Is a 0.01s time base really useful in a frequency counter?

I am currently designing a frequency counter based on the PIC16F628A. It is working as expected but I would like to add some useful features.

According to the images found by googling, I noticed that many commercial frequency counters have the 0.01s time base feature. I know 10 seconds is for 0.1Hz resolution. 1s for 1Hz. 0.1s is 10Hz resolution but it is very useful when adjusting an oscillator frequency and see the changes immediately on the display. 0.01s is 100Hz resolution but what is it really useful for?

• @Arsenal. Thanks for your reply. With 0.1s time base the frequency count and display will be done 10 times per second. The display is refreshed 10 times in a second. I think it is enough to see any fluctuations. With 0.01s time base, the display will be refreshed 100 times per second. It is beyound the human eye capabilities to see fluctuations in 0.01s. Maybe I am wrong. Feb 7, 2018 at 10:56
• @Arsenal. Yes, but 0.1s time base is still good enough for that too. Feb 7, 2018 at 12:41
• @Arsenal. I am a little confused about what your are saying regarding jitter. 0.1s time base has one more useful digit than 0.01s time base. The former gives better resolution hence more information. So it should be better for jitter observation and analysis. Could you please provide an example proving that 0.01s time base is better for jitter observation? If so, this would be a good answer for my main question. Feb 8, 2018 at 10:11
• I've added an answer and removed my comments as they are now obsolete. Feb 8, 2018 at 10:33

My thinking goes along these lines:

With 0.01 s the display will update so rapidly, that you cannot perceive the single values anymore. But if you are looking at a clock source you might not be interested in the exact value of the frequency, but the frequency stability over a short time.

For this a 0.01 s gate is a nice addition as you will be able to see up to which digit you will get changes.

For example: a 2 MHz resonator provides 2.000 MHz +- 0.001 MHz if you look at it with the 1 s setting. Looking at it with 0.1 s you might see some fluctuation already, let's say 2.000 MHz +- 0.003 MHz. But at the 0.01 s setting you might suddenly see 2.000 MHz +- 0.1 MHz because there is a problem with the resonator which gets averaged out otherwise.

It's the same basically with a multimeter reading speed. I can set it so high that I can't read more than 1 digit, but I can see how much the reading varies (qualitative) and judge if this is abnormal or not. Also an additional feature like math functions providing a minimum, average and maximum reading come in handy with faster readings. The instrument will keep the values of those spikes I am not able to perceive and the average will be pretty close to the value you get with a longer gate time.

Experiment I've done just now:

Frequency generator generates a rectangular signal with 4 MHz, on top of that is a frequency modulation of 100 kHz, so the signal has either 3.9 MHz or 4.1 MHz. The modulation frequency is 1 Hz. My frequency counter actually doesn't have 0.01 s as a gate time, so I can't prove my point for that, but the concept is the same.

Setting the gate time to 0.1 s it displays 4.09998 MHz and 3.90001 MHz, if I set it to 1 s it just displays 4.000163 MHz or 3.999864 MHz. I would have had no idea that I'm looking at a signal which has +- 100 kHz.

On a less serious note, this could be just a marketing feature which no one uses but is needed to sell the product over a competitor.

• Many thanks for your convincing answer. I now understand that the key point is just to see at which digit fluctuations start rather than actually reading the digit value. Feb 9, 2018 at 9:27
• After reading your answer, I did some experiments which confirmed all what you have said. I always tought that jitter is a small variation of the frequency due to some noise (EMI for examlpe). I now realize that modulation is a wanted (or controlled) jitter. It is much bigger depending on the modulation depth. Feb 9, 2018 at 9:28
• I even went further in my experiments by applying a sweeper output signal to my frequency counter. You are right. At 1s gate time, the displayed frequency is an average value. At 0.01s, I can see that the displayed value is continually changing. I will add an other feature to my counter to choose whether the displayed value is average, maximum or minimum. As you said, this will be very handy in some situations. So the answer to my question is YES. 0.01s time base is very useful. Many thanks again for your help. I greatly appreciate. Feb 9, 2018 at 9:30
• @racboni glad I was finally able to write my thoughts up in a way that they came through. Feb 9, 2018 at 9:32

I can think of a practical reason....

Your counter (and display) has a limited maximum count so, you use a faster time base to ensure that that maximum count is not exceeded. If it is exceeded you might get a false count or maybe (based on how it's designed) you get an overrun error.

So, it's for counting pulses that have the highest frequency.

• thanks for your quick answer. My counter goes up to 99.999999 MHz. The PIC microcontroller, by software, makes two counts. The first one has 0.1s time base to make sure the input frequency is below 100MHz. The second count is done with 0.1s or 1s or 10s time base. The result of this second count is then displayed. This way, I think, an overrun error could never occur. Maybe the 0.01s time base solution is useful only for non software controlled counters. Feb 7, 2018 at 10:35

One use of frequency meters is to monitor operations while adjustments are made. So, when tuning to a target frequency, you might adjust a knob, turn a screw, position a Geiger tube, while watching a frequency display. The feedback from that display is confusing if it represents the value over a preceding one-second period, so it makes sense to update the displayed value frequently and with a shorter delay time.

It makes even more sense, to use a short averaging time AND to display the result in analog form (like a bar-graph or moving needle) rather than a series of decimal digits. Analog conversion of (F_measured - F_target) might be displayed on a moving-needle meter.

• Thanks for your answer. You are right. As I said in my question, 0.1s time base is enough to see any changes since the count and display are refreshed 10 times per second. Can anyone see changes made 100 times per second? Feb 7, 2018 at 11:43
• @racboni It is possible (every pianist does it) to respond at tenth-second time scales, and a tenth-second window time has to be added to the display update time (whatever that is). 100 times per second (about the fastest frame rate for video output) is non-flickering, but 24 Hz (old-style motion picture rate) is why movies are called 'flicks'. 48 Hz light pulsing fixed that. Feb 7, 2018 at 11:53
• Yes you are right. But this is not the case for a frequency counter. We just need to see how the frequency changes. We do not have to do any action as fast as 1/100 second. So, 0.1s time base is still good enough. Feb 7, 2018 at 12:37
• @racboni 1.0s is clearly awkward, 0.01s is clearly faster than human reaction requires. At 0.1s, I'd be inclined to try an experiment. I've seen programmed test beds blink up a dozen displays in seconds, and bench techs don't need more time than that to assimilate and respond. Feb 7, 2018 at 22:23