[K3] Serial interface noise issues

[Thomas Horsten]

Hi,

I use a cheap USB to serial dongle to connect my Eee netbook (running
Linux) to the K3, but I have a couple of issues with it, and I would like
to solicit suggestions to sort it out.

On 10/12/15m and if the bands are relatively quiet, I get bursts
of S1-S2 static when the PC is talking to the K3, this happens at any baud
rate (at least down to 4800, haven't tried going lower than that). The
problem gets massively worse, S4-S5, if the netbook's power supply is
plugged in, but it's still discernible even without it unless the bands are
noisy. I haven't heard it on 20m and below, but it could just be because of
higher QRN blanking it out.

The other problem is that the serial dongle resets itself when I transmit
on 40m, this doesn't happen on any of the other bands but almost invariably
when I tx on 40m at anything more than 25-30W, even just sending a long dah.

I guess both of these problems could be down to bad/no screening of the
little chip in the dongle, I tried enclosing it and the cable in kitchen
foil but with no discernible difference. Any suggestions, or should I just
get a better dongle and cross my fingers?

73, Thomas M0TRN
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[KV5J]

I use a KUSB from Elecraft and have an increase in noise of 2-3 s-units when used on my lap top.

Keith

[Jim AB3CV]

I noticed some noise from the KUSB as well but after playing around a bit
discovered that it was being picked up by my nearby temporary antenna setup
rather than being conducted or radiated into the K3 directly. Disconnecting
the antenna coax killed the noise completely.

YMMV

Jim ab3cv
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Jim Brown-10]

On 11/1/2011 4:51 AM, Thomas Horsten wrote:
> I guess both of these problems could be down to bad/no screening of the
> little chip in the dongle, I tried enclosing it and the cable in kitchen
> foil but with no discernible difference. Any suggestions,

The more likely cause is a Pin One Problem on one or both ends of the
cable. The only proper termination of a cable shield is to the shielding
enclosure at both ends. A Pin One problem, first acknowledged and
described in the world of pro audio by Neil Muncy, ex-W3WJE, occurs when
the cable shield goes PAST the shielding enclosure to wiring inside the
box. This causes any common mode current flowing on the shield to wander
around inside the box until it eventually finds a path to the chassis,
and from there to "ground."  The Problem with this is that, as the
current wanders around circuit common inside the box, voltage is dropped
across the inductance of the conductors (usually circuit traces) that
make up circuit common, and that voltage is injected into the signal
path, detected, and amplified.  THAT is the Pin One Problem, and it is
the most common cause of RFI in today's world.

One solution is to FIX the improper wiring, and sometimes that is
possible, but often it is not.  The other solution is to "put a
band-aid" on the Pin One Problem by killing the current that excites
it.  In the case you describe, the solution is to form a common mode
choke by winding 4-6 turns of the serial cable around a #31 or #43
ferrite core made by Fair-Rite.  This forms a low-Q parallel resonance
in series with the common mode circuit, which looks like a high
resistance near resonance.  Like any other coil, we put the resonance
where we want it by using the "right" number of turns (which sets the
inductance) and the "right" spacing between them (which sets the
capacitance).  The nature of the two ferrite materials I've recommended
is that a lot of RESISTANCE is coupled from the lossy ferrite core, so
the resonance is VERY low Q (typically about 0.5), and thus very broad.

You should also bond between the computer chassis and the K3 chassis
using a short conductor. In the US, I would recommend  #10 AWG, which is
0.1 inch diameter. This minimizes the hum and buzz produced by leakage
currents in the mains power system that also excite the Pin One
Problem.  The chassis of most laptop computers can be accessed at the
retaining screws of a D-connector.

For (much) more information on these issues see my various tutorials at
http://audiosystemsgroup.com/publish.htm

73, Jim Brown K9YC
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[k6dgw]

On 11/1/2011 9:54 AM, Jim Brown wrote:
> On 11/1/2011 4:51 AM, Thomas Horsten wrote:
>> I guess both of these problems could be down to bad/no screening of the
>> little chip in the dongle, I tried enclosing it and the cable in kitchen
>> foil but with no discernible difference. Any suggestions,
>
> The more likely cause is a Pin One Problem on one or both ends of the
> cable. The only proper termination of a cable shield is to the shielding
> enclosure at both ends.

I had the same problem when using my K2/100 for RTTY.  I had the
headphones bridged across the audio output feeding the sound card and I
could hear the polling.  It sounded like a low pitched "rush ... rush
... rush ..." sort of similar to the ET signal in the movie "Contact" --
only not in prime numbers :-)  It was louder when I powered the laptop
from the mains, almost disappeared on the battery.  It did not show up
on the transmitted signal.

I switched to the K3 for RTTY, and don't have the problem anymore.

73,

Fred K6DGW
- Northern California Contest Club
- CU in the 2012 Cal QSO Party 6-7 Oct 2012
- www.cqp.org
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[tomb18]

Do you think that this issuee could still occur with the computer off but the cable still attached?  I have some issues with noise but have attributed them so far to the power lines.  I mean if the cable is picking up interference from somewhere else, then perhaps this could be an issue?
I'll have to try disconnecting the cable during a noise spell.

[Jim Brown-10]

On 11/1/2011 3:33 PM, tomb18 wrote:
> Do you think that this issuee could still occur with the computer off but the
> cable still attached?  I have some issues with noise but have attributed
> them so far to the power lines.  I mean if the cable is picking up
> interference from somewhere else, then perhaps this could be an issue?

Unlikely, but not impossible.  Do the chassis bonding that I recommended
-- this is good engineering practice that never hurts, is well known to
drastically reduce hum and buzz, and to minimize RFI issues.

73, Jim Brown K9YC
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Rick WA6NHC]

Wall wart power supplies are notorious for creating RF noise (and audible
hum).  My stock HP laptop power supply generates a vast spectrum of noise
but the triple source (wall, 12v, aircraft) replacement is mute on all
frequencies, same brand.

So part of the test should be removing ALL power sources, not just a shut
down or power off.  With all the remote controls used these days, it is rare
that electronics are really off unless unplugged [also stops parasitic
consumption of energy if folks are sensitive about that].

Note that on some computers, chassis ground is NOT audio ground and I'd
suspect the same is true of some radios (true for at least one of my
Batwings).  Chassis to chassis bonding may increase noise in that case.

Rick WA6NHC

-----Original Message-----
From: Jim Brown

On 11/1/2011 3:33 PM, tomb18 wrote:
> Do you think that this issuee could still occur with the computer off but
the
> cable still attached?  I have some issues with noise but have attributed
> them so far to the power lines.  I mean if the cable is picking up
> interference from somewhere else, then perhaps this could be an issue?

Unlikely, but not impossible.  Do the chassis bonding that I recommended
-- this is good engineering practice that never hurts, is well known to
drastically reduce hum and buzz, and to minimize RFI issues.

73, Jim Brown K9YC

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Jim Brown-10]

On 11/1/2011 4:57 PM, Rick Bates wrote:
> Note that on some computers, chassis ground is NOT audio ground and I'd
> suspect the same is true of some radios (true for at least one of my
> Batwings).

It SHOULD be.  If it is not, the mfr goofed. THAT'S what a Pin One
Problem is!  Further, unless the audio interface is BALANCED, audio
return will always be referenced to the chassis, but it may be a lousy
path.

>   Chassis to chassis bonding may increase noise in that case.

WRONG.  This is another of those widely held misconceptions that is
WRONG. Chassis to chassis bonding is ALWAYS good practice, and ALWAYS
minimizes buzz and RFI.

73, Jim Brown K9YC
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Rick WA6NHC]

Not always.  It may be considered poor design but if the in/output goes
directly to a differential amp with little or no 'padding' to chassis
(creating the comparison to 'different'), it may have a floating ground and
actually cause problems if connected to ground.  

In the case of Motorola radios, you'd have to ask them why, but never assume
that either side of a speaker (for example) is chassis grounded, or you'll
blow up a very expensive audio amp (like on the popular GM300).  A careful
resistance test should be done before connecting audio ground to chassis.

In general, we agree that chassis to chassis bonding SHOULD be encouraged to
reduce noise (and is electrically safer having everything at the same ground
potential).  But it does NOT always solve problems.  

I've seen cases where one (presume a computer to radio) chassis-chassis
connection made the problem worse (ground looped hum), because it introduced
more noise into the radio, even if plugged into the same power source
(should have been equal grounding already).  Audio isolation transformers
were the solution, pass only the audio.

It might be against common practice, but reality trumps theory.  Chassis to
chassis grounding may have any range of results (but certainly try it).

And in the case of RF, remember what may be good DC ground isn't always RF
ground.  Ask any mobile HF operator (striving to get very low RF resistance
to ground to improve antenna efficiency).  ;-)  With RF, more grounding
(overkill) is often MUCH better and always use plenty of low RF resistant
metal because of skin effect (braid or at least multi-strand, not solid
wire).  A lot of the crud generated by electronics is actually low level RF
so it should be treated as such.  Fortunately, it also works well at DC to
audio frequencies as well.

Rick WA6NHC

-----Original Message-----
From: Jim Brown

On 11/1/2011 4:57 PM, Rick Bates wrote:
> Note that on some computers, chassis ground is NOT audio ground and I'd
> suspect the same is true of some radios (true for at least one of my
> Batwings).

It SHOULD be.  If it is not, the mfr goofed. THAT'S what a Pin One
Problem is!  Further, unless the audio interface is BALANCED, audio
return will always be referenced to the chassis, but it may be a lousy
path.

>   Chassis to chassis bonding may increase noise in that case.

WRONG.  This is another of those widely held misconceptions that is
WRONG. Chassis to chassis bonding is ALWAYS good practice, and ALWAYS
minimizes buzz and RFI.


______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Jim Brown-10]

On 11/1/2011 6:58 PM, Rick Bates wrote:
> Not always.  It may be considered poor design but if the in/output goes
> directly to a differential amp with little or no 'padding' to chassis
> (creating the comparison to 'different'), it may have a floating ground and
> actually cause problems if connected to ground.

You have described a BALANCED input.  I'm specifically talking about
UNBALANCED inputs and outputs.

> In the case of Motorola radios, you'd have to ask them why, but never assume
> that either side of a speaker (for example) is chassis grounded,

I'm talking about UNBALANCED outputs and inputs for line level
interconnects, not loudspeakers.

> or you'll
> blow up a very expensive audio amp (like on the popular GM300).  A careful
> resistance test should be done before connecting audio ground to chassis.

Such an output should NOT be used to interconnect equipment.

> In general, we agree that chassis to chassis bonding SHOULD be encouraged to
> reduce noise (and is electrically safer having everything at the same ground
> potential).  But it does NOT always solve problems.

You are badly mistaken. Indeed, NEC REQUIRES that every chassis
connected to mains power be bonded to ground, and that ALL grounds be
bonded together.

> I've seen cases where one (presume a computer to radio) chassis-chassis
> connection made the problem worse (ground looped hum), because it introduced
> more noise into the radio, even if plugged into the same power source
> (should have been equal grounding already).

I suggest that you study my tutorial on audio interfacing that is part
of my RFI tutorial, and also the Power Point on Ham Interfacing, both of
which show WHY power-related noise is coupled into unbalanced
interfaces, and WHY chassis-to-chassis bonding kills it, and without the
need for a transformer.

> Audio isolation transformers
> were the solution, pass only the audio.

They were YOUR solution, but they are not the only solution, they are
not the simplest solution, and they are not the lowest cost solution.

> It might be against common practice, but reality trumps theory.  Chassis to
> chassis grounding may have any range of results (but certainly try it).

Not if you do it thoroughly.

> And in the case of RF, remember what may be good DC ground isn't always RF
> ground.

It is a MAJOR mistake to treat audio, RF, and power differently.  It is
this fuzzy thinking that CAUSES  the Pin One Problem, and fails to get
to the heart of what is really going on.  Again, I suggest that you
study my tutorials on this.  I'm not just "some guy with an opinion,"
I'm chair of the Tech Committee on EMC of the Audio Engineering Society,
vice-chair of the AES Standards Committee WG on EMC, and principal
author of all existing AES Standards on EMC. I had my first job in
broadcasting in 1959, and I've designed and used a very wide variety of
audio systems, from the very large to the very small, in a wide variety
of situations, including VERY high RF environments. In other words, I
know quite a bit about the real world.

73, Jim Brown K9YC
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Don Wilhelm-4]

I will only comment on the part reproduced below - audio return is NOT
audio ground.  In the case of the  Motorola amp cited, I suspect the
audio amp output is a bridged amplifier, and for those, neither speaker
lead should be grounded (it shorts half of the audio amplifier.

BUT, Jim Brown was talking about unbalanced audio inputs, not speaker
outputs.  In the ham world and the world of consumer audio, most audio
inputs are unbalanced, with one conductor (the shield) connected to
"ground".  That ground used to be the equipment enclosure, but more
recent implementations circumvent that and put the connection to the
shield onto the PC board rather than the enclosure - and often the
enclosure is plastic which creates yet another set of potential problems.

We used to connect our audio (and RF) jacks directly to the enclosure
and then inside the enclosure, we would route that connection to the
circuit contained in that enclosure - no "pin 1" problems because the
shield of connected cables were connected first to the outside of the
enclosure..  In today's implementations, pin 1 problems abound because
jacks for any and every purpose are designed to connect to the PC board
and may or may not have any connection to the equipment enclosure.
That allows noise and other undesired signals to be introduced onto the
device circuit board without being first stripped off by the (grounded)
metallic enclosure for the device in question.   We have come a long way
in convenience for manufacturers, but several problems have occurred
along the way, and they are not easy to correct.

73,
Don W3FPR


On 11/1/2011 9:58 PM, Rick Bates wrote:
> In the case of Motorola radios, you'd have to ask them why, but never assume
> that either side of a speaker (for example) is chassis grounded, or you'll
> blow up a very expensive audio amp (like on the popular GM300).  A careful
> resistance test should be done before connecting audio ground to chassis.
>
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

[Rick WA6NHC]

Thank you Don, that was clear and concise.  

I'm not an expert on radios or RF like some on this list.  I just try to go
with what works.  I was a firefighter in my previous life.  We adapt and
overcome.  I make things work, then I make them 'pretty'.  (Pretty in this
case means safer, more reliable and sometimes putting a 'bow on it.)  Simple
is good; it's often easier and elegant too.

We're basically saying the same thing but coming in from different points.
Chassis to chassis bonding is wanted, whenever possible.  It's not always
possible.  My old shack laptop has a two wire (ungrounded) mid line wart
power connection and bonding the laptop chassis to the station is an
exercise in futility (no grounding point on the laptop; the third wire if
any, grounds the power supply not the laptop).  A desktop chassis should be
bonded to the station (yes, as required by code but one size doesn't always
fit).

And yes, audio 'ground' (return) is rarely chassis ground these days.  We
agree again though I may have phrased it poorly or used the wrong terms
(floating ground) or examples, it was something I was trying to point out.
But that is a very common mistake to make too.  The two 'grounds' aren't the
same and connecting one to the other can cause huge problems, balanced or
not.  Don't assume that audio ground is chassis ground, it's not that way
anymore.

Considering the floating pin one problem, that makes many 'ground' wires,
(audio return, power, chassis and 'pin 1') each with a different meaning.
It'd be easy to presume they're all the same, but they aren't.  It makes
noise reduction even more difficult.

To the original problem of serial data noise:  With so much electronics
generated RF crud floating around a shack from numerous sources, it's often
difficult to find the cause(s) or cure(s) for that noise.  Better cables,
ferrites, common bonding and similar help but it's a long process to have a
quiet shack.  A product may meet FCC spec, but be ham lousy with noise.

What is working for noise reduction for me (K3, bonded to P3 bonded to
KPA500) was moving the antenna farther from the shack and using common mode
chokes to keep as much crud off of the feed as I could (reduced by more than
75%).  Next was finding quieter laptop power supplies or running off of
batteries.  I'm still chipping away at the rest (mostly Ethernet CAT 5 wires
radiating, switching to WiFi).  It has limited success with noise sources
not on my property, but its better.

And lastly, (then I'll go back to lurking since I'm no expert) when bonding
a station to ground (or at least to each other per NEC for those of us in
upper stories, not close to dirt or plumbing in the shack), while solid wire
meets the NEC code, we're also talking about RF.  So braid is best, followed
by multi-strand wire, and if nothing else at least a solid #10 wire.  Guess
what; if you bond for high RF environments, it works to help keep RF in its
place and meets the code too (you've exceeded the code with wide braid).

Hopefully I was more clear this time.  I get the code, concepts and means
though I may not state it clearly.  Comments, if any, can be made off group,
I'm sure the others are tired of this thread.

73,
Rick WA6NHC

-----Original Message-----
From: Don Wilhelm

I will only comment on the part reproduced below - audio return is NOT
audio ground.  In the case of the  Motorola amp cited, I suspect the
audio amp output is a bridged amplifier, and for those, neither speaker
lead should be grounded (it shorts half of the audio amplifier.

BUT, Jim Brown was talking about unbalanced audio inputs, not speaker
outputs.  In the ham world and the world of consumer audio, most audio
inputs are unbalanced, with one conductor (the shield) connected to
"ground".  That ground used to be the equipment enclosure, but more
recent implementations circumvent that and put the connection to the
shield onto the PC board rather than the enclosure - and often the
enclosure is plastic which creates yet another set of potential problems.

We used to connect our audio (and RF) jacks directly to the enclosure
and then inside the enclosure, we would route that connection to the
circuit contained in that enclosure - no "pin 1" problems because the
shield of connected cables were connected first to the outside of the
enclosure..  In today's implementations, pin 1 problems abound because
jacks for any and every purpose are designed to connect to the PC board
and may or may not have any connection to the equipment enclosure.
That allows noise and other undesired signals to be introduced onto the
device circuit board without being first stripped off by the (grounded)
metallic enclosure for the device in question.   We have come a long way
in convenience for manufacturers, but several problems have occurred
along the way, and they are not easy to correct.

73,
Don W3FPR


______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html