A diode for RF power meter (1.8-470MHz, 100W)

[iMo]

Hi, I got two old parts from scrap - a dummy load (50W up to 2.7GHz, N-connector) and a diode detector (a T-shaped module with 2x N-connectors and 1 BNC voltage output). Inside the detector was a diode (looks like the old germanium point-contact one in glass, DO7 package I guess, it was broken however, and a capacitor at diode's output against GND, DC signal wired to the BNC). No special markings on that part.
I wanted to make a simple RF power meter from that parts, and I replaced the diode with a vintage GA204 in the same package (a Ge detection diode, Vr=100V).
It worked fine at aprox 100W output from my transceivers, showing stable and reasonable DC voltages (88-107V) on my DMM in the 1.8-50MHz bands.
At 145MHz (50W) the voltage was only a half of the expected, at 440MHz (35W) it got crazy showing around 0V. After that I again tried with HF and all voltages were lower by some 4-5V. So it seems the diode got degraded after it was exposed to those aprox 60Vpeak @440MHz.

Is there an available diode to be recommended for this application as the replacement?
People wrote about 1N5711 (Vr=70V), 1N4148 (Vr=100V), all other diodes I saw are for lower reverse voltages or not easily available.

I think the Vr of the diode should be at least 100V as the Vpeak is 100V with 100W at 50ohm load. And it should be fast enough such I get it working till 70cm  ..

[Uunoctium]

I tried to build an sensorhead for my Boonton 4200. There are tons of examples to diy’ing such RF-powersensor/powermeters. One used BAT62 with success. Another recommendtions are1N5712, HP5082-2800, HP5082-2811, HP5082-2851 …

5082-2820 did’nt work in this application, because it is not an zero biased diode at all. At least i was fine with HP5082-2835 (an zero biased shottky)

[iMo]

It seems that in my application the min Vr of the diode should be 200V, not 100V as I mentioned above. The peak reverse voltage at the diode is 2*Vp. That limits the choices.. 
@Uunoctium: yea, but all those diodes are low voltage, afaik..
Not sure my above design is feasible for 100W at the dummy load..

[bd139]

Use a 40db power tap before the 1n5711 detector (3x 820 and 1x 51 ohm resistor)

https://ae6pm.com/SCCARA-GRAM_Articles/A_40_dB_Power_Tap.pdf

[Uunoctium]

From the 80'? R&S (System Neuwirth) used BAT16-046

[iMo]

@Uunoctium: yea, but that is a different design - they are using coupler, thus the voltage at those diodes is lower.
I've been using full voltage of the signal there..

@bd139: thanks, I was also thinking to put a divider inside, but the space there is pretty limited.
I think no need to care on the matching in my case (as in your example) as the output off my BNC is high impedance and DC.
I will try with some smd parts, perhaps 2 smd resistors fit inside the diode module somehow..

Something like this:

[Uunoctium]

flat black round = diode (unfortunately without marking). R's are 100/68/100
(did'nt proof their math, but should meet Z0 = 50)

[iMo]

 
Even my schematics with the divider above will not work as we talk 100W on the 50ohm.
The smd divider Rd1 and Rd2 will evaporate with 100W on the load.. 
A capacitive divider could be a better choice, imho..

[bd139]

Hence 40dB tap. Less melty.

[iMo]

Hmm, even his 40db tap with that 2k5 load (and 50W) would be pretty hot..
50W on 50ohm is 140Vpp on the 2k5 resistor.. (1Watt).

[Uunoctium]

i still can't believe the the original  purpose of your device is an straight through powersensor, capable 100W+.

Can you post a picture?

[iMo]

Nope, it is not the original purpose. I took 2 random scrap yard telco parts - a dummy load 50ohm (50W continuous, up to 2.7GHz, with a big heavy heat-sink) and the diode detector (unknown rating, but UHF), both with N-connectors and I put them simply together in series. The schematics as on my first picture above - 50ohm load, diode, capacitor - that is all. Having a 200V UHF diode I am a happy ham 
I will make a shot tomorrow..

[iMo]

Here is the setup..

[Uunoctium]

I dont know any advice.
But I also don't know any diode sensor that can directly measure 100W + RF.
All diode based ones work in a range of ~ -70...+15 dBm. The output is from a few µV to a few mV and logarithmic, so you need a log-amp over 4,5,6 decades. Or by software.

Direct measurement is best via a coupler.
(the one shown in the picture above has an attenuation pad and was still connected via an attenuator. I own an similar with coupler for return loss and output circulator. Both are ~ 2140Mhz = LTE from a BTS).

[iMo]

OK, I've made the measurements again, as the GA204 recovered somehow - I've learned a short measurement like 2 seconds max will give a stable and repetitive results, after 2 secs the voltage drops as the diode starts to heat up (or something else happens). Measurements with voltages up to 50V give stable results for an "unlimited time". That corresponds to the fact the GA204 (point contact Ge diode) has got Vr_max=100V. A miracle the diode can somehow survive 2x higher Vr.. 
Here are the results with the measurements with my two transceivers. VHF/UHF frequencies at 50/35W do not work with this diode.
Next week I will replace the GA204 with 3x 1N5711 in series (provided I will be able to put all the stuff inside the probe).

PS: the Volts measured at the detector to the PA's Watts conversion (with the 50ohm load and the Ge diode) is as follows:

P = (Vdet + 0.3)^2/100  [Watt, Volt]

[iMo]

I've replaced the Ge diode with 3x 1N5711 in series (see the shot below) and it seems it is a clear winner.. 
With Vr=210V it works fine up to 150W on HF, rock stable readings.
At 2m and 70cm the power shown fits my expectation too..

And the calculation with 3x 1N5711 in series:

P = (Vdet + 1.05)^2/100  [Watt, Volt]

[bd139]

Interested to see how long those 1n5711's last. I've found them to be quite fragile managing to blow them up in fairly low level mixers. Respectable unknown ST branded ones from from China via a respectable distributor as well, not unknown ones from China from Aliexpress 

[iMo]

I've bought 20 of them today (0.12E each in a local shop) so enough material for a replacement ..
It survived 250V for 10secs..

[TheMG]

Any particular reason you want to directly measure such high RF power levels? A far better approach is to use an in-line attenuator, sampler/sniffer, or directional coupler to get the signal down to a more manageable level.

[iMo]

No specific reason except that to build that pretty precise wattmeter for all bands till UHF took me 20minutes and cost me 36 cents

[joeqsmith]

At 432MHz, 100pf is under 4 ohms.   Add the cable.   

With it driving a meter, do you even need the cap?   If you did use the resistor divider you show and up the values to get the power down, would it still be linear?   You would need to handle the voltage as well but I don't think that would be a problem.   If you wanted to run lower power levels you would have a problem. 

[iMo]

Out of curiosity I've made a simulation of the measurement system, based on the ST model for the 1N5711 (there is a lot of models for the 1N5711 with quite large differences in the params).
This is what comes out:

Code: [Select]


200Vpp input signal at 50ohm

Freq [MHz]       DC Voltage drop [V] with 3x 1n5711 (simulation, ST model)

 1 1.03
 10 1.05
 50 1.18
 150 1.45
 450 2.35
So I have to correct my power calculation tables above a little bit, with the biggest difference at 2m and 70cm - the output power being higher than calculated (I was using 1.05 Vd_drop).

Power = (Vdmm + Vd_drop)^2/100  [Watt, Volt, with 50ohm load]

At 432MHz, 100pf is under 4 ohms.   Add the cable.   

With it driving a meter, do you even need the cap?   If you did use the resistor divider you show and up the values to get the power down, would it still be linear?   You would need to handle the voltage as well but I don't think that would be a problem.   If you wanted to run lower power levels you would have a problem.

The capacitor and the BNC cable are at the DC part of the circuit, the Xc plays no role here (a detector).
What is important is to use DMM with 10Mohm input (or 1Mohm with larger cap).
The smd capacitor in the probe looks like something small like 10-100pF (forgot to measure) for higher voltage. Perhaps you do not need one with coax at the BNC..
Nope, I do not use a resistive divider, signal goes straight to the 3 diodes in series (see the schematics below).
Lower power levels - tried with my 2m/70cm handheld and got 16.5/17V with quite empty batteries (did SOTA on Sunday), that is 3.22W/3.74W output - so it works with QRP too..

For really precise measurements you have to calibrate the power meter from say 3Vpp up to 200Vpp (90mW to 100W) input signal and from 1-450MHz in several steps.

[gf]

Out of curiosity I've made a simulation of the measurement system, based on the ST model for the 1N5711 (there is a lot of models for the 1N5711 with quite large differences in the params).

I wonder what's the current through D1/D2/D3?
Can the initial current hurt the diodes or the smoothing capacitor?
How much AC curent is still flowing though the diodes (due to their capacitance) after settling to a stable state? (particularly at UHF frequencies)
I also wonder if the source V1 is really supposed to have zero output impedance?

[iMo]

Out of curiosity I've made a simulation of the measurement system, based on the ST model for the 1N5711 (there is a lot of models for the 1N5711 with quite large differences in the params).

I wonder what's the current through D1/D2/D3?
Can the initial current hurt the diodes or the smoothing capacitor?
How much AC curent is still flowing though the diodes (due to their capacitance) after settling to a stable state? (particularly at UHF frequencies)
I also wonder if the source V1 is really supposed to have zero output impedance?

Below the current through the diodes and the capacitor (red) at 450MHz. The current surge lasts for 2us with a 1nF or 200ns with a 100pF loading, so perhaps it does not hurt the diodes much
The smaller the capacitor the safer life for the diodes..
I need 200Vp-p on the 50ohm resistor (100W on the dummy load R1), so it does not matter how it is created in this scenario, imho.. But sure the impedance of the V1 in the real situation is something like 50ohm (TRX output Z is something like XX+/-jYY).

[joeqsmith]

The capacitor and the BNC cable are at the DC part of the circuit, the Xc plays no role here (a detector).

Sorry, I had just woke up and no coffee.

[gf]

Below the current through the diodes and the capacitor (red) at 450MHz.

150mA_pp - that's quite a bit of current. Or about 3.5W reactive power if I'm not wrong.
And I guess also ~100...200mW (?) real power, heated by the diodes.
That's not so little for the fact that it's "just a high-Z detector"

[iMo]

From the simulation: With my max output powers
35W at 432MHz     Idiodes = 32.6mA rms
50W at 145MHz     Idiodes = 12.1mA rms
100W at 50MHz     Idiodes =   5.5mA rms
155W at 3.75MHz  Idiodes =   0.5mA rms