As I always have been a friend of the high quality standard of the former HP products I oriented my design goals as close as possible to the spirit the HP 5316B were based on.
These main objectives I kept in mind during design:
- built the device to have a look and feel like an original part
- keep the size of the original oscillator board to prevent space problems
- use the original connectors and mounting
- add some protection circuitry so that failure of the OCXO-assembly will not harm the HP 5316
- build a device that can be send to other people with a clear conscience
Therefore, this is the result. On the left side you can see the original standard crystal oscillator HP 05316-60008, on the right side you see my ovenized board.
On the backside there are some surface mounted parts. In my first design, I tried to avoid using them as this would break the goal of designing with the look and feel of an original part. However, SMD parts are smaller, less expensive and more reliable (if soldered correctly).
All of the soldering is done with lead-free solder. The SMD-parts are soldered with a correct Thermal Profile using a reflow-process (staged IR-preheating up to 220°C in combination with a hot-air gun to spike up to 240°C and a controlled cool-down afterwards). Through the hole parts are soldered conventional by using a soldering iron.
The parts are strictly handled in a static free environment to eliminate every possibility of an ESD damage. Keep in mind that research has shown almost half of premature circuit failures are the result of ESD events a circuit got during assembly.
Most types of ESD damage will not result in an immediate failure of a device, but in a failure after weeks or months of usage.
The front side shows two potentiometers for adjustment. One of this is a single turn trimmer to set the 10 MHz output level; the other is a 25 turn trimmer to set the 10 MHz frequency.
Output level limiting
The output signal of the Morion MV85 OCXO parts I got on the used parts market do have high variations in their signal level. I could measure Ueff signal levels between 400 mV and 600 mV. The data sheet does only specify the minimum level at 225 mV.
The maintenance manual of the HP 5316 does not specify the maximum allowable input level for the 10 MHz signal that is fed into a custom made (HP-specific and not available as a spare part) integrated circuit. Blowing out this circuit by feeding a signal out of its spec would immediately make the whole HP 5316 unrepairable.
There are several oscillograms in the HP 5316 maintenance manual showing a signal level not more than 800 mVpp, that equals to 283 mVeff. Measurement in my HP 5316B shows that the the injection-locked multiplier that is used for clock forming when feeding in an external input signal will produce a slightly higher value of up to 1,2 Vpp. So I defined the maximum allowable output value for my circuitry at 800 mVpp. At this level, the device will surely not overload the following input and can be protected against a wrong adjustment of the output level with only two diodes (a similar protection can be found in the external input stage of the HP 5316).
One of the design objectives had been that it should be possible to calibrate the assembly after assembly, give it a burn in time outside of a HP 5316 and send it to other electronic enthusiasts to give them the chance of just plugging the device in and having a perfectly adjusted HP 5316 device.
The Morion MV85 (as most of the OCXOs on the market) only has a limited stability against output load changes. To decouple the OCXO from load changes i added a simple common collector circuit on the output stage.
The 25-turn potentiometer will set the output frequency of the OCXO. Its frequency pulling range is 5 x 10−7 (± 5 Hz). With this, an adjustment up to an accuracy of 10 mHz is achievable. This accuracy is in the range of the weekly frequency drift due to aging of the OCXO crystal.
Power and fusing
The power rail is fused by a 1 Amp picofuse. As this OCXO-assembly is connected to power as long as the HP 5316 is plugged into the wall socket i decided to add this fuse.
The OCXO is powered with 5V from an LM2940 regulator. Using the +5V sourced from the HP 5316 main board were not an option because it’s current limit is not known and there is a lot of digital noise on this power supply. The OCXOs frequency stability has a dependency to the supply voltage stability.
As OCXOs are expensive devices I added a suppressor diode on the 5V rail to protect the OXCO against a voltage regulator failure. Therefore, any overvoltage due to a defective LM2940 should blow the picofuse or at least be limited to a non-damaging level.
On the unregulated power side there is a 1500µF Capacitor to keep the rectifying ripple on a similar level as without the new ovenized assembly.