Networking issues
Russ K2TXB
it for a few years and have been somewhat following it's progression.
Leif, you may recall meeting me at the EME conference in Trenton, NJ a
few years ago.
Anyway I bought a SDR-IQ and got it working. Then installed Joe's
MAP65 package which includes linrad. After a day of playing with it I
got it working on my laptop where the SDR-IQ is connected. In fact it
works very well. Next I wanted to network it so I could run Linrad +
the IQ on the laptop in the basement (where the antenna is), but run
Map65 upstairs on my main machine. That went fairly well too and
works fine, however it leaves me unable to hear the audio. Now I know
there are other ways to get audio up here, in fact I've done it with
HRD and NetMeeting, but the audio is choppy when doing that. So I
read about Linrad networking and thought I could run a networked slave
copy of Linrad up here and run Map65 up here too. That is where I
have run into some problems.
The first one is just a small textural error in the setup menu but I
thought you might like to know about it. In the network menu, if you
use option #3 to set the receive IP address, you get a message that
says "Set receive address to -1 to read from file par_netsend_ip...".
That stumped me for a while. I finally found in online notes that I
readlly needed to use file "par_netrec_ip", but the wrong setup
message did cause a problem. I spent a while trying to add a secone
IP address to the netsend file. Not a big deal and easily fixed, I am
sure.
Once I got the addresses fixed, the two instances of Linrad started
talking, but I have some problems. The main one is that the
frequencies of the two instances do not match. For example if I set
the master frequency to 14.075, the slave gives me a frequency range
below that frequency. So If I tune the master to 14.076, the slave
actually receives signals about 50 KHz lower than that. It is
impossible to tune the slave above about 14.050 in this example. The
same thing holds true for any other frequency or band.
Also, I note that when I change frequencies on the master (by entering
the center frequency in the type-in box), the slave does not know it
and continues to think it is on the original frequency.
Finally it seems that I have to increase the gain of the master
drastically in order to get enough signal in the slave to be heard or
make the waterfall signals visible.
I wonder what I am doing wrong here? I played with it for hours today
but was unable to affect any change. The version of Winrad I am using
is 3.05B, and both master and slave are running on Pentium PC's under
Windows XP (SP3).
Thanks for any insight.
73, Russ K2TXB
Leif Asbrink
> Hi All. I'm new to this group and to Linrad, although I've known about
> it for a few years and have been somewhat following it's progression.
> Leif, you may recall meeting me at the EME conference in Trenton, NJ a
> few years ago.
Yes:-)
> Anyway I bought a SDR-IQ and got it working. Then installed Joe's
> MAP65 package which includes linrad. After a day of playing with it I
> got it working on my laptop where the SDR-IQ is connected. In fact it
> works very well. Next I wanted to network it so I could run Linrad +
> the IQ on the laptop in the basement (where the antenna is), but run
> Map65 upstairs on my main machine. That went fairly well too and
> works fine, however it leaves me unable to hear the audio. Now I know
> there are other ways to get audio up here, in fact I've done it with
> HRD and NetMeeting, but the audio is choppy when doing that. So I
> read about Linrad networking and thought I could run a networked slave
> copy of Linrad up here and run Map65 up here too. That is where I
> have run into some problems.
>
> The first one is just a small textural error in the setup menu but I
> thought you might like to know about it. In the network menu, if you
> use option #3 to set the receive IP address, you get a message that
> says "Set receive address to -1 to read from file par_netsend_ip...".
> That stumped me for a while. I finally found in online notes that I
> readlly needed to use file "par_netrec_ip", but the wrong setup
> message did cause a problem. I spent a while trying to add a secone
> IP address to the netsend file. Not a big deal and easily fixed, I am
> sure.
My memory is not quite what it used to be, but I think this error
was corrected long ago. The version of Linrad that you are using is
probably rather old. The current version is 03-08
> Once I got the addresses fixed, the two instances of Linrad started
> talking, but I have some problems. The main one is that the
> frequencies of the two instances do not match. For example if I set
> the master frequency to 14.075, the slave gives me a frequency range
> below that frequency. So If I tune the master to 14.076, the slave
> actually receives signals about 50 KHz lower than that. It is
> impossible to tune the slave above about 14.050 in this example. The
> same thing holds true for any other frequency or band.
Presumably you have only half the spectrum. Use the arrows to see a
wider spectrum. The frequency that you have entered in the master
instance of Linrad should be the center frequency of the slave as well
as of any other instance.
> Also, I note that when I change frequencies on the master (by entering
> the center frequency in the type-in box), the slave does not know it
> and continues to think it is on the original frequency.
This is an error. I suggest you try to run the master with the Linrad
default sendto address and then let the slave in your second computer
listen to it on the default receive address. Then make the slave
send to 127.x.x.x or where MAP-65IQ wants to pick up the data.
That would mean that you send one set of 16 bit data between the
computers.
> Finally it seems that I have to increase the gain of the master
> drastically in order to get enough signal in the slave to be heard or
> make the waterfall signals visible.
>
> I wonder what I am doing wrong here?
I do not know. Something is wrong because the data should be identical
is the slave and in the master.
You may have detected a bug or you may have done something that is
incorrect. I do not know.
> I played with it for hours today
> but was unable to affect any change. The version of Winrad I am using
> is 3.05B, and both master and slave are running on Pentium PC's under
> Windows XP (SP3).
Linrad-03.05b ;-)
I suggest you make a fresh install of Linrad-03.08. That means create
a new directory where you unpack wlr03-08 and start Linrad.exe on your
first computer (to which the SDR-IQ is connected.) That would be
your Linrad master. Configure the network with default addresses
and set the master to send 16 bit raw data.
Then make a second directory which is a copy of the new installation
on the same computer. Start linrad.exe from the second directory to get
a Linrad slave with its own parameters. Configure the network with
default addresses and set the slave to receive 16 bit raw data.
With the same parameters in Linrad you should see identical spectra
and waterfalls in the master and slave.
Then move the slave to the other computer. Just copy everything in the
slaves directory. It should work exactly the same.
Use the parameters that Joe supplied in the slave and set the slave
to transmit FFT2 to MAP65-IQ. That would implicate a par_netsend_ip
file.
I think you would end up with something that works.
Then, I would appreciate if you can analyze what you did and that
does not work. The Linrad installations you currently have would be
left as they are now so you can go back to a non-working system.
Regardless of the reason I think your experiences motivate a change
somewhere to avoid others from running into the problems you have
experienced. BUT, before doing anything else you should upgrade to
03-08 to see if the problems are still present.
73
Leif / SM5BSZ
stan...@aol.com
I have a question, why?
I am using Linrad-03.08 with Perseus. I have a 144 to 28 converter in front of the Perseus. I have discovered something that surprises me. When pointing my antenna in 1 direction the A/D amplitude margin starts in the 60's but in a short period of time, (less than a minute) it drops to 0.00. All of the other Amplitude margins remain good and I see no trace on the waterfall display indicating a strong signal had appeared and caused the A?D to drop to zero. Everything appears to be working fine. Hitting the "Z" key rerstors the A?D margin but in less than a minute it drops to 0.00. This is directional. When pointing mu antenna in other directions the A?D does not drop..
Stan, KA1ZE/3, FN01xt/EL87
Leif Asbrink
The Perseus is saturated occasionally. Every time the A/D
converter saturates the Perseus will set a flag and Linrad will
tell you it happened by setting the A/D margin to zero.
I suggest you disable the second FFT and set the sampling rate
to 2 MHz with no waterfall averaging. You would then have no
noise blanker and you would see wideband pulses as white horizontal
lineas. Change the frequency in steps of 1.5 MHz or so and see
if you can locate a pulsed narrowband signal.
Occasional saturation of the A/D is not necessarily harmful.
Every strong signal in the passband will get an incorrect
amplitude value and therefore produce large wideband "keying
clicks" The Linrad noise blanker might remove them completely.
In case saturation lasts a little longer you might find false
responses and it could be a good idea to reduce the total gain.
In case your amplifier chain is already optimized for gain
(the preamplifier dominates and increases the noise floor by
about 17 dB) the only improvement you can add is in the form
of filters.
In case saturation is caused by wideband pulses there is no reason
to worry about it. You could get rid of the problem by reducing
the bandwidth into the Perseus. If there is a pulsed transmitter
somewhere outside the band you should reduce it with a suitable
filter. Maybe a cavity filter on 144 because the signal level
required to saturate the Perseus is pretty high and probably not
within the linear range of your 144 to 28 converter.
73
Leif / SM5BSZ
stan...@aol.com
Thanks for the ideas. I suspect it is wideband. I see the horizontal lines in the waterfall but looking 2 mhz wide from 27-30 I could find no narrow spikes. I think the noise blanker eliminates it good. I will probably be on a holiday until spring and won't be able to put the cavity in front of the preamp. I have used the cavity but it needs looser coupling as the response is too narrow.
Thanks again for the great program, Linrad.
73, Stan
From: Leif Asbrink <le...@sm5bsz.com>
To: lin...@googlegroups.com
Sent: Mon, Oct 19, 2009 5:18 pm
Subject: [Linrad] Re: A/D Margin 0.00
Russ K2TXB
master and slave (but made a copy of both linrad folders before updating.
Then I copied linrad.exe and the two .lir files into the two linrad folders
and built new par_userint files for both. I set the network send and
receive addresses to default values as well.
The end result is pretty much the same as before. Attached are two zip
files, for master and slave, each containing a screen shots and some of the
configuration files. You can see that the frequency ranges and levels are
significantly different. It clearly does not seem to work right but I am
clueless at this point...
73, Russ K2TXB
PS: if these attachments do not come through, please send me an email
address to send them to.
Jeremy Alexander
I am in same boat. But lack the computer skills of even Russ. I hope to see some helpful resolve to this thread.
Jeremy
www.w7eme.org
--- On Tue, 10/20/09, Russ K2TXB <k2...@dxcc.com> wrote:
Leif Asbrink
> Leif, thanks for the advice. I have now updated to version 3.08 on both
> master and slave (but made a copy of both linrad folders before updating.
> Then I copied linrad.exe and the two .lir files into the two linrad folders
> and built new par_userint files for both. I set the network send and
> receive addresses to default values as well.
>
> The end result is pretty much the same as before. Attached are two zip
> files, for master and slave, each containing a screen shots and some of the
> configuration files. You can see that the frequency ranges and levels are
> significantly different. It clearly does not seem to work right but I am
> clueless at this point...
You have zoomed waterfall graphs. Put the mouse on the frequency scale
of the main spectrum/waterfall and press F1.
When you set the frequency range the same in both instances of Linrad
you should see the same spectrum/waterfall. The parameters for the
main spectrum/waterfall are stored in par_ssb_wg These files must be quite
different.
73
Leif / SM5BSZ
Russ K2TXB
tried making the frequency range the same on both versions as you described.
I had tried that before but missed the idea that I had to hit 'apply' after
making the changes. So when I tried it, I first set the lower value to 14.2
and the upper to 14.8. But when I hit apply, the range changed. The bottom
stayed close, but the top reduced down to around 14.5. Tried that a few
times to make sure.
Then I opened both par_ssb_wg files and made the slave one be equal to the
master. After doing that, I was able to get the range to approximate that
of the master, but could not get it to be exactly the same. However I was
able to see the same signals in the range displayed. But I still have the
problem where the level on the master has to be at least 10 db higher than
normal in order to get a reasonable signal on the slave.
And I still see that when I change center frequency on the master, the slave
does not know it. It continues to show the old frequency range - even if I
change to a new band. If I x-out twice on the slave and then hit D again,
the range will then reflect the new frequency. However after changing from
20 to 40 meters and then to 80 meters, I noticed that the slave range was
somewhat expanded again. The master showed 3.82 to 3.86 while the slave
shows 3.825 to 3.85. I looked at the slave's par_ssb_wg file again and see
that some of the values had changed back to what they were before.
I attempted to make the files identical again, and now the slave always
errors out when I click on the waterfall to start audio out, with a message:
[1058]Output can not start because the parameter "output delay margin ms"
is too small.
Maybe the max DMA rate is set too low.
The max DMA rate was set to 100, but I tried changing it to 500 with no
improvement. Something else must have changed but I am sure I did not make
any changes except in the par_ssb_wg file. Replacing that file with the
original still does not cure the problem, so I cannot make any more tests
tonight. Hopefully you can tell me how to fix this new problem.
Thanks, Russ K2TXB
> -----Original Message-----
> From: lin...@googlegroups.com
> [mailto:lin...@googlegroups.com] On Behalf Of Leif Asbrink
> Sent: Tuesday, October 20, 2009 8:58 PM
> To: lin...@googlegroups.com
> Subject: [Linrad] Re: Networking issues
>
>
Leif Asbrink
> Hi Leif. Well I am making some progress, but still some issues. First I
> tried making the frequency range the same on both versions as you described.
> I had tried that before but missed the idea that I had to hit 'apply' after
> making the changes. So when I tried it, I first set the lower value to 14.2
> and the upper to 14.8. But when I hit apply, the range changed. The bottom
> stayed close, but the top reduced down to around 14.5. Tried that a few
> times to make sure.
Parameters in Linrad are of three kinds.
1) The first kind relates to hardware: Soundcard, (or SDR-IQ etc),
screen, fonts, network, parallel ports etc. These parameters can
only be changed from the main menu in the setup functions.
S for screen, U for input, N for network etc. These parameters
are stored in par_userint, par_sdr14, par_network and similar.
2) Fixed parameters for a particular receive mode. Things like first
FFT bandwidth, use or not use the second FFT, decimation rate, use
16 or 32 bit arithmetics etc. These parameters are stored in par_cw,
par_ssb and similar. One for each rx mode.
3) Parameters that can be changed on the fly while operating.
These are changed with mouse or keyboard while receiving is
running and they are stored in par_ssb_wg, par_ssb_bg, par_cw_wg
and so on. There is one file for each window in each receive mode.
When you first set the lower value by typing in the digits and then
enter Linrad will check whether the frequency is within range. You
the lowest possible frequency is your center frequency-48 kHz.
(Assuming your SDR-IQ is still configured for MAP65-IQ) The
sampling rate (near 96kHz) belongs to type 1 above and can not be
changed on the fly. The fact that you could set 14.2 sucessfully
means that your center frequency was 14.248 or below. You can then
type 14.8 into the upper frequency box, but when you hit enter the
test would change the number to 14.096 (or below.) when you hit
return.
You report you got the limits 14.2 and 14.5 which means a 300 kHz span
so I conclude you have set a much higher sampling rate than 96 kHz
for the SDR-IQ. 300 kHz is very high for this hardware that uses
USB 1.0. It would be a good idea to use a less extreme speed when you
are new to everything in Linrad. The manufacturer specification for SDR-IQ
says the highest sampling rate is 196 kHz. You can go higher in Linrad
but at some point there will be problems. I suggest you start with a sampling
speed of 150 kHz or less.
> Then I opened both par_ssb_wg files and made the slave one be equal to the
> master. After doing that, I was able to get the range to approximate that
> of the master, but could not get it to be exactly the same. However I was
> able to see the same signals in the range displayed. But I still have the
> problem where the level on the master has to be at least 10 db higher than
> normal in order to get a reasonable signal on the slave.
The wide graph and its place on the screen are parameters in par_xxx_wg.
To get identical results, everything has to fit inside the screen. If your
master screen is bigger Linrad will contract the graph to make it fit and
that changes the endpoint frequencies.
> And I still see that when I change center frequency on the master, the slave
> does not know it.
OK. This is a bug. Every UDP package contains the frequency and a change should
be noted by the slave. I have to look at that but it will take some time because
to become 3.09 is in a messy state right now.
> It continues to show the old frequency range - even if I
> change to a new band. If I x-out twice on the slave and then hit D again,
> the range will then reflect the new frequency.
OK. That means that the bug is in the slave. You may type 'X', then 'B'
to get the new frequency.
> However after changing from
> 20 to 40 meters and then to 80 meters, I noticed that the slave range was
> somewhat expanded again. The master showed 3.82 to 3.86 while the slave
> shows 3.825 to 3.85. I looked at the slave's par_ssb_wg file again and see
> that some of the values had changed back to what they were before.
Linrad has few restrictions on what the user can do. You seem to do extreme
things....
The total span you have is 40 kHz only in the main spectrum. Why?
The purpose of the main spectrum is to display the entire frequency
range of your hardware. (That is needed for you to monitor red/white
points in the main spectrum so you can set blanker thresholds correctly.)
You may have many reasons for the small frequency span. Maybe you have
set the sampling speed of the SDR-IQ to about 40 kHz in order to place
strong signals outside the frequency range that is transmitted via USB.
The USB channel has a limited dynamic range of 16 bits and SDR-IQ uses
24 bits internally.
> I attempted to make the files identical again, and now the slave always
> errors out when I click on the waterfall to start audio out, with a message:
>
> [1058]Output can not start because the parameter "output delay margin ms"
> is too small.
> Maybe the max DMA rate is set too low.
This error may be caused by many kinds of timing problems.
> The max DMA rate was set to 100, but I tried changing it to 500 with no
> improvement. Something else must have changed but I am sure I did not make
> any changes except in the par_ssb_wg file. Replacing that file with the
> original still does not cure the problem, so I cannot make any more tests
> tonight. Hopefully you can tell me how to fix this new problem.
I have no idea what you are trying to do. Maybe you have set small value
for the first mixer bandwidth reduction in combination with a baseband
filter in many points and also to run the baseband filter and resampler
in the time domain. That would cause CPU overload and maybe this error
message.
I suggest you play a bit more with a single instance of Linrad with your
SDR-14 and get an understanding of the differend processing blocks that
you use. For use with MAP65 you do not need to click any signal in Linrad
and then you would not need to know anyting about the baseband. But you
obviously clicked the waterfall and then baseband processing would start.
Use the mouse and the F1 help to read what all the different buttons
in the baseband window do. (F1 with the mouse outside all windows, on black
screen, will show where the buttons are.)
Once you are happy with the processing you can do in the master and have
become familiar with the arrows in the upper corners and so on you can
copy all the par_ssb*.* files to the slave directory. Provided that you
have specified at least the same screen size (in pixels) for the slave
it should then become identical to the master.
73
Leif / SM5BSZ