HP 3588A Spectrum Analyzer Repair (Relays + Memory)

[Smoky]

I bought this HP 3588A Spectrum Analyzer a couple of years ago to help with future radio repairs.



I rebuilt its power supply early-on along with the 35665A, and 35660A power supplies since they were the same. All three supplies work and measure perfectly.

The 3588A spectrum analyzer started to show "calibration error" messages during operation. The spectrum analyzer self-calibrates while it's in use and it began to show error messages every now and then on the screen.

I ran Long-Confidence tests and found random failures occurring during the process (Receiver Autorange, Source Flatness, Receiver ADC, Receiver 2nd IF, etc.). Sometimes, the errors were different. Other times, the analyzer would pass the complete test. Something wasn't right:



I read a few repair articles about this spectrum analyzer on the web and most relate to the chasing-down of errors all over the machine. I wasn't going down that hole since I know very little about these machines.

Last year someone from the Youtube channel "ExclusivelyHP" posted a video suggesting that these errors were due to faulty relays. He traced the signal through the A11 board's input relays. He demonstrated how the signal would drop-out in between the relays. He would tap the relay and the signal would reappear

I contacted ExclusivelyHP through e-mail and he hooked me up

Just in time too, because when I went back to his Youtube channel recently, it was gone! A statement reads "This account has been terminated due to a Trademark claim by a third party."

[Smoky]

Here are the original relays. The three dark brown relays are RG2ET-12V-H11 and are the troublemakers. The H11 designates them as a custom HP relay and unobtanium:



The new relays are much smaller and look nothing like the originals. But the specs on these little guys best the original parts:



The seven smaller ones are RK1E-12V relays, and still available, so I replaced them just for good measure.

Here's a shot from the underside of the A11 board just beneath the relays:

[Smoky]

I cleared away the old relays using my new Quick hot air station and a Hakko 808 de-soldering gun. The copper is heavy on both sides of these boards and they need pre-heating:



ExclusivelyHp designed these adapter boards to allow the Panasonic relays to work perfectly with the A11 board:



To secure the adapter, I used gold-plated pins from Mouser. I used my smooth pliers to create a flat on them to keep them above the surface for easy soldering:





Once installed, silver-plated Teflon jumpers were attached to provide power:

[Smoky]

I was extremely critical with every single solder joint under a microscope (Leaded/Silver/Kester only too).

I ran the Long-Confidence tests a few times over and it's absolutely error free.

Even the calibration correction numbers are tighter.

I put up a video on Youtube today of the 3588A going through its Long-Confidence test:





Now it's off to the memory upgrade attempt

[Smoky]

Here I'm treading into no-man's land. The memory board uses Texas Instruments TMS4C1024-10N 100ns 1Mx1 RAM. 16 pieces are installed and 16 slots are solder filled. I scoured the internet and looked all around and could not find a picture of this memory board ever fully populated. Can it be done?



The pictures are large so that you can see the details. All of the traces cascade down from the populated RAM through the un-populated areas to where is collects towards the end of the PC board. Does this mean RAM can be added without any modification elsewhere? The factory installed yellow capacitors are there to make me think that's the case.

I cannot find a menu within the spectrum analyzer that states the amount of factory RAM installed. All I see is a message on start-up reading "testing RAM" and that's about it. The spectrum analyzer then goes on its merry way.

Here's a picture of the board where I carefully removed all of the solder in the un-populated areas. I also started to install 18-pin Augat sockets:



...and here's a picture showing showing all of the sockets installed:



The memory board slides out of the analyzer with ease. It has just one edge connector. My plan is to slide the board back in without any extra memory put into the sockets. I want to make sure that I didn't disturb anything and that the analyzer functions normally. I will also try one more time to find some hidden menu relating to the memory.

Next, and if all is good, I will pull the memory board back out of the analyzer and install the 16 extra pieces of RAM. I'll power the analyzer back up and see how it reacts

Get a load of this. Back in 1990, adding 2MB's of RAM to this machine would cost you $1,500



This machine has the high-stability reference and it was half the cost of a memory upgrade!

[Smoky]

This should wrap up the repair and modification to this analyzer

Here's a picture of the new RAM installed:



The part number of this memory board is 35672-66587:



Unbelievably, the schematic of this exact memory board was printed in the service manual and it shows all of the RAM positions populated. I put a red dot next to the memories I installed. The location of every piece of memory is printed on the PC board:



I uploaded a new video on Youtube of the memory test:



It appears everything is working properly

Here's one thing I did notice. I went back and watched the first video of this analyzer after the relay repair. I timed how long the memory test took to "Pass" with the 16 pieces of factory-installed memory. It took ~16 seconds.

I then timed how long the memory test took after the added 16 pieces of memory were installed in the new video. It took ~31 seconds.

I wanted to thank ExclusivelyHP for sending me the relay adapter boards

Next is Mr. Paul Carlson's custom made DC Block for the input