Re: [Elecraft] [K3] Rx/Tx equalization
True story related to your comment below ... I worked my way through law school in the 1970's as a transmitter engineer at a 50 KW AM directional station in the Detroit market. The studio was 15 miles from the transmitter, connected by main and backup broadcast quality telephone lines. One afternoon I'm watching the meters and listening to the monitor speaker when I hear a series of beeps, sounds like a string of Morse dots and then the main telco line goes dead. Switch to the backup line and let the Chief Engineer know. A half hour or so later, same thing happens to the backup line, so crossing my fingers, I switched back to the main and it was working. Later that day, the jocks started complaining that the audio sounded muddy. All the guys at the transmitter thought it sounded fine to them. For those who have not worked in this environment, a 50 KW transmitter room is not silent, as it has several multi-horsepower blowers to keep the transmitter cool. And, the monitor speaker was not what one would use for serious audio reference. We then ran a sweep of the lines that night and sure enough, instead of the 15 KHz response the station paid for, the line cut off around 3.5 KHz. Both of them. It turns out that the beeps were the pair identification tones the telco splicers use when working on cables and what they had done was add loading coils to the program lines, converting them to a standard toll grade quality line. That experience taught me my ears are not a precision calibrated instrument and that I should have more faith in spectrum analyzers. Jack K8ZOA With ANY audio adjustments, the final test is what it SOUNDS like. Good operators ALWAYS get reports from good ears on the other end of the QSO after they've done the first round of tweaking. The broadcast guys tweak and listen, tweak and listen, tweak and listen. ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
Re: [Elecraft] [K3] Rx/Tx equalization
On Mon, 25 Aug 2008 13:09:06 -0400, Jack Smith wrote: > What do we measure and what do we tweak? The best answer I can give is to suggest that you study (not just read) the tutorial. Then study the sections of the K3 manual that describe the equalization and the processing. Once you really understand the concepts (the tutorial), it becomes obvious that Elecraft has done an excellent job of providing all the equalization and processing you need to make your audio sound great and be very competitive. It's also very easy to set up, once you understand the concepts. BTW -- there's also some additional graphs of how mics behave in http://audiosystemsgroup.com/HamInterfacing.pdf I'm currently working on adding them, along with the associated discussion, to the tutorial. With ANY audio adjustments, the final test is what it SOUNDS like. Good operators ALWAYS get reports from good ears on the other end of the QSO after they've done the first round of tweaking. The broadcast guys tweak and listen, tweak and listen, tweak and listen. 73, Jim Brown K9YC ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
Re: [Elecraft] [K3] Rx/Tx equalization
Jim: Very interesting discussion. For those of us who are mostly data mode and CW operators and might make a handful of SSB contacts in a good year, how should we go about adjusting the transmit frequency shaping parameters? What do we measure and what do we tweak? At one extreme would be adjust until it sounds good, but that rubs against my analytical side. Looking at the audio output in monitor mode with a spectrum analyzer misses parts of the audio chain ahead of the transceiver. Ultimately, of course, some "make it sound good" tweaking is necessary, but I'm looking for some numerical guidance. E.g., equalize so that the spectra below 200 Hz is X dB down from midrange peaks or whatever. Jack K8ZOA ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
Re: [Elecraft] [K3] Rx/Tx equalization
On Mon, 25 Aug 2008 07:04:21 -0700, Mike Scott wrote: >Jim (K9YC), what part of the 50 Hz spectrum will get by a 2.7 KHz roofing >filter at any equalization setting? At least on my system I could put 50 Hz >at +16 dB gain and 50 Hz will still be down 20 dB. The DSP skirts put it >lower. I ask this question because I have been staring at my actual roofing >filter plots. >So I don't quite understand where you are coming from. Perhaps the devil is >in the details and breath-pop energy which would be cut off by the roofing >filter/DSP combination is still affecting the transmit-audio gain in >deleterious ways. Maybe it is just my lack of understanding of how the K3 >works. In this area it is a bit of a black box to me, the price we pay for >not soldering. Hi Mike, You're looking at the total signal path from mic in to RF out, but you're not considering what happens in the audio blocks that precede the transmitter. Good audio processing for communications uses careful equalization to shape the audio response for maximum use of the transmitter bandwidth, as well as compression and peak limiting to bring softer parts of the transmitted speech up to nearly full modulation level. There's also VOX, that looks at the transmit audio and turns the transmitter on when it sees audio. The compressor, limiter, and VOX all look at transmit audio to decide how to do their thing. If your microphone is producing lots of output on low frequency energy, it will cause the compressor and/or limiter to turn down the gain, reducing your modulation. That LF energy can also trigger the VOX prematurely. Yes, the transmit crystal filter imposes some hard limits on the transmitted bandwidth, but all of that is FAR up the signal chain from the audio processing noted above. The audio chain can suffer from these problems, and it can also produce distortion if over driven. Another point -- proximity effect is a LARGE effect when it is present, easily 15-20 dB of LF boost. It's only present in directional mics, but most pro mics suffer from it. If you're using one of those mics, it's easy for breath pops and other LF noise to be 10-20 dB louder than your voice. So your audio gain may be just fine for your transmitted speech, but pops and LF noise may be distortiing in the audio chain. The value of equalization and dynamics processing for ham radio are discussed in an appendix of http://audiosystemsgroup.com/RFI-Ham.pdf. Good audio processing -- equalization and compression/limiting -- can increase your effective talk power by 6-10 dB! That's equivalent to adding a 400W - 1,000W amp to your K3. Broadcasters learned this MANY years ago -- all of them are carefully equalized and processed. 73, Jim Brown K9YC ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
[Elecraft] [K3] Rx/Tx equalization
>50 Hz and 100 Hz octave bands are VERY important -- they allow equalization to correct for proximity effect in directional microphones, and to reduce the effects of breath pop. Jim (K9YC), what part of the 50 Hz spectrum will get by a 2.7 KHz roofing filter at any equalization setting? At least on my system I could put 50 Hz at +16 dB gain and 50 Hz will still be down 20 dB. The DSP skirts put it lower. I ask this question because I have been staring at my actual roofing filter plots. So I don't quite understand where you are coming from. Perhaps the devil is in the details and breath-pop energy which would be cut off by the roofing filter/DSP combination is still affecting the transmit-audio gain in deleterious ways. Maybe it is just my lack of understanding of how the K3 works. In this area it is a bit of a black box to me, the price we pay for not soldering. Mike Scott - AE6WA Tarzana, CA (DM04 / near LA) K3-100 #508/ KX1 #1311 ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
Re: [Elecraft] [K3] Rx/Tx equalization band selection
On Sat, 23 Aug 2008 16:03:40 -0700, Mike Scott wrote: >band centers of: 150, 250, 416, 693, 1154, 1921 and 3200 wit The centers of audio filters in octave, half-octave, one-third octave, and one-sixth octave increments were standardized many years ago by scientists and engineers working in audio and acoustics. Unlike the communications standard for the audio bandwidth of communications systems, these are excellent standards that take into account how humans HEAR sound. We tamper with these standards at our own risk. >It would appear that the bottom equalizer band is beyond useful. >Even with maximum equalization gain the 50 Hz spectrum is still >way down. This tells me that we don't really have an 8-band >equalizer as only 7 bands have utility. The 100 Hz band is >questionable but barely within the zone. 50 Hz and 100 Hz octave bands are VERY important -- they allow equalization to correct for proximity effect in directional microphones, and to reduce the effects of breath pop. The octave bands below 500 Hz contribute little to communications. If allowed to modulate our transmitters, they waste transmitter power. The best designers of sound systems for reverberant spaces know this well -- we carefully roll off the low end of speech systems for big churches beginning somewhere between 200 Hz and 300 Hz. 73, Jim Brown K9YC ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
Re: [Elecraft] [K3] Rx/Tx equalization band selection
Mike, All that effort may be an exercise in futility. If I read your report correctly, you measured the roofing filter only. The full story is with both the roofing filter and the DSP filter engaged. That will tell the story about what you hear, not just what the input to the DSP looks like. 73, Don W3FPR Mike Scott wrote: After plotting filter response on my 2.7 KHz filter I have been thinking about the Rx/Tx eq settings. The current equalization scheme is 7 octave bands at: 50 Hz, 100 Hz, 200 Hz, 400 Hz, 800 Hz, 1600 Hz, and 3200 Hz plus an interloper non-octave band at 2400 Hz. I can guess why these bands, they match a popular external microphone equalizer product. My filter is down 20 dB at approximately 100 Hz (from zero beat.) This number is from graphical construction where I only took data points every 100 Hz and plotted the results. It would appear that the bottom equalizer band is beyond useful. Even with maximum equalization gain the 50 Hz spectrum is still way down. This tells me that we don't really have an 8-band equalizer as only 7 bands have utility. The 100 Hz band is questionable but barely within the zone. Perhaps Wayne is going to open the lower end some more and then my issue could become moot and everybody would have to go buy a woofer. However whenever I try to equalize waterfall speckling I wish I had more granularity in the useful part of the audio spectrum. If we wanted to start at 100 Hz and split the band between 100 Hz and 3200 Hz into 7 equal fractional octave bands then sub octave processing would do that with band centers of 100, 178, 317, 566, 1008, 1796, 3200; then keeping with the theme, throw in an interloper band around 2400 or someplace useful to make an 8-band processor. The octave band centers above are 100 Hz multiples of the 6th root of 32 (3200/100.) The current scheme is 50 Hz multiples of the 6th root of (3200/50) which is the magical 2 factor. Starting at 150 Hz instead of 100 (my favorite) may also make sense leading to band centers of: 150, 250, 416, 693, 1154, 1921 and 3200 with an extra one at 2500. Since these are broad anyway, why not call them 150, 250, 400, 700, 1200, 1900, 2500 and 3200 and give people the option of maintaining the current scheme or shifting to one of these? I kind of like having one of the filters centered on 700 Hz anyway. My 2 cents Mike Scott - AE6WA Tarzana, CA (DM04 / near LA) K3-100 #508/ KX1 #1311 ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
[Elecraft] [K3] Rx/Tx equalization band selection
After plotting filter response on my 2.7 KHz filter I have been thinking about the Rx/Tx eq settings. The current equalization scheme is 7 octave bands at: 50 Hz, 100 Hz, 200 Hz, 400 Hz, 800 Hz, 1600 Hz, and 3200 Hz plus an interloper non-octave band at 2400 Hz. I can guess why these bands, they match a popular external microphone equalizer product. My filter is down 20 dB at approximately 100 Hz (from zero beat.) This number is from graphical construction where I only took data points every 100 Hz and plotted the results. It would appear that the bottom equalizer band is beyond useful. Even with maximum equalization gain the 50 Hz spectrum is still way down. This tells me that we don't really have an 8-band equalizer as only 7 bands have utility. The 100 Hz band is questionable but barely within the zone. Perhaps Wayne is going to open the lower end some more and then my issue could become moot and everybody would have to go buy a woofer. However whenever I try to equalize waterfall speckling I wish I had more granularity in the useful part of the audio spectrum. If we wanted to start at 100 Hz and split the band between 100 Hz and 3200 Hz into 7 equal fractional octave bands then sub octave processing would do that with band centers of 100, 178, 317, 566, 1008, 1796, 3200; then keeping with the theme, throw in an interloper band around 2400 or someplace useful to make an 8-band processor. The octave band centers above are 100 Hz multiples of the 6th root of 32 (3200/100.) The current scheme is 50 Hz multiples of the 6th root of (3200/50) which is the magical 2 factor. Starting at 150 Hz instead of 100 (my favorite) may also make sense leading to band centers of: 150, 250, 416, 693, 1154, 1921 and 3200 with an extra one at 2500. Since these are broad anyway, why not call them 150, 250, 400, 700, 1200, 1900, 2500 and 3200 and give people the option of maintaining the current scheme or shifting to one of these? I kind of like having one of the filters centered on 700 Hz anyway. My 2 cents Mike Scott - AE6WA Tarzana, CA (DM04 / near LA) K3-100 #508/ KX1 #1311 ___ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
