Computer generated and decoded Morse Code seems to be a not-uncommon operating practice in the HF bands. Given this, and that there's no reason why Morse/CW signals can't be synthesized and decoded at at many 100s or even 1000s of WPM (perhaps with Gaussian pulse shaping at the higher rates), is there a maximum legal (in the U.S.) CW/Morse QRQ WPM in the HF bands? How about even higher WPM rates in the higher amateur UHF bands? (10e6 WPM?)

Or does CW plausibly have to be human copiable?

Also, does QRQ CW require the operator to slow down to 20 WPM every 10 minutes to ID?

  • 1
    $\begingroup$ I have never heard of a maximum speed. And, why slow down to 20, I have signed at 30 wpm. $\endgroup$
    – K7PEH
    Commented Mar 15, 2020 at 3:28
  • 1
    $\begingroup$ At a glance, the hardest limit I can find is 97.307(b), which says that you have to confine your signal to the band you're licensed for. That limits you to a few hundred kHz or less on most of the HF bands ;) $\endgroup$ Commented Mar 15, 2020 at 5:21
  • 3
    $\begingroup$ Isn't there an FCC-mandated maximum data throughput? Something like 300bps on HF. That would limit you to about 360 wpm $\endgroup$
    – Scott Earle
    Commented Mar 15, 2020 at 11:12
  • $\begingroup$ @ScottEarle I remember the same, and by Shannon's non-existent beard, that's a stupid rule. $\endgroup$ Commented Mar 15, 2020 at 13:01
  • 2
    $\begingroup$ @ScottEarle this? ham.stackexchange.com/questions/10556/… $\endgroup$ Commented Mar 15, 2020 at 15:35

3 Answers 3


I have noticed that hams using CW, PSK31, RTTY, etc, use different equations to describe their bandwidth requirements.
Over the years, I have used an equation which combines the "k" and the "baud" and the "freq shift" : BW = (k) (baud) (delta freq) . This has worked well with various forms of pulse signals, and matches spectragraph samples of RF signals.

(1) "k" is the squareness shape from 1 (sine) to above 10 (requires much odd harmonic energy to produce a very 'square' shaped waveform ). Sharply shaped Squared CW pulses require much Odd Harmonics energy and this produce sideband 'splatter' for CW pulses.

(2) "baud" is the repitition rate, the number of square pulses produced per second. The more square pulses per second, the more 'sideband' required.

(3) "freq shift" related directly to the FSK signals of PSK31 and RTTY. Here we should note that even a Pierce crystal oscillator has 'jitter' of 3 to 10 Hz, but this is so close to zero shift that we notate it as "1".

Thus, my personal CW bandwidth is calculated at BW= (3) (15) (1) = 45 Hz. Speeds of a K=10 squared signal at 360 wpm would be BW= (10) (360) (1) = 3600 Hz. We should note that at higher speeds, the 'ear' requires a crisper defined wave-form in order to be 'heard'. A soft ( K=1.2 as per PSK31 ) CW waveform would require a slower wpm rate. A harder ( K=5 ) CW waveform would be intelligible at 25 wpm. A harder ( K=7 ) CW waveform would be intelligible at 45 wpm.

Using a SDR approach and DSP, software phase-locked loop algorithms, a computer could 'copy' CW at very high speeds, but might also require some clocking signals. PSK31 has clocking markers embedded in the waveform design ( this is the +/- shift of 15.25 Hz between elements. ) The JT65, et al, also rely on inherent clock signals, either for reasons of speed in decoding or in identifying a weak signal below noise level.

Still, to the question at hand, Faster Speeds require More Band-Width, and there are good equation descriptors of this electrical phenomenom.

I have calculated this for PSK31 hams and inspected RF spectrographs , and it fits well. PSK31 at a K= 1.2 always occupies a BW of 67 Hz, measured -27dB down (per FCC).

Same goes for RTTY, where the FreqShift is d=200 or d=1200 ( depending on your country and operating system. ) In RTTY, the delta(F) swamps the other factors.

So, with regards to CW band-width restrictions, we can , in general , see that high speed and squared-wave-shape can possibly require a potentially illegal bandwidth.
The suggested limit of 360 wpm fits in this equations description.

These are generalizations, based on my engineering and personal experience, and 62 years of CW experience. Please note the Caveat " Y M M V " [ Your Milage May Vary ] always applies when evaluating our own experiences. We do NOT invite a casual debate on the issues of CW pulse shape and human 'hearing' in this forum, at least NOT with me.

Note: Mike Waters has some good insight into the electrical requirements of producing a 'clear' CW signal. My first experience 'hearing' fast CW was ... In our Navy Ham Club, N.A.S. Norfolk, 1957, our chiefs from WWII operated CW at 65 wpm !!! You have not heard "real CW" until you have listened in awe as real Navy CW ops play ball at high speed ! Bugs & head-copy & a rattling mill ... ... amazing. Just imaagine how very short the 'dit' signal was, and imagine just how much 'crispness' that 'dit' must have to be 'heard' ! My 'dit' length is about 18 mS at 17wpm, and would be about 4 mS at 65 wpm. At 65wpm, the RF bandwidth could be more than 320 Hz at -27 dB (FCC).

Good luck with your discussions, I am going back to listening to CW on 20M and see who is up today !
Glen Ellis, K4KKQ ( see QRZ.com )

  • $\begingroup$ CW of typical QSO text with "proper" 1:1:3:7 timing is self clocking, and can be made even more so by embedding a percentage of known text. ITU recommendations for human-copyable includes the 3rd and 5th harmonic (K=5?), but computer copied CW can be modulated with K<1.5. $\endgroup$
    – hotpaw2
    Commented Mar 17, 2020 at 22:23

Yes, there is a maximum legal WPM for CW.

The higher the data rate, the more bandwidth a signal occupies. Thus, the tag.

That also applies to CW. As the WPM is increased, the wider the signal will be. And at some point, the bandwidth of the signal will be so great that it violates FCC regulations. According to Scott, the maximum legal speed would be about 360 wpm.

Or does CW plausibly have to be human copiable?

Also, does QRQ CW require the operator to slow down to 20 WPM every 10 minutes to ID?

The answer to both of those points is no. Think of all the CW signals on the bands that are 40 WPM and higher. There are no present laws that require them to periodically ID at a lower speed.

  • 2
    $\begingroup$ For future readers: "the higher the data rate, the more bandwidth": naaaaah, that's a bit oversimplified. Given free choice of modulation, and given sufficient SNR, one could also increase the number of symbols you can pick your symbol from per channel use. BUT: Here, "CW" in this context means "morse", so that's something relatively rectangularly shaped with fixed alphabet, no choice there, so yes, we're dealing with Morse words per minutes as the only way to increase data rate, so yes, that increases bandwidth. $\endgroup$ Commented Mar 15, 2020 at 19:46
  • $\begingroup$ However, if one wanted to make the most out of the limited bandwidth and limited power given in terms of data rate, Morse wouldn't be the mode of choice. $\endgroup$ Commented Mar 15, 2020 at 19:47
  • $\begingroup$ Unless there was a legal distinction between the maximum allowed data rate of CW/Morse versus a (or another) digital or data modulation mode. $\endgroup$
    – hotpaw2
    Commented Mar 16, 2020 at 4:13
  • 1
    $\begingroup$ There is a legal limit to the speed of a US repeater's Morse code identification, but of course that doesn't apply to ordinary Morse code QSOs. $\endgroup$
    – rclocher3
    Commented Mar 16, 2020 at 14:43
  • $\begingroup$ The rise and fall times of CW dits and dahs have to be shaped to prevent key clicks. That is, we cannot use square waves; sharp corners on each side of the "plateau" (as seen on an oscilloscope) have to be rounded (have a slight radius). Otherwise, the bandwidth increases, causing QRM to nearby QSOs. At some point, we reach a speed where shaping must present a problem, resulting in key clicks. But I can't quite wrap my mind around the specifics. Any thoughts? $\endgroup$ Commented Mar 16, 2020 at 16:37

300 BAUD or elements per second on HF

Part 97.307(f) says that for a data emission (which would include our QRQq communications, since it would be unlikely you could send/receive such high speeds by yourself) the fastest rate must be 300 baud.

So if a dit takes up less than 1/300 of a second, you're breaking the law. 1 dit per second roughly translates to 2.4 wpm (source) so 2.4×300=720 wpm is the most legally allowed by FCC data rate limits.

I may not be correct, the other answers seem more heavily researched. I just want to put this out as a possibility. If you really want to know, you could always write to the FCC and ask. 73 AI7OW


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .