HP 4275A ERR1 on 10Khz

[ffranzini]

I found my HP 4275A failing Zero Cal with ERR1 on 10khz and erratic short value on almost all frequency.

Looking on the service manual, after some investigation i ended up on A6 Board (Oscillator), input of the filter(TP2) was OK, on TP1 and TP9 i found the same problem, 10khz amplitude way too high.

After further investigation i found both filter not working as expected because lack of amplitude level control, at this point i was a little bit confused, every thing on both filter seemed to work just fine and i started to look at the Photocell CP1 and CP2, as far i understood from the schematic and the block diagram discussion the amplitude level control is performed adjusting the frequency roll off through the variable resistor on the photocells CP1 and CP2, automatic level control work with the feedback feeding the led of the photocells, brilliant design !!!

Measuring the photocells i found CP1 resistance value high and variable around 200kohm, CP2 resistance value around 20kohm, photocells are marked MCD-521 YOKOGAWA brand, of course no way to find the datasheet, since is there no power on the resistor of the photocells i guess is this issue is just related to the aging.

Even if with strange value i got good response feeding the led, the low side of the resistance was OK (high side frequency roll off), at this point was clear to me that the problem was related to the too low resistance on the photocells with no light from the led (low side frequency roll off).

With complete understanding of the issue has been pretty easy to fix it working on the filter capacitor components C6, C8, C26 and C28, looking amplitude and settling time with the scope i ended up adding 0.02uf on C6-C26 and 1000pf on C8-C28, to further improve the high ranges i added 100pf from the photocells exit to ground.

Finally my 4275A pass all of the tests and the Zero Cal is perfect !!!

Of course i am still concerned, i consider my fix as temporary and i am looking to replace the photocells, it would be very useful if someone have some knowledge of the spec of this photocell type

I hope this repair could help someone!!

[picburner]

The only thing I found about this opto-coupler is this skinny piece of paper:

[ffranzini]

From your "skinny piece of paper" i got exactly what i wanted, Roff 3Mohm, this confirm my finding, on both photocells dark value are out of spec, light on value seems to be Ok.

Now i can start to find a suitable replacement, any advice on that??

[picburner]

Fluke here has replaced two neon lamp/Cds opto-couplers with two H11F1 (opto-fet) with many modifications to the circuit but your case is simpler: you already have LEDs instead of neon lamps.
If you don't find a suitable replacement you could try this way.

[ffranzini]

Very interesting read!

I did some search and i found the LDR NSL-32SR2 on Farnell look very similar but i will try first with your advice, H11F1 seems to fit perfectly the needs of A6 board, is very fast, with better spec and the pinout is just right!!

Thank You !!

[ffranzini]

Some update...

I replaced both MCD-521 with two H11F1 (opto-fet), from level stand point and setting time they behave extremely well and 4275A is performing both zero cal flawlessy.

Unfortunately seems they are introducing on the circuit some additional condition because, investigating with the oscilloscope, especially on the low frequency settings, i do not see sine wave but something more "triangular".

During my search i remember some comments about the need of DC bias on this type of application (variable resistor) may be the case? 

Francesco

[picburner]

The H11F1 is not a direct drop-in substitute, even though it was born as an analog variable resistor, some adaptation is certainly necessary.
Also check that, at the extremes of the frequency band, the current generated by Q6 and Q12 is either too high or too low for the H11F1's internal LEDs.

[ffranzini]

I think the problem is on the photo-fet side, checking at the output of the photo fet (capacitor filter side) i see, at the inversion on the top of the curve a pretty big step going down, this behavior is frequency dependent.

As plan B I bought some cheap 5539 LDR, with some surgery on both MCD-521 I replaced the original photo cell in it, the electrical result seems to be perfect, amplitude very stable, sine wave just perfect, good settling time and no issue on zero cal, i think this is the best way to permanently fix the issue even because is replicating almost exactly the original configuration of the circuit, now i need to find the right way to seal the "modified" MCD-521

You can check this modification on the attached pictures.

Francesco