[time-nuts] Pulse width for start signal for HP5370

[Magnus Danielson]

On 04/09/11 09:07, Tom Van Baak wrote:
> Paul,
>
> Do you have an oscilloscope handy? When doing precision
> time interval measurements it's useful to take a look at what
> the counter inputs actually look like. Use a BNC tee and a 1M
> scope probe (so as not to artificially load the counter input).

I tend to do the same. Also, some errors is not due to the wrong 
frequency but due to interference, and you can't see that very good with 
a counter, but on a scope.

Here a short-hand of observations and probable causes:

1. too high frequency (unstable)

May be trigger error, usually bad slew-rate on the slope or slope of 
other direction. In really bad cases double frequency can be seen.
Injected noise has also been seen.

2. too low frequency (unstable)

May be trigger error, usually a trigger on the bad slew-rate peak, so 
moving tricker back to safe ground high slew-rate is needed.

3. unstable

Beyond the above cases, you can also have ringings or other similar 
features of the signal which can cause bad timing triggers etc. High 
slew-rate is where you want to trigg, if you just can avoid other 
trigger problems.

Low-pass filters can help to clean up trigging.

It is a good exercise to train the trigging skills on oscilloscope, 
since you will get fuzzy and/or double/tripple/whatever images.

So, in all these a good scope gives you clarity and a good feel for the 
signal.

> For pulse inputs it should help you set the DC trigger level
> appropriately. I usually use 0.5 or 1 or 2.5 VDC. But seeing
> the actual waveform helps me decide.
>
> For CW inputs you will probably want to set the counter to
> AC coupling and zero volt trigger.
>
> Some counters give you a choice of 50R or 1M termination.
> Be careful with that. In some cases the waveform is much
> cleaner with 50R. In other cases 50R puts way too much
> load on the source and you trigger level or risetime suffers.

All of the above matches my experience.

Doing measurements properly is an art. While I am sloppy, it is mostly 
from lazyness. I can even get called in to the lab just to "find" the 
signal with the scope they already hooked up (usually they made the 
error of going for higher frequency rather than lower frequency 
offending signals... so that I find within minutes with just turning 
time-base and tuning trigger).

Also, if you have it, at some times a spectrum analyzer can help you to 
understand what happens... such as unexpected harmonics. The spectrum 
analyzer has better dynamics than the scope.

Cheers,
Magnus