4
\$\begingroup\$

I am thinking of what way would be the best to layout my system in order for it to have the best SI and pass EMC tests. My system consists of two PCBs joint by a flat flex cable (up to 7cm) and these are some requirements:

  • The bottom board must have an analogue sensor and a DC brushed motor controlled with an h-bridge (20kHz PWM).
  • The power supplies must be on the bottom board.
  • There is a uC that can be at the top and at the bottom boards and its 12 bit ADC reads the analogue sensor output which is the main purpose of the device.
  • There is GPRS module that can consume up to 2A and, in theory, could go on the top and on the bottom board.
  • The top board must have a display, and the uC uses SPI to send the display data. Alternatively, the display data can be sent over 8 lines but the number of lines in the flat cable is limited/pins available in the processor are limited.

In order to be CE marked, the product must pass the EMC tests. On the other hand, the main objective is not passing EMC tests but to read an analogue signal. They both might collide with each other. I have never had to pass an EMC test so I do not really have an educated intuition in terms of how good/bad is a design in that respect a priori.

Some more info:

  • The resulting measurement is derived from an average of several sensor reads to improve the SNR.
  • The GPRS is never going to transmit/receive at the same time as the sensor so I am not really worried about it.
  • The DC motor will be operating at the same time as the sensor. The motor is small 10k rpm at 5V, 200-500mA at a steady speed.

Solution 1: uC on the bottom board, GPRS on the bottom board.

In order to have the best signal integrity for the sensor signal, it makes sense to have the uC as close as practical to the analogue sensor.

For space reasons, the GPRS is close to the analogue sensor. The ground is a solid plane that everyone shares but the GPRS module is idle when not used (consumes less than 50mW). The GPRS and antenna connections would be on the other side of the board and it is possible to create a small slit in the power planes between them.

Then we have the display data lines, SPI in the worst case, that will radiate.

SI: good EMC: might be bad, or not. I guess I can do this to improve:

  • reduce the SPI speed or use the 8 lines interface, if I have enough available lines
  • reduce the rise time of fastest signals with resistors

Problems: Really difficult/maybe not possible to route

Solution 2: uC on the top board, GPRS on the bottom.

Having the uC on the top board means that the analogue sensor signal must go a long way and share the flat flex cable with SPI and PWM signals. Having the analogue signal wandering around the board in a longer path and in the flex cable maybe is not the best idea. It would mean more chances for it to pick up radiation from the motor, SPI, PWM or anything else.

The GPRS is at the bottom and close to its supply so looks good.

SI: Possibility of degrading the analogue signal with noise and interference. Unknown though. EMC:

  • Probably good.
  • Short tracks for fast signals like display SPI.
  • GPRS close to the supply so short high current paths/loops.
  • Longer path for PWM signals.

Really easy to route, good space on the boards.

Solution 3: GPRS on the top and uC on the bottom.

This option is unlikely to happen, for several reasons, but I have one doubt.

The GPRS demands up to 2A impulse bursts. As the power supplies are on the bottom the 2A impulse current will be travelling through some distance; up to 9cm maybe. I guess it is not the best idea to have a 90mm loop with an impulse current through it. Even if it is not as efficiently tuned as an antenna, I am sure it will increase the radiations quite a bit. Probably, due to different geometry, it would radiate better or emphasize different frequencies than the antenna but surely it will increase emissions.

In terms of emissions, there is already a radio whose purpose is, well emitting RF so… will the test be less strict or different?

I am not a fan of this option but I would like to have some comments just out of curiosity, to learn.


In my opinion, the best option is the number 1 but is very unlikely I can fit everything I want on the bottom board and probably I will need the option 2 but I am scared it could compromise the analogue measurements which are the most important function. What are your thoughts and why? Am I reasoning anything wrong? Ask if you need any more information.

\$\endgroup\$
  • \$\begingroup\$ Is there any problem with the question that no one answers? \$\endgroup\$ – jorpese Jul 28 '16 at 13:08
  • \$\begingroup\$ You're just talking about components' approximative position, we don't know the size of the boards, nor we have any layout. In fact you're talking about system "optimization" whereas EMC is a lot of hardware issues. A good layout and following good SI rules will provide great confidence in passing EMC validation. Also if you have a conductive case, EMC may not be a problem. About the analog acquisition you never mention which precision you need, if you need 0.1 mV precision or 10 mV, this is not the same. Your question and thinking is nice but we lack essential values and maybe some graphics. \$\endgroup\$ – zeqL Aug 20 '16 at 13:23
  • \$\begingroup\$ @jorpese It is amazing that no one has answered this .If it remains unanswered then the only way to make sure something gets done is to do it yourself.Look on the bright side it has not been closed . \$\endgroup\$ – Autistic Mar 4 '17 at 12:12
  • \$\begingroup\$ Thank you for your answer zeqL and sorry if I did not reply earlier. I understand there was information missing but I could not post layouts and other reserved information, sorry. On the other hand, talking about approximate positions was because the layout was not yet decided and things were changing depending on other considerations like mechanical, marketing and other and not purely electrical. The case was not conductive, otherwise I would not be worried about radiation and the whole question would not have been asked in the first place as the option 1 is then obvious. \$\endgroup\$ – jorpese Mar 5 '17 at 14:25
  • \$\begingroup\$ The reality is that the motor was not electrically affecting the analogue measurement, neither the PWM controlling it as the currents to drive the H bridge are very small. In any case, it was not the factor affecting the repeatibility of the measurements. \$\endgroup\$ – jorpese Mar 5 '17 at 14:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.