+++++++++++++++++ From: "Gregg R. Lengling" To: , Subject: Re: [Elecraft] PSK31 -- ghost image: Date: Fri, 14 Feb 2003 06:00:51 -0600 No there is nothing wrong...what you found was a poorly setup transmitting station. If you were to watch his IMD when he idles with no characters sent you would have seen a very poor IMD figure....chances are his signal was very wide on the waterfall also. He is overdriving his radio and causing such a wide signal and distortion that he is actually splattering. I've helped out a couple hams that didn't realize they were doing this.....a good clean PSK31 signal on the waterfall should be clean and less than 50 Hz wide...anymore and they defeat the purpose of the mode. Gregg R. Lengling, W9DHI PSYCHOLOGY: The science of inventing words for things that do not exist. ----- Original Message ----- From: "Martin AC6RM" To: Sent: Thursday, February 13, 2003 11:11 PM Subject: [Elecraft] PSK31 -- ghost image: > I've just started playing with PSK31 with the K2 and a rigblaster Nomic. > I noticed something strange today. A station appeared strongly just above > 14071 kHz, but the station also appeared almost 2kHz higher in the band as > well. I'm using digipan 1.6 which allows you to watch two QSOs at the > same time. They were both the same station -- character-for-character. > Is there something wrong with the way that I have set up my equipment? > > Tks in advance, and 73, > > Martin AC6RM K2#3021 ++++++++++++++++++ From: "Sandy W5TVW" To: "Gregg R. Lengling" , , Subject: Re: [Elecraft] PSK31 -- ghost image: Date: Fri, 14 Feb 2003 07:30:52 -0600 Some of this "ghost image" stuff seems to be a quirk of the computer sound card. Or even possibly some distortion in the receiver's audio system causing accentuation of the harmonics of the audio signal. I have noticed this even when a VERY clean CW signal is generated and it's spectrum is watched carefully. The "ghost signals" appear on the spectrum display in exact multiples of the fundamental audio tone, even if the audio level into the sound card is carefully adjusted until it "just appears" on the display. Usually when the transmitted PSK signal is being overdriven, "sidebands" appear on either side of the "railroad track" signal. I have also seen wide bands of spurious signals that appear over the portion of the spectrum being monitored at times. This is also sometimes accompanied with harsh sounding signal or a "hissing" note. Must PSK operators are unaware of this unless you TELL THEM that their signal is "dirty", but I hasten to add, that the "ghost signal" images you are viewing are not usually the fault of the transmitted signal, but flaws in the actual receiving setup/software/hardware. I was somewhat alarmed in first seeing this until I ran some tests and discovered what I have conveyed above! I would be curious to find out what happens to the audio signal while it is being processed in the sound card! To date, I am not brave enough to go probing around in the audio card with an oscilloscope! 73, Sandy W5TVW ----- Original Message ----- From: "Gregg R. Lengling" To: ; Sent: Friday, February 14, 2003 6:00 AM Subject: Re: [Elecraft] PSK31 -- ghost image: +++++++++++++++++ From: "Rich Lentz" To: , Subject: RE: [Elecraft] PSK31 -- ghost image: Date: Fri, 14 Feb 2003 16:28:45 -0600 Not an expert on this issue but have been told by informed PC-Audio people that this is caused by the PC sound card "oversampling." Ask some that knows more than me but is something about the analog to digital conversion technique and trying to provide a higher frequency response out of the available binary bits. Rich KE0X ++++++++++++++++ From: "Rich Lentz" To: Cc: "Elecraft Reflector" Subject: RE: [Elecraft] PSK31 -- ghost image: Date: Fri, 14 Feb 2003 19:23:04 -0600 Agreed, with a high quality soundcard the majority of any (all?) ghosts would be caused by overdriving the sound card. However, that does not explain the phenomena that I have observed with = my built-in (chip on motherboard) sound card and other inexpensive = soundcards. And I am looking at a ghost with the volume cranked down to the point = the main trace is barely readable right now. For me this usually occurs = during high traffic periods and when the tones are at certain combinations of frequencies. Rarely does it happen when there is only one or two = signals.=20 Do a search on "high fidelity digital audio conversion" ( or this = concept) on the internet. You will find some interfaces that cost any where from $500 to $2500 with specifications as to their quality and statements = that "no oversampling is used." With frequency responses up to 40 to 50 kHz. = To do that the sample rate must be no less than twice that - 64, 88.2, & 96 = kHz sample rates (higher if you want to pay MORE. These interfaces also = provide 32, 64, 128, and higher bit in the AD / DA converter.=20 Now look at the average PC soundcard and the PSK programs - digipan, = MixW, and the others. They are using 11025 Hz sample-rate. (And few go up to = 32 bit DAC.) That is just over twice the bandwidth displayed and about four times the K2 BW. That means that a 5 kHz tone is a stepped square-wave = and a 2 kHz tone is smoother. Speakers/headphones don't care - but digital displays don't lie - and show all of the artifacts. Rich KE0X ++++++++++++++++ From: "Rich Lentz" To: "'Jim W7RY'" Cc: "Elecraft Reflector" Subject: RE: [Elecraft] PSK31 -- ghost image: Date: Sun, 16 Feb 2003 09:16:44 -0600 I will agree that when you overdrive the transmitter you will produce = ghosts and all kind of other spurious signals. But, if the person(s) = transmitting causes all the "ghost(s)" that I observe how does that explain my observations. I receive ghosts that have the EXACT same frequency DIFFERENCE of = signals that are within the pass band of the K2, Icom 706, etc., HOWEVER, the = ghosts are more than 2 kHz outside of the pass band of the receiver. Right = now I am observing a pair of ghosts at 4.6 kHz and 4.9 kHz caused by two = signals at 1.1 kHz and 1.4 kHz. (Do you realize how strong of a signal would be required to receive a signal at 4.9 kHz (audio) on the K2?). The MixW = IMD reading on the signals causing the ghosts is -23 and -31, which leads me = to conclude neither of them are overdriving their rig. Further, when I = observe these ghosts (when caused by only two signals in the pass band), when = either one of the transmissions stops, BOTH of the ghosts disappear. =20 The same phenomena happens with three or more ghosts, with the exception that when the strongest transmission stops all of the ghosts disappear, = when either one of the weaker transmissions stop the position now depends = upon the new sum of audio frequencies. With my cheap sound card I can observe this with the volume of the K2 set at 8:00 or less (the main signals are barely visible and the audio from the PC speaker is also barely = audible). =20 =20 Rich KE0X ++++++++++++++++ Date: Sun, 16 Feb 2003 18:28:51 +0000 From: "Julian (G4ILO)" Subject: Re: [Elecraft] PSK31 -- ghost image: To: "Elecraft List" Many sound cards have sound "enhancement" features intended to give a pseudo-3D effect and suchlike. On some of the cheaper ones it may not be possible to turn this off. I've never encountered this myself but I've read about it. Goodness knows how they go about creating these effects but I can't imagine it does much good to a PSK signal. On top of that, the average sound card is not a high quality product and the A-D conversion may be anything but linear. Either of these things could be the cause of "ghosts" on the transmitted signal, even if the user has correctly adjusted the level, as well as on the received signal. The only real way to tell if you are in the clear is to transmit into a dummy load and monitor yourself. I did that once, and was impressed at how clean the K2 was. However I no longer have a second ham receiver so it is no longer possible. For most people, I guess, the difficulty in doing this will be not having a second computer. 73, -- Julian, G4ILO. (RSGB, ARRL, G-QRP, K2 #392) G4ILO's Shack: http://www.qsl.net/g4ilo Rich Lentz wrote: I will agree that when you overdrive the transmitter you will produce = ghosts and all kind of other spurious signals. But, if the person(s) = transmitting causes all the "ghost(s)" that I observe how does that explain my observations. ++++++++++++++++ From: "Rich Lentz" To: "'Don Wilhelm'" , "Elecraft Reflector" Subject: RE: [Elecraft] PSK31 -- ghost image: Date: Sun, 16 Feb 2003 15:07:42 -0600 Don: My responses on the reflector were to someone else's question as to why there are ghosts. Several have stated that the ghosts are caused by the original transmission. I find this hard to believe. How can the transmission cause a ghost at 4.9 kHz? A real signal here would be more than 90 db down after going through the K2 band pass. Also, If I = re-center the received frequency on this ghost - there is nothing there. = Therefore it is caused by the soundcard. As far as level, I have the slider for the aux input set at about 25%. = With this setting and default MixW or Digipan colors and a divider pot in my interface set at about 10% going to the PC, it takes the volume control = on the K2 set at 12:00 to get a red trace on the waterfall. I really don't think I am overdriving the PC sound card. And I can get ghosts with the = K2 volume set to less than 8:00. =20 My opinion is that the sampling rate of the sound card generates false artifacts of the original tones. This explains why the ghosts are the = same distance apart as two of the stronger signals in the pass band. These = are rarely observed with a PURE sine wave. However, when you have multiple = PSK signals these will introduce false signals caused by the nonlinearity of = the DAC. You are only using 12 - 24 bits to represent the signals and these = are taken at an 11025 Hz sample rate. Get out your college math book on = Fourier analysis and do the math. (I am a math major) You will find that the = DAC act as a nonlinear device and generates these false signals. That's why high quality CD's use 96 kHz (and higher) sample rates and 64, 128, or higher bit DAC's. Rich KE0X +++++++++++++++++