I have recently been and got myself some programming cables from
MicroControls. They may be quite helpful during development if the
grounding is suspected as a source of trouble. Instead of pluging
and un-pluging the "Prog" and "Diag" connectors onto the header there
is blue box with a switch to do the same, therefore the ground is
permantly connected. So as long as you hook up the first time
correctly everything should be ok.

I have not ever purchased a lot of RCM's but I have had a dudd one
that I could'nt program. Its a pitty theres no spare board space for
some proper rs232 or rs485 buffers on the programming port, that
would make them more robust.

Neil

--- In r..., "Alexis" wrote:
>
> LOL
>
>
>
> The irony is the best battery chemistries still contain lead. Ni-
cads are
> great, look after them and they can last years.
>
>
>
> These new types (NiMh) start decaying the day they are made. They
only last
> about 2 years from date of manufacture. So when purchasing one, ask
when it
> was made. Imagine if it lay on the shelf for 18 months..
>
>
>
> Of course there is the hype about the RoHS solder and fluxes being
more
> toxic than agents used in leaded products. So the poor personnel in
the
> assembly plants are now exposed to more hazardous chemicals.
>
>
>
> Another point about device failures. I have an interesting theory.
Some of
> you will laugh at me, but here goes.
>
>
>
> On the Rabbit CPU's there are the programming lines that are routed
to a
> programming header. Personally I have never had a CPU fail. I use
the
> modules and a custom design with their CPU's. In the field that is
another
> issue. I have had many cards returned for repair. In each and every
case I
> had to replace the CPU. It took me a while to figure this out - and
since
> then the problem seems to have gone.
>
>
>
> Both my clients were using laptops to program the devices. I think
that
> there was a potential difference between the cards and the laptop.
When
> connecting the programming cable the Rabbit was damaged. After
issuing an
> instruction that both devices must share a common ground, that the
> programming connector be connected to the rabbit first and then the
laptop
> the problem has gone. I believe the programming header on the
boards should
> have a ground pin that is a little longer than the others. When
connecting
> the programming cable, this pin will make contact first.
>
>
>
> Alexis.
>
>
>
> _____
>
> From: r... [mailto:rabbit-
s...] On
> Behalf Of neil2452
> Sent: Monday, August 13, 2007 5:14 AM
> To: r...
> Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
>
>
>
> Its realy hard to fathom the logic that the EU RoSH burocrats used
> when they dreamt up the idea that to much lead is put in the
> environment by the electronics industry. They want a little guy
like
> me who uses less than 500g a year not to use lead solder but Ford &
> GM etc can use as much as they like because there in the exempt
auto
> industry !! Its bananas, they should start with the big guys first.
>
> Also its now illegal to repair church organ pipes made of lead here
> in the EU but it seems to be ok to slap 50Kg patch of lead on the
> church roof to fix a leak. Its bananas, do they assume the vicar is
> going to be licking his organ and die from lead poisoning, but the
> verger can use the rain water off that lovely lead roof to water
his
> cabbages and thats safe !!
>
> --- In rabbit-semi@
yahoogroups.com,
> "Alan Matheson"
> wrote:
> >
> > Many thanks Neil.
> > Youre comments about lead are also interesting. For various
reasons
> we are using lead-tin rechargable batteries in our product. So in
> Europe we will have to meet rohsys standards and use lead free
> components but they will be powered by batteries containg maybe 500
> grams of lead and thats all legal.
> >
> > Regards
> > Alan Matheson
> > Atech Research Ltd
> > New Zealand
> >
> >
> >
> > ----- Original Message -----
> > From: neil2452
> > To: rabbit-semi@
yahoogroups.com
> > Sent: Monday, August 13, 2007 1:35 PM
> > Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
> >
> >
> > You need answers from Zworld I think. The questions I would be
> > asking
> > are what Q&A procedures are they using, and what testing is
> happening
> > to the product before it is shipped.
> >
> > When I used to work with Lucas some years ago each and every ECU
> unit
> > they made would go through a burn in test. This was literally
> > running the boards in an oven for several hours hooked up to a
> peice
> > of ATE that simulated the car and checked for the correct
> operation
> > of all the I/O pins. We made ATE (Automatic Test Equipment) for
> > them.
> >
> > As an example of PCB assembly yeild rates I have just had about
> 250
> > not too sophisticated SMT PIC circuits built in LeadFree. I had
> 25
> > out of the batch that failed on my test rig. Most fail due to
> simple
> > faults like bridged pins, or toomstoning, or occasionaly missing
> > componenets. The raw PCBs should normally have gone through an
> > automatic bare board test so there should be less to worry about
> > there. Also I can't remember the last time I had duff chip from a
> > semiconductor manufacturer so I am sure they do rigourous QA and
> > testing. I think you have a bigger chance of winning the lotto
> than
> > finding a new duff part from them.
> >
> > The RCM modules I would think are quite deemanding circuits to
> build
> > given a multilayer PCB design and the realy small components they
> use
> > and now LeadFree (Why are we crazey about not using lead solder
> in
> > europe, I bet theres more lead on are church roofs than in our
> tv's
> > etc). If they just drop the boards in a box as they roll off the
> end
> > of the line then duff boards will be shipped for sure. However if
> > are they are doing a proper burnin test with ATE that checks
> > everything then how can they explain your 5% failure rate ?
> Perhaps
> > the answer is somewhere between the two.
> >
> > Regards
> > Neil
> >
> > --- In rabbit-semi@
> yahoogroups.com, "Alan Matheson"
> >
> > wrote:
> > >
> > > We are beginning to notice problems with Zworld RCM3100
> modules. A
> > small percentage, around 5% I'm told of brand new units won't
> accept
> > a
> > 300kb program. Tend to give up half way through the program load.
> > Given that we are looking at using 200 per month this is very bad
> > news.
> > Has anyone else come across this problem.
> > > Alan Matheson
> > >
>

LOL

The irony is the best battery chemistries still contain lead. Ni-cads are
great, look after them and they can last years.

These new types (NiMh) start decaying the day they are made. They only last
about 2 years from date of manufacture. So when purchasing one, ask when it
was made. Imagine if it lay on the shelf for 18 months..

Of course there is the hype about the RoHS solder and fluxes being more
toxic than agents used in leaded products. So the poor personnel in the
assembly plants are now exposed to more hazardous chemicals.

Another point about device failures. I have an interesting theory. Some of
you will laugh at me, but here goes.

On the Rabbit CPU's there are the programming lines that are routed to a
programming header. Personally I have never had a CPU fail. I use the
modules and a custom design with their CPU's. In the field that is another
issue. I have had many cards returned for repair. In each and every case I
had to replace the CPU. It took me a while to figure this out - and since
then the problem seems to have gone.

Both my clients were using laptops to program the devices. I think that
there was a potential difference between the cards and the laptop. When
connecting the programming cable the Rabbit was damaged. After issuing an
instruction that both devices must share a common ground, that the
programming connector be connected to the rabbit first and then the laptop
the problem has gone. I believe the programming header on the boards should
have a ground pin that is a little longer than the others. When connecting
the programming cable, this pin will make contact first.

Alexis.

_____

From: r... [mailto:r...] On
Behalf Of neil2452
Sent: Monday, August 13, 2007 5:14 AM
To: r...
Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues

Its realy hard to fathom the logic that the EU RoSH burocrats used
when they dreamt up the idea that to much lead is put in the
environment by the electronics industry. They want a little guy like
me who uses less than 500g a year not to use lead solder but Ford &
GM etc can use as much as they like because there in the exempt auto
industry !! Its bananas, they should start with the big guys first.

Also its now illegal to repair church organ pipes made of lead here
in the EU but it seems to be ok to slap 50Kg patch of lead on the
church roof to fix a leak. Its bananas, do they assume the vicar is
going to be licking his organ and die from lead poisoning, but the
verger can use the rain water off that lovely lead roof to water his
cabbages and thats safe !!

--- In rabbit-semi@ yahoogroups.com,
"Alan Matheson"
wrote:
>
> Many thanks Neil.
> Youre comments about lead are also interesting. For various reasons
we are using lead-tin rechargable batteries in our product. So in
Europe we will have to meet rohsys standards and use lead free
components but they will be powered by batteries containg maybe 500
grams of lead and thats all legal.
>
> Regards
> Alan Matheson
> Atech Research Ltd
> New Zealand
>
> ----- Original Message -----
> From: neil2452
> To: rabbit-semi@ yahoogroups.com
> Sent: Monday, August 13, 2007 1:35 PM
> Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
> You need answers from Zworld I think. The questions I would be
> asking
> are what Q&A procedures are they using, and what testing is
happening
> to the product before it is shipped.
>
> When I used to work with Lucas some years ago each and every ECU
unit
> they made would go through a burn in test. This was literally
> running the boards in an oven for several hours hooked up to a
peice
> of ATE that simulated the car and checked for the correct
operation
> of all the I/O pins. We made ATE (Automatic Test Equipment) for
> them.
>
> As an example of PCB assembly yeild rates I have just had about
250
> not too sophisticated SMT PIC circuits built in LeadFree. I had
25
> out of the batch that failed on my test rig. Most fail due to
simple
> faults like bridged pins, or toomstoning, or occasionaly missing
> componenets. The raw PCBs should normally have gone through an
> automatic bare board test so there should be less to worry about
> there. Also I can't remember the last time I had duff chip from a
> semiconductor manufacturer so I am sure they do rigourous QA and
> testing. I think you have a bigger chance of winning the lotto
than
> finding a new duff part from them.
>
> The RCM modules I would think are quite deemanding circuits to
build
> given a multilayer PCB design and the realy small components they
use
> and now LeadFree (Why are we crazey about not using lead solder
in
> europe, I bet theres more lead on are church roofs than in our
tv's
> etc). If they just drop the boards in a box as they roll off the
end
> of the line then duff boards will be shipped for sure. However if
> are they are doing a proper burnin test with ATE that checks
> everything then how can they explain your 5% failure rate ?
Perhaps
> the answer is somewhere between the two.
>
> Regards
> Neil
>
> --- In rabbit-semi@
yahoogroups.com, "Alan Matheson"
>
> wrote:
> >
> > We are beginning to notice problems with Zworld RCM3100
modules. A
> small percentage, around 5% I'm told of brand new units won't
accept
> a
> 300kb program. Tend to give up half way through the program load.
> Given that we are looking at using 200 per month this is very bad
> news.
> Has anyone else come across this problem.
> > Alan Matheson
>

Its realy hard to fathom the logic that the EU RoSH burocrats used
when they dreamt up the idea that to much lead is put in the
environment by the electronics industry. They want a little guy like
me who uses less than 500g a year not to use lead solder but Ford &
GM etc can use as much as they like because there in the exempt auto
industry !! Its bananas, they should start with the big guys first.

Also its now illegal to repair church organ pipes made of lead here
in the EU but it seems to be ok to slap 50Kg patch of lead on the
church roof to fix a leak. Its bananas, do they assume the vicar is
going to be licking his organ and die from lead poisoning, but the
verger can use the rain water off that lovely lead roof to water his
cabbages and thats safe !!
--- In r..., "Alan Matheson"
wrote:
>
> Many thanks Neil.
> Youre comments about lead are also interesting. For various reasons
we are using lead-tin rechargable batteries in our product. So in
Europe we will have to meet rohsys standards and use lead free
components but they will be powered by batteries containg maybe 500
grams of lead and thats all legal.
>
> Regards
> Alan Matheson
> Atech Research Ltd
> New Zealand
>
> ----- Original Message -----
> From: neil2452
> To: r...
> Sent: Monday, August 13, 2007 1:35 PM
> Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
> You need answers from Zworld I think. The questions I would be
> asking
> are what Q&A procedures are they using, and what testing is
happening
> to the product before it is shipped.
>
> When I used to work with Lucas some years ago each and every ECU
unit
> they made would go through a burn in test. This was literally
> running the boards in an oven for several hours hooked up to a
peice
> of ATE that simulated the car and checked for the correct
operation
> of all the I/O pins. We made ATE (Automatic Test Equipment) for
> them.
>
> As an example of PCB assembly yeild rates I have just had about
250
> not too sophisticated SMT PIC circuits built in LeadFree. I had
25
> out of the batch that failed on my test rig. Most fail due to
simple
> faults like bridged pins, or toomstoning, or occasionaly missing
> componenets. The raw PCBs should normally have gone through an
> automatic bare board test so there should be less to worry about
> there. Also I can't remember the last time I had duff chip from a
> semiconductor manufacturer so I am sure they do rigourous QA and
> testing. I think you have a bigger chance of winning the lotto
than
> finding a new duff part from them.
>
> The RCM modules I would think are quite deemanding circuits to
build
> given a multilayer PCB design and the realy small components they
use
> and now LeadFree (Why are we crazey about not using lead solder
in
> europe, I bet theres more lead on are church roofs than in our
tv's
> etc). If they just drop the boards in a box as they roll off the
end
> of the line then duff boards will be shipped for sure. However if
> are they are doing a proper burnin test with ATE that checks
> everything then how can they explain your 5% failure rate ?
Perhaps
> the answer is somewhere between the two.
>
> Regards
> Neil
>
> --- In r..., "Alan Matheson"
>
> wrote:
> >
> > We are beginning to notice problems with Zworld RCM3100
modules. A
> small percentage, around 5% I'm told of brand new units won't
accept
> a
> 300kb program. Tend to give up half way through the program load.
> Given that we are looking at using 200 per month this is very bad
> news.
> Has anyone else come across this problem.
> > Alan Matheson
>

Many thanks Neil.
Youre comments about lead are also interesting. For various reasons we are using lead-tin rechargable batteries in our product. So in Europe we will have to meet rohsys standards and use lead free components but they will be powered by batteries containg maybe 500 grams of lead and thats all legal.

Regards
Alan Matheson
Atech Research Ltd
New Zealand

----- Original Message -----
From: neil2452
To: r...
Sent: Monday, August 13, 2007 1:35 PM
Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
You need answers from Zworld I think. The questions I would be
asking
are what Q&A procedures are they using, and what testing is happening
to the product before it is shipped.

When I used to work with Lucas some years ago each and every ECU unit
they made would go through a burn in test. This was literally
running the boards in an oven for several hours hooked up to a peice
of ATE that simulated the car and checked for the correct operation
of all the I/O pins. We made ATE (Automatic Test Equipment) for
them.

As an example of PCB assembly yeild rates I have just had about 250
not too sophisticated SMT PIC circuits built in LeadFree. I had 25
out of the batch that failed on my test rig. Most fail due to simple
faults like bridged pins, or toomstoning, or occasionaly missing
componenets. The raw PCBs should normally have gone through an
automatic bare board test so there should be less to worry about
there. Also I can't remember the last time I had duff chip from a
semiconductor manufacturer so I am sure they do rigourous QA and
testing. I think you have a bigger chance of winning the lotto than
finding a new duff part from them.

The RCM modules I would think are quite deemanding circuits to build
given a multilayer PCB design and the realy small components they use
and now LeadFree (Why are we crazey about not using lead solder in
europe, I bet theres more lead on are church roofs than in our tv's
etc). If they just drop the boards in a box as they roll off the end
of the line then duff boards will be shipped for sure. However if
are they are doing a proper burnin test with ATE that checks
everything then how can they explain your 5% failure rate ? Perhaps
the answer is somewhere between the two.

Regards
Neil

--- In r..., "Alan Matheson"

wrote:
>
> We are beginning to notice problems with Zworld RCM3100 modules. A
small percentage, around 5% I'm told of brand new units won't accept
a
300kb program. Tend to give up half way through the program load.
Given that we are looking at using 200 per month this is very bad
news.
Has anyone else come across this problem.
> Alan Matheson
>

You need answers from Zworld I think. The questions I would be
asking
are what Q&A procedures are they using, and what testing is happening
to the product before it is shipped.

When I used to work with Lucas some years ago each and every ECU unit
they made would go through a burn in test. This was literally
running the boards in an oven for several hours hooked up to a peice
of ATE that simulated the car and checked for the correct operation
of all the I/O pins. We made ATE (Automatic Test Equipment) for
them.

As an example of PCB assembly yeild rates I have just had about 250
not too sophisticated SMT PIC circuits built in LeadFree. I had 25
out of the batch that failed on my test rig. Most fail due to simple
faults like bridged pins, or toomstoning, or occasionaly missing
componenets. The raw PCBs should normally have gone through an
automatic bare board test so there should be less to worry about
there. Also I can't remember the last time I had duff chip from a
semiconductor manufacturer so I am sure they do rigourous QA and
testing. I think you have a bigger chance of winning the lotto than
finding a new duff part from them.

The RCM modules I would think are quite deemanding circuits to build
given a multilayer PCB design and the realy small components they use
and now LeadFree (Why are we crazey about not using lead solder in
europe, I bet theres more lead on are church roofs than in our tv's
etc). If they just drop the boards in a box as they roll off the end
of the line then duff boards will be shipped for sure. However if
are they are doing a proper burnin test with ATE that checks
everything then how can they explain your 5% failure rate ? Perhaps
the answer is somewhere between the two.

Regards
Neil
--- In r..., "Alan Matheson"

wrote:
>
> We are beginning to notice problems with Zworld RCM3100 modules. A
small percentage, around 5% I'm told of brand new units won't accept
a
300kb program. Tend to give up half way through the program load.
Given that we are looking at using 200 per month this is very bad
news.
Has anyone else come across this problem.
> Alan Matheson
>

We are beginning to notice problems with Zworld RCM3100 modules. A small percentage, around 5% I'm told of brand new units won't accept a 300kb program. Tend to give up half way through the program load. Given that we are looking at using 200 per month this is very bad news. Has anyone else come across this problem.
Alan Matheson

I assume the problematic core modules do work, aside from the
Ethernet-related failure you are seeing. If you have not already, you
should give special attention to the solder connections on the Ethernet
connectors. If any of them appear to be disconnected, or if the
connector is a loose that could be the culprit.

Also, are you noticing these issues with units that were manufactured
around a particular time period?

Omid

________________________________

From: r... [mailto:r...]
On Behalf Of Mariss Freimanis
Sent: Tuesday, August 07, 2007 10:17 AM
To: r...
Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues

Numbers 01 thru 07 are RoHS compliant modules. Numbers 08 thru 11 are
the green, non-RoHS boards.

Visual inspection with a microscope reveals no problems associated
with lead-free solder such as poor wetting of joints, fractured solder
or tin dendrite growth.

Mariss

--- In r...
, "Fournier, Pete"

wrote:
>
> Would the latest (failing) ones be the Lead-Free versions (red PCB)?
> -Pete
>
> -----Original Message-----
> From: r...

[mailto:r...
]
> On Behalf Of Mariss Freimanis
> Sent: Tuesday, August 07, 2007 12:04 PM
> To: r...

> Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
>
> Thank you for the reply. I'm not sure it can be heat related in this
> case:
>
> 1) Input power is 230mA at 12VDC or 2.76 Watts. The regulator
dissipates
> 1.61 Watts leaving 1.15 Watts for the entire circuit.
>
> 2) The package is aluminum, 2.5" by 1.5" by 7" (63mm by 38mm by
178mm).
> The measured temperature rise is 2.8C on the package surface, the
> measured internal temperature rise is 3.6C above ambient.
>
> 3) The RCM3720 thermal "exclusion zone" requirements have been
followed.
>
> 4) The RCM3720 is rated for operation at a 70C ambient. It sees a
> maximum ambient of 44C inside the package (for 40C external).
>
> 5) 2 to 3 RCM3720's out of 100 ordered are DOA right out of the box.
> The refuse to flash with either RFU.exe or Flashit.exe on initial
> installation.
>
> The Rabbits are terrific little modules at a great price and we are
> otherwise very pleased with them. I have to assume we are doing
> something wrong to have these problems. I would like to know what it
may
> be.
>
> Mariss
>
> Below is a sample of failed RCM3720s
> ---------------
> 01. CE84AC Flashes OK. Get "TCP/IP error 735". Field failure.
> 02. CE84C1 RFU "Error reading bytes". Initial failure.
> 03. CE84C3 RFU "Flash parameters not found". Initial failure.
> 04. CE84B7 RFU "No Rabbit detected". Field failure.
> 05. CE84C4 Same as (02). Field failure.
> 06. CE8416 Same as (03). Field failure.
> 07. CE95FF Flashes OK. No Ethernet. Field failure.
> 08. C8C1DF RFU "Target comm error". Initial failure.
> 09. CBBFF2 Flashes OK. Port A failed. Field failure.
> 10. CBBEC5 Same as (09). Field failure.
> 11. CBBEC2 Flashes OK. DS1 LED 'on' w/o Ethernet cable. Field failure.
>
> In all cases replacing the RCM3720 fixed the problem. There were no
> problems with the rest of the hardware.
>
> --- In r...
, "neil2452"
wrote:
> >
> > I had similar problems with a few years ago with the RCM3400. I had
> > it all in a seal box with a 20w switch mode power supply on my own
PCB
>
> > with about 9 PIC's on it. The rabbit was effectly a TCP/IP to serial

> > converter. I found that once they were installed they over heated
and
>
> > seem to loose TCP/IP connection. They were installed high up in the
> > room where there was a lot of heat from lights and crowds of peope
> > below. The PIC's never crashed but once the ambient temp got to
about
>
> > 25c the rabbit got to hot inside the box.
> >
> > I solved the problem by squeezing in a realy small DC fan to direct
> > air onto the rabbit. They don't seem like the heat !
> >
> >
> > --- In r...
, "Mariss Freimanis"
> > wrote:
> > >
> > > Is anyone having reliability problems with the RCM3720?
> > >
> > > We have used several hundred RCM3720s in a product we build. After
a
>
> > > few days to a few months the RCM3720 will lose the ability to
> > connect
> > > via Ethernet, the on board Ethernet controller gets hot and the
> > > RCM3720 will not respond even to RFU.exe when we get the product
> > back
> > > from the field.
> > >
> > > The RCM3720 operates in a benign environment: Sealed enclosure,
25C
> > > ambient, LM7805 supply regulator and no RCM3720 I/O pin is exposed
> > to
> > > the outside world. All internal I/O circuitry interfacing with the

> > > RCM3720 is 3.3V logic.
> > >
> > > Our failure rate is approaching 10% of units shipped. Invariably
it
> > is
> > > the RCM3720; once replaced the FPGA and 74HC series logic
> > constituting
> > > the rest of the circuit runs flawlessly.
> > >
> > > Mariss Freimanis
> > > Geckodrive Inc.
> > >
> > Yahoo! Groups Links
>

Numbers 01 thru 07 are RoHS compliant modules. Numbers 08 thru 11 are
the green, non-RoHS boards.

Visual inspection with a microscope reveals no problems associated
with lead-free solder such as poor wetting of joints, fractured solder
or tin dendrite growth.

Mariss

--- In r..., "Fournier, Pete"
wrote:
>
> Would the latest (failing) ones be the Lead-Free versions (red PCB)?
> -Pete
>
> -----Original Message-----
> From: r... [mailto:r...]
> On Behalf Of Mariss Freimanis
> Sent: Tuesday, August 07, 2007 12:04 PM
> To: r...
> Subject: [rabbit-semi] Re: RCM3720 Reliabilty Issues
>
> Thank you for the reply. I'm not sure it can be heat related in this
> case:
>
> 1) Input power is 230mA at 12VDC or 2.76 Watts. The regulator dissipates
> 1.61 Watts leaving 1.15 Watts for the entire circuit.
>
> 2) The package is aluminum, 2.5" by 1.5" by 7" (63mm by 38mm by 178mm).
> The measured temperature rise is 2.8C on the package surface, the
> measured internal temperature rise is 3.6C above ambient.
>
> 3) The RCM3720 thermal "exclusion zone" requirements have been followed.
>
> 4) The RCM3720 is rated for operation at a 70C ambient. It sees a
> maximum ambient of 44C inside the package (for 40C external).
>
> 5) 2 to 3 RCM3720's out of 100 ordered are DOA right out of the box.
> The refuse to flash with either RFU.exe or Flashit.exe on initial
> installation.
>
> The Rabbits are terrific little modules at a great price and we are
> otherwise very pleased with them. I have to assume we are doing
> something wrong to have these problems. I would like to know what it may
> be.
>
> Mariss
>
> Below is a sample of failed RCM3720s
> ---------------
> 01. CE84AC Flashes OK. Get "TCP/IP error 735". Field failure.
> 02. CE84C1 RFU "Error reading bytes". Initial failure.
> 03. CE84C3 RFU "Flash parameters not found". Initial failure.
> 04. CE84B7 RFU "No Rabbit detected". Field failure.
> 05. CE84C4 Same as (02). Field failure.
> 06. CE8416 Same as (03). Field failure.
> 07. CE95FF Flashes OK. No Ethernet. Field failure.
> 08. C8C1DF RFU "Target comm error". Initial failure.
> 09. CBBFF2 Flashes OK. Port A failed. Field failure.
> 10. CBBEC5 Same as (09). Field failure.
> 11. CBBEC2 Flashes OK. DS1 LED 'on' w/o Ethernet cable. Field failure.
>
> In all cases replacing the RCM3720 fixed the problem. There were no
> problems with the rest of the hardware.
>
>
>
> --- In r..., "neil2452" wrote:
> >
> > I had similar problems with a few years ago with the RCM3400. I had
> > it all in a seal box with a 20w switch mode power supply on my own PCB
>
> > with about 9 PIC's on it. The rabbit was effectly a TCP/IP to serial
> > converter. I found that once they were installed they over heated and
>
> > seem to loose TCP/IP connection. They were installed high up in the
> > room where there was a lot of heat from lights and crowds of peope
> > below. The PIC's never crashed but once the ambient temp got to about
>
> > 25c the rabbit got to hot inside the box.
> >
> > I solved the problem by squeezing in a realy small DC fan to direct
> > air onto the rabbit. They don't seem like the heat !
> >
> >
> > --- In r..., "Mariss Freimanis"
> > wrote:
> > >
> > > Is anyone having reliability problems with the RCM3720?
> > >
> > > We have used several hundred RCM3720s in a product we build. After a
>
> > > few days to a few months the RCM3720 will lose the ability to
> > connect
> > > via Ethernet, the on board Ethernet controller gets hot and the
> > > RCM3720 will not respond even to RFU.exe when we get the product
> > back
> > > from the field.
> > >
> > > The RCM3720 operates in a benign environment: Sealed enclosure, 25C
> > > ambient, LM7805 supply regulator and no RCM3720 I/O pin is exposed
> > to
> > > the outside world. All internal I/O circuitry interfacing with the
> > > RCM3720 is 3.3V logic.
> > >
> > > Our failure rate is approaching 10% of units shipped. Invariably it
> > is
> > > the RCM3720; once replaced the FPGA and 74HC series logic
> > constituting
> > > the rest of the circuit runs flawlessly.
> > >
> > > Mariss Freimanis
> > > Geckodrive Inc.
> > >
> >
>
> Yahoo! Groups Links
>

Mariss,

Please contact Technical Support or your Salesperson and we would be
happy to work with you on the issues you are seeing.

Thanks
Joel Baumert
Director, Engineering
--- In r..., "Mariss Freimanis"
wrote:
>
> Is anyone having reliability problems with the RCM3720?
>
> We have used several hundred RCM3720s in a product we build. After a
> few days to a few months the RCM3720 will lose the ability to
connect
> via Ethernet, the on board Ethernet controller gets hot and the
> RCM3720 will not respond even to RFU.exe when we get the product
back
> from the field.
>
> The RCM3720 operates in a benign environment: Sealed enclosure, 25C
> ambient, LM7805 supply regulator and no RCM3720 I/O pin is exposed
to
> the outside world. All internal I/O circuitry interfacing with the
> RCM3720 is 3.3V logic.
>
> Our failure rate is approaching 10% of units shipped. Invariably it
is
> the RCM3720; once replaced the FPGA and 74HC series logic
constituting
> the rest of the circuit runs flawlessly.
>
> Mariss Freimanis
> Geckodrive Inc.
>

I read elsewhere in this group that there have been reliability
issues with the RoHS changes. It is very possible that the no-lead
reflow solder is marginal and develops bad joints over time or after
the boards have been flexed during insertions / extractions on the
user boards. We've seen quite a few DOAs with the RCM37xx modules
(all red boards), also, but we haven't sold enough product at this
point to know if there are any field failures. Our beta units have
held up well, but most of them are mounted in the open in air
conditioned areas (no enclosures).
--- In r..., "Mariss Freimanis"
wrote:
>
> Thank you for the reply. I'm not sure it can be heat related in
this case:
>
> 1) Input power is 230mA at 12VDC or 2.76 Watts. The regulator
> dissipates 1.61 Watts leaving 1.15 Watts for the entire circuit.
>
> 2) The package is aluminum, 2.5" by 1.5" by 7" (63mm by 38mm by
> 178mm). The measured temperature rise is 2.8C on the package
surface,
> the measured internal temperature rise is 3.6C above ambient.
>
> 3) The RCM3720 thermal "exclusion zone" requirements have been
followed.
>
> 4) The RCM3720 is rated for operation at a 70C ambient. It sees a
> maximum ambient of 44C inside the package (for 40C external).
>
> 5) 2 to 3 RCM3720's out of 100 ordered are DOA right out of the box.
> The refuse to flash with either RFU.exe or Flashit.exe on initial
> installation.
>
> The Rabbits are terrific little modules at a great price and we are
> otherwise very pleased with them. I have to assume we are doing
> something wrong to have these problems. I would like to know what it
> may be.
>
> Mariss
>
> Below is a sample of failed RCM3720s
> ---------------
> 01. CE84AC Flashes OK. Get "TCP/IP error 735". Field failure.
> 02. CE84C1 RFU "Error reading bytes". Initial failure.
> 03. CE84C3 RFU "Flash parameters not found". Initial failure.
> 04. CE84B7 RFU "No Rabbit detected". Field failure.
> 05. CE84C4 Same as (02). Field failure.
> 06. CE8416 Same as (03). Field failure.
> 07. CE95FF Flashes OK. No Ethernet. Field failure.
> 08. C8C1DF RFU "Target comm error". Initial failure.
> 09. CBBFF2 Flashes OK. Port A failed. Field failure.
> 10. CBBEC5 Same as (09). Field failure.
> 11. CBBEC2 Flashes OK. DS1 LED 'on' w/o Ethernet cable. Field
failure.
>
> In all cases replacing the RCM3720 fixed the problem. There were no
> problems with the rest of the hardware.
>
>
>
> --- In r..., "neil2452"
wrote:
> >
> > I had similar problems with a few years ago with the RCM3400. I
had
> > it all in a seal box with a 20w switch mode power supply on my
own
> > PCB with about 9 PIC's on it. The rabbit was effectly a TCP/IP
to
> > serial converter. I found that once they were installed they
over
> > heated and seem to loose TCP/IP connection. They were installed
high
> > up in the room where there was a lot of heat from lights and
crowds
> > of peope below. The PIC's never crashed but once the ambient
temp
> > got to about 25c the rabbit got to hot inside the box.
> >
> > I solved the problem by squeezing in a realy small DC fan to
direct
> > air onto the rabbit. They don't seem like the heat !
> >
> >
> > --- In r..., "Mariss Freimanis"
> > wrote:
> > >
> > > Is anyone having reliability problems with the RCM3720?
> > >
> > > We have used several hundred RCM3720s in a product we build.
After a
> > > few days to a few months the RCM3720 will lose the ability to
> > connect
> > > via Ethernet, the on board Ethernet controller gets hot and the
> > > RCM3720 will not respond even to RFU.exe when we get the
product
> > back
> > > from the field.
> > >
> > > The RCM3720 operates in a benign environment: Sealed enclosure,
25C
> > > ambient, LM7805 supply regulator and no RCM3720 I/O pin is
exposed
> > to
> > > the outside world. All internal I/O circuitry interfacing with
the
> > > RCM3720 is 3.3V logic.
> > >
> > > Our failure rate is approaching 10% of units shipped.
Invariably it
> > is
> > > the RCM3720; once replaced the FPGA and 74HC series logic
> > constituting
> > > the rest of the circuit runs flawlessly.
> > >
> > > Mariss Freimanis
> > > Geckodrive Inc.
> > >
>