[Elecraft] High Speed CW issue with dropped characters/delay QRQ CW

Fri Aug 7 08:48:20 EDT 2015

Executive Summary

Issue: Dropped characters during high speed (over 60wpm) CW sent via
Keyboard connected to the P3 and K3. Approximately 4s Pause in transmission
at around each approximately 130 character mark.

Testing: 

1.   Several different keyboards (at least 4) have been used to test. All
keyboards exhibit the same issues.

2.   Lower the port speed on the P3 - should have NO effect on
communications between the P3 and K3, however, the issue with dropped
characters appears to have resolved. I believe the P3 firmware is suspect as
a result.

3.   Reseat all internal/external connections. No apparent effect.

Next Steps:

1.   Leave the P3 port speed settings low if that works around the dropped
character issue. This shouldn't affect the P3 to K3 communications.

2.   Due to the high cost of shipping parts back and forth to CA from Marc's
QTH I'm suggesting he try a new cable. 

3.   IF we are not able to try a new cable OR a new cable doesn't work then
I'm suggesting we snoop/monitor the serial data steam between the P3 and K3
to determine where the delay is being initiated from.

Sending Marc's email to the reflector..

>From Marc>

I saw that comment, too, but Eric is completely wrong.  If he saw the
problem in action he'd know that it occurs long after the keyboard is fully
active.

Paul and I have been in extended e-mail conversations.  First thing he said
is that the baud rate setting on the K3 and on the P3 has no effect on the
baud rate used for inter-device communication.  Those baud rates only apply
to the "host" connection, that is a pc connected to one or the other of the
devices.  So changing the baud rate on the P3 to 4800, which had a dramatic
effect on dropped character rates, did not change the comm rate from P3 to
K3.  That rate is always 38,400.  

We also confirmed that a "macro" stored in the SVGA board by saving it from
the keyboard will be dumped to the K3 without error.  However, a one-letter
macro will be dropped when repeatedly sent from the keyboard (by typing the
macro name, for example [ALT]m).  Thus we're sure that the character drop
occurs between keyboard and SVGA board.

I use 4 different keyboards for testing, 2 wireless and 2 wired.  One of the
wired boards, commonly used for Raspberry Pi projects, is pretty junky and
tough to use, but none of the keyboards avoids the dropped character
problem.  

All this suggests that there may be simply poor contact at the USB connector
on the P3.

The RS-232 data flow is almost undoubtedly the culprit.

When I started programming full time (I leased from Hewlett Packard a HP9820
and the single pen plotter) in 1972, I pretty much put away the soldering
iron.  Only many years later when we started using a bank of modems for
Internet connections and started using Ethernet internally did I do any sort
of building and that was all just cable work.  Oh, once in a while I'd
change boards in a box or clean finger contacts, but nothing like the early
days.

So this retirement gig is different.  In fact just this morning I ordered a
set of Xcelite drivers.  One more step backwards toward my youth.

I did own and manage a consumer electronics repair business way back when,
but ended up hiring repair people and training them.  I was more interested
in writing programs to manage bookkeeping than I was repairing stuff.  It
did teach me that the best tools are really good investments, though.  Prior
to that I used cheap tools, but our repairs were done faster, more
profitably, when we used the best tools and the best repair parts.  

Thanks for the cell number.  I'll put it in my contacts list and try to keep
you up-to-date.

But ignore Eric's statement.  It was completely out of context and may even
refer to some other thread.  More likely, though, he was just remembering
conversations among their techs when my P3/SVGA was in the shop and they
couldn't reproduce any problems *except* the periodic delay.  

-- 

Marc

From: Jerry Moore [mailto:jermo at carolinaheli.com] 
Sent: Friday, August 07, 2015 8:10 AM
To: 'elecraft at mailman.qth.net'
Cc: 'eric.swartz at elecraft.com'
Subject: [Elecraft] Re: High Speed CW issue with dropped characters/delay
QRQ CW

I've not heard from Marc on this but sent him an email asking if he had a
standard keyboard to try. From our earlier conversations it sounded like he
had tried at least two different keyboards already. One wireless and one
wired. I'm not doubting you Eric. Just wanting to look at this more closely.
If you've worked with Marc and the issue is resolved that's great!  Were we
able to find out if the keyboard is also the cause of the dropped
characters? Lowering the port bitrate resulted in fewer/no dropped
characters but that shouldn't be necessary (from reading and searching it
didn't sound possible to change the bitrate but apparently it is). 

For some reason I'm not getting emails from the reflector so I'm having to
get them on the web..

Jer

AE4PB

>It turns out the delay in this case is due to the particular keyboard in
use as 

>it has a sleep mode and wakes up slowly.

> 

>73,

> 

>Eric

>/elecraft.com/

>On 8/6/2015 10:58 AM, ae4pb at carolinaheli.com
<http://mailman.qth.net/mailman/listinfo/elecraft>  wrote:

> Re: High Speed CW issue with dropped characters/delay at high speeds with

> K3/P3 (Jerry Moore)

> 

> Some thoughts and conjecture:

> Dropped Characters:

> At the default port speed of 38400 any errors in transmission could
possibly

> result in dropped data rather than a resend depending on error

> handling..etc.. I've suggested to Marc to re-seat all physical connections

> in the signal path. Lowering the port speed resulted in no dropped

> characters. I don't have the gear to actually look at this but it could be

> caused by something as simple as a bad cable, bent cable..etc.. serial
data

> is pretty simple but doesn't take much to trip it at high data rates.
That's

> why there's usually some error detection/correction going on with some

> applications.

> 

> Delay:

> My thoughts on this are that the P3 is probably polling the radio for
status

> and the K3 interrupts the TX data stream to send status,

> frequency..etc...The P3 buffers the sending data until the polling cycle

> completes and goes back to sending.

> 

> Without a lot more information and gear all of my ramblings are little
more

> that guesses. The Delay could perhaps be affected by prioritizing the

> traffic, knowing how long a polling cycle takes, and maybe splitting the

> polling cycle in to smaller, more often sent, bits, rather than a chunk

> stream. Of course this only currently affects communications where a

> keyboard is sending high speed characters/CW to the P3 to be sent via the

> K3. The port speed should support those rates easily but HOW the

> conversation takes place may need to be tweaked a bit.

> 

> Fixes for the Dropped Characters may be as easy to fix with a new cable or
a

> shorter one.

> 

> Just thinking out loud here. I enjoy helping solve technical issues.

> 

> Jer / AE4PB

> 

> 

> ______________________________________________________________

> Elecraft mailing list

> Home: http://mailman.qth.net/mailman/listinfo/elecraft

> Help: http://mailman.qth.net/mmfaq.htm

> Post: mailto:Elecraft at mailman.qth.net
<http://mailman.qth.net/mailman/listinfo/elecraft> 

> 

> This list hosted by: http://www.qsl.net <http://www.qsl.net/> 

> Please help support this email list: http://www.qsl.net/donate.html

> Message delivered to eric.swartz at elecraft.com
<http://mailman.qth.net/mailman/listinfo/elecraft> 