PM666x Frequency Counter: OCXO & Reset MOD

I recently got a collection of 3 PM666x-series frequency counters (two PM6669, one PM6666). The previous owner had already started building the OCXO upgrade board from https://www.jackenhack.com/philips-fluke-pm6665-pm6669-ocxo-pcb/, so it was my task to finish the project and make the counters useable for our local hackerspace.

When checking the stability of the two already assembled OCXO boards, I found the stability worse than expected – the frequency drifted a lot more than it should, and trimming was sensitive to supply voltage variations. Upon closer inspection, it turned out that the MAX6198A that generates the reference for the OCXO trim voltage was oscillating wildly – it seems it did not like the 1μF capacitor on its output. After substituting a more appropriate 1nF capacitor (the datasheet specifies a maximum load capacitance of 2.2nF!), the oscillation was gone, but the voltage was far from the expected 4.096V — more like 3.7V and drifting wildly. I don’t know if the oscillations killed both MAX6198A or if the damage had other reasons. Before ordering spares, I searched through my parts and found MAX874s which are pin-compatible in this application. Using the MAX874 with 1nF output capacitor, the control voltage is now stable and the OCXO performs as it should.

Stable, but not starting?

Some of the PM666x counters seem to have problems with their reset circuit: the CPU reset is generated by a simple capacitor+diode combination (C123 + GR111), and this does not always produce a clean reset for the CPU on power-up. The fix is to remove both parts and replace them with a supervisor IC (5V, push-pull active-high output). The MCP101T-475 nicely fits the existing layout, so the fix looks like this:

PM6666 reset mod with MCP101

The PM9604 GPIB board has the same reset circuit (with slightly different component values). Although I did not experience any problems, I decided to apply the same fix here. However, that board uses THT components. A TO-92 MCP101 or similar would probably fit fine with some bending of pins, but I had none of these at hand, so I bodged in a DIP-8 ADM707AN on the bottom side.

PM9604 reset mod (top view)
PM9604 reset mod (bottom view)

The CPU part of the PM6666 is quite similar to the PM6669 (and it uses the same GPIB board), so the instructions apply with minor variations due to a different layout.

Jack Zimmermann has User- and Service Manuals for the PM6665/PM6669 at https://www.jackenhack.com/upgrading-philipsfluke-pm6665-frequency-counter-oven-controlled-crystal-oscillator-ocxo/ – those helped a lot, thanks!

Update 2020-09-16:

I needed another OCXO for the third counter, and the ISOTEMP OCXO131-100 was difficult to get, so I looked for alternatives that come with a decent datasheet. I chose a FOQ PTOC32227 and can now report that it is pin-compatible and works as a 1:1 drop-in replacement for the ISOTEMP part. The PTOC32227 has a wider trimming range (±2ppm compared to ±0.4-0.9ppm on the Isotemp), so trimming is a bit more sensitive.

Update 2020-10-16: Capacitor Replacement

On a 25-year old instrument, electrolytic capacitors may start to deteriorate. While the two big electrolytics still measured fine, I did not like the use of tantalum capacitors directly on power rails (without much voltage derating). This kind of (ab)use is known to lead to spontaneously combusting capacitors, leaving behind burnt FR4 (this happened on my HP 34401A). Looking through the schematics, there is nothing critical about these caps requiring tantalum, so I looked for non-smoking replacements that fit the footprints. Philips used two sizes of tantalum: 3216 and 6032.

While at it, I also replaced the two big electrolytics:

With these new parts, the counter should be good for some more years of operation. The following pictures show the locations of the replaced SMD capacitors:

new capacitors at voltage regulators
new capacitor C118, near power switch
new capacitors in counter circuit

2020-10-16 by rincewind