Keyboard Shortcuts
Likes
- N2adr-Sdr
- Messages
Search
Re: Modify SSB send buffer size with AC2YD remote?
Dear Ben,
Thanks for taking the time to respond. It seems that I will have to do some deeper hacking. It is nice to know that I'm not missing an obvious setting somewhere. I will investigate with wireshark, but my knowledge of UDP is quite basic (so a great learning opportunity). So far I did some basic hacking that improved my situation somewhat, but still not enough to be usable: 1. I was getting "Tx hermes buffer underflow" messages on the remote site. This was fixed by increasing "HERMES_TX_BUF_SAMPLES" in microphone.c. Perhaps my need to increase this has some implications of the limitations of my setup. The buffer underflow happened right after I connect from the control-head, without any transmitting involved. 2. I increased the size of the UDP-packages by setting UDP_SEND_INT16 to 600 in ac2yd/remote.c, I believe this has decreased the choppiness of the transmit audio. I believe this is the maximum size I can get the packages before they start to fragment, but I might be wrong. I will try listening to my signal via an appropriate websdr. Mike: Thanks for the suggestion, it seems to be that this is the default setting. Ultimatley I intend to use this for SSB and CW, but the digital modes are convenient for testing. It might be a multitude of different things. I will report back to this thread if I manage to get it to work better. Best regards, Ludde |
Re: Modify SSB send buffer size with AC2YD remote?
¿ªÔÆÌåÓýHello Ludde, I think Jim has already reduced the sample rate for the remote audio channels (both directions) to 8 KHz. I'll suggest some analysis via Wireshark to see what's arriving
at the Remote Radio (server) from the Control Head (client).? The
channels are UDP, so there's a possibility that some packets are
being dropped over the network somewhere.? Or, your internet
service provider may have some "interesting" behavior on data you
upload (Tx, choppy), vs. download (Rx, which you say works okay).?
Wireshark can help diagnose the network behavior, looking at
packet timestamps and continuity ... tedious, but it should tell a
story. Also, if WiFi is involved anywhere, be aware that many devices will periodically "scan" for other WiFi access points, for "roaming" purposes.? This briefly tunes away from the WiFi channel being used, and therefore may delay or drop data.? It can take some internet research to figure out how to keep your device(s) from "scanning", but scanning can definitely cause problems.? I still haven't figured out how to keep Windows 10 laptops from scanning, so I use ethernet between laptop and AP when I can, or just tolerate the interruptions if using WiFi (but I am a CW-only user).? I wrote some information about the scanning problem in a document I wrote about my Remote Quisk proof-of-concept prototype ... Jim has included this in the ac2yd directory of the Quisk distribution, as "Design_2022_0531.pdf". CAVEAT ABOUT Design_2022_0531.pdf:? This was written before Jim
integrated remote capability into Quisk, and before the Quisk user
community helped to find some bugs ... so some details differ
between the document and what Jim actually implemented, including
the network behavior of the remote audio channels. If UDP data is being dropped, you could try experimenting with
TCP ... I've had troubles, though, using TCP with higher
bandwidths (i.e. receiving I/Q samples from an SDRPlay radio via a
WiFi connection), because the network would get choked trying to
replace dropped data *caused by my Windows laptop periodically
scanning*!!? I used Wireshark to figure that out.? Perhaps the
Quisk bandwidths are low enough that TCP could be helpful. Another thought, perhaps try receiving your Tx signal on an
entirely separate receiver, in case FDX is somehow clogging things
up. Regarding the buffer implementation(s) ... it might be worth taking the time to study the code and try some hacking, so you can do some experiments yourself in your unique situation. I hope some of these thoughts may be helpful ... Good luck! -- Ben, AC2YD -- On 4/1/23 08:30, Michael Black via
groups.io wrote:
|
Re: Modify SSB send buffer size with AC2YD remote?
Michael Black
For JS8Call you probably only need 8KHz of bandwidth (that will give 4KHz of audio bandwidth).? Can you lower your data rate? Mike W9MDB
On Saturday, April 1, 2023 at 05:45:42 AM CDT, Ludde <luddek@...> wrote:
Dear group, I'm setting up a remote Hermes Lite 2 in an solar powered off-grid location with 4G internet connection. The connection is rather high jitter, with a round-trip of around 50ms and occasional spikes upwards 200ms. I manage to get a good and stable receive signal. However when I transmit my signal seem to chatter and chop, I test by monitoring using full-duplex (FDX). The CW-spot signal is quite reasonable. I've tried transmitting JS8Call test-messages, and those signals get heavily distorted and chopped. I don't understand how the buffers are setup, and I wish to increase the buffers on the remote end of the AC2YD package. The settings I'm currently trying to use are: Quisk version 4.2.17 PTT Hangtime msec: 200 Tx buffer msec: 500 Play latency msec: 1000 Start SSB delay msec: 1000 Start CW delay msec: 150 I also tried changing RX_BUFFER_SIZE in remote.c, and then re-compiling quisk. I don't know exactly what that implies, but I'm open to modifications of the code if needed. I'd be happy if someone could enlighten me with how to modify the send-buffers, or if there is something else that you think I might have missed. Best regards and 73 de Ludde |
Modify SSB send buffer size with AC2YD remote?
Dear group,
I'm setting up a remote Hermes Lite 2 in an solar powered off-grid location with 4G internet connection. The connection is rather high jitter, with a round-trip of around 50ms and occasional spikes upwards 200ms. I manage to get a good and stable receive signal. However when I transmit my signal seem to chatter and chop, I test by monitoring using full-duplex (FDX). The CW-spot signal is quite reasonable. I've tried transmitting JS8Call test-messages, and those signals get heavily distorted and chopped. I don't understand how the buffers are setup, and I wish to increase the buffers on the remote end of the AC2YD package. The settings I'm currently trying to use are: Quisk version 4.2.17 PTT Hangtime msec: 200 Tx buffer msec: 500 Play latency msec: 1000 Start SSB delay msec: 1000 Start CW delay msec: 150 I also tried changing RX_BUFFER_SIZE in remote.c, and then re-compiling quisk. I don't know exactly what that implies, but I'm open to modifications of the code if needed. I'd be happy if someone could enlighten me with how to modify the send-buffers, or if there is something else that you think I might have missed. Best regards and 73 de Ludde |
Re: Quisk Version 4.1.67 July 2020
The code was removed in version 4.1.58:
? ?if sys.platform == 'win32':
? ? ? self.Bind(wx.EVT_ENTER_WINDOW, self.OnEnter) ? def OnEnter(self, event):
? ? if not application.w_phase:
? ? ? self.SetFocus() # Set focus so we get mouse wheel events
After that, the wheel stopped working in 32-bit Quisk. Fine with 64-bit Quisk. I tried Windows 64-bit. |
Re: Quisk with Rasbian
Thanks Michael, you are right. I made "sudo apt purge quisk" then "sudo pip install quisk". It worked very fast and now I have a Quisk v.4.2.17
I guess I shall have to do something similar with the other softwares I have installed like SparkSDR in order to upgrade them from the os-repos version. 73 - Pierre - FK8IH |