Jus tbefore new year I found a ebay-Kleinanzeige that offered this HP3577A vector network analyzer. When it appeared on the market in the mid 80s up till today this analyzer set the benchmark regarding linearity and dynamic range. It spans the bandwidth from 5Hz to 200MHz. Also it features three identical input channels so that all four S-parameters can be measured. A optional S-parameter test set allowed for convenient measurement procedures. Unfortunately the test set was not part of the offer. I can add it later on or use something DIYed.

On my request the seller performed the ´confidence tests´, a row of automated self-test routines that check the basic functionality of the VNA. All three input channels reported ´Failed Receiver Impedance test´ and the main R-channel also a ´Log- and Linear Sweep signal level test´.

Initial research suggested a rather small issue, like roasted input stage resistors. But it could also have been a more serious issue with high follow up costs. So I made a reduced offer and won the bid

First a couple of pictures.

The analyzer is built really very elaborate which adds to the impressive weight of 28kg.

The different plug-in boards are situated in their own screened chambers made from 4mm thick aluminum. There is even additional scrrening on some of the boards itself.

The traces are mainly gold plated. Everything appears excessively solid and oversized

Now I undertook the input boards a closer inspection

The three round silver casings of TO-5 size are the HF-Input-Relays (>50€ each)

Luckily all of them are fully ok. Close to them one finds the resistors and capacitors of the input attenuators for 1MOhm and 50Ohm input impedance. Quite often the resistors get roasted by too large input signal levels. The resistors here measured absolutely ok. So the failure must be something different.

The error message ´Receiver impedance test´ comes up when at least one of two test procedures fails. For one the level between switching from 1Mohm to 50Ohm shall differ by 6dB and no more than max. +-0.2dB tolerance and for second shall the display level between switching from 0dB to 20dB attenuation differ by less than max. +-0.1dB.

When performing these tests it showed that the inputs A and B slightly surpassed the +-0.2dB limit. Input R though showed a difference of more than 3dB.

There are blue trimpots at the top side of each plug-in module. R78 is marked ´High Z Gain ADJ´ and R71 is marked ´50 Gain ADJ´. A deeper study of the service manual confirmed that here the 1MOhm and 50Ohm levels could be adjusted.

Initially though I checked the level of the output board against my function generator.

The required -30dBm matched very precise.

Then I adjusted the levels of the input boards with the two potis. When I merely touched the associated poti of the R-channel with the tuning stick the level jumped from ca. 3dB to almost the required 6dB. (note: the wiper of the poti could have mild corrosion) After the restart the analyzer bootet up properly and passed all confidence test without reporting any errors ... so it´s a complete success.

Been lucky again and got a great measurement device for a small price

Presumably one or the other parameter has to be adjusted in the refurbishing process, but it gives a nice feeling already that the analyzer is basically functioning properly.

More than one third of the casing is filled with the dispay tube and it´s asssociated complex electronics. If one day this assembly dies, then there is a replacement possible in form of the Newscope5 LCD-display. The LCD-display is a colour display and is much sharper in its presentation of traces and signs than the original green-cloured tube display. It also builds much smaller so that the very loud fan might be replaced by a smaller and softer type. Apart from the display, the motherboard and Memory board all the screened subassemblies are almost closed so that the fan won´t be able to cool them anyway.