Network, cables & switches

Mark_WS7M

Good Saturday Morning all,

Recently I was browsing the community and came across a thread on radio disconnects.  I can't seem to find the recent thread. However in that thread there was a mention by a gent about finding a bad network switch and once he replaced it the radio issues disappeared.

I have had occasional radio disconnects as well.  I've not worried too much about it but after reading that post I became curious.

So yesterday I decided to check out all the switches in my home network.  I have quite a few due to when the house was remodeled I routed CAT5e everywhere so there are like 5 switches.

I found several interesting things.  At the point where my internet cable modem feeds my IQ router, the IQ router was feeding two managed 8 port switches.   These turned out to be 100mb switches.  I thought they were gig but it turned out they were not.

This was interesting because a couple months ago I had received the email from Comcast, my internet provider, that they had magically and at no cost to me upgraded my network speed.  Multiple speed tests from my computer still gave me a fairly lousy 40 mbps download.  But my IQ router would test and report over 500 mbps.

So I logged into the first managed switch and low and behold it reported its maximum capable rate of 100 mbps.  So why still was my computer not able to get close to 100!?  I began to dig into the managed switch menus and finally stumbled upon a statistics page which showed a whopping error stat of about 35%.  I checked the other managed switch and it was similar with actually a higher rate.

I tried a reboot of the switches and watched the stats pages during network download tests and both error rates rapidly incremented up to the 35% range.   These things were HP switches, not that old!  I went and checked my amazon order and clearly I tried to order gigabit switches but what I received were older models.  They looked the same but were 100 mb swtiches.

So first point:  When you order new equipment, especially these days of massive online ordering be very careful to check what you get.  I'm long past the return date on these switches by well over 3 years.

So I had a gigabit 8 port switch.  I swapped it in for one of the managed switches that feeds my desk computer.   Speed test.  Still 45 mbps.   So something still not right.

I had two trendnet gigabit switches with fiber sfp ports and I had fiber.  So as a test I ran one of those off the back of my IQRouter, then ran fiber across the floor to my computer desk and plugged the computer into the trendnet directly.

Speed test:  648 mbps download!   Wow...   I plugged the flex radio into this same trendnet and got my iPad out (cellular) and connected to my radio over smart link.  Suddenly things on the ipad were nice and snappy.  It was actually much more usable!

I ran a network cable to a WiFi access point from the trendnet.  I then fired up my maestro which had been kind of spitting and popping audio before.   I went outside, about 60 feet away to a nice table in the back yard and fired up my maestro on battery power.

I set the refresh rate to 25 fps and made the waterfall nice and fast.  Suddenly just nice clear audio.  No pops or clicks!

I logged into the WiFi access point and it was reporting very high throughput to the maestro.

So I continued to look at various wiring and switches and found another switch which was somehow degraded.  I replaced it with another small 8 port gig switch I had and suddenly Apple TV was MUCH happier as well.

So sorry this post got so long, but my point is simple.  If you have hardwired stuff or if you have multiple devices it pays if you are not sure of things to do some snooping around.

I ordered some new gig switches from amazon, all 12V DC powered so I can power them from nice quiet supplies but with my several patches and making use of the fiber channel I have all my WiFi access points, my desk computer, the radio all getting nice fast network throughput.

I left SmartSDR running last night and no disconnect.  I keep it running all the time and I will monitor to see if I get dropouts.

So check your gear!  

Mark_WS7M

I meant to add one other thing.  Based on a post by time about WiFi channels I did some reorg of my WiFi access points to separate channels and that has made a huge difference in how well the maestro performs on WiFi.

N2TU

Mark...Excellent! Thanks....

Ken W9IE

Mark,

I have also fought problems like this and as you know the solution can
sometimes be very elusive.

Case in Point......    For about a year I had slow and drop out problems between
my Maestro and Flex 6700.  Not a solid problem but one none the less.

I won't go through all my testing, changes, and problem solutions I tried
but I, in my mind narrowed it down to the Flex 6700.  
Remember the problem was not SOLID.

I called Flex, and went through the Service process and gave them the
situation that I knew it would fail with.

Flex put the 6700 on the bench and in a few days informed me there was
nothing wrong with the radio.

On the phone I Pleaded with them to keep testing....

They did and FINALLY I got the Email...   It failed, so now we can fix it.

The problem was a bad, NETWORK connector in the back of the Flex.

Problem solved and fixed.

My Point is that the same can occur in SWITCHES AND ROUTERS.

Changing Ports on a switch or router has fixed problems for me in the past.

73's  and Good Write UP on your Solution

Ken  W9IE

Ted VE3TRQ

Mark, I'm the "gent" you referred to above :-) Your experience about mirrors what I found. Never accept performance you cannot explain!

Mark_WS7M

Awesome Ted... For some reason this morning the article eluded me but yes your comments got me going on this and quite a difference!

I'm a software developer by trade and working at home now.  So having 600mbps speeds will certainly help over the 50 I was getting!

Mark_WS7M

Ken I remember your woes on that.  Glad it is all better.

Gremlins they call these things I think!

In my EE career for a short time doing circuit design before I went hard into computers I had designed a solid state guitar amp.  I used direct coupling to avoid capacitors everywhere.  The amp was flat from almost DC to the highest my little signal generator could make at the time.  That was about 50 khz.

I posted a note at our local guitar shop that I would custom make these amps if anyone was interested.  I got about 50 orders.

So I made circuit boards manually, ordered components and built like mad.  Tuning and alignment wasn't too much since it really was based on the component tolerances and if I found some bad components I just used different ones.

The problem was that the final darlington transistor pairs I needed to get to the desired 150W were very expensive at the time.  Close to $80 a pair.  After blowing a few up I decided to build a test jig that would simulate (of course with out the full gain) the final pair.

I built all 50 of my amps and tested them against the test jig.  Everything looked great on the scope.   So I assembled the finals and built 5 amps.  They performed perfectly.  I set them aside and moved on to the next 5 and so on.

I think I was on my final 5 and one of the amps in there when I turned it on with the finals connected went up in a puff of smoke taking the $80 finals with it.  

Visually I could see nothing wrong.  I finished the remaining 4 just fine and went back to the smokey one.  I had no choice but to rebuild it from scratch.   I got everything removed, put new components in, setup the test jig, it tested fine, frequency response good.   Built a new final set, hooked it up and filled the room with smoke again.

$160 of my meager profit was no gone.  I gave up for the day.

A few days later I took the components off and examined the circuit board.   Something just didn't feel right.  I measured the inductance and capacitance of a brand new circuit board, some extras I had and then measured them on this one that had blown and found they were substantially different.

So I loaded up the inexpensive components again, setup the test jig and started poking around and low and behold I found a major oscillation.  I mean big time.  

Why I hadn't see it before was because I was too set in my ways testing using my little jig.   I never looked at the power supply which was drawing almost 8 amps on this circuit board with absolutely no signal.  It should have been way down below 100ma.

My best guess is that went I put the finals on I drove them to something like 150% of their rated power and because of the direct coupling when they went they took out almost everything back down the line.

Gremlin.  Unseen, not obvious but expensive.

Tim N9PUZ

Just curious, what 12VDC 1GB switches did you select for your network upgrade?

Tim N9PUZ

Santiago Mejia HI8O

Following. Was going to ask the same.

Ted VE3TRQ

When I put in a switch with SFP connectors so I could link with fiber, all I could find for a reasonable price was a PoE switch that had a 48V power requirement. I bought a (relatively) inexpensive 12V-to-48V converter, tested it for RF noise, and was off to the races. Everything in the shack runs on 12V, including the Intel NUC that comes with an 18V brick, but runs just fine on 13 to 14V.  Ted  VE3TRQ

Mark_WS7M

I've had good luck with the consumer grade Netgear switches.

I get that they are just that, consumer grade, but they are simple, 12V DC and they seem to work very well:

https://www.amazon.com/gp/product/B07PFYM5MZ/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

$18 for 8 ports, 

I have a couple of the 16 port models but I tend to buy the 8 port models, two of them cost less than a 16 port.