Hi,
I don't use LEDs for overall general or bench lighting, I have overhead Fluorescents for that. I do have a magnifier that has 144 LEDs on it that are variable in intensity. I purchased one with the Electronics at the bottom of the Boom, not near the Magnifier or Lights to keep it away from my prototyping area.
DISCLAIMER *
THE FOLLOWING IS OFFERED TO INFORM AND PROVIDE SOME STARTING SUGGESTIONS OF HOW TO LOOK AND WHERE TO LOOK FOR THE PROBLEMS ASSOCIATED WITH NOISY LED LIGHTING STRIPS (OR OTHER NOISY PRODUCTS). AGAIN, THIS IS JUST A STARTING POINT.
IT IS EXPECTED THAT KNOW HOW TO WORK WITH AC LINE POWERED ELECTRONICS AND VOLTAGES THAT EXCEED 30 VOLTS. IF NOT, DO NOT ATTEMPT TO MODIFY ANYTHING. YOU CAN STILL EMI CHARACTERIZE THE LED LIGHTING PRODUCT, THOUGH.
So, with the inexpensive LED strips there are two main issues with them Conducted EMI and Radiated EMI. You need to find out which one (or both) are the problem. It is recommended that you have an 3cm EMI Magnetic Loop Probe (purchased or home-made),? Interconnect Cable and a Spectrum Analyzer set to display frequencies from 100kHz through 1000MHz as appropriate. I am only giving information of Conducted to Radiated EMI Conversion. To provide an exact characterization, you need a Dual AC, Line Impedance Stabilization Network (LISN), and measurement in an EMI Chamber with real antennas and Receiver/Spectrum Analyzer. What is presented here is usually sufficient to EMI characterize many products.
EMI Characterization
----------------------------
You can start by using the EMI Loop Probe to probe the local environment watching the Spectrum Analyzer.
Separate the LED Lighting product from the installed location. You need at least a foot or three clearance from all other cables (plug it into a separate power strip and move it away from your bench).
0. Spectrum Analyzer Set up.
Set the Spectrum Analyzer up for a frequency range of 100kHz to 1 MHz, continuous sweep, Resolution Bandwidth of 30kHz, Reference Level 0 dB, No Attenuation (the emissions shouldn't be that high - if they are either back the probe away an inch or add attenuation).
1. Ambient EMI Emissions scan.
Turn the Light Strip Off and watching the Spectrum Analyzer Display,? just move the EMI loop Probe across (sweep), all of the surfaces in both Horizontal and Vertical orientations of the Loop Probe around the LED Light String and the Power Cord.
It should be dead with no emissions unless it uses a keep-alive circuit. Be aware that other instrument cables. If you see a lot of emissions, disconnect the Power Cord of the LED Strip, wait a few minutes for it fully discharge any bulk capacitance in the Power Supply. With it Powered off re-sweep to see a real Ambient environment.
Be aware that you may see other emissions on the Spectrum Analyzer even with everything powered-off. Those are Ambient Emissions and can be TV, Radio, other noisy equipment, room lighting - especially AC Light Dimmers, etc. emissions.
You just need to see where those emissions are, and know that they are there, so that you can mentally subtract them out when you see real Emissions from the Equipment Under Test (EUT).
Set the Spectrum Analyzer up for a frequency range of 1MHz to 10 MHz, same Resolution bandwidth and repeat an ambient sweep.
Set the Spectrum Analyzer up for a frequency range of 1MHz to 10 MHz, 100kHz Resolution Bandwidth and repeat an ambient sweep.
You now have an idea of what emissions ARE NOT coming from the LED Light Strip.
2. Product EMI Emissions.
Turn the Light String On and observe what happens on the Spectrum Analyzer as you sweep it again with it powered-on. Make note of emissions you see of what emissions are associated with physical areas of the LED Light Product. Using your phone to take pictures is a good idea, especially hot spots you see.?
You now see what the offending emissions are. Some may matter, some may not. Some of those are going to be seen as single peaks (Narrowband Emissions), that rise above the base line, some will be quite high. Some emissions will appear as groups of peaks or as large areas of the screen rise above the noise floor (Broadband Emissions - depending on the Resolution Bandwidth set on the Spectrum Analyzer). All of these emissions matter, but it is likely the Broadband Emissions are the source of the offending noise. Be aware that some emissions may occupy the same frequency space that the Ambient Emission Scan showed.
Set the Spectrum Analyzer up for a frequency range of 1MHz to 10 MHz, same Resolution bandwidth and repeat step 2.
Set the Spectrum Analyzer up for a frequency range of 10MHz to 100 MHz, 100kHz Resolution Bandwidth and repeat step 2.
Set the Spectrum Analyzer up for a frequency range of 100MHz to 1000 MHz, 1MHz Resolution Bandwidth and repeat step 2.
Same comments for Narrowband and Broadband Emission results seen on all of the Spectrum Analyzer display.
Results Evaluation.
-------------------------
You should have an idea what emissions are where now. The following is offered as suggested areas to look at.
If they are mostly below 50MHz Broadband Emissions on the Power Cable, see below discussion about the Trip-Lite EMI Filter.
If they are above 1 MHz and below 200MHz on the LED Strip Power Supply Area of the product, see below discussion about Shielding and Ferrite Cores.
If they are above 100 MHz on the LED Strip Wires or PCB that holds the LEDs, see the AC and DC LED Discussion below.
For the Conducted EMI problem...
One is that they use a Switch-mode Power Supply from AC Mains (120V/240V) to DC for the LED Controller Electronics and LED Power Source. The AC to DC Power Supply can cause multiple problems.
Those problems can be mitigated through various methods.
If it is an inexpensive Led Strip, it may or may not have adequate Power Line Filtering to keep the noise within their design and packaging. If it has an FCC B or CE-Mark it should have adequate filtering to meet the requirement - but a lot of the China stuff is marked and includes no design for compliance. Depending on their type of EMI Filter design or quality of components, and how it was tested or not, makes a big difference.
If it does have a Conducted EMI issue, that noise is impressed in Common-Mode (impressed on both Line and Neutral pins), or Differential Mode (impressed on Line or Neutral with reference to Ground).
If the noise is Common Mode, it follows through all of that power Cabling, Power Strips, etc., back to the Lowest Impedance Source. That source can be another well-designed EMI Filter (like the ones in you Spectrum Analyzers, Network Analyzers, Signal Generators and Receivers). It does not have to go back to your Power Wall Plug. And, we all have a lot of power cabling, right?
The Common Mode Conducted emissions couple into the cables and radiates off of the cables (unshielded cables can be very efficient radiators - read that as antennas - and are dependent on cable length, frequency, Impedance, coupling distance between cable, how long they are coupled, etc.). The coupled emissions become Radiated EMI which also couples into cables and wires that we are breadboarding, ungrounded, isolated or poorly grounded inputs to sensitive equipment, etc.. The Radiated Emissions E-Fields couple Capacitively to equipment cabinets and to some extent other cables, but mostly H-Field couple to other cables Magnetically into other cables and structures.
Many times, my EMI Filters designed in products have been "magnets" for other crappy designed EMI Filters in other products on the AC or DC Power Bus.
What to do?
The easiest way to diagnose if this is a problem is to use a well-designed EMI Power Filter (like a Trip-Lite ISOBar 1 or 4 or an EFI Brand Power Strip), and plug the LED Strip into it and the EMI Filter into the Power Source Outlet. The Trip-lite ISOBar device includes X and Y Capacitors as well as a Common Mode Choke in addition to Surge MOVs. A Surge Power Strip without the EMI Filtering in the ISOBar or the EFI unit will not diagnose the problem. If the emissions go down, there is a problem there. If it fixes the problem, you can't do anything with it or you will violate its Product Safety Certification - if it actually had it. But, you can use an external EMI Filter like the ISOBar and not violate Product Certification. The Trip Lite ISOBar products can be used to isolate AC EMI issues quickly.
A note about System Leakage Current:
----------------------------------------------------
With all of our equipment plugged into a common power wall outlet(s) - I usually have 27 instruments, we need to consider the leakage current induced into our "System of Systems" and that it can become too high for "Touch Current".
Each instrument has (should have), an EMI Filter composed of X and Y Capacitors to control Common-Mode and Differential-Mode Conducted EMI. All of those currents do a Vector-add since not all EMI Filters have the same values of Y Capacitors.
Many Equipment have EMI power Entry Filters that are not Powered Off at all - they stay on resident whether that instrument is creating EMI or not. Those circulating currents are on all of the chassis of our equipment that is common in the racked or shelved equipment.
Some equipment, have very aggressive EMI Filtering resulting in high value Y Capacitance in their EMI Filters, and when an Y Capacitor goes bad, how would you know it? I made a special Tektronix AC Cable Breakout TM500 Fixture with an AM503B and A6322 Current probe with an SC503. I use it periodically to check each instrument Leakage Current fore my safety. That's how I found a bad EMI Filter in my HP-4192A years ago. It tripped my GFCI Breaker on the bench Power. With all of the other instruments in parallel, it was k=just bad enough to trip it and remove the hazard it saw.
In some circumstances, all of those Vector Summed Currents may interfere with low level measurements - causing poor noise floor or erroneous measurements. Seen that happen by the way.
No man is an island.
That may or may not be a part of the problem.
The other issue is that the Power Cable itself radiates because of Conducted Emissions on it. If that is the case, wrap some Tin Foil around the whole cable length and Ground it. If that helps, the best thing to do is Z-Fold the Cable. That means that you make the total cable length about 12" with folding it back on top of itself as 12" sections. This has the EMI effect of making the cable look only 12" inches long. It does this by reversing the emission currents back on top of themselves causing the majority of those currents - and radiation strength and coupling efficiency to be reduced. Zip-tie around the ends of the cables about 2" from the ends to capture the cable loops.
You might also use a Ferrite Core on the Power Cable. A good choice for broadband power noise is a Fair-rite Corporation #44 Material or Steward Corporation #28 Material Snap-on Ferrite Core. Get one that the opening is big enough so that you can wrap the Power Cord a couple of times through the hole before you close it and lock it into place. The Ferrite Core is a low Q, crude Common Mode Choke on the cable. It looks like a low impedance to the Conducted Emissions and they are dissipated as heat in the Ferrite Material - no, they don't get hot, the currents are small.
For those of us with a lot of equipment on standby and large array of equipment, we should be Z-Folding all of our cables. Do not make them a coil of cable or let them droop, certainly, do not bundle groups of coils or groups of drooping cables.
That should address Power Line AC Conducted EMI issues in my experience. Radiated EMI issues can be problematic to solve depending on the design.
Since they are likely using a Switch-Mode Power Supply, the Outputs to the LED Array are likely unfiltered. They may even be Pulse Width Modulated (PWM) to control Color and brightness. It should be obvious that the base frequency of the PWM and we are having issues with the harmonics of the PWM edges on the wires going to the LEDs. Those wires are long of course and are likely the radiating antenna.
ALL OF THE BELOW APPLIES TO A DC ONLY OUTPUT TO THE LED STRIP ONLY.
-----------------------------------------------------------------------------------------------------
Normally, you would just add X and Y Capacitors or possibly a Common-Mode Choke to the output wires if they were straight DC. If they DC (you can control the brightness of LEDs by varying the current), great.
The DC is probably being trashed by the Switch-Mode Power Supply Output Conducted Emissions not being adequately filtered. That can happen as Common Mode or Differential Mode.
If it is Common Mode. We are trying to fix a High Frequency problem here (Switching harmonics from the Power MOSFET, Rectification Catch Diode, and Transformer Leakage Inductance), so, the values can be small and unpolarized. The X Capacitor should be 10X the value of the Y Capacitors. 1uF X Mylar Capacitor and 0.1uF for the Y Ceramic Capacitors. Those should be Probably a 100uH Common Mode Choke. Make sure the Voltage Values of the components are 3 to 4X the working Voltage of the All of this is great if it doesn't cause additional Power Dissipation in the MOSFET and Catch Diode or cause additional Conducted Emissions on the Power Line Side of the Power Supply - which they may have minimal Output Capacitance to the LED String to meet Conducted Emission Certification in the first place. You also have to make sure that the Y Capacitors do not affect the Common-Mode path back to the source - a double edge sword as a Capacitor is bi-directional and allows you to send Conducted Emissions away from the LED Strip, but also accepts Power Source Noise into the LED Strip.
If it is Differential Mode, you can use an L-C Filter on the Power Supply Output to LED Strip. Here again, the Inductor could be in the 50 to 100uH range, the Capacitor value is a little tricky here, ESR and dissipation in the Capacitor package is really important though. You need a low ESR Capacitor. You have to determine the value either by calculation or empirically. Same concerns apply for the Component Safety (including the MOSFET and Catch Diode and Product Safety). It may also create a problem with the Loop Bandwidth of the Switching Regulator.
Once again, you might also use a Ferrite Core on each wire or both wires from the Output of the Power Supply to the LED Strip. Use the same Cores except small ones that will go onto the wires - they don't have to be snap-on type, the solid ones are actually better. The longer, the better, the thicker the better (higher insertion attenuation).
ALL OF THE ABOVE APPLIES TO A DC ONLY OUTPUT TO THE LED STRIP ONLY.
FOR PWM OUTPUT DESIGNS THE FOLLOWING APPLIES.
--------------------------------------------------------------------------
For PWM Based LED Strip design, the only real thing you can do is shield it. You can't fix the output because it is AC Pulsed in nature and the chances of you causing compliance issues, damaging an output MOSFET or BJT driving the LED String is high. You will likely need to find a good way to shield the product, the internal wiring, or the PCB.
Any Output Filtering removes harmonics, for sure, but will likely increase Power Dissipation in the MOSFET Driving the LED Strip resulting in bad things up to and including ***FIRE***. The MOSFET has to stay in conduction much longer as it charges and discharges any Capacitance. It may also create a problem with the Loop Bandwidth of the Switching Regulator.
IT IS NOT RECOMMENDED THAT YOU FIX THE PWM DESIGNS FOR RELIABILITY, REGULATORY COMLIPANCE AND *** SAFETY ***. JUST SHIELD IT OR GIVE UP IN MY OPINION, DESIGN A CURRENT MODE DESIGN IF YOU WANT TO VARY THE INTENSITY.
Sorry for the long email.
Ross
-----Original Message-----
From:
[email protected] <
[email protected]> On Behalf Of John Griessen via groups.io
Sent: Thursday, September 19, 2024 6:34 PM
To:
[email protected]Subject: Re: [HP-Agilent-Keysight-equipment] Introduction
On 9/19/24 15:53, Radu Bogdan Dicher wrote:
> when you align FM tuners (uV of signal) or do metrology applications, this noise can be a determining factor.
Yeah,? I've been thinking of a zero crossing and filtered and shielded and low frequency switcher supply to make and sell for LEDs in labs.
The usual products are designed for "don't care" how much EMI, and low cost.
Paul Bicknell
Dave McGuire