RDXC 2007 Log Errors

Totally there were 111803 log errors discovered, that is 6.1 % of overall QSO number. It is funny, this time percentages of CW and SSB errors are absolutely the same. Log errors by-band distribution looks like this:

  • 1.8 - 4.9%,
  • 3.5 - 5.6%,
  • 7 - 6.2%,
  • 14 - 6.4%,
  • 21 - 6.7%,
  • 28 - 5.3%.

Some QSO are not false, but they were not credited for another reasons (B4, check-log, etc.). Some QSO had two and more errors.

As usual, RDXC 2007 log errors can be separated into a few typical groups, against their origin.

1. Prefix errors percentage increased, compared with previous years. This is explained by the fact that many new prefixes were issued in some countries, and many participants used special prefixes in RDXC 2007. Common prefix errors are still here, such as: UA-RA, RA-RN, RK-RX-RZ, G-M, SP-SQ, IK-IZ, JA-JF, etc.

2. CW Audio Errors.
Most often there are "S" instead of "H" (785 times), and vice-versa, "H" instead of "S" (392). Such errors increased due to new prefixes number (OS, SA, etc).

Here are more typical errors: "D" instead of "B" (360), "I" instead of "S" (257), "B" instead of "D" (216), "U" instead of "V" (209), "V' instead of "U" (205), "H" instead of "5" (156).

Hence, there are group errors, such as,

     OH5N instead of OS5N (150);
     YT6A instead of CT6A (48);
     HI3A instead of 5I3A (45);
     IN9U instead of SN9U (39);
     9K2SN instead of 9K2HN (36);
     HA6G instead of SA6G (35);
     BV7E instead of 6V7E (33).

Another CW audio errors happen also. Say, AP0L instead of UP0L logged 46 times. A reason was T/R switching problem, and the first dit transmission delay.

Here is another error, curious at the first glance: OK4CU, instead of DK4CU, was copied 34 times out of his 125 QSO. It turned out that DK4CU operated a straight key, and skewed his callsign in such way.

Russian oblast ID errors (in CW) were totally 1563. All these errors may be referred to audio errors. In most cases they were: SM instead of HM (52 times), KK instead of KR (28), and YA instead of YR (24). The rest group errors happened due to unnecessary software confidence.

CW Numerical Errors.
Numerical errors have similar rate both in callsign reception, and in exchange reception. Almost all of them are audio-close, and keyboard-close errors. Therefore it is not always possible to determine audio or keyboard origin of the error:

     2-1 and 1-2 - 1238 times;
     8-7 and 7-8 - 1183 times;
     2-3 and 3-2 - 1021 times;
     3-4 and 4-3 - 994 times;
     1-0 and 0-1 - 899 times;
     5-4 and 4-5 - 896 times;
     9-8 and 8-9 - 853 times;
     7-6 and 6-7 - 737 times;
     0-9 and 9-0 -  714 times;
     6-5 and 5-6 - 707 times.

"No-close" numerical errors number did not exceed 2-3 dozens. Say, 4 and 8 were mixed up 14 times only.

In general, numerical errors did not cause group errors, which become the worst reason of the particular participants (the victim) score decrease. Such group error number was not more than 4, for example, YL8W instead of YL9W.

Errors like "number-letter" are considered as letter errors.

It is not true to guess that error fault lies on the receiving side. The two participants with equally "uncomfortable" calls, RU4SS and RN4SS, operated in the same entry, SOAB-CW-LP, and scored equal QSO number (by the way, they are in the same QTH). The first operator succeeded to be a winner in this entry, the other op happened to be a few steps down because his call was copied wrong 24 times.

Correct transmission speed and keying quality (in case of manual keying) strongly influence at your call copy-quality. Some participants used to merge last symbols of their calls. For example, RA3DTH was repeatedly logged as "RA3DB", UA6LAM was logged as "UA6LJ", and UA3DAM logged as "UA3DJ".

3. SSB Audio Errors.
The most part of these errors happened due to phonetic alphabet poor knowledge. As usual, there are "W" instead of "V" (457 times). Vice-versa, "V" instead of "W" - 131 times. The rest SSB typical error "places" scored like this: "G" instead of "J", and vice-versa - 209 times; "O" instead of "A", and vice-versa - 204 times; "Z" instead of "X" ("znak" in Russian), and vice-versa - 140 times; "S" instead of "Q" - 48 times. "C" (Canada) and "Q" (Quebec) transferred into "K" 86 times. Many participants did not know correct phonetic interpretation of "Y" and "J" (51 times).

The most common SSB "group" errors are:

     RW6AWM instead of RV6AWM (84 times);
     RW9JE instead of RV9JE (39);
     LZ9W instead of LZ9V (37);
     ON5JQ instead of ON5GQ (25);
     IW3SKB instead of IV3SKB (20).

In CW, typical errors prevail, but in SSB the most errors are impossible to understand, and not always explainable. The fault in such cases is not always on the receiving side. In SSB, this depends on operator pronunciation, transmit path quality, DVP quality (if in use). Besides that, wide bandwidth is a reason of greater QRM possibility.

Russian oblast ID errors (in SSB) were totally 1237. Very often, YA was copied instead of YR (24 times). Hams from Kuban suffered the most: their KR copied as KK (7 times), as MA (9 times), as SR (7 times), and as ST (7 times). Almost half of errors happened due to unnecessary software confidence, but not to your own ears.

SSB errors and CW errors have different nature. Many neighbor numbers differ in pronunciation. Though, Number One here was sweet pair 2 and 3, that was incorrectly copied 586 times.

     6-7 and 7-6 - 355 times,
     8-9 and 9-8 - 353 times,
     6-5 and 5-6 - 322 times,
     4-2 and 2-4 - 310 times,
     2-1 and 1-3 - 278 times,
     4-5 and 5-4 - 270 times,
     2-0 and 0-2 - 261 times,
     1-0 and 0-1 - 249 times,
     4-3 and 3-4 - 224 times, the rest combinations are more seldom.

But numerical group errors are much more higher in SSB than in CW. For example, TM2Y logged as "TM3Y" 26 times.

4. Missing and Extra Characters.
Such errors in calls are met equally both in CW, and in SSB. Most often, the last letter in the call is missing:

     HB9DA instead of HB9DAX (10 times in CW);
     RA2FF instead of RA2FFE (7 times in CW);
     K1TT instead of K1TTT (5 times in CW);
     LZ1CL instead of LZ1CLR (6 times in SSB);
     RX3RX instead of RX3RXX (5 times in CW, and 6 times in SSB).

Quite frequent errors are missing numbers in 4-character control exchange (usually the last character). Perhaps, not all participants realized that QSO number in RDXC could exceed 1000. In 2007, this limit was exceeded by 397 stations. This fact honors our RDXC.

Extra characters also arrive in the end of callsigns. Operator especially try to prolong short ones, such as W4VV instead of W4V (5 times). But extra characters are rather seldom.

In CW, there are special cases when operator merges his callsign with code phrase. For example, one can get UT5SAK out of UT5SA if there is no space between UT5SA and K transmission (8 times). ON6PQ could be ON6PQT, if there is no space between the call and TEST (7 times). If there is no space between your own call transmitted twice, one can get LY3MLY out of LY3M.

5. The next error group appears both in CW, and SSB. These errors happen due to your high level of confidence in data bases contained in logging programs, and to the callsigns you contacted on other bands. In this way, we can explain such single, and even group errors like:

     RA3TT instead of RN3TT;
     RK3DG instead of RK6DG;
     UR7M  instead of YR7M, and others.

The same approach with Russian oblasts ID can be used. Software gives you a prompt, but the operator himself has to make a decision. For example, U1BA lives in Leningradskaya Oblast (LO), but U1BB is in St.Peterburg (SP); RW3DA and UA3GM live in Moscow (MO); UA3WI is in Vladimirskaya Oblast (VL), and UA3YH is in Kaluzhskaya Oblast (KG).

Besides that, many programs do not put oblast IDs for portable stations, or identify them as home oblast. After the contest, operators failed to check this. In all such cases contacts were credited to participants.

6. We noted a lot of keyboard errors. They have no explanations but incorrect PC entry. There are character transpositions, and neighbor button unintentional pushes:

     LY6W instead of YL6W;
     VR2MXT instead of VR2XMT;
     UA6IYU instead of UA6YIU;
     UA6RAT instead of UA6ART;
     JA2TSB instead of JA2TBS.

Many numerical errors happen due to the same reason, but not of the wrong copy.

Here we can add errors during PC log entry from your paper logs, or log editing in post-contest mode. Only this reason explains "T" instead of "Y", and "O" instead of "D" in your logs!

7. Errors Time Distribution.
From year to year, errors time distribution remains about the same. There is a high callsign and Not-In-Log error rate in the beginning of the contest. In 2007 this number increased because 28 MHz band was not open for the most participants. Therefore they moved band or two down, thus increasing stations density there.

During night time (local for particular operator) control exchange error rate increases certainly. Error percentage also grows during the last contest hour, when operators try to make up leeway.

8. Systematic Log Errors.
Systematic errors are those that appear either in complete participants log, or in any part of it. In 2007 such errors discovered in 193 logs out of total 3059 received logs. Detection of systematic errors, and their correction significantly extend scoring times.

The most reason of such errors are shifted times in your logs. Most often, QSO time is wrong (local or unintelligible at all). If you did not switch PC time, you have to note this fact in your log summary page.

Incorrect band or mode in successive QSO also cause problems with scoring. Particularly, such systematic errors are difficult to discover during initial scoring stage.

The most difficulty for scoring are shifted control exchanges that start from the definite place in your log. The reasons for that are different, but participants anyway have to take care about this.

9. About Log Formats.
Initial log processing is not difficult if they are in standard Cabrillo format. Standard does not mean exact anchoring of characters, but only each QSOs specified data sequence.

Pseudo-Cabrillo formats with changed data sequence (for example, RST after contest exchange) or without space (for example, between RST and exchange, like in RW0LBM log) are troublesome.

Classic N6TR, K1EA, EI5DI and others are processed with no problems.

Homemade formats, like where the band is placed randomly, and even not shown at all in multiband entry, slow log processing and the whole scoring time.

The most time we loose if we process your logs in WORD, EXCEL, and UEE formats. In such cases we strongly recommend you not to separate your log data into columns with vertical lines.