Diode RF limiters

[ironcurtain]

Anyone around has PCB designs in Kicad and part numbers suitable for building RF diode limiters? Preferably for 10MHz-1GHz.

[ironcurtain]

A good reference so far:

https://www.skyworksinc.com/-/media/SkyWorks/Documents/Products/1-100/200480C.pdf

I'm interested in the 2 stage detector limiter design in page 11. Trying to select parts for decent performance 0-2GHz, for 0dBm and 5dBm.

[RoV]

Did you give a look at https://www.minicircuits.com/WebStore/Limiters.html?
They are not very costly and they may save you some headache 
Example https://www.minicircuits.com/WebStore/dashboard.html?model=RLM-33H%2B

[ironcurtain]

Hola :-)

A more suitable model:
https://www.minicircuits.com/pdfs/CLM-83-2W+.pdf

Or:

• https://www.minicircuits.com/pdfs/RLM-23+.pdf
• https://www.minicircuits.com/pdfs/RLM-23-1WL+.pdf

The ones with ESD protection explicitly mentioned could also help removing the BAV99 in some front-ends, which I think is likely a worse performer than their solution.

The one you linked has a rated output power (leakage power) of +18 dBm, which will positively destroy most front-ends in many SDR receivers (HackRF being infamously weak/prone to damage at anything near or above 5 dBm).

I would like to come up with my own design (or "our" design). I don't think opening the "can" of the Mini-Circuits parts will reveal much, as they are likely integrated silicon/don't have discrete parts from other manufacturers.

I also looked at Macom parts (MADL-011021-14150T, https://www.mouser.es/datasheet/2/249/MADL_011021_14150T-1316677.pdf) but again the output power is too much (15dBm). They are much cheaper than Marki's or Mini-Circuits.

If I can find a suitable part with 0-5dBm output or a 0 dBm threshold, that would be great.

[kleiner Rainer]

Hello all,

the BAV99-based limiter is not so bad - a comparable circuit is the core of the hp 11947A Transient Limiter that is specified from 9kHz to 200MHz, schematic on page 11:

https://xdevs.com/doc/HP_Agilent_Keysight/HP%2011947%20Operations%20&%20Service.pdf

The one in my lab states: "Max input: +-12VDC / 10kW for 10us"

A homebrew project, complete with measurements is available here:

https://www.w0qe.com/Projects/rf_clipper_2016.html

Both are useful to protect equipment from bursts of power and ESD, especially sensitive spectrum analyzer inputs.

Greetings,

Rainer DG1SMD

[ironcurtain]

I'm interested in a two stage limiter, not a "clipper". They are very different animals (check my linked app. note "PIN Limiter Diodes in Receiver Protectors" by Skyworks). The clipper design is nice and simple for the one you linked, though, but it does not offer the same protection (we want to divert all RF input above the threshold to ground, more or less).

[kleiner Rainer]

Since you did at first not specify your requirements, I assumed you wanted to protect a sensitive RF input - in this case, the clipper is the better option. As stated in the paper, the limiter has a widely varying return loss when active - it depends on your application whether you can live with that. My application is EMC work, and there you want a controlled impedance, otherwise your measurements are not acceptable. The other advantage of the clipper in my application is that it does not suffer from "spike leakage" (figure 12) and is not frequency sensitive (PIN diode carrier lifetime!). Especially when measuring the EMC behaviour of mains circuits or SMPS, you want something that limits the maximum input voltage to your input regardless of frequency or source impedance. The spikes you encounter there are nasty...

Greetings,

Rainer DG1SMD

[ironcurtain]

Hi,

I did mention the two stage limiter:

"I'm interested in the 2 stage detector limiter design in page 11. Trying to select parts for decent performance 0-2GHz, for 0dBm and 5dBm."

:-)

The use case is not EMC for me, it is indeed to protect a sensitive RF front-end for a SDR receiver (not hamradio equipment also, so it is not designed to tolerate inputs higher than 5-10dBm, 5dBm seems to be the sweet spot for most of the hobbyist-accessible SDR devices around, including the ubiquitous HackRF).

We can talk EMC later in a separate thread, I intend to build two LISNs, one for DC and another for AC, for some future work. I did some volunteer work for the local "ARRL" analyzing some light fixtures sold at scale (they function as fantastic wideband jammers especially at HF) years ago.

Controlled impedance is not necessarily crucial for my use case, obviously you don't want to completely mismatch the entire setup since that will cause loss. I need fast action/clamping as soon as the threshold is reached. Low part count is also ideal. I might design afterwards a combined PCB with a LNA immediately behind the limiter, and perhaps an adjustable attenuator). For now I want to keep it simple. People are being charged quite a bit, even from Aliexpress, with limiters that have dubious design and components.

[MartinL]

Mike Ossmann has done some work in this area as part of our URTI project. Specifically he was evaluating limiters with a Schottky diode anti-parallel to a PIN diode. This is a topology with several advantages, explored in a number of publications by Chin-Leong Lim.

Notes and references on this work can be found in this section of the 2023-06-02 URTI project report.

VNA test results can be seen in this lab report, and the KiCad designs associated with this work are here.

[RoV]

+5 dBm are only 0.56 V peak on 50 ohm, it's difficult to have an hard limit at this level with diodes. I think a schottky+PIN design would just kick-in, its hard limit would be higher. Perhaps you can use two anti-parallel schottkies as a clipper, but be prepared to have plenty of distortion, because their knee is very gradual and they will distort even much weaker signals, producing mostly odd harmonics.
You could perhaps try to slightly forward-bias the PIN in the schottky+PIN design. Bias should be low impedance, otherwise strong signals would change it by rectification.

[glenenglish]

Suggest planning for a limit of around +10dBm. That is more practical
the CLA4605-***  devices from skyworks have a useful package, also, read the app notes. Understand the operation, exactly how they work.
I design the input stage (transformers, termination) of my SDRs for around +13 peak (0dBfs) , so the limiter has a chance of doing something.
In order for high IP3 to be able to be realised , any amplifiers ahead will need P1 of around +16 to +20, which means you certainly need a limiter. I see this error in lots of designs.    If using  MMICs , some are much better than others like ADF5535 beats the hell out of MCL for IP3. But at low freqs, transformer feedback discrete push pull amps are far superior in IP3, IP2,
Do the sims and measure what happens when the amplifier proceeding the limiter sees limiter diode action. It might not like it.
If narrowband, a 90 deg network between the amplifier device and the limiter may assist stability.

[mag_therm]

Another approach? Using Triode as limiter for Ham HF Band SDR.

There have been frequent thunderstorms here in Michigan this year.
There are two single board SDR Txcvr in use, a Hermes14 and a HackRF1

A home brew active Rx antenna tuner here was receiving almost continously for 3 years.
It was connected during most thunderstorms. In 3 years there was one failure during a thunderstorm of the input FET (2n4416)
which has a maximum gate rating Vgss of 30V. There is no limiter in this circuit

Modeling in qucs to try to add shunt TVS to the fet input, led to looking at a grounded grid triode as a way to limit the through energy
without reducing the wanted signal.

A suitable triode still available NOS is 6GK5, being a VHF single triode with  grid-plate capacitance of 0.52 pF, 180mA heater current,
and a cathode-heater maximum voltage of 100V. The data sheet is useful, having remote cut-off curves right down to zero plate voltage,
 along with gm and Rp curves. The tube is powered from 13.8Vdc

There are some freeware papers (Nasa etc) about lightning in HF spectrum. Thse two were useful:
Info on the RF Spectrum of lightning:
NASA Technical Memorandum 87788 Review of Measurements of the RF Spectrum of Radiation from Lightning David M. L e Vine
Info on field intensity with distance:
V.A. Rakov CHARACTERIZATION OF LIGHTNING ELECTROMAGNETIC FIELDS AND THEIR MODELING

The following circuit:
https://app.box.com/s/p8ndefcgfmtkp8ugaieinjmdnzteflh3
 was partially modeled in Qucs to design the LC impedance transformers, one at input for 50 Ohm :Rk,
 and one at output for Rp:50 Ohm. As the transient solver shows, the grid and plate coupling capacitors can be reduced in value to act a high pass filters, limiting the long duration voltage of the pulses.
Qucs of the input transformation between input sma and cathode:
https://app.box.com/s/geevbpqqu03gqtveqst87ppqodk0tz98
 There are antiparallel clamping diodes on 50 Ohm output.When they start clamping, the output transformation of plate signal becomes a mismatch to the high impedance plate resistance, and the level stays solidly at about 0.65V regardless of input level. This allows use of ordinary 1N4148W with Cd=1.5pF

The prototype build  was adjusted for cathode bias to provide +3dB of S21 gain and tested for S11, S21, and S22 using nanoVNA, and then tested for compression.(Gain needs to be approximately unity for accuracy of SDR signal levels.)
With the old S.A. here,
At levels below ~ -30dBm, S21 gain is linear with no noticable 2nd Harmonic increase over the level from generator.
In -10 dBm range the (+) signal peaks approach grid conduction which pushes grid voltage toward cut-off, and S21 compression starts.
At about +5 dBm the output clamping diodes start to conduct, holding level at +5dBm.

S11:
https://app.box.com/s/nl4qhcxy4eljd0z7d4i2hlouxiyhf177
S21: Add +30dB to this (below) curve to compensate for the -30dB attenuator used to reduce the level from nanoVNA to below limiting level
Off band attenuation is provided by the input and output LC impedance transformers.
https://app.box.com/s/56se1dkpcox27l0snv4ee3j299psjcz6
The prototype was tested to destruction. At approximately +47dBm (50Watt, 50Vrms)  the cathode arced and tube failed.
Also the 50V mlcc input capacitors failed, which led to uprating those to 500V NPO MLCC.

Layout: to allow fitting into existing Linear Amplifier compartment, the tube axis was laid parallel to the surface mount board, and short flex wires run up to the 7 pin socket.
https://app.box.com/s/ysreoroiro7i94r6cde0y63n8lzvva32

[glenenglish]

An interesting approach mag_therm !
I build ionosonde devices and antennas, and so being connected during the thunderstorm is necessary.

* The Skyworks PIN limiter diodes have a good /useful range down to about 30-50 MHz, then their I regions are a little insufficient.
* BAR64  are not bad down to a meg with limited dissipation .
* chip ESD suppressors are very useful, most of them are << 1pF, and have plenty of instananeous capability.
* 1N4007s and 1N5408s make excellent PIN diodes , and very reasonable limiter diodes 40 MHz down to 100kHz.  Their large capacitance is not a big problem, design a network to use that capacitance (IE put the diode as the capacitor in the  shunt leg of a simple T low pass filter, pad as required)  (dont forget a choke across the diode so it has DC conditions) .
* 70V GDTs are good as the  high energy last resort, although on an 8/20 waveform  they dont fire over until 600V
glen