Maybe the clock option is working just fine, but not applied properly.
It could be that the driver output is an open-collector, meant to pull a
stepper coil or such down to ground for each pulse. Without a load,
you'd see just some small collector voltage due to the base current in
the output transistor - the forward base-emitter voltage, minus the
forward base-collector junction voltage could be in the tens of mV into
high a impedance (like a scope) measurement.

So, try adding a pull-up resistor to draw at least a few mA, and see if
it looks different. Good luck.

Ed

As always on this subject type, I must comment on the utility of
managing ground loops with common-mode chokes on cables between
equipment. I think this is the simplest, most beneficial thing you can
do for good quality signal distribution.

Good cables, shielding, and grounding are all fine and good, but do not
address the ground loop currents that inevitably flow between any
interconnected pieces of gear. Adding CM chokes raises the common-mode
impedance of the cable(s), greatly reducing the loop currents. You can
see these parts all the time, inside equipment, or added or built into
cable assemblies and such - almost always to fix EMC issues. The
simplest are the clip-on type ferrites that are easily added to almost
any cable type.

These are mainly to suppress RF and HF emissions from SMPSs and digital
circuits and such, but they can also be effective even for mains
frequency (and harmonics) ground loop control, with the right cores and
some simple tricks. The reason is that the voltage levels and impedances
are quite low (millivolts and milliohms), so adding even a small
inductive reactance can be quite effective, percentage-wise.

If you have mains frequency ground loop problems, you can try cores with
as high an A sub L as possible, and even better, multiple turns to get
the squaring benefit. There are ferrites available that should do well,
depending on the situation. I have even used regular silicon steel wound
or laminated line frequency transformer cores as CM chokes in certain
situations where very large line rejection was needed.

For narrow-band distribution of 10 MHz for instance, another option is
to use transformer coupling of the signal, so the low frequency ground
current and interference is eliminated, and a high frequency CM choke
can take care of the rest.

Ed

A cold oven is nearly the only and best explanation for that much error.
The good news is that these can be readily opened up and worked on.
There's plenty of info out there.

Ed

Wow - that is scary-burnt. Maybe it's a good idea to have some form of
thermal cutout, even if a nuisance sometimes.

Ed

It's a bit tricky to use prescalers for arbitrary input signals. As
others have mentioned, they self-oscillate when there isn't a valid
input signal. This is not only a possible, but nearly certain
characteristic due to the (typically) ECL input amplifier being biased
right at the logic threshold, followed by lots of gain, then divided by
flip-flops. These were typically used in PLL synthesizers over limited
frequency ranges, with known, adequate power levels.

If you are counting strong signals at well defined DC levels, you can
use ECL directly with proper biasing. For a general purpose counter, you
want lots of sensitivity and wide frequency range, so the usual
prescaler with self-biasing and AC coupling is a simple way to go. You
just have to be aware of the limitations. You can eliminate the
oscillation tendency by giving up some sensitivity, but it's easiest to
just ignore it unless it causes trouble.

I'm quite familiar with the MB series and other prescalers, and have
used them in many projects. All the prescalers I have were salvaged from
old equipment, and I believe that most are long obsolete, since their
function has been mostly absorbed into the modern PLL chips.

Nowadays you can get 100-series ECL parts that can toggle to several
GHz, and in various gates, FFs, and counter types. You could build say,
a divide by 10 or other values with a programmable counter, I believe up
to a couple GHz. I don't recall all the parts and specs, but they're out
there. If you want to make it for arbitrary inputs, you'd first buffer
it with an AC coupled, self-biased line receiver cascade before
counting, and you'd be right back to a self-oscillating - but much more
versatile - prescaler.

So, if you can get this fully assembled, connectorized, ready to go
MB506 module for seven pounds, just do it - maybe get several - if the
specs are suitable for your needs. It beats the hell out a DIY one,
except for not having a nice convenient decade divide ratio. You should
look up the MB506 datasheet to see the actual capabilities - you can
usually extend the low end, for instance, by adding more input coupling
capacitance. Anyway, I don't know if the MB506 is even made anymore, but
if it is, I'd guess it alone would cost more than this module.

There used to be all sorts of prescaler ICs with various ratios,
including 10, so you might find NOS parts, but with the grief of
building something. There's also an old trick in doing counters with
binary prescalers, which is to change the counter clock frequency so the
net overall divide ratio is a nice decade value, but making the clock
and working the decimal points gets tricky. Also, if the binary divide
is big, the counter is correspondingly very slow.

Of course, if you want to get fancier, you can make a PLL/VCO based
converter that replicates the unknown input frequency at say 1/10th. But
you'll still likely have the same sort of prescaler deal involved,
inside of a PLL chip.

Ed

Looks like the classic MC12080 is still in production by Onsemi in the
USD7 range. It can divide by 10, 20, 40, or 80 up to 1.1 GHz spec and
likely can reach over 1.5 GHz. You'd have to build a circuit for it,
unless someone offers a module of some sort.

Ed