Fixing GH2462 Directional Power Meter
2022/4/1
GH2462 is almost a 1:1 china replica of R&S NAUS-80. Identical performance and struct. I got 2 of them, both broken. The fist one was easier to fix, even though the investigation is little bumpy.
The second one had a broken sensor, is obviously challenged to fix , and I have no identical detector diode as replacement.
Fault finding
The sensor is well constructed, RF shield everywhere. The detector show a very flat frequency response, from 30Mhz to 1Ghz, the detector output almost no change(visual check the needle).
This is unbelievable, the detector diode is not possible response like this flat.
A obvious fault is the N connector, one of them is broken, need fix. It's luck the connector could be take off from the body. let the hot-gun blow for a very long time, heat all the connector up to melt the solder, fixed it.
The detector diode is in the right hole
everything is silver coated
This diode is broken, no replacement available. I don't even know the model number.
In circuits check the diode is almost impossible, cause both of the 2 diodes DC shorted to ground, and kind of parallel together.
For a temporarily fix, and also check it's frequency response, a 1N60 installed.
Things not that easy.
Fault afte Fault
The detector had a cold joint, that's because the central rod which sens the RF energy, is too good for heat conductive. Soldering that joint need to be very patient and take very long time.
This silver coated board form a capacitor as part of RF detector
But, the under layer meca is broken, short it to ground.
And the left joint, connect the silver belt to central rod, is cold joint.
Beside the cold joint, a more serious problem is the short circuits cause by a broken mica sheets. Fix this is impossible. There is no way i could had a available mica sheet. The polyimide tape could do same thing, might alter the capacitance a little, and maybe temperature stability worse than Mica. Anyway, I gonna do that.
The Longest Faulty fight
Yeah, seem I fixed everything till now. Repack test it. En..., frequency response is shit, there is almost 20% change from VHF to UHF band. And the detector is so sensitive, almost 300x larger amplitude response than the original one.
I kind of forgot why the response is so shit, and how I fix it. I remember changing the diode to every kind i had, even the HSMS-2850. None of them work properly.
Wait a moment before i tell you the awesome story. See the load resistor on the right? One of them is broken, shit.
Soldering that one is nightmare, big silver metal.
I finally change both diode to 1N60, and the response is identical for both arm. I kind of guess the diode sensitivity could be adjust by replace the sensen capacitor, the 2 block on the sense rod.
One of the block is tear down.
I had to say, yes , it works somehow. One month elapsed, no progress at all, beside the frequency response, I hit another severe problem, the 2 arm loss the directional functionality!
Success
I finally understand how the detector works. The 2 gold coated rod slit by the 2 block. These 2 block form a couple capacitor.
The whole sensor is kind of 'balance mode' of the right one. BTW, the 4 port block is detailed explained by math in << microwave project>>, that's too unbelievable complex to state by a simple explain why it is directional.
Anyway, without the control block, both end detect same amplitude signal even under 50 ohm dummy load terminated, lost it's all directional.
Balance mode: the central couple set up a virtual reference point
unbalanced mode
While under why it lost the directional, everything seem quick on track: I find a broken 1n60 (mysteries broken... ), and everything then seems work!
Frequency response is also back online. Sensitivity still high, but not that dramma, only 3x times than original diode, It's acceptable, now it's full scale is 300W, instead of 1000 Watt.
Sensitivity of 2 arms is little unbalance, but it's easy to adjust the diode load resistor to correct. I highly suspect that the way too high sensitivity lead by miss test method: the cover is not totally seal while test the response.