The Tekbox LISN Mate includes circuitry that will split differential mode and common mode signals.
Evaluating power supplies for conducted emissions requires a line impedance stabilization network, or LISN. Tekbox Digital Solutions (Reference 1) had introduced an AC line LISN and DC LISN a couple years ago and I reviewed these previously (Reference 2). The company also released a new device called LISN Mate (Figure 1), which is based on a design by Mark Nave, an EMC consultant based in Florida, and described in his book on designing switched-mode power supplies (Reference 3). LISN Mate includes circuitry that will split differential mode (DM) signals from common mode (CM), which is valuable for evaluating filter circuits. It is specified from 30 kHz to 110 MHz, but is characterized by Tekbox up to 150 MHz.
Figure 1 The LISN Mate by Tekbox splits differential mode and common mode signals, allowing better spectral analysis and appropriate filter design.
In order to demonstrate how the LISN Mate works, I’ll test a DC-DC converter demo board from Texas Instruments, which uses the TPS54525 converter IC. I’ll use a couple filter types to show the effect on both CM and DM; a ferrite choke and a conventional DM/CM discrete filter. All the testing will be performed at 12.6 VDC and using a pair of the Tekbox TBOH01 DC LISNs and the LISN Mate.
Note that LISN Mate also works with line-operated LISNs, but you’ll either need two of them or one with dual (line and neutral) outputs.
The test setup is as shown in Figures 2 and 3. A Siglent SPD3303C power supply, Rigol DL3021 electronic load, and Siglent SSA 3032X spectrum analyzer were used for the testing.
Figure 2 The LISN Mate is used with a pair of LISNs, in this case Tekbox TBOH01 DC models. Courtesy of Tekbox.
Figure 3 The test setup shows the power supply, electronic load, and spectrum analyzer. The LISNs and LISN Mate, along with the EUT are placed on the conducting plane.
The whole test setup should be placed on a conducting plane and I just used aluminum foil taped down to my test bench. Both LISNs should be bonded at the minus input terminal to the plane. I used copper tape for this (Figure 4). I also taped the minus terminals on the LISN outputs, but this proved unnecessary.
Figure 4 A closeup showing the connections between the two LISNs and LISN Mate.
The LISN Mate has two inputs (LISN1 and LISN2) and two outputs; one for DM and one for CM. These outputs go to the spectrum analyzer. The unused output should be terminated in 50 ohms (supplied with LISN Mate).
I used a couple means to filter the DC-DC converter input; a low-frequency ferrite choke (Fair-Rite #75 material) and a conventional line-style topology filter made up from the Würth Elektronik (WE) 744 998 filter design kit, which was reviewed earlier by my colleague, Arturo Mediano (Reference 4). I used the WE “Example 1” in their guide book, selecting component values specifically to reduce conducted emissions in the 10 kHz to 4 MHz range, which is usually problematic for many designs. The X-capacitor was 0.15 μF, the Y-capacitors were 2200 pF each, and the CM choke was 10 mH (Figure 5).
Figure 5 This is the filter board from the Würth Elektronik filter kit with components added to suppress conducted emissions in the range 10 kHz to 30 MHz.
The published insertion loss curves from Fair-Rite for the ferrite choke is reproduced in Figure 6. I chose this because it has some effectivity in the 10 kHz to 4 MHz region.
Figure 6 Here is the impedance plot for the Fair-Rite #0475167281 (#75 material). Courtesy of Fair-Rite.
The published DM and CM attenuation for the WE filter board is reproduced in Figure 7.
Figure 7 These are the attenuation curves of the filter design used. Courtesy of Würth Elektronik.
The TI board had high conducted emissions, so I’ll use that as an example of how well the LISN Mate can separate out the DM from CM emissions. Figure 8 shows the unfiltered results.
Figure 8 This shows the ambient noise floor (yellow), DM (violet), and CM (aqua). We can see the DM emissions are dominant. We’re looking from 10 kHz to 30 MHz.
Now, knowing that the DM emission is dominant, we can be sure to include appropriate DM filtering, typically a capacitor across the DC input. But CM is also relatively high, especially in the low kHz region, so that indicates we’ll need either a ferrite choke or standard CM (wired) choke, which should perform much better than the ferrite.
The following two figures show the difference between filtering and no filter for both DM and CM. The DM emission is almost completely suppressed, while the CM has been reduced 10 to 20 dB.
Figure 9 This shows the ambient noise floor (yellow), DM unfiltered (violet), and DM filtered (aqua). We can see the DM emissions are almost completely suppressed across the range 10 kHz to 30 MHz.
Figure 10 This shows the ambient noise floor (yellow), CM unfiltered (violet), and CM filtered (aqua). We can see the CM emissions are reduced 10 to 20 dB over much of the range of 10 kHz to 30 MHz.
The ferrite choke was installed at the power input of the TI demo board. It worked much less effectively than the WE filter board and Figure 11 shows the CM results looking from 10 kHz to 4 MHz, typically the noisiest portion of the spectrum for DC-DC converters. We can see an overall 5 to 15 dB improvement and this is about all you can expect from a ferrite choke – even one designed for these lower frequencies. Of course, the CM ferrite choke would not have helped the DM emissions.
Figure 11 This shows the ambient noise floor (yellow), CM unfiltered (violet), and CM filtered (aqua). We can see the CM emissions are reduced 5 to 15 dB over much of the range of 10 kHz to 4 MHz.
Switch-mode power supplies can produce large amounts of both differential-mode and common-mode emissions. The LISN Mate from Tekbox can help the designer by displaying either the DM or CM emissions. Depending on which appears dominant, designers can determine whether the filtering requires DM or CM filters, or a combination.
—Kenneth Wyatt is president and principal consultant of Wyatt Technical Services.