Hello,
when shielding sensitive eletronics is the thickness of the metal shield important?
As I'm gonna pot my small circuit into epoxy, I'm tempted to use cheap aluminium
foil to shield the circuit (which seems necessary since I increased the clock
speed (PLLx16), and the environment is disturbed, at 1/16s speed it never crashed
although all the EMI).
I believe that the thickness does matter to some degree. Also, it
greatly depends on what you are trying to do - prevent radiation FROM
your device, protect your device from external EM radiation, protect
your device from external low-frequency fields, etc. Can you describe
your application further?
Shielding is a tough subject - it is difficult to get it right the
first time. I would suggest that you get your circuit working
properly, including the shielding aspects, before you pot it.
>
> Hello,
> when shielding sensitive eletronics is the thickness of the metal shield important?
>
> As I'm gonna pot my small circuit into epoxy, I'm tempted to use cheap aluminium
> foil to shield the circuit (which seems necessary since I increased the clock
> speed (PLLx16), and the environment is disturbed, at 1/16s speed it never crashed
> although all the EMI).
>
> Greets,
> Mario
>
On Sun, Nov 6, 2011 at 8:02 AM, Electron <.....electron2k4KILLspam@spam@infinito.it> wrote:
>
> Hello,
> when shielding sensitive eletronics is the thickness of the metal shield important?
Thickness helps with low frequency magnetics. A good resource for this
would be any of the vendors of Mu-Metal.
They have application notes that discuss where Mu-Metal works, and
where copper/aluminum works, and how thickness impacts things.
I would look for other sources for your problem though.. From what
little you've said, it sounds more like a PLL stability problem
That all depends on the structure you want to protect as well as the
dispersion parameters of the foil you use. When you cover it with
aluminium foil you simply cover it with a Faraday cage where grid
mesh size goes to limit 0. Thickness will be a problem if you are
running high frequency components in your circuit due to skin effect
(not penetration depth). If frequency of the core is 16 times the
peripherials, it would be an unexpected behavior if one/some of the
peripherials malfunction where core operates flawlessly. Because high
frequency components will be more prone to environmental effects. It
ends up evaluating the operation environment of your gadget. I have
encountered cases where the device works %100 at a location and goes
crazy at another location due to magnetic and electric field
components of those specific locations - even it was shielded. It is
NOT about the amplitude of the EM around -directly-, it is about the
vectoral components of the disturbing field/wave.
Model the structure and expose to random(vector) far fields then you
can extract em response of your system. It is also even possible to
determine the electrical level effects of the environmental em
radiation in your circuit via more advanced approaches.
>
> Hello,
> when shielding sensitive eletronics is the thickness of the metal shield important?
>
> As I'm gonna pot my small circuit into epoxy, I'm tempted to use cheap aluminium
> foil to shield the circuit (which seems necessary since I increased the clock
> speed (PLLx16), and the environment is disturbed, at 1/16s speed it never crashed
> although all the EMI).
>
> Greets,
> Mario
>
On 11/6/2011 12:15 PM, David VanHorn wrote:
> On Sun, Nov 6, 2011 at 8:02 AM, Electron<.....electron2k4KILLspam.....infinito.it> wrote:
>> Hello,
>> when shielding sensitive eletronics is the thickness of the metal shield important?
>
> Thickness helps with low frequency magnetics. A good resource for this
> would be any of the vendors of Mu-Metal.
> They have application notes that discuss where Mu-Metal works, and
> where copper/aluminum works, and how thickness impacts things.
>
>
> I would look for other sources for your problem though.. From what
> little you've said, it sounds more like a PLL stability problem.
Dear Sean,
I need to protect my application from the external environment, namely
from the interference caused by a spark plug. It's a motorbike application.
The spark at the oscilloscope looks like an impulse with 1uS rising edge
and 20-25uS falling edge.
>Hi Mario,
>
>I believe that the thickness does matter to some degree. Also, it
>greatly depends on what you are trying to do - prevent radiation FROM
>your device, protect your device from external EM radiation, protect
>your device from external low-frequency fields, etc. Can you describe
>your application further?
>
>Shielding is a tough subject - it is difficult to get it right the
>first time. I would suggest that you get your circuit working
>properly, including the shielding aspects, before you pot it.
>
>Sean
>
>
>On Sun, Nov 6, 2011 at 10:02 AM, Electron <EraseMEelectron2k4spam_OUTTakeThisOuTinfinito.it> wrote:
>>
>> Hello,
>> when shielding sensitive eletronics is the thickness of the metal
>shield important?
>>
>> As I'm gonna pot my small circuit into epoxy, I'm tempted to use
>cheap aluminium
>> foil to shield the circuit (which seems necessary since I increased the clock
>> speed (PLLx16), and the environment is disturbed, at 1/16s speed it
>never crashed
>> although all the EMI).
>>
>> Greets,
>> Mario
>>
At 18.15 2011.11.06, you wrote:
>On Sun, Nov 6, 2011 at 8:02 AM, Electron <electron2k4spam_OUTinfinito.it> wrote:
>>
>> Hello,
>> when shielding sensitive eletronics is the thickness of the metal
>shield important?
>
>
>Thickness helps with low frequency magnetics. A good resource for this
>would be any of the vendors of Mu-Metal.
>They have application notes that discuss where Mu-Metal works, and
>where copper/aluminum works, and how thickness impacts things.
>
>
>I would look for other sources for your problem though.. From what
>little you've said, it sounds more like a PLL stability problem.
Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
crystal in XT mode and and PLLx16. The part can reach 30 MHz.
> Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
Without knowing what part you are using, your crystal frequency X 16 >>>> 30
MHz. Unless the crystal is divided before PLL, that would be a problem.
Sergey Dyrga http://beaglerobotics.com
At 03.52 2011.11.07, you wrote:
>Electron <electron2k4 <at> infinito.it> writes:
>
>> Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
>> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
>
>Without knowing what part you are using, your crystal frequency X 16 >>>> 30
>MHz. Unless the crystal is divided before PLL, that would be a problem.
This is a dsPIC30F2011, very basic PLL (x4 or x8 or x16)
Em 7/11/2011 15:18, Electron escreveu:
> At 03.52 2011.11.07, you wrote:
>> Electron <electron2k4 <at> infinito.it> writes:
>>
>>> Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
>>> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
>> Without knowing what part you are using, your crystal frequency X 16 >>>> 30
>> MHz. Unless the crystal is divided before PLL, that would be a problem.
> This is a dsPIC30F2011, very basic PLL (x4 or x8 or x16)
Beware, many PLLs accept the input frequency only in a very narrow band
(say, 4MHz to 8MHz).
If your crystal frequency divided by the input divider is outside this
range, the PLL may not work.
> Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
PLL circuits can be very interesting to get working properly, under
all conditions.
If you are taking a 7.3728 Mhz clock and multiplying by 16, then you
are trying to run the part at 118 MHz, which seems like it's going to
be a problem. :)
What I would do:
First, get the basic system up on direct crystal oscillator, and
provide an output pin that indicates clock speed by using a timer or
pin toggle, or whatever you can easily diagnose with the tools at
hand.
Then read very carefully the PLL section of the data sheet, including
the step-by-step sequence to transition from the crystal to the PLL.
Implement per data sheet with NO shortcuts. You may need to implement
an analog PLL filter depending on the particular part you're using.
Verify that the system is now running correctly at the proper speed.
Given that you're working in an automotive environment, are you using
regulation and bypassing designed for this?
Don't look for complicated problems when you have simple ones. :
At 20.03 2011.11.07, you wrote:
>Em 7/11/2011 15:18, Electron escreveu:
>> At 03.52 2011.11.07, you wrote:
>>> Electron <electron2k4 <at> infinito.it> writes:
>>>
>>>> Thank you. How can I investigate if it's a PLL issue? I'm using a
>7.3728 MHz
>>>> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
>>> Without knowing what part you are using, your crystal frequency X 16 >>>> 30
>>> MHz. Unless the crystal is divided before PLL, that would be a problem..
>> This is a dsPIC30F2011, very basic PLL (x4 or x8 or x16)
>
>
>Beware, many PLLs accept the input frequency only in a very narrow band
>(say, 4MHz to 8MHz).
>If your crystal frequency divided by the input divider is outside this
>range, the PLL may not work.
Thank you, but this is not the case, I had checked the errata too to make sure
(indeed there're even bugs on the PLL, but only for 4x and 8x modes, while I'm
using the 16x mode).
A futher examination of the errata made me get doubts about this though, do you
think it could be the cause?
---
15. Module: PLL Lock Status Bit
The PLL LOCK Status bit (OSCCON<5>) can occasionally get cleared and generate an
oscillator failure trap even when the PLL is still locked and functioning correctly.
Work around: The user application must include an oscillator failure trap service
routine. In the trap service routine, first inspect the status of the Clock Failure
Status bit (OSCCON<3>). If this bit is clear, return from the trap service routine
immediately and continue program execution.
---
Unfortunately my app is strictly real-time and I cannot enable any interrupt. But
is this possibly causing the malfunction at near full speed?
At 21.56 2011.11.07, you wrote:
>> Thank you. How can I investigate if it's a PLL issue? I'm using a 7.3728 MHz
>> crystal in XT mode and and PLLx16. The part can reach 30 MHz.
>
>
>PLL circuits can be very interesting to get working properly, under
>all conditions.
>If you are taking a 7.3728 Mhz clock and multiplying by 16, then you
>are trying to run the part at 118 MHz, which seems like it's going to
>be a problem. :)
It's within specs, the dsPIC30F2011-30 can run up to 120 MHz, the system
clock (Fcy) runs at only 1/4 of that frequency (i.e. 29.4912 MHz). :-)
>What I would do:
>
>First, get the basic system up on direct crystal oscillator, and
>provide an output pin that indicates clock speed by using a timer or
>pin toggle, or whatever you can easily diagnose with the tools at
>hand.
>
>Then read very carefully the PLL section of the data sheet, including
>the step-by-step sequence to transition from the crystal to the PLL.
>Implement per data sheet with NO shortcuts. You may need to implement
>an analog PLL filter depending on the particular part you're using.
>
>Verify that the system is now running correctly at the proper speed.
>
>
>Given that you're working in an automotive environment, are you using
>regulation and bypassing designed for this?
Well, yes, but currently it's on my bench, not on the vehicle. However,
when I generate sparks (I got the coil and all on my bench, it's a CDI
that I'm designing) then every n (random) seconds the MPU crashes. If
instead of the spark I use a resistor (~same discharge current of the
coil) then no problems arise, hence I thought it was an EMI problem.
>Don't look for complicated problems when you have simple ones. :)
I don't wish to complicate my life, don't worry. :-) Just to make things
reliable, as I'll race on the woods with my CDI, and a failure is gonna
cause me some problems. I may not become the enduro world champion for
example. :P
> Without knowing what part you are using, your crystal frequency X 16 >>>> 30 MHz.
> Unless the crystal is divided before PLL, that would be a problem.
The datasheet does say that you can use XT crystal mode with x16 PLL, and crystals over the range of 4 - 7.5MHz, provided the maximum operating frequency of 120MHz is met (ref table 17-1 of the Rev C data sheet). So he is within the operating parameters of the PLL.
-- Scanned by iCritical.
> Well, yes, but currently it's on my bench, not on the vehicle. However,
> when I generate sparks (I got the coil and all on my bench, it's a CDI
> that I'm designing) then every n (random) seconds the MPU crashes. If
> instead of the spark I use a resistor (~same discharge current of the
> coil) then no problems arise, hence I thought it was an EMI problem.
So now you're down to impedance, decoupling and layout issues.
Have you checked your design to see that all circuits are built with
the lowest practical impedance?
Decoupling can be critical. 0.1uF is not always the best value.
Build a noise probe, using a relay wired as a buzzer and a series LC
across the contacts, with the L being a small (pencil-eraser sized)
coil. Use this to probe for weak points and evaluate. Change design
as needed
> Build a noise probe, using a relay wired as a buzzer and a series LC
> across the contacts, with the L being a small (pencil-eraser sized)
> coil. Use this to probe for weak points and evaluate. Change design
> as needed.
Any schematics for a noise probe out there?
It sounds interesting, but I have never heard of one before, I am
used to working with off the shelf hardware, but we are at the stage
where we need our own boards. The current design has been farmed out
to a hardware house, but it would be great to build up my own test
bench so I can QC their QC.
At 00.48 2011.11.09, you wrote:
>> Build a noise probe, using a relay wired as a buzzer and a series LC
>> across the contacts, with the L being a small (pencil-eraser sized)
>> coil. Use this to probe for weak points and evaluate. Change design
>> as needed.
Thanks, but I'm not sure I understand the implementation, nor if this
noise probe should detect EMI or should be connected to various points
in the circuit.
The noise probe is a small localized source of magnetic noise.
When the contacts open on the buzzer-relay, the coil inductance has
some fair energy stored that MUST go somewhere. The series RC network
that is the probe itself provides a convenient place. If you build
it, you can take a thruhole LED, short its leads and bend them into a
hoop, and light the LED from energy radiated up to about an inch away.
At 15.30 2011.11.09, you wrote:
>The noise probe is a small localized source of magnetic noise.
>
>When the contacts open on the buzzer-relay, the coil inductance has
>some fair energy stored that MUST go somewhere. The series RC network
>that is the probe itself provides a convenient place. If you build
>it, you can take a thruhole LED, short its leads and bend them into a
>hoop, and light the LED from energy radiated up to about an inch away.
>
>I'll try to post a schematic today.
Interesting, thanks.
So the purpose of the noise probe (on my circuit at least) would be to
deliberately cause interference, i.e. to test the reliability of my circuit?
> So the purpose of the noise probe (on my circuit at least) would be to
> deliberately cause interference, i.e. to test the reliability of my circuit?
Yes. The small probe coil allows you to do this in a targeted manner.
Alternatively, you can spend some serious bucks illuminating your PCB
in a chamber at various frequencies.
I have a PDF that I can email you directly, but I don't have a way to
post it anywhere convenient
At 21.38 2011.11.09, you wrote:
>> So the purpose of the noise probe (on my circuit at least) would be to
>> deliberately cause interference, i.e. to test the reliability of my circuit?
>
>
>Yes. The small probe coil allows you to do this in a targeted manner.
>Alternatively, you can spend some serious bucks illuminating your PCB
>in a chamber at various frequencies.
>
>I have a PDF that I can email you directly, but I don't have a way to
>post it anywhere convenient.
Megaupload maybe? I don't know any other free hosting sites.
Meanwhile, thank You for the eMail.. please feel free to send it! :-)
infinito.it> wrote:
