Help me repair an Advantest R3261 Spectrum Analyzer (first time repairing a SA)

[Ghanou]

Hello everyone,
I recently got an Advantest spectrum analyzer from a friend of mine for a decent price, the unit shows a low input level of about -40 dBm for a 0dBm signal up to 1 GHz (did not have a higher frequency gen at the time) (check first attachement), i thought at first it surely must be an attenuator problem or some small resistor or diode at the input, so i ordered a tinySA Ultra and a NanoVNA V2 plus 4 to investigate further.

fast forward to when they arrived, i firstly connected the nanoVNA to the input of the SA and i found out that it indeed shows a 50 ohm impedance through the input with minimal return loss, so that confirms its not a mechanical problem such as a connector/solder joint problem (check attachement 2). i then followed the maintenance instructions from the service manual, and i checked the signals from IF, 3rd and 2nd converters, as well as the YTO (LO), they were all lower level except for the YTO which was around -7 dBm, which is a bit low because they specify it at +8dBm, but its not so bad that it would create this fault. right!

i then opened the unit up and started testing stage by stage, i firstly tested the attenuator by inputtting a signal and probing the output with different attenuation settings from front panel, i found out it's working correctly, next is the LPF, also works correctly, i then desoldered the 4 GHz LPF and measured it's frequency response using the NanoVNA, i firstly thought that it was the problem because it had a weird response at lower frequencies but looking at around 4GHz where it actually operates, it shows an insertion loss of about 10 dB which seemed fine, i believe the NANO is a bit limited at the 4.4Ghz so it doesn't show its roll off. i then so i soldered in its place a COAX cable to simulate a straight through, but then an error/warning apeared on the SA saying something along the lines of "calibrated response altered" or something like that i can't really remember, so i understood that the calibration coefficients stored in it's memory do actually match the LPF response, then it must be working correctly, i then resoldered the LPF and now the signal dissapeared completely  .

i thought to myself, well a completely non working thing is better, then a half assed one ,right  .

I opened the RF shields completely to see if there are any burn marks or anything unusual visually, and everything seemed fine, i then wanted to test the first mixer/coupler, so i set the SA to 700 MHz center freq, and zero span, and inputted a 700 MHz -25 dBm signal with no attenuation, i measured the signal right at the mixer input and it looked okay. i then measured the LO right at the diode input and it also seemed fine @ 4.766 GHz  (a bit down but it should work), and then i measured the output IF signal @ 4.066 GHz and it was way way down @ -56 dBm right at the diode output and -67 at the 2nd converter input. (check attachement 3 for measurements summary)

this led me to believe that the mixer diode is bad since the mixer should output around -10 dBm or so. which explains the 40dB drop, i then measured the diode using my HP34401 in diode mode, and it always gives me 0.11 V no matter the orioentation, expect for the top and bottom pins which are connected at DC by a ground plane that give a short.

since this is my first RF instrument repair i'm a bit hesitant for the next step wheather it is to desolder/ replace the diode or to try something else first, so any suggestions / tips are welcomed.

[feedback.loop]

So you are saying that LO should be +8dBm, you are getting -7dBm, and this is "a bit low" and "not so bad"?
LO is 15dB down. No wonder the mixer does not mix much.

[Ghanou]

So you are saying that LO should be +8dBm, you are getting -7dBm, and this is "a bit low" and "not so bad"?
LO is 15dB down. No wonder the mixer does not mix much.

thanks for the reply,
well i'm using a homemade "probe" which is just an RG58 cable with an SMA connector,  so i'm loading the outputs when i probe down, plus i'm measuring the signals at 4 GHz so i suspect the attenuation from the probe to be atleast 10 dB at these frequencies, i will try to inject an LO from my NanoVNA without powering the SA and check again (i just wish i ordered a second tinySA to use it as Gen), if there is any output. will update you shortly.

[Ghanou]

update:
sadly, the NanoVNA can only output a maximum of -13.5 dBm around 4 GHz, plus the signal coming out of it isn't the cleanest, so i won't be able to test with it.

[Ghanou]

update :
i desoldered the diode and it measures 0.6 V correctly , here is the schematic for the mixer.

[Ghanou]

update:

i've tested the YTO output directly from the SMA connector and it neasures a good +15 dBm, so no problem there, so if the LO is good, the input is good and the diode is good, why else would the output of 1st mixer be so low ?