I don't want any of my energy going outside the filter or coming in from outside. I would swap 1,2 and always do. It probably makes a tenth dB difference altogether but I don't want even that error.

Bob
N4HY



Gerald Youngblood wrote:

--
Laziness is the number one inspiration for ingenuity. Guilty as charged!

Thanks for the offer, it would be nice to compare.

I'm not going around buying sound cards for the fun of it, I own several sound cards (most free) and I would like to know how well they do in my setup. I bought the Delta-44 because that one is the card that is supposed to have really good qualities and can be used for both receive and transmit, which I intend to experiment with. I have the following cards;

5 SB1024I Pci 24 bits
4 SB Live Pci 16 bits
2 SB MP+ USB 16 bits
1 SB Live 24 Pci 24 bits
1 SB Live 24 USB 24 bits
1 SB Audigy 2 ZS Pci 24 bits
1 Delta-44 Pci 24 bits, on it's way
1 CHAINTECH AV-710 Pci 16/24 bits, just got it today.
1 iMic USB 16 bits? it's on my Mac but it can move
1 Philipps 7:1 Pci 24 bits, took me forever to find the right drivers.
1 Weirdo Brand Pci 32 bits 192KHz claims 129 dB range, eats CPU cycles like there is no tomorrow, if I can find the darn thing, it's in a box somewhere.
Various built in sound cards, most are AC97 based 20 bits

I'm sure if I look around in the garage I can find some additional models packed in boxes, but this gives you an idea.

The Delta-44 is going in the main Dell PC(2) on my workbench next to my radios, but I would like to play with SDR on the PC's I have in the den(4), and in the living room(2). I have a lot of PC's, that is not counting two MAC', nor my son's PC's(3). No! I don't have them all turned on at the same time, usually no more than three at a time.

Crazy huh?


At 04:25 PM 10/27/2005, you wrote:

If you are interested, Gerald may be willing to share info we have here
on several sound cards. I'm sure he has MDS numbers for at least a few
of them.

For what it's worth, you could probably just measure the MDS of the
SDR-1000 and use those numbers combined with MDS figures of soundcards
that are well documented. That might keep you from having to buying one
of each card for your tests.

Eric

-----Original Message-----
From: KD5NWA [mailto:kd5nwa@...]
Sent: Thursday, October 27, 2005 4:17 PM
To: eric@...
Subject: RE: [Flexradio] Turning off the AGC

That is precisely what I'm trying to measure, the MSD of the entire
system, when you listen to a signal you are using the entire system,
front end, sound card, PC and all. If I was designing a new front
end, I would be interested in the figures for it alone, but that is
for another day.

I can then compare the readings of various sound cards and see which
one makes the system as a whole behaves the best. I have a Delta-44
coming and hopefully will be here by Tuesday, hopefully it will get
the best numbers since it's the most expensive.

Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the same results every time is insanity: I've almost proved it isn't; only a few more tests now and I'm sure results will differ this time ...

开云体育

Yes and NO.

It turns off all variable gain but it turns on a FIXED gain.

Go the AGC/ALC setup tab and set this fixed gain to ZERO 0dB and now you are all set.

Bob
N4HY


KD5NWA wrote:

--
Laziness is the number one inspiration for ingenuity. Guilty as charged!

No. Each of the capacitors is charged around the 0-2pi circle at different turn on and different turn off phases. 0,90,180,270 degrees but each capacitor is turned on and off only ONCE per carrier frequency cycle. If you use a Johnson counter to drive this sampling clock then it needs to be 4X to turn each cap on and off at the correct phases.

Bob
N4HY


John H. Fisher wrote:

--
Laziness is the number one inspiration for ingenuity. Guilty as charged!

--- In softrock40@..., KD5NWA <KD5NWA@c...> wrote:

What I would need to know what is the maximum gain that the software
would use while listening to weak signals.

I think it is settable in the agc section of the setup screen.

73 de Phil N8VB

I'm not trying to measure the noise floor, I'm trying to measure the lowest discernible signal "MSD"

A voltage reading on a meter by itself means nothing since it's not being compared to a known quantity.

The procedure is;

Run the radio at maximum gain with a 500 Hz filter
Measure the Audio output with a RMS meter, and a 1 micro-volt signal in the pass-band, set for maximum audio output.
Remove the 1 microvolt signal and measure the Audio output with the meter
Do some simple math with the two figures and you come up with MSD rating of the radio.

But for this to be accurate the radio must be running at maximum gain, not minimum gain. Once you have that figure you can figure out the noise floor in dB.

What I would need to know what is the maximum gain that the software would use while listening to weak signals.

At 02:35 PM 10/27/2005, Eric Wachsmann - FlexRadio wrote:

Cecil,

If you are trying to measure the noise floor of the unit, the easiest
thing to do is to just connect a dummy load, open the software and put
it in 500Hz BW. Look at the multimeter (might help to put it in Sig
Avg. mode). This is the effective noise floor of the unit (assuming
that it is higher than the sound card noise floor).

To answer your question more specifically, turning the AGC to Fixed and
setting the Fixed Gain on the Setup Form to 0dB will effectively set the
processed audio to have no gain, variable or otherwise. The spectrum
and multimeter are both before the AGC however.


Eric Wachsmann
FlexRadio Systems

-----Original Message-----
From: FlexRadio-bounces@...

[<mailto:FlexRadio-bounces@flex->mailto:FlexRadio-bounces@flex-

radio.biz] On Behalf Of KD5NWA
Sent: Thursday, October 27, 2005 12:50 PM
To: Flex Radio
Cc: softrock40@...
Subject: [Flexradio] Turning off the AGC

I want to try to measure the noise floor of my SR-40 setup and will
need to have the AGC completely turned off.

If I turn off the AGC of the Flex software will it really completely
turn off all forms of variable gain?


Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the
same results every time is insanity: I've almost proved it isn't;
only a few more tests now and I'm sure results will differ this time

...

_______________________________________________
FlexRadio mailing list
FlexRadio@...

Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the same results every time is insanity: I've almost proved it isn't; only a few more tests now and I'm sure results will differ this time ...

Hi Milt,

Sounds interesting... Do you have a circuit diagram of this
arrangement anywhere?

Thanks.

73 de Phil N8VB


--- In softrock40@..., Milt Cram <w8nue@a...> wrote:

Here's a question and a comment for Phil, Alex, Tony, et al--

In the SoftRock40 the transformer provides both non-inverted and
inverted RF to the switches. The outputs of the FFs are gated so that
the switches are operated only over one quarter cycle each (90
degrees). /Why not operate the four switches as two SPDT switches and
eliminate the gating of the two FF outputs?/ This would provide two
"full wave rectified" outputs, instead of two "half wave rectified"
outputs, and therefore a 6dB gain improvement. I use such an
arrangement in my "PC-Less" PSK31 interface to generate my I and Q

outputs.

I'm also puzzled at the use of the inverting inputs of the audio
amplifiers. Since these are very low impedance points in the circuit,
the drive to the amplifiers is essentially a current. However, the
level of the current is very dependent on the total

resistance/impedance

in the circuit. In this case, it is the sum of the two 10 ohm
resistors, plus the on-resistance of the switch, plus the transformed
source impedance. I haven't looked at the specs for the switch, but if
there is much variation in on-resistance between the individual
switches, it could contribute to gain inbalance between the

channels. Is

the intention of this to give a good impedance match to the 50 ohm

source?

Regards,
Milt, W8NUE


--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.362 / Virus Database: 267.12.5/150 - Release Date:

10/27/2005

Here's a question and a comment for Phil, Alex, Tony, et al--

In the SoftRock40 the transformer provides both non-inverted and inverted RF to the switches. The outputs of the FFs are gated so that the switches are operated only over one quarter cycle each (90 degrees). /Why not operate the four switches as two SPDT switches and eliminate the gating of the two FF outputs?/ This would provide two "full wave rectified" outputs, instead of two "half wave rectified" outputs, and therefore a 6dB gain improvement. I use such an arrangement in my "PC-Less" PSK31 interface to generate my I and Q outputs.

I'm also puzzled at the use of the inverting inputs of the audio amplifiers. Since these are very low impedance points in the circuit, the drive to the amplifiers is essentially a current. However, the level of the current is very dependent on the total resistance/impedance in the circuit. In this case, it is the sum of the two 10 ohm resistors, plus the on-resistance of the switch, plus the transformed source impedance. I haven't looked at the specs for the switch, but if there is much variation in on-resistance between the individual switches, it could contribute to gain inbalance between the channels. Is the intention of this to give a good impedance match to the 50 ohm source?

Regards,
Milt, W8NUE


--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.362 / Virus Database: 267.12.5/150 - Release Date: 10/27/2005

SWCad will allow you to output a wave file from the simulation.

.wave c:&#92;mysound.wav 16 48k V(left) V(right) is the directive.

The cool part is that you can play the wave files in PowerSDR. As a
test I generated two sine 10khz sine waves, one at 0 deg and the other
at 90 deg. Then I created a phase error by increasing the 90 deg
signal to 110 deg. Using the manual I/Q correction I found that
PowerSDR was able to correct for this 20 deg phase imbalance.

It should be posssible to do all kinds of simulations and then
actually test them in PowerSDR.



73 de Phil N8VB

Now that softrock-40 mania has passed and softrock Mark 5 mania is approaching, will be soon had SDR-908 mania??

George, how can you even begin to estimate how many of these to make available in the first run??? I can't even imagine the demand, it's a little frightening to even think about.

GL.

Art



At 09:04 PM 10/26/2005, you wrote:

In a real radio yes, but I want to make sure that it is running flat out full gain on the sound card.

At 01:04 PM 10/27/2005, Tayloe Dan-P26412 wrote:

It seems to me that if you use MDS measurements (3 db S+N/N audio shift) to measure the noise floor, the AGC will take care of itself.

- Dan, N7VE

-----Original Message-----
From: FlexRadio-bounces@... [mailto:FlexRadio-bounces@...] On Behalf Of KD5NWA
Sent: Thursday, October 27, 2005 10:50 AM
To: Flex Radio
Cc: softrock40@...
Subject: [Flexradio] Turning off the AGC

I want to try to measure the noise floor of my SR-40 setup and will need to have the AGC completely turned off.

If I turn off the AGC of the Flex software will it really completely turn off all forms of variable gain?


Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the
same results every time is insanity: I've almost proved it isn't;
only a few more tests now and I'm sure results will differ this time ...



_______________________________________________
FlexRadio mailing list
FlexRadio@...

Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the same results every time is insanity: I've almost proved it isn't; only a few more tests now and I'm sure results will differ this time ...

I want to try to measure the noise floor of my SR-40 setup and will need to have the AGC completely turned off.

If I turn off the AGC of the Flex software will it really completely turn off all forms of variable gain?


Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the same results every time is insanity: I've almost proved it isn't; only a few more tests now and I'm sure results will differ this time ...

--- In softrock40@..., KD5NWA <KD5NWA@c...> wrote:

Are you emulating real switches or perfect instantaneous switches?

I have models for both ideal and real.

The reason is fairly simple, real switches do not turn on and off
instantaneously, if you have switches turn on while another turns
from on to off, the timing difference will cause the switches to have
a short circuit for a brief period. By having less than 25% on the
waveform you are insuring that the switches are all off before
turning one on, and thereby avoid that brief short.

If it were only that simple...

For example in the modeled QSD circuit that I am using, even if both
switches are on at the same time there is no "short circuit".

As a matter of fact, let's just forget about the other switch and deal
with the case of ONE switch and ONE sample integrating capacitor.
This removes any discussion on switch turn on/off times causing
overlap/short circuits. For even this circuit the simulation seems to
indicate that a 20-25% on time for the sampling clock is best.

This would translate to 20-25nS on time, 75-80nS off time ratio on a
100 nS period clock (10MHz).

Obviously this has something to do with the integration time of the
sampling capacitor and I am sure there is a mathematical equation
somewhere that predicts this. It is basically a sample and hold
circuit.

73 de Phil N8VB

P.S. LOL... I hope I know the difference between real and ideal
components after 26+ years in the field :-)

Hi Kees,

It 'tis a good idea. Take a look at


Cecil - WA6RDY

KD5NWA wrote:

[snip]
The reason is fairly simple, real switches do not turn on and off
instantaneously, if you have switches turn on while another turns
from on to off, the timing difference will cause the switches to have
a short circuit for a brief period. By having less than 25% on the
waveform you are insuring that the switches are all off before
turning one on, and thereby avoid that brief short.
[snip]

That's one of the reasons why I chose the ADG704 for my QSD. From the datasheet :

PRODUCT HIGHLIGHTS
.................
7. Break-Before-Make Switching Action.
.................

73 Alberto I2PHD

Milt Cram wrote:

[snip]
I don't have the exact numbers at the tip of my tongue, but I
you are interested, I can do a quick MathCad model. A rough estimate is
that a 1 degree phase error (1/60th of a radian) will give an image
rejection of no better than -36dB, and this assumes that the I and Q
channel gains are matched perfectly. A 0.1 degree phase error will
result in an image rejection of the order of -55dB. Likewise, a 1 db
gain error between the two channels, with perfect phase, results in an
image rejection of no better than -38dB.
[snip]

Quite true. If it can be of help, I saved this page from a site of
Advanced Microwave Inc. where the relevant math is laid down.



73 Alberto I2PHD

Are you emulating real switches or perfect instantaneous switches?

If you are emulating real switches the your conclusions are correct in that a slightly less that 25% waveform is the most efficient. I'm in the process of creating my own QSD and it will use 20% waveform.

The reason is fairly simple, real switches do not turn on and off instantaneously, if you have switches turn on while another turns from on to off, the timing difference will cause the switches to have a short circuit for a brief period. By having less than 25% on the waveform you are insuring that the switches are all off before turning one on, and thereby avoid that brief short.


At 08:45 AM 10/27/2005, Phil Covington wrote:

Hi Tony,

Let me take a look at this and report back.

I reran the simulation with a 4 switch QSD using two signal
integrating capacitors. I am taking the 4 switch QSD with 4
integrating capacitors and 50% clock duty cycle as the reference
(0dB). Here is what I have so far:

4SW4 QSD is the 4 switch qsd circuit with 4 signal integrating
capacitors. I(Q) and /I(/Q) are summed with an op amp. This is like
the SDR-1000 QSD.

2SW2 QSD is the 2 switch qsd circuit with 2 signal integrating caps.
This would be like SR V5

4SW2 QSD is the 4 switch qsd circuit with 2 signal integrating caps.
The caps are selected twice each cycle like SR V4.

50% duty cycle clock:

4SW4 QSD -> 0 dB (reference)
2SW2 QSD -> -5.86 dB
4SW2 QSD -> -5.62 dB

25% duty cycle clock:

4SW4 QSD -> +2.45 dB
2SW2 QSD -> -3.26 dB
4SW2 QSD -> -2.63 dB

15% duty cycle clock:

4SW4 QSD -> +2.14 dB
2SW2 QSD -> -3.45 dB
4SW2 QSD -> -2.22 dB

The 4SW2 QSD seems to have the least loss at around 12 - 15% duty
cycle clock.

73 de Phil N8VB

--- In softrock40@..., "Tony Parks" <raparks@c...> wrote:

Hi Phil,

The question I just ask about gain reduction as a function of phase

clock

skew is not the right question to ask. What we really are

interested in is

the gain from RF input to demodulated signal output. In the case of

CW I

would think this would just be the vector sum of the I and Q signals.

Another important question is if clock skew causes undesired signal
response.


73,
Tony KB9YIG
----- Original Message -----
From: "Phil Covington" <p.covington@g...>
To: <softrock40@...>
Sent: Wednesday, October 26, 2005 1:04 PM
Subject: [softrock40] QSD Models

Hi all,

I've placed some results of simulation of a two switch and four switch
QSD circuit on my blog at:

Check it out if you are interested in the QSD circuits.

73 de Phil N8VB







Yahoo! Groups Links

Yahoo! Groups Links

Cecil Bayona
KD5NWA
www.qrpradio.com

I fail to see why doing the same thing over and over and getting the same results every time is insanity: I've almost proved it isn't; only a few more tests now and I'm sure results will differ this time ...

Tony Parks wrote:

Hi Phil,

The question I just ask about gain reduction as a function of phase clock
skew is not the right question to ask. What we really are interested in is
the gain from RF input to demodulated signal output. In the case of CW I
would think this would just be the vector sum of the I and Q signals.

Another important question is if clock skew causes undesired signal
response.


73,
Tony KB9YIG

Tony,

If I might add my two cents--

Your second question is the more important one. The loss of gain due to clock skew is not very significant for desired signal detection. However, undesired signal response, such as poor image rejection, is a significant problem. For example, if you note the adjustments required for several of the software applications that have been made available for SoftRock40, good image rejection requires adjustments to a degree or better. I don't have the exact numbers at the tip of my tongue, but I you are interested, I can do a quick MathCad model. A rough estimate is that a 1 degree phase error (1/60th of a radian) will give an image rejection of no better than -36dB, and this assumes that the I and Q channel gains are matched perfectly. A 0.1 degree phase error will result in an image rejection of the order of -55dB. Likewise, a 1 db gain error between the two channels, with perfect phase, results in an image rejection of no better than -38dB. Consequently, gains should be matched to withing a fraction of a dB, and phase error should be within a few tenths of a degree. These can be significant problems at high frequencies.

I have been designing an interface for PSK that does not involve a PC. Instead, it uses a pair of DDSs that provide two LOs that are 90 degrees apart. I am using two SPDT switches for my QSD. The interface uses three '908 uC for control, display driver, and the amplitude modulator that is needed for PSK31. I have done quite a bit of modelling of this interface, but have not really examined the effect of phase errors on image rejection as I have filtered the input ahead of the QSD. My problems are relatively minor, because I am working with only audio input frequencies (the audio out of my HF rig).

73,
Milt W8NUE


--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.361 / Virus Database: 267.12.5/149 - Release Date: 10/25/2005

Hi Tony,

Let me take a look at this and report back.

I reran the simulation with a 4 switch QSD using two signal
integrating capacitors. I am taking the 4 switch QSD with 4
integrating capacitors and 50% clock duty cycle as the reference
(0dB). Here is what I have so far:

4SW4 QSD is the 4 switch qsd circuit with 4 signal integrating
capacitors. I(Q) and /I(/Q) are summed with an op amp. This is like
the SDR-1000 QSD.

2SW2 QSD is the 2 switch qsd circuit with 2 signal integrating caps.
This would be like SR V5

4SW2 QSD is the 4 switch qsd circuit with 2 signal integrating caps.
The caps are selected twice each cycle like SR V4.

50% duty cycle clock:

4SW4 QSD -> 0 dB (reference)
2SW2 QSD -> -5.86 dB
4SW2 QSD -> -5.62 dB

25% duty cycle clock:

4SW4 QSD -> +2.45 dB
2SW2 QSD -> -3.26 dB
4SW2 QSD -> -2.63 dB

15% duty cycle clock:

4SW4 QSD -> +2.14 dB
2SW2 QSD -> -3.45 dB
4SW2 QSD -> -2.22 dB

The 4SW2 QSD seems to have the least loss at around 12 - 15% duty
cycle clock.

73 de Phil N8VB

--- In softrock40@..., "Tony Parks" <raparks@c...> wrote:

Hi Phil,

The question I just ask about gain reduction as a function of phase

clock

skew is not the right question to ask. What we really are

interested in is

the gain from RF input to demodulated signal output. In the case of

CW I

would think this would just be the vector sum of the I and Q signals.

Another important question is if clock skew causes undesired signal
response.


73,
Tony KB9YIG
----- Original Message -----
From: "Phil Covington" <p.covington@g...>
To: <softrock40@...>
Sent: Wednesday, October 26, 2005 1:04 PM
Subject: [softrock40] QSD Models

Hi all,

I've placed some results of simulation of a two switch and four switch
QSD circuit on my blog at:

Check it out if you are interested in the QSD circuits.

73 de Phil N8VB







Yahoo! Groups Links