Re: [wsjt-devel] Weird USB issues with Icom rigs and WSJT-X - a debug journey

[Bill S. <g4...@cl...>]

On 27/03/2019 11:39, Magne Mæhre wrote:
> A couple of my club mates have written about a weird issues which 
> showed up when trying to use WSJT-X with an IC-9100.   I've seen the 
> same issues with IC-7100 as well.  Even though WSJT-X is not the real 
> culprit here, I thought some of you might find it amusing...
>
> https://www.la1k.no/?p=4219
> "Exploring why WSJT-X didn’t want to decode – an illustrated guide to 
> printf-debugging"
>
>
> Perhaps some of you have deeper USB knowledge, and can shed a light on 
> what's _really_ going on?
>
> --Magne / LA1BFA 

Hi Magne,

this is a known issue with external USB hubs. Most are poor at handling 
streaming data (audio) and interrupt driven data (serial port emulation) 
simultaneously. The problem being seen is due to dropped audio packets, 
it is not a change in sampling rate but lost data that means WSJT-X 
never receives enough data to complete a full received message. The 
decoding only starts once a full receive period of audio samples has 
been captured.

I don't think the problem is specific to the hub alone, after all the 
USB connections on the PC are a hub as well. The problem seems related 
to how the USB protocol is handled when there is more than one hub 
between the communicating end-points.

Bottom line, don't use external USB hubs when connecting a rig that has 
a USB connection that provides both audio and CAT.

The extensive detective work listed in the link above does appear to 
show that USB 3.0 hubs do not show this problem, or at least it seems 
less prevalent.

It should be noted that there is no point in modifying WSJT-X to somehow 
work around this problem by accepting more audio samples to make up form 
the missing ones, this is because the dropped samples mean that the rx 
period data is pretty much useless.

73
Bill
G4WJS.