Re: [Elecraft] I/Q details for the KX3.
But I think that SDRs that digitize at the VFO frequency itself (I believe this is generally called "direct sampling" on RX and "direct digital synthesis" on TX?) can get away with a single channel, since there's no mixer to cause the "you mix A and B and get both A + B and A - B even though you only wanted one of them" problem. Yes, but... You still need to reduce the sample rate to a lower value that the DSP can handle. Either you need to mix down to a low-frequency IF or all the way down to an I/Q signal at a zero Hz center frequency. Either way, the mixer needs to be quadrature to avoid the image. The output of the ADC goes to a quadrature digital downconverter (DDC), which consists of two digital mixers (that multiply the ADC signal times the digital local oscillator), with the local oscillator to each mixer 90 degrees out of phase. In this P3 this is done in an ASIC (application-specific IC) and in the K4 it is done in an FPGA. Alan N1AL On 6/22/22 20:45, Julia Tuttle wrote: Mike, I think you're right that SDRs that digitize at a zero IF (like the KX3) need I and Q channels to distinguish the AF sidebands. That is, if the radio mixes 7030.4 kHz down to a sidetone of 0.4 kHz, it also catches signals at 7029.6 kHz. And I think that SDRs that digitize at a non-zero and 'properly intermediate' IF (that is, neither 0 nor the VFO frequency) also need I and Q channels to distinguish the IF 'sidebands'. That is, if the radio mixes 7030 kHz down to an IF of 455 kHz, it also catches signals at 6120 kHz. But I think that SDRs that digitize at the VFO frequency itself (I believe this is generally called "direct sampling" on RX and "direct digital synthesis" on TX?) can get away with a single channel, since there's no mixer to cause the "you mix A and B and get both A + B and A - B even though you only wanted one of them" problem. Does this match your understanding? Thanks, Julie On Wed, Jun 22, 2022 at 2:04 PM Mike Markowski wrote: Jerome, This answer is for generic SDRs, since I don't have a KX3. An SDR always must use both I and Q, even for CW. I and Q are two streams of samples of the same signal. One of many advantages to i/q sampling is the ADC's can run slower (cheaper). In the extreme and ignoring the intrusion of real life components on sampling, a 10 kHz wide signal could be sampled with I and Q ADCs running at 10 kHz. Nyquist is satisfied because you have 2 streams, or 20 kHz sampling in this example. Aliasing prevents you from using just I in that case. The fact that Q is 90 deg from I, means we can plot them on axes 90 deg apart. That allows re-appropriating the entire field of complex mathematics to work with the samples. I and Q are both physical - trying to avoid the word 'real'! - streams of samples, but I can be called real and Q can be called imaginary. Using Euler's Formula, they can be bundled into a concise exp(j phi) formulation to work with. I/Q imbalance is always a concern. Maybe others can quantify it better than me saying smaller is better. :-) 73 es GL! Mike ab3ap On 6/21/22 11:08 PM, JEROME SODUS wrote: Hello, Although I have asked some of these questions on another forum, here I have expanded on them a bit. When the KX3 transmits a CW-signal, are both I and Q used? If so, why would that be necessary? (My guess is only the I is needed.) Same questions too for a SSB-signal. But, for any digital transmission like RTTY, FT8, or Olivia, wouldn't both I and Q would be active? The LO is used for the "In Phase" and is delayed by 90 degrees for the "Quadrature"; how is that delay done? How much can the delay deviate from 90 degrees and still be useful? My reason for asking is that I'm to give a talk to the Radio Club in August about how a KX3 works and want to mention these details. TIA for any replies. 73 jerry km3kKX3#6088 __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to mike.ab...@gmail.com __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to ju...@juliatuttle.net __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to a...@elecraft.com
Re: [Elecraft] I/Q details for the KX3.
Mike, I think you're right that SDRs that digitize at a zero IF (like the KX3) need I and Q channels to distinguish the AF sidebands. That is, if the radio mixes 7030.4 kHz down to a sidetone of 0.4 kHz, it also catches signals at 7029.6 kHz. And I think that SDRs that digitize at a non-zero and 'properly intermediate' IF (that is, neither 0 nor the VFO frequency) also need I and Q channels to distinguish the IF 'sidebands'. That is, if the radio mixes 7030 kHz down to an IF of 455 kHz, it also catches signals at 6120 kHz. But I think that SDRs that digitize at the VFO frequency itself (I believe this is generally called "direct sampling" on RX and "direct digital synthesis" on TX?) can get away with a single channel, since there's no mixer to cause the "you mix A and B and get both A + B and A - B even though you only wanted one of them" problem. Does this match your understanding? Thanks, Julie On Wed, Jun 22, 2022 at 2:04 PM Mike Markowski wrote: > Jerome, > > This answer is for generic SDRs, since I don't have a KX3. > > An SDR always must use both I and Q, even for CW. I and Q are two > streams of samples of the same signal. One of many advantages to i/q > sampling is the ADC's can run slower (cheaper). In the extreme and > ignoring the intrusion of real life components on sampling, a 10 kHz > wide signal could be sampled with I and Q ADCs running at 10 kHz. > Nyquist is satisfied because you have 2 streams, or 20 kHz sampling in > this example. Aliasing prevents you from using just I in that case. > > The fact that Q is 90 deg from I, means we can plot them on axes 90 deg > apart. That allows re-appropriating the entire field of complex > mathematics to work with the samples. I and Q are both physical - > trying to avoid the word 'real'! - streams of samples, but I can be > called real and Q can be called imaginary. Using Euler's Formula, they > can be bundled into a concise exp(j phi) formulation to work with. > > I/Q imbalance is always a concern. Maybe others can quantify it better > than me saying smaller is better. :-) > > 73 es GL! > Mike ab3ap > > On 6/21/22 11:08 PM, JEROME SODUS wrote: > > Hello, > > > > Although I have asked some of these questions on another forum, here I > have expanded on them a bit. > > > > When the KX3 transmits a CW-signal, are both I and Q used? > > If so, why would that be necessary? > > (My guess is only the I is needed.) > > > > Same questions too for a SSB-signal. > > > > But, for any digital transmission like RTTY, FT8, or Olivia, wouldn't > both I and Q would be active? > > > > The LO is used for the "In Phase" and is delayed by 90 degrees for the > "Quadrature"; how is that delay done? > > How much can the delay deviate from 90 degrees and still be useful? > > > > My reason for asking is that I'm to give a talk to the Radio Club in > August about how a KX3 works and want to mention these details. > > > > TIA for any replies. > > 73 jerry km3kKX3#6088 > > __ > > Elecraft mailing list > > Home: http://mailman.qth.net/mailman/listinfo/elecraft > > Help: http://mailman.qth.net/mmfaq.htm > > Post: mailto:Elecraft@mailman.qth.net > > > > This list hosted by: http://www.qsl.net > > Please help support this email list: http://www.qsl.net/donate.html > > Message delivered to mike.ab...@gmail.com > __ > Elecraft mailing list > Home: http://mailman.qth.net/mailman/listinfo/elecraft > Help: http://mailman.qth.net/mmfaq.htm > Post: mailto:Elecraft@mailman.qth.net > > This list hosted by: http://www.qsl.net > Please help support this email list: http://www.qsl.net/donate.html > Message delivered to ju...@juliatuttle.net > __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to arch...@mail-archive.com
Re: [Elecraft] I/Q details for the KX3.
Jerome, This answer is for generic SDRs, since I don't have a KX3. An SDR always must use both I and Q, even for CW. I and Q are two streams of samples of the same signal. One of many advantages to i/q sampling is the ADC's can run slower (cheaper). In the extreme and ignoring the intrusion of real life components on sampling, a 10 kHz wide signal could be sampled with I and Q ADCs running at 10 kHz. Nyquist is satisfied because you have 2 streams, or 20 kHz sampling in this example. Aliasing prevents you from using just I in that case. The fact that Q is 90 deg from I, means we can plot them on axes 90 deg apart. That allows re-appropriating the entire field of complex mathematics to work with the samples. I and Q are both physical - trying to avoid the word 'real'! - streams of samples, but I can be called real and Q can be called imaginary. Using Euler's Formula, they can be bundled into a concise exp(j phi) formulation to work with. I/Q imbalance is always a concern. Maybe others can quantify it better than me saying smaller is better. :-) 73 es GL! Mike ab3ap On 6/21/22 11:08 PM, JEROME SODUS wrote: Hello, Although I have asked some of these questions on another forum, here I have expanded on them a bit. When the KX3 transmits a CW-signal, are both I and Q used? If so, why would that be necessary? (My guess is only the I is needed.) Same questions too for a SSB-signal. But, for any digital transmission like RTTY, FT8, or Olivia, wouldn't both I and Q would be active? The LO is used for the "In Phase" and is delayed by 90 degrees for the "Quadrature"; how is that delay done? How much can the delay deviate from 90 degrees and still be useful? My reason for asking is that I'm to give a talk to the Radio Club in August about how a KX3 works and want to mention these details. TIA for any replies. 73 jerry km3kKX3#6088 __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to mike.ab...@gmail.com __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to arch...@mail-archive.com
Re: [Elecraft] I/Q details for the KX3.
On receive, if you only mix the RF signal with only the I phase of the VFO,
you receive both sidebands, and you can't observe them separately. On
transmit, you have the same problem in reverse -- you transmit both
sidebands, and you can't control them separately.
For modes with a single sideband (SSB, DATA besides RTTY), the KX3 must
output the AF signal on both I and Q with the proper phase offset.
If it output the AF signal on I but left Q at 0 V, the TX mixer would
generate a mirror image (DSB-SC) signal.
(These are the "SSB and some data modes" covered by the TXSBNUL calibration
procedure on page 34.)
For modes that transmit a single frequency (CW, RTTY), the KX3 uses
*neither* I nor Q -- it leaves them at 0 V and controls the CW frequency
entirely with the VFO.
If it synthesized a constant sidetone frequency on I but left Q at 0 V, the
TX mixer would generate two CW signals, at the VFO frequency +/- the
sidetone frequency.
If it synthesized a constant sidetone frequency on I *and* Q with the
proper phase offset, it would also get a CW signal, but that would be
unnecessarily complex.
(These are the modes not covered by TXSBNUL.)
Digital modes like FT8 and Olivia (and RTTY if sent from a computer) are
handled just like SSB signals, just with some potentially distorting
features (compressor, equalizer, etc.) disabled.
The LO generates both the I and Q signals ("The synthesized, digitally
controlled local oscillator (LO) provides quadrature signals to the
transmit and receive mixers", page 49) -- there's no separate delay, so I
expect the phase difference would be quite accurate. I'm not sure how
accurate it actually is, nor how much it can deviate before it causes
problems.
If you haven't yet, I'd take a look at the Theory of Operation (pages
49-50) and KX3 Block Diagram (page 51) in the manual -- that's what I'm
basing my statements here on.
Related to how the KX3 handles RTTY internally, you might find the QRP Labs
QDX Digital Transceiver (https://qrp-labs.com/qdx.html) an interesting
architectural contrast. The KX3 handles CW and RTTY by simply leaving the
VFO output unmixed; the QDX handles *all* single-frequency digital modes
that way. It exposes an audio output over USB, but instead of running the
TX AF signal through a DAC and a mixer, it analyzes it in software to
determine which single frequency it contains, and adjusts the VFO to follow.
I hope this is helpful!
73,
Julie
On Wed, Jun 22, 2022 at 11:56 AM JEROME SODUS wrote:
> Hello,
>
> Although I have asked some of these questions on another forum, here I
> have expanded on them a bit.
>
> When the KX3 transmits a CW-signal, are both I and Q used?
> If so, why would that be necessary?
> (My guess is only the I is needed.)
>
> Same questions too for a SSB-signal.
>
> But, for any digital transmission like RTTY, FT8, or Olivia, wouldn't both
> I and Q would be active?
>
> The LO is used for the "In Phase" and is delayed by 90 degrees for the
> "Quadrature"; how is that delay done?
> How much can the delay deviate from 90 degrees and still be useful?
>
> My reason for asking is that I'm to give a talk to the Radio Club in
> August about how a KX3 works and want to mention these details.
>
> TIA for any replies.
> 73 jerry km3kKX3#6088
> __
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Elecraft@mailman.qth.net
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
> Message delivered to ju...@juliatuttle.net
>
__
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:Elecraft@mailman.qth.net
This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to arch...@mail-archive.com
[Elecraft] I/Q details for the KX3.
Hello, Although I have asked some of these questions on another forum, here I have expanded on them a bit. When the KX3 transmits a CW-signal, are both I and Q used? If so, why would that be necessary? (My guess is only the I is needed.) Same questions too for a SSB-signal. But, for any digital transmission like RTTY, FT8, or Olivia, wouldn't both I and Q would be active? The LO is used for the "In Phase" and is delayed by 90 degrees for the "Quadrature"; how is that delay done? How much can the delay deviate from 90 degrees and still be useful? My reason for asking is that I'm to give a talk to the Radio Club in August about how a KX3 works and want to mention these details. TIA for any replies. 73 jerry km3kKX3#6088 __ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to arch...@mail-archive.com
