Transient Limiter for Spectrum Analyzer Protection

[Jay_Diddy_B]

Hi group,
Time to share another project with the group. This time it is a transient limiter designed to protect spectrum analyzer inputs from damage, especially when doing EMC tests. The design is based on the discontinued HP 11947A.

The device is a combination of:

9kHz High Pass Filter
10dB attenuator
Limiter than limits the output to 1.4V p-p  (about 5mW)

LTspice
Like many of my projects I started with an LTspice model:



and the modeling results:




The modeling shows 10dB of attenuation a nice 9kHz filter with steep roll-off.

Schematic

The schematic was drawn in KICAD:



PCB Design

The circuit is laid out using 50 coplanar waveguide. The board is designed to replace the lid on a Hammond 1590G diecast box.



Mechanical Design

I reviewed the mechanical design using Fusion 360:



Prototype Build

Here is a picture of the prototype board. The board was made on my LPKF c60.




And the assembled unit:




In the next message I will present the measured test results.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,

Measurements



This shows the transmission. There is a 9kHz high pass filter. There is an additional 30dB of attenuation at 1kHz.
The transmission is very flat from 100kHz to 200MHz.



The HP 8714A shows the transmission id 0.8dB down at 500MHz.



The transient limiter was terminated with a 50 and the reflection was measured from the input. The measurement shows that the input impedance is within 1 of 50 over the frequency range 300kHz to 200MHz.



This is the same reflection information plotted on a Smith chart. The trace is a very small circle at 50 , the center of the Smith chart.


Limiting



This trace shows the input and output traces. The output is limited to 1.4V p-p.

Regards,
Jay_Diddy_B

[TimNJ]

Thanks! What’s the maximum power input? I know the original HP can go up +34dBm. These components look smaller, but maybe it’s still okay for majority of spectrum analyzer work?

[langwadt]

Thanks! What’s the maximum power input? I know the original HP can go up +34dBm. These components look smaller, but maybe it’s still okay for majority of spectrum analyzer work?

1.4Vpp is ~7dBm, 34dBm is ~32Vpp

[Jay_Diddy_B]

Hi,

30dBm = 1W
33dBm = 2W
34dBm approximately 2.5W

Here is a thermal image with 10V p-p at 1MHz applied:



There is equal power dissipation in R10 and R6. This was confirmed in the LTspice model.

10V p-p = 3.5V RMS

Power =  0.25W

Temperature rise is about 12 degrees C.

If we allow for a 50C temperature rise the maximum power input is 1W or 30dBm


Increasing the size of the resistors will raise the maximum power.

However if left as it is, the spectrum analyzer is protected if the Transient Protector fails.

Regards,
Jay_Diddy_B

[TimNJ]

By the way, what part number did you use for the 1mH inductor and what was the criteria you used to make the selection? (If I may pick your brain.) 

[Jay_Diddy_B]

Hi TimNJ and the group,

I have used several different inductors in that location.

I am currently using TDK NL45323T-102J-PF

Digikey 445-6411-1-ND

The only selection criteria is that impedance needs to be greater about 200 over the frequency range of interest.

Regards,
Jay_Diddy_B

[Noy]

Hi, nice thingy.

But because i'm new to SA/VNA:
Do i need such a transient limiter only at the SA input or additionally on the TG output too?
I have a SVA1032X so i think the TG is also an input in VNA mode?

[jonpaul]

Hello: A  blown SA, VERY long and costly front end repair.


With any substantial transient, The tiny 1n4148 will easily blow open and the gas protector may not fire in time.

I would explore low capacitance transient absorbing Zeners and at least reoplace the 1N4148 with a fast low C higher I diode.

To respond we need to know the spectrum analyzer model and the applications eg monitor broadcast/ham transmitter, data line/cable/internet, EMI test and compliance, transient testing.

The expected energy of the transient needs to be defined as well.


Finally a simulation must have a model of the expected transient source.

Here is a simple test: Use a LISN, with both Line and Neut switched out.

Imagine the switch is moved L>>N.

The AC coupling caps may be 470 nF. Thus a charge cap will have a 120V or 240V RMS to zero transient.

Calculate the Joules.

Apply charged cap to the protector.

Finally you can find HP and TEK recommended protectors.

Bon Chance,

Jon

[Noy]

My device is a Siglent SVA1032X .
I will use it as general input protection.
Already build / are building a LISN (last post):
https://www.eevblog.com/forum/projects/5uh-lisn-for-spectrum-analyzer-emcemi-work/75/

But there is the same circuit used. Maybe i should add in parallel to the 1mH inductor something like this:
https://www.mouser.de/ProductDetail/MACOM/MADL-011021-14150T/?qs=%2F9lm7XkpNBiyHlfXH%2Fm8aA%3D%3D


I already have this one:
https://de.aliexpress.com/item/4000839596986.html?spm=a2g0o.productlist.0.0.200ab9bd1hNnDg&algo_pvid=12ce58d1-08cb-4c6f-a999-c163115c95df&algo_expid=12ce58d1-08cb-4c6f-a999-c163115c95df-2&btsid=0b0a182b16086857307735005e308c&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

I can use it between the LISN / SA but I'm unsure if this is useable for the lLISN frequency. Starting frequency of this PIN Diode Limiter is above the LISN frequency..

I will use the LISN only as a DC (60V / 5A) one..

I will build these transient limiters for use with antennas and than maybe i should put such a PIN Diode part additional on the pcb? I think this would give a better  / higher frequency useability instead of using this unmodified Transient Limiter  + the external Aliexpress PIN Diode Limiter in series.. ?
Unfortunally the Aliexpress PIN Diode Limiter has no metal enclosure (only for >30€ or i must build an enclosure by myself for this device..?)

[Jay_Diddy_B]

Noy and the group,

The transient limiter described here is designed for use in conducted EMC measurements below 100MHz.

The PIN diode limiter sold on AliExpress, is probably based on the Macom design:





The Macom application circuit uses small value, 100pF coupling capacitors and is suitable for applications 10MHz to 6GHz.

Two different devices for two different applications.

The LISN I designed has a transient limiter built in. There is no need to modify it or change it.

Jay_Diddy_B

[Noy]

I know that your design here is for conducted one. But you build this already in your LISN design so i don't need "seperated" one.
My intention was to build 2 seperated ones which can cover the whole "spectrum" 9kHz up to (in my case 3.2Ghz) so i thought about using your design and add a "higher" frequency limiter like the macom ones. Could this work? Simply add these PIN diodes or another kind of these MACOM /skyworks PIN diodes in parallel to the inductor? Is the inductor suitable as DC back path if i simply add a PIN Diode without the build in schottky?

The Aliexpress ones are even worse... I opened one up (unfortunally after i ordered a second one.. :-() it is a 0 Ohm resister , 2 caps, and a "outdated" Skyworks/Alpha SMS7621-005 (Marking: SH2)

Maybe i can solder a real pin diode / skyworks limiter on this pcb then i don't lost all the money. And maybe I'm lucky and get my money back, contacted Aliexpress ..

[Noy]

Hi,
i bought now:
https://cdn.macom.com/datasheets/MADL-011021-14150T.pdf (example page 5)
https://cdn.macom.com/datasheets/MADP-011029.pdf

I will use it as limiter (together with a SMA attentuator screwed in front) which i will use when I am using my swr bridge / directional coupler to measure antennas.
And in general when I'm using Antenna / VNA features.

My issue is now;
1. Which value should i pick for the ac couple caps if i would have a pass frequenca from 10MHz to (3.5 Ghz)? Is 100nF suitable? I will use 0402 and 4 Layer JLCPCB stackup to get a 50 Ohm CPWG. All mounted inside a 25x25x50 alu profile case from aliexpress.
2. Whats the diode behaviour below the 10MHz?
3. Do i need a highpass with fg >=8MHz at the input? For example 4nH, 100nF? This is not inside the example schematic, but maybe worthy?
4. The resistors drawn inside the example schematic, are this internal / pcb traces or dedicated ones? No information inside the datasheet.
5. Here is a information: https://www.richardsonrfpd.com/docs/rfpd/PIN_Limiter_Design_Guide.pdf and there is stated you should place the diodes (2 stage limiter) with lambda/4  between them. But which lambda if i want a limiter for a wide spectrum ?If possible <=10MHz to >=3.5 GHz ?

[Noy]

Did a pcb for Hammond 1590A (cheap from china) files attached..

[ironcurtain]

I opened a thread recently precisely related to RF limiting diodes (but for RX frontend protection) here: https://www.eevblog.com/forum/rf-microwave/diode-rf-limiters/msg5518564/#msg5518564