6-gang FM stereo tuner heads

[regenfreak]

Last year I downloaded LTSpice but never got to learn to use it. Tonight I have spent an hour learning to create my first ever simulation: a lumped element model of a 10K SMD 1206 resistor. According to W0QE, the stray inductance can be assumed to around 600pH but the exact value is not important. So the model is a 10k resistor in series with a 600pH inductor and both of them are connected to a shunt capacitor of 50fF. Here is the S21 from 10M to 4G in linear scale. It shows the same trend as my nanoVNA measurement that the S21 is about -40db at low frequencies and approaches -20db at 4Gb. Looking at my Xp and Cp measurements closely, my data still dont agree with both W0QE's and Jeroen Belleman's measurements at the lower frequencies.

If I remember correctly, the baluns for the directional coupler of the NanoVNA does not work very well either at very low frequency or very high frequency. Hence you see ripples in my S21 at the GHz top end.

Here is W0QE's video on the modelling of a 10k resistor using SimSmith:

https://youtu.be/0JglLhrisOs

I tried to model the 10K resistor with SimSmith last night but the wirings of Ruse block function is so user-unfriendly that I aborted.

[regenfreak]

I have received the cheapo "0dbm RF PIN limiter 10MHz-6GHz".

It consists of a PIN diode ( with back-to-back diodes inside the SMT package) in series with a 100nH inductor and two 10nF DC blocking capacitors.

The S21 looks ok with 0db to maximum -0.1db insertion loss up to 500MHz.

Hard clipping is supposed to occur at around 630mV peak-to-peak. I tested it with the sig gen and scope at 10MHz. Instead of seeing clipped sine waves above the clipping threshold, I only get "soft clipping"; distorted and attenuated sine waves as I increase the power. I have used a cheapo power meter to measure the input and output power through the PIN protector. As shown in the power measurement, it does not do a good job of clipping. The power output should be roughly flat once it reaches above the threshold. Oh well ,life of full little disappointments, I can classify this product as "junk" status. meaning it cannot be trusted to protect valuable spectrum analyzers.

Ref source of PIN diode clipping figures:
https://www.skyworksinc.com/-/media/...00/200480C.pdf

PS In my LTspice model I added a duplicated 50 ohm load for the AC voltage source but it would not make much difference to the result as here is a big mismatch anyway.

[regenfreak]

Clarification; PIN diode on its own consists of P-I-N layers and not back-to -back diodes. But they do make version of pin limiter with a pin diode back to back with a Schotty diode in an integrated package.

[G0HZU_JMR]

The single diode limiter (with a bias choke across it) probably isn't a good choice as a protection limiter. It won't offer very good isolation for large signals. A better limiter design will use four diodes arranged as a classic clipping limiter.

A typical silicon diode will have a forward voltage of about 0.8 to 0.9V when biased hard so the clipping limiter will produce a rounded square wave output at about 1.8Vpkpk when hard limiting. Hiding inside this 1.8Vpkpk rounded square wave will be a fundamental sine wave of about 2.3Vpkpk (or 1.15Vpk)

This will produce a through leakage power of (1.15^2)/100 watts into a 50R load when hard limiting. This is about +11dBm at the fundamental frequency. That's what you can expect from a classic 4 diode limiter. Some limiters only use two diodes, and this can help extend the bandwidth to >2GHz. The 2-diode limiter won't be as rugged though.

I'm a bit out of date when it comes to the capability of modern limiter diodes, but 15-30 years ago a well-designed silicon (4) diode limiter could withstand a 5-10W incident power (without damage) whilst limiting the leakage to less than about +13dBm. The limiter should be transparent to signals that have a maximum amplitude that is in the order of <0.6Vpkpk. i.e. the limiter should not cause any significant distortion to waveforms smaller than this.

[regenfreak]

Originally Posted by G0HZU_JMR

The single diode limiter (with a bias choke across it) probably isn't a good choice as a protection limiter. It won't offer very good isolation for large signals. A better limiter design will use four diodes arranged as a classic clipping limiter.

A typical silicon diode will have a forward voltage of about 0.8 to 0.9V when biased hard so the clipping limiter will produce a rounded square wave output at about 1.8Vpkpk when hard limiting. Hiding inside this 1.8Vpkpk rounded square wave will be a fundamental sine wave of about 2.3Vpkpk (or 1.15Vpk)

This will produce a through leakage power of (1.15^2)/100 watts into a 50R load when hard limiting. This is about +11dBm at the fundamental frequency. That's what you can expect from a classic 4 diode limiter. Some limiters only use two diodes, and this can help extend the bandwidth to >2GHz. The 2-diode limiter won't be as rugged though.

I'm a bit out of date when it comes to the capability of modern limiter diodes, but 15-30 years ago a well-designed silicon (4) diode limiter could withstand a 5-10W incident power (without damage) whilst limiting the leakage to less than about +13dBm. The limiter should be transparent to signals that have a maximum amplitude that is in the order of <0.6Vpkpk. i.e. the limiter should not cause any significant distortion to waveforms smaller than this.

Thanks. The 4 rectifier diode idea is still in my mind but I am concerned about the large diode junction capacitance which may call for a matching network to tweak the S11 and S21 at high frequencies.

I have taken another potshot at the PIN diode. I ordered a Skyworks SMP1330-005LF last night. It is a PIN-Schottky pair, having a threshold about +10dbm, C=0.7pF, 14db attenuation at +30dbm working up to 4GHz (see power characteristics attached). I speculated that the cheapo pin-limiter does not work well because it uses only a single diode PIN diode.

https://www.skyworksinc.com/-/media/...es_200050N.pdf

I am going to add two 1kV 1206 10nF and 250mA SMT fuse plus a 70V 4532 SMD ceramic GDT (capacitance = 0.5pF) to safeguard against "catastrophic event". I will house them in RF shield instead of using a metal box.

I have been looking at the integrated SKY1662-632LF datasheet:

https://www.skyworksinc.com/-/media/...LF_202945L.pdf

It is kind of interesting to see how they put a simple series LC network to optimize the S11 and S21. With L = 1.8nH, C=15pF for 2.45GHz, they transpose the peak and trough for S11 and S21 to the right. The equivalent circuit of a PIN diode (in open state) is like a small inductor in series with the paralleled capacitor and a very large resistor. The S21 and S11 look a bit like a very broadband bandpass filter.


PS: I have received the ATC caps and other goodies that you sent. Thank you so much. I am building a "Q detector probe" with your ATC 1pF caps. It will use FET MMBF310 and MMBF3906 as impedance convertor. The device will be used to find approx Q and resonance frequency of air variable tuning gangs and IF transformers for MW, SW and FM radios in unpowered state.
The ATC100 series caps have Q about 10,000..quite badass!

[G0HZU_JMR]

Thanks. Glad the caps are of use

One thing to watch out for with the single diode limiter (with a bias choke across it) will be the carrier lifetime of the PIN diode. The Skyworks SMP1330 series appear to only have a carrier lifetime of about 4ns. This limits their use to UHF. At lower frequencies, the PIN diode won't act as a resistor for the whole of the RF cycle as the carrier lifetime will be too short for this. This will spoil its performance as an effective limiter at lower frequencies.

[regenfreak]

Originally Posted by G0HZU_JMR

Thanks. Glad the caps are of use

One thing to watch out for with the single diode limiter (with a bias choke across it) will be the carrier lifetime of the PIN diode. The Skyworks SMP1330 series appear to only have a carrier lifetime of about 4ns. This limits their use to UHF. At lower frequencies, the PIN diode won't act as a resistor for the whole of the RF cycle as the carrier lifetime will be too short for this. This will spoil its performance as an effective limiter at lower frequencies.

Thanks. I almost got caught out. (I didnt know the choke across is for self-biasing.) The SMP1330-005LF datasheet only states it works up to 4GHz. It shows some parameters from 1M, 100Mhz and 1GHz. This carrier lifetime business seems very complicated. The link below states that as a rule of thumb, the operating frequencies of PIN diodes should at least 10 times the reciprocal of the carrier lifetime:

https://www.infineon.com/dgdl/AN034....13de8d362503df

For 4ns, it is 250MHz, 10 x 250MHz = 2.5GHz which seems an over-conservative estimate.

Now this suggests a two-tier protection system: conversional diodes for low frequencies to VHF and PIN-Schottky for UHF. I have ordered some 1N4148 silicon diodes. I have some BAV99 switching diodes and 1SS86 Schottky diodes are on their way.

The power output curve for the SMP1330-005LF indicates that the Schottky diode clamps first and the PIN diode wakes up later at a higher power level. It seems it does not need any biasing choke as in SKY1662-632LF?


The SKY1662-632LF is for 200MHz to 4GHz. I guess the sharp dip of the S11 in the datasheet suggests the coupled resonance of parasitic elements from the PIN-Schottky and the transmission lines. It is a bit counter-intuitive that they put a LC series-resonance network in shunt with the diodes to shift the dip of S11 and plateau of S21 to higher frequencies. I would expect they put a LC series-resonance network in series with the diodes to increase the resonance frequencies. That's the black art of microwave...as the frequencies go to 1 to 2 GHz, the lumped element models fall apart.

[G0HZU_JMR]

I think the Schottky diode is there in place of the choke, and it will act as a detector to bias the PIN diode more efficiently than the choke.

For 4ns, it is 250MHz, 10 x 250MHz = 2.5GHz which seems an over-conservative estimate.

This strict requirement is probably to allow the PIN diode to be used as a variable resistor in a variable RF attenuator. The resistance of the PIN diode needs to remain linear across the whole RF cycle and there needs to be plenty of margin for the carrier lifetime. Otherwise, intermodulation distortion can begin to appear if the resistance misbehaves even slightly over the RF cycle.

The 1SS86 Schottky diodes are going to be a bit weedy for use in a general- purpose protection limiter. A lot depends on what your goal is though.

In my case, I use a 4 x 1N4148 based limiter that incorporates a dc block and a fuse, and this is good to about 250MHz. The 4 x 1N4148 diodes are OK up to about 5W incident power but the fuse will blow above about 2W incident power. Looking back through my old test notes, the leakage is typically 14dBm across 14MHz through 145MHz at 5W incident power. The insertion loss is about 0.2dB and this is due to the ESR of the inline RF fuse.

For UHF and higher, I use a custom limiter made by Cobham. This has SMA connectors and protects across 20MHz to 6GHz at cw power levels up to 10W. It can cope with narrow (low duty cycle) pulsed power levels up to 1kW according to the datasheet. The leakage level is typically <13dBm at 1W incident power. I very rarely use it.

[regenfreak]

Originally Posted by G0HZU_JMR

I think the Schottky diode is there in place of the choke, and it will act as a detector to bias the PIN diode more efficiently than the choke.

This strict requirement is probably to allow the PIN diode to be used as a variable resistor in a variable RF attenuator. The resistance of the PIN diode needs to remain linear across the whole RF cycle and there needs to be plenty of margin for the carrier lifetime. Otherwise, intermodulation distortion can begin to appear if the resistance misbehaves even slightly over the RF cycle.

The 1SS86 Schottky diodes are going to be a bit weedy for use in a general- purpose protection limiter. A lot depends on what your goal is though.

In my case, I use a 4 x 1N4148 based limiter that incorporates a dc block and a fuse, and this is good to about 250MHz. The 4 x 1N4148 diodes are OK up to about 5W incident power but the fuse will blow above about 2W incident power. Looking back through my old test notes, the leakage is typically 14dBm across 14MHz through 145MHz at 5W incident power. The insertion loss is about 0.2dB and this is due to the ESR of the inline RF fuse.

For UHF and higher, I use a custom limiter made by Cobham. This has SMA connectors and protects across 20MHz to 6GHz at cw power levels up to 10W. It can cope with narrow (low duty cycle) pulsed power levels up to 1kW according to the datasheet. The leakage level is typically <13dBm at 1W incident power. I very rarely use it.

Thanks. I will try your suggestions of using 4 x 1N4148 diodes. One wonders if the diodes are too resilient to high power input is a good thing..It would be ideal if the diodes fail shorted. But Murphy Law states" screw you, you never had it so good". If there is lots of energy through the diodes, it would be shorted initially, and the heat vaporizes the diodes making them open. Without the fuse, the energy will dump to the spectrum analyzer...

[Radio Wrangler]

The Schottky-PIN combination is used because the PIN alone has problems.

At high frequencies, the PIN is a lousy rectifier because of excessive storage time stopping it doing any dioding.

At low frequencies, it starts to work as a rectifier, but the resistive PIN action is gone (not enough storage time)

PIN diodes are good because they can pass high RF currents, much larger than the DC current turning them on, and they reverse conduct large currents for the storage time, when they eventually start thinking like a diode and turn off.

So, in a feat of demarcation worthy of some politician, the Schottky rectifies the RF and the DC turns the PIN on. The PIN can carry more RF current than the DC in the loop, and it can carry both halves of the RF cycle.

When the overload burst stops, you still get the limiter hanging on shorted untl the carriers recombine.

David

[G0HZU_JMR]

Yes, the Schottky diode + PIN combo is quite neat and can also allow the limiting to begin at a lower drive level.

It's worth studying the spec sheets carefully for all limiter types. There is usually a compromise somewhere. Some can limit at lower drive levels, but they can end up having higher leak through power. Other limiters might only be rated for 1W incident power.

This evening, I'll lift the cover off my fused limiter and show you the circuit inside. I still have my old design notes here. It uses four 1N4148 diodes, a 100nF DC block cap and a 160mA 20x5mm quickblow fuse from Littelfuse. The 20mm fuse has some inductance, so I fitted 3.9pF shunt caps either side of it to absorb this into a 500MHz LPF. I arranged the two pairs of 1N4148 diodes such that there is a small amount of inductance between them. This also makes up a pi LPF, but this has a cutoff at about 1GHz.

I'll post up some images of the insides and some plots later. It's the limiter on the far left in the image below. I've been using it for many years and so far, I've never blown the input fuse. I still have the prototype version on some copper PCB material and I tested the prototype many times to make sure the fuse blows rapidly at power levels above a few watts.

I think anyone who has a spectrum analyser and who is also interested in CB/ham radio up to 146MHz should have a limiter like this. This might not be the best limiter available, but it is cheap, and the performance is predictable in terms of signal handling and through leakage.

[G0HZU_JMR]

Here's an internal view of the 250MHz fused limiter.

The 20mm fuse is soldered in place like a surface mount component. This is a bit of a fudge, but it seems to work OK.

There's also a recent plot of insertion loss and also the return loss. Although the performance looks OK up to 300MHz or so, I try and use it only with transceivers that can transmit up to about 145MHz.

The idea is that the 1N4148 diodes should be reliable at about 3W incident power and the fuse will blow at around this power level. This fuse may or may not provide protection against incident power levels above 10W. The fuse might not blow quickly enough at (say) 25W incident power. The diodes may fail before the fuse blows.

[G0HZU_JMR]

Another option is to download the old HP app note AN1050 and try making one of the PIN diode limiters described in it. I think there are several versions of AN1050 and the oldest one has the most info.

There is a 4-diode limiter in the old version of AN1050 and it is claimed to be able to withstand 10W incident power and operate up to 1.5GHz. The HSMP-482x (and HSMP-382x) PIN diodes used in these limiters are all obsolete but you may be able to find them online somewhere. They might be listed as HP or Avago or Broadcom parts.

I've still got lots of HSMP-482B diodes, but these are single diodes in a SOT-23 package. I think the PIN diode is the same as the PIN diode in the dual diode HSMP-4822.

I've also got quite a few HSMP-389Z PIN diodes, but I think these are for switching rather than limiting. I've got plenty more PIN diodes from Macom and Microsemi, but these are all high-power diodes used in RF switches.

I'm not sure how rugged the HSMP-4822 diodes will be at low frequencies. That's why I've stuck with the 1N4148 diodes. A limiter made with the HSMP-4822 or HSMP-3822 diodes will probably have less leakage and it will probably be more reflective (a good thing), but these diodes probably won't be as rugged as the 1N4148 diodes at low frequencies.

[regenfreak]

Originally Posted by Radio Wrangler

The Schottky-PIN combination is used because the PIN alone has problems.

At high frequencies, the PIN is a lousy rectifier because of excessive storage time stopping it doing any dioding.

At low frequencies, it starts to work as a rectifier, but the resistive PIN action is gone (not enough storage time)

PIN diodes are good because they can pass high RF currents, much larger than the DC current turning them on, and they reverse conduct large currents for the storage time, when they eventually start thinking like a diode and turn off.

So, in a feat of demarcation worthy of some politician, the Schottky rectifies the RF and the DC turns the PIN on. The PIN can carry more RF current than the DC in the loop, and it can carry both halves of the RF cycle.

When the overload burst stops, you still get the limiter hanging on shorted untl the carriers recombine.

David

Thanks. It seems the Schottky and PIN diode work in synergy in an intricate manner. The plot thickens when I think of the PIN diode as a rectifier diode. Maybe it is less confusing to think of them as "current controlled resistor" or high speed Rf switches instead. It has good "linearity" in term of series resistance vs forward bias current in the log scales at high frequencies. It excels at passing RF current:

https://www.microsemi.com/sites/defa...PENDIX%20B.pdf

But the PIN limiters seem to very difficult to design and not always behaves what you want it to ....

G0HZU_JMR Here's an internal view of the 250MHz fused limiter.

The 20mm fuse is soldered in place like a surface mount component. This is a bit of a fudge, but it seems to work OK.

There's also a recent plot of insertion loss and also the return loss. Although the performance looks OK up to 300MHz or so, I try and use it only with transceivers that can transmit up to about 145MHz.

The idea is that the 1N4148 diodes should be reliable at about 3W incident power and the fuse will blow at around this power level. This fuse may or may not provide protection against incident power levels above 10W. The fuse might not blow quickly enough at (say) 25W incident power. The diodes may fail before the fuse blows.

Thanks thats very helpful. My 250mA 1206 SMD fast fuse has about 50nH and 0.6 Ohm dc resistance. Anyway, I will build one based on your design to see how it goes. I normally never go above 120MHz in spectrum analyzer sweep with FM broadcast receivers anyway. I can live with the undesirable characteristics at UHF. At some point, I will receive some cheapo VCO through the shipment. It can go beyond 1GHz that may allow me to test the SMP1330-005LF with a RF amp and attenuators in a limited manner. Probably my cheapo power meter can't cope with a higher power level.

I have found the AN1050 application note:

http://www.hp.woodshot.com/hprfhelp/...lit/an1050.pdf

It mentions the integration of the low-cost PIN diode with the coplanar waveguide (CPW) which is not for the fainted hearted. I often romanticize the magical kingdom of microwave circuits fabricated with lots of sparkling silver and gold plating.

[Radio Wrangler]

Yes, the PIN diode is a two headed beast, it segues from being a less than impressive rectifier diode into being a DC current controlled resistor at a frequency set by the carrier lifetime.

I've had to do induced lightning transient protection for receivers and transmitters RF lines as well as power supply inputs and control lines.

Often you have to split the problem into several layers, a fast protector without much endurance close to the device being protected, and then moving towards the source of the threat progressively beefier but slower protection devices. The fast but light device takes the first sting, then its beefier pals get in on the act and reduce the power hitting the small but fast protection.

I don't have any protectors, I'm just VERY careful to use sensible amounts of attenuation... Big fat attenuators outside the analyser. Beware of modern microprocessor controlled analysers, some have latching internal attenuators and even with the analyser off may leave input attenuation and switching as it was last used. If someone had all attenuator sections out, the mixer can be destroyed all too easily before you even press the on button.

David

[G0HZU_JMR]

But the PIN limiters seem to very difficult to design and not always behaves what you want it to ....

Based on the AN1050 app note and the HSMP-48x datasheet, it should be possible to crudely predict the thermal performance of the 4-diode limiter when used at frequencies where the diodes have adequate carrier lifetime.

If they behave as pure resistors and the leakage power at 5W incident power is (say) +12dBm as indicated in AN1050 then the resistance of the diodes can be approximated to be about 6 ohms each.

This forms a potential divider with a 50 ohm source. If the four diodes are modelled as a single diode, then the resistance will be 1.5 ohms. This is a VSWR of about 33:1.

The forward power transfer to the load will be just over a tenth of the incident power in this case. At 5W incident power this means that 0.6W gets dissipated in the diodes and about 15mW of this ends up in the 50R load after them. So, each of the four diodes would be dissipating 0.6/4 = 150mW as heat.

The datasheet for the HSMP-482B states that the SOT323 packaged version has a Jc thermal resistance of 150degC/W.

So, the diode junction might end up 23degrees hotter than the case/pin temperature at 5W incident power. This is quite impressive! I think the 1N4148 diodes will run hotter than this. I'm not sure what happens to the HSMP-482x diodes when this test is repeated at much lower frequencies where the carrier lifetime isn't adequate.

[regenfreak]

Originally Posted by Radio Wrangler

Yes, the PIN diode is a two headed beast, it segues from being a less than impressive rectifier diode into being a DC current controlled resistor at a frequency set by the carrier lifetime.

I've had to do induced lightning transient protection for receivers and transmitters RF lines as well as power supply inputs and control lines.

Often you have to split the problem into several layers, a fast protector without much endurance close to the device being protected, and then moving towards the source of the threat progressively beefier but slower protection devices. The fast but light device takes the first sting, then its beefier pals get in on the act and reduce the power hitting the small but fast protection.

I don't have any protectors, I'm just VERY careful to use sensible amounts of attenuation... Big fat attenuators outside the analyser. Beware of modern microprocessor controlled analysers, some have latching internal attenuators and even with the analyser off may leave input attenuation and switching as it was last used. If someone had all attenuator sections out, the mixer can be destroyed all too easily before you even press the on button.

David

Modern spectrum analyzers often use ESD or PIN protection diodes + DC blocker cap as first line of defense, then you have the RF PIN diode switch for the pre-amp and another switch for attenuator acting as "sacrificial buffers" for the failure modes.

Murphy Law 1st corollary: the first and second lines of defense usually fail simultaneously as a consequence of someone's stupidity or complacency.

The Murphy's 2nd corollary is that the third line of failure protection occasionally fails as a result of Force Majeure, human error or design thinking-e.g. the P-63 smashing into the B-17 because the Bell engineers decided to move the P-63 engine into the middle, cramping the avionics panels because of the side doors, tricycle landing gears, making the nose long and thin to reduce drag, and reducing the cockpit downward visibility. Then the pilot flew too fast in a turn...

The Siglent SSA 3921X plus uses GaAs FET digital step attenuators instead of resistive attenuators:

https://www.mouser.com/catalog/specs...mc307qs16g.pdf


I built a high voltage RF protection filter called Terry filter for neon sign transformers in tesla coils:
http://www.hvtesla.com/terry.html

It uses multiply layers of protection as the neon sign transformers are expensive and known to be "snowflakes". I have some experience using metal oxide varistors and GDTs when i built tesla coils.

So, the diode junction might end up 23degrees hotter than the case/pin temperature at 5W incident power. This is quite impressive! I think the 1N4148 diodes will run hotter than this. I'm not sure what happens to the HSMP-482x diodes when this test is repeated at much lower frequencies where the carrier lifetime isn't adequate.

The HSMP-482B is obsolete and hard to get now.
5W is about 37dbm, a fair bit of power. At the moment, I am not the kind of person who would like to measure the harmonics and output from transmitters and transceivers with a spectrum analyzer. I would leave this kind of adventure to the enthusiastic HAM hobbyists. If I do, I would probably use a properly rated directional coupler.

[G0HZU_JMR]

Yes, I'd always recommend using a decent RF attenuator or a coupler when measuring a high-power transmitter.

For an added layer of security, I also fit a limiter right at the analyser input in case I get lazy and make a mistake with the coupler or attenuator connections or if the coupling factor of the coupler is not known for out of band signals. If the coupling factor is lower out of band then this can cause damage from out of band spurious oscillations from any amplifier under test.

At work, most damage occurs from the application of a DC voltage at the analyser input rather than high power RF. This can happen if a blocking cap fails or is missing or if it gets shorted by clumsy probing. I've never damaged an analyser throughout my career but plenty of my colleagues have...

[regenfreak]

Originally Posted by G0HZU_JMR

Yes, I'd always recommend using a decent RF attenuator or a coupler when measuring a high-power transmitter.

For an added layer of security, I also fit a limiter right at the analyser input in case I get lazy and make a mistake with the coupler or attenuator connections or if the coupling factor of the coupler is not known for out of band signals. If the coupling factor is lower out of band then this can cause damage from out of band spurious oscillations from any amplifier under test.

At work, most damage occurs from the application of a DC voltage at the analyser input rather than high power RF. This can happen if a blocking cap fails or is missing or if it gets shorted by clumsy probing. I've never damaged an analyser throughout my career but plenty of my colleagues have...

I guess it would be prudent to sweep the unknown coupler with a VNA first and use a directional coupler with the coupled power at least 20-40db below the input power. The high directivity and isolation offer a layer of protection. However. the directivity only maintains if the port 2 is properly terminated. For 3-port couplers, the isolation port is usually terminated internally with a 50 ohm load with maximum power normally rated much less than an external dummy load. If a muppet accidentally switches on the transmitter before he attaches a dummy load to port 2 of a directional coupler, all the power will be reflected back to port 1 and 3. Potentially, you would have a double whammy; toasted transmitter and spectrum analyzer.

No directional coupler has perfectly flat frequency response in coupling. I anticipated that it is necessary to obtain the coupling S31 sweep with a VNA for power measurement correction (or maybe just use the datasheet for the coupler if it is available).

If a coupler has high isolation, it is desirable to use them for two-tone testing of receivers instead of using a power combiner. However, the coupled port requires lots of juice from the signal generator to compensate the attenuation through coupling. As you excitedly pump more power from the sig gen, it may produce more intermodulation junks to screw up your measurement. So no matter what, you never win.

[regenfreak]

A while ago, I acquired the RF and front IF modules for both Revox B285 and B261 FM tuners. It is challenging to power them up as they require awkward input voltage; -15V, +15Vand +33V for the B261, and +16.5V, -16.5V and +36V for B285. For example, I would have to use my Farnell Triple Tops 3D for the -15/+15V and get the +33V using piggy bank of batteries instead of taking the risk to use my Heathkit HT regulated power supply. On top of that I would need a frequency synthesizer for the LO input. My FY6800 arbitrary function generator can only go up to 100MHz.

So i have constructed a no-frills 90MHz-210MHz VCO signal source using Mini-Circuits, POS-200+. Both Vcc and Vtune are powered by battery packs to reduce phase noises. Tuning is controlled by a 10-turn 10k pot. The spec states around 10dbm output. phase noise -102dBc/Hz at 10kHz and 2nd harmonics -28dbc at 100MHz.
I will add a cheapo digital attenuator, RF amp module and a DIY low pass filter, then I will have a simple frequency source to test and study FM broadcast receivers.