[Deefoo]

For a customer on a ship in a remote location I've build a COTS 160MHz radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box). The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions. There is also an extra microcontroller board in the same box
with Ethernet connections on the same screw terminal as the VHF receiver
antenna. All cables (antenna, Ethernet & power) come in through cable
glands.

The VHF modem contains a PIC (10MHz) and it doesn't look like any special
effort has been done to prevent interference from the PIC to reach the input
stage except maybe that it is as far away as possible from the antenna
input. So I've mounted the extra microcontroller (22MHz) on the PIC side of
the modem.

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

Unfortunately I can't do any easy measurements and even if I could, I
probably do not have the right equipment (basically just a multimeter and an
oscilloscope).

What may be the problem? The aluminium box? The extra microcontroller board?
The Ethernet connection? The antenna connection? A combination? Other?

Any help will be greatly appreciated.

Thanks,
--DF

 

[Phil Allison]

"Deefoo"

For a customer on a ship in a remote location I've build a COTS 160MHz
radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box). The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions.

** 1st bad move.

There is also an extra microcontroller board in the same box

** 2nd bad move.

with Ethernet connections on the same screw terminal as the VHF receiver
antenna.

** Huh ?

All cables (antenna, Ethernet & power) come in through cable
glands.

** Watertight - maybe.

RF tight is another matter.

The VHF modem contains a PIC (10MHz) and it doesn't look like any special
effort has been done to prevent interference from the PIC to reach the
input
stage except maybe that it is as far away as possible from the antenna
input. So I've mounted the extra microcontroller (22MHz) on the PIC side
of
the modem.

** All has to do with the actual frequencies being created by the uP.

None at all can be allowed on 160 MHz or else the receiver is compromised.

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

** Many reasons are possible - but RFI from your handiwork is top of the
list.

Unfortunately I can't do any easy measurements and even if I could, I
probably do not have the right equipment (basically just a multimeter and
an
oscilloscope).

** Oh dear.




.......... Phil

 

[PeteS]

"Deefoo"


** 1st bad move.




** 2nd bad move.




** Huh ?




** Watertight - maybe.

RF tight is another matter.





** All has to do with the actual frequencies being created by the uP.

None at all can be allowed on 160 MHz or else the receiver is compromised.





** Many reasons are possible - but RFI from your handiwork is top of the
list.




** Oh dear.




......... Phil

I assume this is a standard Marine band setup. Apart from Phil's
comments (which I concur with - obviously, if the current setup is
worse, it's something or things that was done in the 'improvement),
there are certain tests you can make to help isolate some things.

If it's the processor, you'll typically find it interferes fairly
narrow band. Is the relative sensititvity worse on some channels (or
channel groups) than others? if so, which channels? That would normally
point to specific interference from harmonics / switching.

Apart from that, the RFI holes you have added should be properly
shielded.

You'll need more than a scope and meter, though, to fully test what's
going on.

Cheers

PeteS

 

[Frank Raffaeli]

For a customer on a ship in a remote location I've build a COTS 160MHz radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box). The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions. There is also an extra microcontroller board in the same box
with Ethernet connections on the same screw terminal as the VHF receiver
antenna. All cables (antenna, Ethernet & power) come in through cable
glands.

The VHF modem contains a PIC (10MHz) and it doesn't look like any special
effort has been done to prevent interference from the PIC to reach the input
stage except maybe that it is as far away as possible from the antenna
input. So I've mounted the extra microcontroller (22MHz) on the PIC side of
the modem.

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

Unfortunately I can't do any easy measurements and even if I could, I
probably do not have the right equipment (basically just a multimeter and an
oscilloscope).

What may be the problem? The aluminium box? The extra microcontroller board?
The Ethernet connection? The antenna connection? A combination? Other?

Any help will be greatly appreciated.

Thanks,
--DF

Changing the antenna connector to a "screw terminal" is unclear, but
it's not recommended. This will probably cause a mis-match. Depending
upon the severity of the mis-match, and coupling of RFI into the
antenna input, the sensitivity will be reduced. assume your antenna is
mounted elsewhere, and comes in via coaxial feed (50 ohms?). I'm not
sure what you mean by "space constraints". so perhaps you can provide a
picture. If the antenna connector is too large, try replacing the
connector with an SMA type on both the radio housing and the antenna
feed cable, using proper RF cabling techniques is essential. Check out
the Radio Amateur's Handbook for a guide to antenna connections.

Frank Raffaeli
http://www.aomwireless.com/

 

[Deefoo]

"Deefoo"


** 1st bad move.



** 2nd bad move.


** Huh ?

screw terminal _block_, not the same pins, obviously.

** Watertight - maybe.

RF tight is another matter.

Well, since the original modem came in an unshielded plastic box RF tight
should not really be an issue, now should it?

** All has to do with the actual frequencies being created by the uP.

None at all can be allowed on 160 MHz or else the receiver is compromised.


** Many reasons are possible - but RFI from your handiwork is top of the


** Oh dear.

The setup is at about 800 km from where I am...

Maybe placing a shield between the modem pcb and the microcontroller?

--DF

 

[Joerg]

Hello DF,

... The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions. There is also an extra microcontroller board in the same box
with Ethernet connections on the same screw terminal as the VHF receiver
antenna. All cables (antenna, Ethernet & power) come in through cable
glands.

If space is an issue use SMA or something, but definitely not a screw
terminal. Not at 160MHz, and most definitely not via the same terminal
block.

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

It isn't too surprising. Most likely crosstalk from the uC and modem
straight into the antenna line.

Unfortunately I can't do any easy measurements and even if I could, I
probably do not have the right equipment (basically just a multimeter and an
oscilloscope).

What may be the problem? The aluminium box? The extra microcontroller board?
The Ethernet connection? The antenna connection? A combination? Other?

Sorry to say but it appears that this might need a complete redesign.
First of all a coax connector. Decoupling of the supplies. Then look at
the materials. Dissimilar metals such as an aluminum box, steel screws
and so on can pose a substantial problem in maritime equipment,
especially on sea-going vessel because of the salt water. The usual
scenario is that noise gets worse over a few months until something
finally quits working altogether.

Regards, Joerg

 

[Deefoo]

I assume this is a standard Marine band setup. Apart from Phil's
comments (which I concur with - obviously, if the current setup is
worse, it's something or things that was done in the 'improvement),
there are certain tests you can make to help isolate some things.

If it's the processor, you'll typically find it interferes fairly
narrow band. Is the relative sensititvity worse on some channels (or
channel groups) than others? if so, which channels? That would normally
point to specific interference from harmonics / switching.

Apart from that, the RFI holes you have added should be properly
shielded.

You'll need more than a scope and meter, though, to fully test what's
going on.

Cheers

PeteS

Could it be that the grounded aluminium box makes matters worse by confining
any internally generated RFI to the box?

--DF

 

[Meindert Sprang]

For a customer on a ship in a remote location I've build a COTS 160MHz radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box)

This wouldn't be a NASA AIS receiver, would it?

The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions. There is also an extra microcontroller board in the same box
with Ethernet connections on the same screw terminal as the VHF receiver
antenna. All cables (antenna, Ethernet & power) come in through cable
glands.

This is really bad practice.

The VHF modem contains a PIC (10MHz) and it doesn't look like any special
effort has been done to prevent interference from the PIC to reach the input
stage except maybe that it is as far away as possible from the antenna
input. So I've mounted the extra microcontroller (22MHz) on the PIC side of
the modem.

The PIC on the board doesn't interfere because the board design is
preventing interference. How is your design with the extra PIC doing? Proper
layout?

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

Unfortunately I can't do any easy measurements and even if I could, I
probably do not have the right equipment (basically just a multimeter and an
oscilloscope).

What may be the problem? The aluminium box? The extra microcontroller board?
The Ethernet connection? The antenna connection? A combination? Other?

The combination of all. But all this is hard to say as we don't see
diagrams, PCB layouts, pictures. Making such a design work requires carefull
EMC proof design. Not to mention the required approvals (compliance with
IEC/EN standards required by the IMO on commercial vessels).

Meindert

 

[JustMe]

What may be the problem? The aluminium box? The extra microcontroller
board?
The Ethernet connection? The antenna connection? A combination? Other?

Any help will be greatly appreciated.

Thanks,
--DF

There are no easy answers to your immediate problem.

A few suggestions though. Ship board equipment is not usually size critical
so this may do the trick.

I assume this modification was intended to protect the modem from the
various forces conspiring to corrode it to a crusty non-functioning lump.
Can I suggest putting a completely unmodified and originally cased modem,
centrally positioned in a LARGER diecast box, and then silicone sealing all
cables where they penetrate the metal enclosure. Five centimetre spacing all
round the modem as a minimum, ten if you can manage it. Silica gel packs to
mop up any moisture. Polystyrene foam to maintain spacing from the walls of
the enclosure.

If the receiver is still desensitised inside the box, then it is almost
certainly the pic micro switching transients. A fallback then could be to
ditch the diecast and do the job with a half dozen TOUGH poly bags sealed
one after the other (ugh!), not forgetting the silica gel packs.

This will certainly not be pretty or compact, but it may just dig you out of
the hole the project is in.

 

[Rene Tschaggelar]

For a customer on a ship in a remote location I've build a COTS 160MHz radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box). The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions. There is also an extra microcontroller board in the same box
with Ethernet connections on the same screw terminal as the VHF receiver
antenna. All cables (antenna, Ethernet & power) come in through cable
glands.

The VHF modem contains a PIC (10MHz) and it doesn't look like any special
effort has been done to prevent interference from the PIC to reach the input
stage except maybe that it is as far away as possible from the antenna
input. So I've mounted the extra microcontroller (22MHz) on the PIC side of
the modem.

Now the customer has done comparative tests with the original setup (VHF
receiver in plastic box without extra microcontroller) and with the
"improved" setup and found that the receiver sensitivity has significantly
diminished for the "improved" setup.

To start with, compare the receiver input sensitivity with the
transients measured from the PIC.

Rene

 

[Fred Bloggs]

What may be the problem? The aluminium box? The extra microcontroller board?
The Ethernet connection? The antenna connection? A combination? Other?

The major problem may very well be the aluminum box if you tied this to
circuit common in several places, such as corner metal standoffs,
creating circulating currents where there were none before. The antenna
screw terminals at 160MHz should cause negligible loss as long as the
launch from the board remains short. But now that you have mentioned the
compact size of the packaging, the aluminum box can introduce a
significant amount shunt capacitance at RF where none existed previously
and this will add to input attenuation as well as possibly seriously
de-tune any input frequency and IF selection circuits. My money is on
de-tuning of the frequency selective circuits as the major cause of the
problem. Somewhere in there is possible an AGC or field strength
indicator which can be used to possibly adjust the tuning for maximum
signal level- with a VOM.

 

[Deefoo]

The major problem may very well be the aluminum box if you tied this to
circuit common in several places, such as corner metal standoffs,
creating circulating currents where there were none before. The antenna
screw terminals at 160MHz should cause negligible loss as long as the
launch from the board remains short. But now that you have mentioned the
compact size of the packaging, the aluminum box can introduce a
significant amount shunt capacitance at RF where none existed previously
and this will add to input attenuation as well as possibly seriously
de-tune any input frequency and IF selection circuits. My money is on
de-tuning of the frequency selective circuits as the major cause of the
problem. Somewhere in there is possible an AGC or field strength
indicator which can be used to possibly adjust the tuning for maximum
signal level- with a VOM.

The only contact between the circuit and the box is through two varistors
and two capacitors on the power lines. Power is isolated by a DCDC converter
and the output is over Ethernet. Does your shunt capacitance hypothesis
still hold in this case?

Thanks,
--DF

 

[RST Engineering \(jw\)]

RF Rule #1. If it works OK as purchased and doesn't work after
modifications, do the modifications one by one until it doesn't work. Then
find out what the problem was with the last mod.

Jim's Rule #2. I don't care if you don't want to top post, but if you want
to interpost or bottom post, learn how to snip.

Jim

 

[Le Chaud Lapin]

For a customer on a ship in a remote location I've build a COTS 160MHz radio
modem into a waterproof diecast aluminium box that is grounded (the modem
originally comes in a plastic box). The antenna used to connect to a BNC
connector but I replaced that by a screw terminal because of space
restrictions.

What exactly did you do with the original antenna configuration? This
would be the first thing to look at.

Secondly, the aluminum box introduces a ground plane that itself is
grounded. If the modem had been purposely built with an internal
ground plane, your aluminum box will distort/kill both transmission and
reception. So first thing you could do is de-ground the box. Don't
let anyone touch the box while testing. Put a towel or something
between the case and the mount and then test. If you don't see
significant improvement, and you didn't do anything strange with screw
terminal to cause impedance mismatch, you're probably going to have to
go back to that plastic box.

-Le Chaud Lapin-

 

[Fred Bloggs]

The only contact between the circuit and the box is through two varistors
and two capacitors on the power lines. Power is isolated by a DCDC converter
and the output is over Ethernet. Does your shunt capacitance hypothesis
still hold in this case?

Thanks,
--DF

Yes- the equivalent circuit is as shown below. There may also be an
issue of proximity effect reducing the selectivity of the tuned tanks:
View in a fixed-width font such as Courier.

 

[Deefoo]

Yes- the equivalent circuit is as shown below. There may also be an
issue of proximity effect reducing the selectivity of the tuned tanks:
View in a fixed-width font such as Courier.

.
.
.
.
.
. RF Node
. |
. ckt gnd v ckt gnd
. ckt brd --------------- - ----------------
. <-capacitor
. .-------+------ --- -------+-------. air dielectirc
. | | | | |
. | '---/\/\-+-/\/\----' |
. | |
. | |
. | Al Box Xsection |
.
.
.

Oh OK, I see. I am planning tests next friday. Thanks to you and the other
repliers I will first concentrate on the antenna connection and the box. It
is easy to take everything out of the box so that will be test number one.

Thanks,
--DF

 

[Deefoo]

Oh OK, I see. I am planning tests next friday. Thanks to you and the other
repliers I will first concentrate on the antenna connection and the box. It
is easy to take everything out of the box so that will be test number one.

Thanks,
--DF

For those who care:

Some tests have been carried out and dramatic improvement of the receiver
sensitivity was observed when connecting the antenna properly (BNC, not on
screw terminal) as close as possible to the original antenna input. The
metal enclosure did not seem to have any influence at all, with or without
it the reception quality stayed the same.

Thanks a lot everyone for your suggestions!

--DF