Project Logs

Collapse

• Comparison with another DIY LISN

  kevarek • 08/19/2021 at 19:59 • 0 comments

  My colleague lent me his DIY version of 5uH CISPR-25 LISN to measure and compare its performance – unit in smaller housing with many inductors in series to form total 5uH inductance.

   

  Output impedance measurement reveals CISPR-25 incompliance below 200kHz as well as earlier occurring incompliance above 30MHz. Due to its 1uF input capacitance it would be also impacted by power supply harness inductance issue as described in simulation project log.

  Insertion losses measurement reveals true weakness of this design. Its response is rather unpredictable. Instead of flat response above 1MHz, there are many depression peaks. I don’t think the measurement could be calibrated reliably for this behavior. Complex PCB with many inductors, that are coupled, is definitely not a good idea if you consider this.

  Channel isolation measurement on the other hand reveals how great idea is to have two separate units instead of one multi-channel. Low frequency cross-talk does not occur at all. Between 20 – 40MHz there is still a cross-talk above noise floor, which was not present on my unit. This could be caused by the same design issue as described in insertion losses measurement above.

  Overall, without tuning insertion losses to be flatter above 1MHz I would consider this construction as not reliable.

• Validation measurement

  kevarek • 08/19/2021 at 19:31 • 0 comments

  Validation measurements were performed using Vector network analyzer (VNA) Rohde&Schwarz ZNB 8, 100kHz to 8.5GHz. This is my first VNA measurement so please be tolerant. Should I evaluate measurement validity and error sources my guess is that any measurement above 100MHz is invalid due to banana plugs being used. Actually I think that propper LISN manufacturers cheat their validations a bit by having SMA connectors directly in the units and connecting measurement equipment there. For test setups see simulation project log.

  Output impedance measurement confirmed that LISN is compliant to CISPR-25 requirements. I’m fairly confident that impedance drop between 90-10MHz is caused by banana connectors. Matches the simulation nicely.

  Measurement setup

  Insertion losses measurement again looks very nice and flat. Measurement below 10MHz does not match the simulation and I have no clue why. I believe there is some issue with units VNA vs LTSpice. This is my homework to explain. Measurement above 90MHz again affected by banana plugs.

  Inter channel isolation measurement explains why it is a bad idea to place two channels into same box and why all respected manufacturers don’t do it. There is quite a high amount of crosstalk below 10MHz. This is caused by magnetic coupling of inductors through thin internal wall separating the channels. CISPR-16 requires less than -40dB though, so I consider this unit as fine.

   

     

  Long story short, after performance validation I consider this home-made CISPR-25 5uH LISN as success and based on measurement results, I release it for hobby use pre-compliance measurements 😊

• Assembly

  kevarek • 08/19/2021 at 18:17 • 0 comments

  Whole LISN was assembled into DIY metal box from solderable metal sheets 1mm thick, 70mm high and 1000mm long (rather permeable and very soft iron). I would recommend to make the box at least 50% bigger to allow greater inductor separation from walls. Reason for this will be clear after validation measurements. Should I build it again, I would create separate housing for each LISN as all the manufacturers do. There is a great reason for it!

  Sorry for very bad dimension drawings… this is what is available.

  Housing body

  Housing cover

  All the parts were cut from sheet using bench scissors and soldered together. This took quite a lot of patience from my wife as quite a lot of solder flux smoke was released across our flat 😊

  Inductors were clamped into their spacers and mounted into the housing.

  Small PCBs for input and output capacitors and resistors were milled out of standard single layer PCB.

  Output side PCB with 100nF capacitor bank and 1K resistor.

  Everything assembled

  Detail of output side PCB. Every centimeter of wire adds significant amount of undesired behavior above ~50MHz.

  Final product ready to be tested

• Air core 5uH inductors

  kevarek • 08/19/2021 at 17:32 • 0 comments

  Inductor L1 has been designed with following online calculator: https://hamwaves.com/inductance/en/index.html#input

  Aim was to achieve up to use copper wire thick enough to allow 30A RMS. Rule of thumb for the design is to keep its diameter D somewhat comparable to its length. Below is the calculator output that was used to model its plastic core for support.

  Just for fun I have plotted the windings effective series resistance considering also skin effect.

   

  Plastic core support to allow easy winding and good tolerances for both wound inductors.

  Plastic spacer and holder that keeps the coils centered within the housing.

  Assembled inductors were marked as 1DOT and 2DOTS and measured using LCR meter Kust LM1050A. Both seem to have almost identical parameters and good enough to be used for CISPR-25 5uH LISN.

• Simulations

  kevarek • 08/19/2021 at 17:11 • 0 comments

  Typically, four basic simulations are performed. This is helpful to understand influence of parasitic properties of electronic components and also helps to identify schematic nodes sensitive to parasitic capacitance to housing etc.

  Output impedance simulation with power input floating.

  
  

  Output impedance simulation with power input shorted to ground.

  These simulations help to verify that LISN complies to CISPR-25 limits to impedance magnitude and phase angle. Both simulation results are in chart below.
  
  

  Output impedance simulation with excessive harness inductance from power supply to LISN

  There is one more output impedance simulation to be done which helps to understand 1uF input capacitor issue and why 10uF or higher capacitance in series with damping resistor is used. In case there is substantial wiring inductance between power supply and LISN input ports it could resonate with input capacitor. It is important that this resonance won’t affect output impedance being within CISPR-25 limits. Placing 10uF helps to push resonance much below 100kHz. If measurement down to 10kHz is required it is possible to increase the capacitance even more.

  Insertion losses simulation

  Insertion losses represent how much of signal generated by DUT (device under test) on LISN power output port is transmitted to signal output BNC connector. Measurement taken by spectral analyzer on this port needs to be compensated by this result before comparison with emission limits.

  Input to output isolation simulation

  This represents how much noise from power supply can pass through LISN power input into power output and affect the measurement. Dominant resonance at 150MHz is caused by parasitic inter-winding capacitance and needs to be kept above measurement range.

  It is worth playing with simulation parameters:

  • inter-winding capacitance C_par of inductor L1 which causes decrease of output impedance in higher frequencies and inductor self-resonance which deteriorates input to output isolation.
  • parasitic capacitance C_parbnc of signal output to GND which causes steep decrease of output impedance above 100MHz and also shifts a bit inductor L1 resonance frequency.

• Schematic

  kevarek • 08/19/2021 at 15:18 • 0 comments

  Schematic is deceptively simple. Basically CISPR25 circuit with exception of input capacitor capacitance which is 10uF ceramic capacitor in series with resistance 0.5 to 1 Ohm instead of just 1uF capacitor. This is helps to prevent "output impedance out of limits" issue when wiring inductance between power supply and LISN is above 0.1uH. This issue and its solution has been proposed somewhere in the mentioned EEVblog forum threads.

  All capacitors consist from multiple units of same capacitance and package to enhance their high frequency properties (less parasitic equivalent series inductance, prevention of resonances and anti-resonances of the capacitors).

  R1 actually consists from four 0805 resistors in parallel to achieve above 1W power dissipation rating. Its final exact resistance does not matter while in range from 0.5 to 1 Ohm.

  All components (except L1) should be surface mount type to lower their parasitic properties impact to high frequency response.
  

  I decided not to implement standard 10dB attenuator to keep it simple, but plan is to build is as external unit later. This also helps to keep the validation and root cause analysis of possible issues simpler.
  

• Intro

  kevarek • 08/18/2021 at 18:41 • 0 comments

  Building 5uH CISPR25 LISN seems like very simple task but evil is in the detail. Without careful evaluation of circuit and housing parasitic properties, it wont pass standard's requirement for output impedance. Furthermore, its unstable insertion losses might render it completely useless.
  

  Hobbyists tent to stuff two channels into single box (like in CISPR-15 50uH LISN used for home appliances), which brings additional issue - channel separation aka cross-talk.

  It is not really possible to evaluate the build quality without access to very expensive instruments (Spectrum analyzer with tracking generator or vector network analyzer).

  To get true feeling how hard the design and build might be please read first:
  https://www.eevblog.com/forum/projects/5uh-lisn-for-spectrum-analyzer-emcemi-work/
  https://www.eevblog.com/forum/projects/5uh-aerospace-lisn-how-dumb-would-i-be-to-throw-one-together/
  https://www.eevblog.com/forum/rf-microwave/50uh-and-250uh-inductor-design-for-lisn/

  Also watch amazing video series from FesZ:
  

  Part 1: https://youtu.be/OEvkRW5vZNA
  Part 2: https://youtu.be/nq-IhHbZdD0
  Part 3: https://youtu.be/M_ab0Oz2ypI

  Aim of this project was to have final working product for daily usage and therefore full testing and its validation across frequencies up to 100MHz is as important as the assembly itself.
  

  
  Everything is finished, I will soon upload build log with simulations, assembly pictures and validation process.
  

View all 7 project logs