Take its frequency accuracy and stability as an example. Its frequency has a 11 digit adjustment range that gives a frequency adjustment resolution down to 0.000001 Hz at frequencies below 100kHz, and 0.001Hz at frequencies up to 60MHz. It has a rock stable, calibrated output, which has an impressive adjustment range of 1mV to 10Vp-p. The low side range is very useful when working with high gain devices. Other function generators require the use of external attenuator to produce a signal this weak.It is an excellent machine, notwithstanding its huge size that takes so much space in my crowded workbench.
The HP3325A is one of my most useful lab equipment. That is why, when it finally conked out, I wasted no time and immediately started repairing it.
I always get mixed feelings whenever one of my test instruments fails. There is of course a feeling of anxiety over the possibility of losing this jewel forever. But, believe it or not, there is also a feeling of excitement, because this gives me a self excuse to explore deeper inside my instrument, gaining practical knowledge along the way.
Round 1, Output level calibration failed error
While in use, the 3325A output suddenly dropped and a “Amplitude Calibration Error” is displayed by the instrument. A huge discrepancy appeared between the displayed output and the actual output level. It stopped producing sawtooth and square wave signals altogether.
Thanks to Agilent’s generous on line library, getting the service manual and schematic of the instrument is not a problem. Armed with this essential info, it took me just a few hours to isolate the failed part of the circuit and pin down the defective component. It is U14 of board A3, part of sine amplitude control circuit.
Getting to the bad component turned out to be the easy part, finding a the replacement gave more challenge. The component is marked with HP part no 1858-0063 (that is actually a standard transistor array whose real part ID is CA3102E) has been out of circulation since like 15 years ago. I was contemplating on substituting the IC with discrete transistors when I noticed that CA3102E actually consists of two identical halves, and the other half left unused. The other half is most probably OK, I could use it as a temporary fix, until I find a permanent replacement. Luckily, the IC pin out is almost symmetrical; to use the good half, I only need to reinstall the IC in the opposite direction, and then use jumper wires to connect the two pins that did not line up to the desired points.
Output level calibration failed because of bad U14. Chip replacement is not available, but a temporary fix was made by using the other half of U14 that is still working.
The instruments worked all right after the fix was made. Experience has taught me however that, when a vintage instrument fails, it always happens in two quick succesion. I mean, after one defect popped out, another one will surface not long after the first one is fixed. So I let the instruments run for hours without reinstalling the covers on, waiting and expecting a second failure.
Round 2, the big crash
Sure enough, after a few hours passed, the instrument went nuts and all it does thereafter is display cryptic characters, and is unresponsive to any keyboard inputs. Nothing worked at all. The control board underwent a big crash. This one got me worried, because I know the control board is using a HP custom processor, coupled with an array of ROM chip. If any one of this chip proved defective, my 3325A is a goner.
The 3325A service manual control board troubleshooting procedure requires the use of a digital signature analyzer- I don’t have one. Armed only with a schematic, I know troubleshooting work this time will require a mighty effort.
 |
| “The 3325A uses three separate RAM chips, all three seemed to have conspired together to fail at the same time.” |
I finally identified the trouble causing ICs; to my relief, it is neither the processor nor the ROM. It is the RAM chips.The 3325A uses three separate RAM chips, all three seemed to have conspired together to fail at the same time. One RAM Chip is 1Kx8 2114 SRAM, and the other two are 2102 256 x 4 SRAM. And as you might already guessed, three obsolete chips. Once again, getting the part took the longer than the troubleshooting work itself. Luckily, I found a 2114 chip on one Tektronix board that I kept for a just in case situation like this. But no 2102 chip. Not a big problem though. I can make a 2102 SRAM equivalent from a bigger capacity SRAM Chips.
I picked a 32Kx8 SRAM in SOIC and solder it on a package converter PCB. And then constructed the whole assembly on a 18 pin HIREL socket so that I can arrange the pin out to match exactly the 2102 layout.
 |
| A 256×4 SRAM replacement made from 32Kx8 chips |
 |
| 2102 RAM Chip replacements mounted on controller board |
The Moment of truth
Keeping my fingers crossed, I flicked the switch on. And while still holding my breath, I checked to see if the display showed something good. It worked! I got my 3325A back!
My fondness with old legacy test instruments baffles some of my friends. Understandably so because they know, if I wanted to, I could replace my old rigs with newer ones (albeit pre-owned, but more recent). I like to use circa 80’s instruments because all their functions can be directly accessed with a flick of a switch. I hate the menu style of newer instruments. Wading through several layers of menus to get to the desired function is not enticing to me. Furthermore, 80’s instruments are built to last a long long time. And if they fail, they are generally repairable because of lack of custom chips in their circuits.
 |
| HP 3325A assembly showing the controller board (largest board), power supply (right), and Function board (top) |
 |
| HP 3325A Mixer board(bottom), attenuator (middle left), and the output circuit board (largest PCB). |
No comments:
Post a Comment