Repair : 3588A Spectrum analyser

[free_electron]

Got this one dirt cheap at Halted during their annual sale. 200$ .. broken. a quikc power up in the store confirmed the machine had a nice and bright picture tube , no burn in , powered up fine but failed self calibration. Putting an end of solder into the bnc and setting input impedance to 1 Meg and the center frequency to 98.5MHz ( local radio station) with a span of 1 MHz and i picked up the transmission just fine. My guts told me : that's a blown 50 ohm path... some dodo sent in too much power and fried the input stage / attenuator. That should be fixable.
so i bought it.

These machines are nice. it does 10Hz to 150MHz AND it has a tracking generator with calibration sweep meaning you can sweep filters with this thing. this particular machine has instrument basic option installed so you can run programs on the machine to post process and visualise data !
attentive 'regulars' will feel familiar with this machine. Dave did a teardown/ repair of a dynamic signal analyser that looks like it could be the twin of this machine. in fact both machines are built from the same chassis. the cpu board, memory board , power supply , fornt panel and monitor are identical between these machines. only the front ends and firmware differ.

Arriving home i ran the extensive self-test of the machine that uses a 'secret' menu. hit special functions , hit second softkey from the bottom ( no label ) , type -99 , hit second softkey again and now you get an additional softkey called 'Service'.

following the test-tree i had my girst failure at the 'input stage'. this confirmed my suspicion that there was something wrong in the input block.
screwdriver time !

Downloading the service manual showed how to open the machine and get access to its guts.

http://www.siliconvalleygarage.com/eevblog/3588/HP_3588A_Service.pdf

A quick check on the output connector J3 , with a scope, of module A11 confirmed there was no signal using the 50 ohm path. Flick to 1Meg and signal passed just fine. Time to whip out the multimeter and schematics and do some checks. first off : the component level schematics of the 3588 are not available. There are some block diagrams and fault trees but even the block diagrams are wrong ! so i hand crafted a diagram of the input section


click here for large version :
http://www.siliconvalleygarage.com/eevblog/3588/3588a.png

Following the traces was easy enough. during this nbit of reverse engineering it became immediately visible what had happened : resistors had popped



the two 100 ohm resistors sit in parallel and form the 50 ohm termination. those were toast.
R102 103 and 104 form a 5 db attenuator leading into the programable attenuator block. the 1 meg amplifier loses 5dB as well. this is compensated by a 5db gain stage before w eleave this input section card.

147 ohm on both sides of the PI attenuater . a quick calc later and this had to be a 38 ohm resistor. The 3 was still readable , and on another part of the board i found an identical 38R3 ohm fusable resistor witht the same font. bingo.

i didnt have fusable ones. so i soldered in 1210 package 200 ohm resistors. stack two on top to get to 100 ohm. since there are now 4 resistors instead of two and they re wider body they will survive a bit more power.
the 147 ohm resistors were swapped out for a 150 ohm and the 38.3 was replace by two 75 ohms in parallel. close enough. the calibration routine of the machine can compensate for that without problems. Of course i am going to order the correct fusable ones for the attenuator. i'll try to find a 2520 package 49r9 or something that can fit the board but dissipate a bit more power. for the 50 ohm input resistors.

power up and self test later the diagnosis was confirmed. all self tests now passed and the machine selfcalibrates without problems. it measures spot on.


the 1 megaohm amplifier. this one has an identical 38r3 resistor.


the boards are clearly labeled which makes tracing them a joy.


the card nest with all the rf interconnects. don't make a mistake here !


the main memory board. this one has a lithium cell. still perfectly fine at 3.6 volts


it's alive !


and instrument basic works too !! yay.

[ludzinc]

Well done!

Now can you fix Dave's  DSA for him?

[HighVoltage]

i didnt have fusable ones. so i soldered in 1210 package 200 ohm resistors. stack two on top to get to 100 ohm. since there are now 4 resistors instead of two and they re wider body they will survive a bit more power.

What is the purpose for not installing such a higher power resistor in the input from the factory?
Then may be more SA would survive a high power burst.

It is nice to read your step by step analysis and learn from it.
Thanks for sharing this repair.
 

[KJDS]

i didnt have fusable ones. so i soldered in 1210 package 200 ohm resistors. stack two on top to get to 100 ohm. since there are now 4 resistors instead of two and they re wider body they will survive a bit more power.

What is the purpose for not installing such a higher power resistor in the input from the factory?
Then may be more SA would survive a high power burst.

It is nice to read your step by step analysis and learn from it.
Thanks for sharing this repair.

The general idea is that you don't apply too much power to an instrument, however if it does happen then you either need to put a limiter and relay in the path or make sure that what dies is cheap and easy to fix.

If the input resistor burns out and nothing behind gets damaged then you have a cheap fix. If a bigger resistor doesn't burn out then the first mixer dies and you have an expensive and rare component to fix.