Folks:
I'm in the process of building an LSK preamp in a HiFi2000 1U Slimline chassis using preamp boards purchased through a group buy and attenuator, source selection and controller boards (and an OLED) purchased from Dantimax. The layout is intended to keep the audio signal boards reasonably far from the microcontroller and the OLED (about 6" to 8" / 15cm to 20cm).
My question: would it be worthwhile to install an RF shield between the signal area and the rest of the preamp? I have 1/8" / 3mm thick aluminum and some thin copper sheeting available, and could sandwich them together to create a barrier if it would help. If not, is there another commonly-available material that I should use? I'm not looking for the ultimate in isolation -- just something to minimize any noise caused by the microcontroller and OLED.
Your advice is appreciated.
Regards,
Scott
I'm in the process of building an LSK preamp in a HiFi2000 1U Slimline chassis using preamp boards purchased through a group buy and attenuator, source selection and controller boards (and an OLED) purchased from Dantimax. The layout is intended to keep the audio signal boards reasonably far from the microcontroller and the OLED (about 6" to 8" / 15cm to 20cm).
My question: would it be worthwhile to install an RF shield between the signal area and the rest of the preamp? I have 1/8" / 3mm thick aluminum and some thin copper sheeting available, and could sandwich them together to create a barrier if it would help. If not, is there another commonly-available material that I should use? I'm not looking for the ultimate in isolation -- just something to minimize any noise caused by the microcontroller and OLED.
Your advice is appreciated.
Regards,
Scott
Attachments
Folks:
Incidentally, "RF" may be the wrong term; what I meant to say was any electronic noise that could be blocked by a grounded barrier.
But you already knew that.
Regards,
Scott
Incidentally, "RF" may be the wrong term; what I meant to say was any electronic noise that could be blocked by a grounded barrier.
But you already knew that.
Regards,
Scott
EMC Shielding is quite complex and its effectiveness is determined by seams...
Done incorrectly it can be worse than no shielding.
What is Electromagnetic Shielding | Why You need EMI Shielding
The Basic Principles of Shielding | In Compliance Magazine
http://ieee.rackoneup.net/rrvs/09/Engineering Aspects of Electromagnetic Shielding.pdf
http://www.analog.com/media/en/training-seminars/tutorials/MT-095.pdf
Done incorrectly it can be worse than no shielding.
What is Electromagnetic Shielding | Why You need EMI Shielding
The Basic Principles of Shielding | In Compliance Magazine
http://ieee.rackoneup.net/rrvs/09/Engineering Aspects of Electromagnetic Shielding.pdf
http://www.analog.com/media/en/training-seminars/tutorials/MT-095.pdf
Strangely, when someone says "RF" I assume he means RF. My friends tell me I take things too literally.SRMcGee said:
To do good shielding you need to know what you are trying to achieve. Stopping 100kHz, 10MHz or 1GHz need different techniques.
I work up to 18GHz immunity on some PCB/Assembly designs...that's fun...
The term airborne EMI may be preferable, RF should be used for real RF i.e. actual RF that you want to transmit, not airborne EMI that you didn't mean to transmit....
My starter references would be:
Henry Ott
Ralph Morrison
The term airborne EMI may be preferable, RF should be used for real RF i.e. actual RF that you want to transmit, not airborne EMI that you didn't mean to transmit....
My starter references would be:
Henry Ott
Ralph Morrison
Well, look at all these wires everywhere.
Two wire loops more or less close to each other make an air-core transformer.
Principle of transformer is that primary winding emits magnetic flux and secondary receives it. You want your noise transformers to be as bad as possible, and that is why you should :
- minimize loop area of transmitting loops
- minimize loop area of receiving loops
- maximize distance
You can do that by using coax, twisted pair, ribbon cable with half of the wires being GND, etc.
Note that "twisted pair" contains the word "pair". If a signal is twisted with anything than its official, private GND return, then it is not twisted pair, it is decoration. Same with coax shield connected at one end only, it shields only against capacitive coupling at low freqs, not anything else.
Also, remember, your microcontroller has a sleep mode. A micro with its clock stopped doesn't emit any noise. Same for an OLED display with its VCC turned off. Your micro board does not have a GND plane so it can't be made clean. Just put it to sleep instead.
The worst is :
- Multiplexed switches & keypads (lots of loop area) : solution is, use clever multiplexing that does not require constant polling of the buttons, rather use a pin change interrupt that is only triggered when a button is actually pressed, then maybe scans the buttons ONCE.
- Matrix displays : the more current draw, the worst. Text LCDs are not so bad, OLED suck, VFDs are the worst by far. Solution : activate the display when something needs to be displayed, then turn it off !
- PWM LED backlights (extra points if driven from a DC-DC)
Two wire loops more or less close to each other make an air-core transformer.
Principle of transformer is that primary winding emits magnetic flux and secondary receives it. You want your noise transformers to be as bad as possible, and that is why you should :
- minimize loop area of transmitting loops
- minimize loop area of receiving loops
- maximize distance
You can do that by using coax, twisted pair, ribbon cable with half of the wires being GND, etc.
Note that "twisted pair" contains the word "pair". If a signal is twisted with anything than its official, private GND return, then it is not twisted pair, it is decoration. Same with coax shield connected at one end only, it shields only against capacitive coupling at low freqs, not anything else.
Also, remember, your microcontroller has a sleep mode. A micro with its clock stopped doesn't emit any noise. Same for an OLED display with its VCC turned off. Your micro board does not have a GND plane so it can't be made clean. Just put it to sleep instead.
The worst is :
- Multiplexed switches & keypads (lots of loop area) : solution is, use clever multiplexing that does not require constant polling of the buttons, rather use a pin change interrupt that is only triggered when a button is actually pressed, then maybe scans the buttons ONCE.
- Matrix displays : the more current draw, the worst. Text LCDs are not so bad, OLED suck, VFDs are the worst by far. Solution : activate the display when something needs to be displayed, then turn it off !
- PWM LED backlights (extra points if driven from a DC-DC)
Last edited:
Gentlemen:
Let's level set expectations all around: yes, there are lots of wires flopping all over the place in the photo I posted of my preamp build in progress. It's a build in progress. The finished preamp will have nicely dressed wires.
RF, EMI, whatever. You guys have expertise that I will never, ever have. I don't know the frequency at which OLEDs and microprocessors operate, but I assume that blocking everything above audio frequencies would be ideal (and yes, I imagine that "audio frequencies" would be above 20kHz).
The noise created by the digital portion of this preamp may be inaudible; I won't know until the project is completed. If there is a relatively simple fence that can be constructed around the audio portion of the preamp to block some or all of the digital, RF, EMI or other bad noise, I'd like to build that fence now when access and flexibility are best. But if need be, I can wait to see if there's anything to fix.
I'm just looking for some simple guidance regarding the viability of the fence idea and the materials that might help minimize the noise. If you think it's a bad idea, say so. If the materials include unobtainium or the project is too involved for a novice, say so. My feelings won't be hurt.
Regards,
Scott
Let's level set expectations all around: yes, there are lots of wires flopping all over the place in the photo I posted of my preamp build in progress. It's a build in progress. The finished preamp will have nicely dressed wires.
RF, EMI, whatever. You guys have expertise that I will never, ever have. I don't know the frequency at which OLEDs and microprocessors operate, but I assume that blocking everything above audio frequencies would be ideal (and yes, I imagine that "audio frequencies" would be above 20kHz).
The noise created by the digital portion of this preamp may be inaudible; I won't know until the project is completed. If there is a relatively simple fence that can be constructed around the audio portion of the preamp to block some or all of the digital, RF, EMI or other bad noise, I'd like to build that fence now when access and flexibility are best. But if need be, I can wait to see if there's anything to fix.
I'm just looking for some simple guidance regarding the viability of the fence idea and the materials that might help minimize the noise. If you think it's a bad idea, say so. If the materials include unobtainium or the project is too involved for a novice, say so. My feelings won't be hurt.
Regards,
Scott
Well, a shield is difficult to do. For example it needs to be conductive on the entire perimeter (cause the problem is the holes : having holes, or slots, is like having no shield).
This means you gonna need to scrape off the anodization/paint on your black metal and somehow make a continuous contact (this includes the top cover). And put filters on the wires. It's a rather large amount of work.
That's why I was saying : it is much simpler to prevent your microcontroller and display from making any noise. This way, no problem. Simply put it in deep sleep mode when it has nothing to do, and power down the display.
You just need to make sure it wakes up when you need it (ie, pin change interrupt). Plus, it's free, just a bit of code.
I have an old amp with a VFD display... you can hear the display scan in the speakers. That is a problem because the display can't be turned off. If it could be turned off, then who cares.
It's much easier to kill the noise at the source than try to fix the mess afterwards with shielding etc.
I've got this board I designed which has a few DC-DC converters on it and in the middle there is a current shunt with something like 30mV on it. There are DC-DC 1 cm away from the current shunt amplifier in all directions 😀. Noise is really low on the ADC because I synchronized the sampling with the DC-DCs, so when it is doing analog stuff, nothing is switching. Just a bit of code, no extra parts, and it works well.
This means you gonna need to scrape off the anodization/paint on your black metal and somehow make a continuous contact (this includes the top cover). And put filters on the wires. It's a rather large amount of work.
That's why I was saying : it is much simpler to prevent your microcontroller and display from making any noise. This way, no problem. Simply put it in deep sleep mode when it has nothing to do, and power down the display.
You just need to make sure it wakes up when you need it (ie, pin change interrupt). Plus, it's free, just a bit of code.
I have an old amp with a VFD display... you can hear the display scan in the speakers. That is a problem because the display can't be turned off. If it could be turned off, then who cares.
It's much easier to kill the noise at the source than try to fix the mess afterwards with shielding etc.
I've got this board I designed which has a few DC-DC converters on it and in the middle there is a current shunt with something like 30mV on it. There are DC-DC 1 cm away from the current shunt amplifier in all directions 😀. Noise is really low on the ADC because I synchronized the sampling with the DC-DCs, so when it is doing analog stuff, nothing is switching. Just a bit of code, no extra parts, and it works well.
Last edited:
#3 some links on shielding, the shield needs to be a full faraday cage with any wires going in and out terminated at the shield for the screen, so coax cable best or ferrites on twisted pair....
its a long and complex subject, the problem being if there is a noise problem it will creep in sneakily if there isn't a noise problem adding an sub optimum shield can create a problem....
- Status
- Not open for further replies.