortunately this only works for S & P; when running, few operators send my call sign and I can't correct them. If one of them spots me as VE3UN (the most common error) I am inundated with dupes. 

The error is so common that VE3UN not only appears in the SCP database, it usually appears above my correct call. When an operator "yanks" a call sign from the database it is often the wrong one. Thus the error persists, perhaps forever. Yet I have never heard VE3UN in a contest. Too many operator trust SCP without reservation. If you make use of SCP to cope with fast CW, be suspicious of what it recommends.

CW Skimmers don't often make this mistake, but they do make mistakes that are uncommon to human operators and apply to all call signs regardless of speed or content. Smart operator treat skimmer spots with extreme caution.

I envy other VE3 contesters with suffixes that are more easily copied because they're not full of dits. Perhaps one day we will be allowed 2×1 contest call signs. Until then I'll have to live with the problem. It is unusual to hear me sending faster than about 34 wpm in a contest because of this, even though I am comfortable at higher speeds. Meanwhile my friends can send closer to 40 wpm and be copied incorrectly less often.

Is there a speed limit?

A few contesters speed recklessly. They rely of spotting networks to distribute their call sign, or are blatantly unconcerned with whether others copy them correctly. All they want is your call in their log. What goes into others' logs, right or wrong, doesn't impact their scores. There are exceptions in a few contests where both parties must copy correctly but they are not the most popular events.

For those in rare multipliers, QRQ works to their advantage. For everyone else it often does not. Don't emulate them. Many operators will be scared off because they are unable to copy north of 40 wpm. How fast the speed demon can copy is irrelevant. The objective should be to be correctly copied without repetition. Moderating the speed delivers results even though each QSO, considered individually, may on average be quicker with extreme QRQ.

There is a tenet of effective communication that applies to all writing, and applies equally well to CW contests:

There are two ways to write. One is to write in a manner that is convenient to the writer. The other is to write so as to be read and understood. To be successful, be the second kind of writer.

When too many potential callers can't copy you or copy you correctly, whether in good conditions or bad, everyone pays the price. As in driving: speed kills; drive according to the prevailing conditions.

I find it both sad and amusing when an operator persists in sending at 40 wpm to a station sending at 18 wpm. There are endless repetitions and frustration on both sides of the QSO. I see it in contests and in everyday DXing. The keyer has a speed control and it's there to be used.

Enjoy the holidays. I can't say whether I'll have time to complete another article before January, so I may end December with one less article than is typical for me.