Frequency counters & 10MHz standards

[hlavac]

I have two questions:

• Does frequency counter always assume the input is a sine wave / square wave of a single frequency? What happens if it is not the case?
• How does the standard external 10MHz synchronization signal that some of the instruments can use look like? Is there a standard for it?

[jpb]

1. my understanding is that the counter has an edge trigger and the level can be set, there is also probably some means of attenuating large signals. So it doesn't assume any shape but if you set the level in a lot of noise it will give erroneous counts and not be very stable.

2. most instruments define the type of signal they expect from an external clock e.g.

for the TTi TF830 counter it just has :

External Standard Socket
An external 10MHz frequency standard (5V/TTL
level) can be applied

while for the Tabor WW5061 AWG it is more specific:
10MHz TTL, 50% ±2% duty cycle

The TTi TG5011 seems quite flexible :
Ref Clock Input
Input for an external 10MHz reference clock
Voltage Range: 1Vpp – 5Vpp
Maximum Voltage: +5V
Minimum Voltage: -1V


so generally they want TTL square waves with 50% duty cycle but I think some accept sine waves.

[olsenn]

Counters only measure the number of times a voltage goes from below a value of 'a' to above a value of 'b'. Sometimes these two values are the same, but there is usually hysteresis (schmitt trigger) to avoid any influence of noise.

Try not to think about it in the frequency domain; instead, consider the time domain analysis, as you would see on an oscilloscope. Also, since it counts the total number of low-to-high (sometimes high-to-low) transistions in a set time (say 1 second for example), the reading that will be displayed is the average frequency over that one second interval.

As for external frequency inputs; it also doesn't depend greatly on the waveform; this reference clock merely clocks the digital logic IC's apon low-to-high transistions, just like the counter itself. My 10MHz rubidium frequency standard for instance produces a sinusoid; however, a TTL square wave will also do the trick.

[jimmc]

JPB has covered your second question well.
Your first question is covered by Agilent's AN200  http://cp.literature.agilent.com/litweb/pdf/5965-7660E.pdf (p9).

Jim

[hlavac]

Excellent appnote, just what i was looking for, thank you!

[KJDS]

Many years ago I designed the reference locking circuitry for the Marconi Instruments 2945 radio test set. Amplitude or wave shape wasn't that important, and it could tell and adapt to a 1MHz, 5MHz or 10MHz input.

[codeboy2k]

to extend on jpb's answer to #1:

I designed a frequency counter input circuit once long ago, it was similar to a scope's input, with 1Meg Ohm impedance, a 22pf cap and an AC/DC coupling switch. After that it was diode clipped to the positive and negative rails, so it would never exceed 600mv peak or so. Then it was gained up about 2000X, so the max 600mv input signal would clip and hit the rails at +/-15V, and a 1mv input signal would be gained up to 2V now, and that was used to feed a schmitt trigger with level set at about 1V, to output a 5V square wave feeding a TTL counter circuit. All of this was discrete transistors. The first input stage after the diode limiting was a common source N Channel JFET.  The following stages and the schmitt trigger were BJTs. No op amps.

The input worked for sine, square, pulses, triangles, ramps and was spec'ed from 1mv to 10V p-p

This falstad circuit shows diode limiting a 5V peak-to-peak sine wave input to 740mv p-p
http://www.falstad.com/circuit/e-diodelimit.html

If I did it again today I would certainly use FET input op-amps, probably can be done in a single 4-device op-amp ( a buffer, 2 gain stages and a schmitt trigger)

[madshaman]

JPB has covered your second question well.
Your first question is covered by Agilent's AN200  http://cp.literature.agilent.com/litweb/pdf/5965-7660E.pdf (p9).

Jim

Another thanks from me.  Trying to design and build my own and happy to see the initial block diagram roughly matches how I was going to implement it (which to me means I'm likely thinking about the problem the right way).

Still a pending digikey order away before I start prototyping, down to picking parts for all the fiddly bits and still waiting for the 10Mhz rubidium reference I nabbed from eBay (tons available), will prolly grab a cheap 10Mhz oscillator in the meantime so I'm not stuck.

For flexibility of the time base, planning on making as flexible as possible, i.e. accepting most waveform inputs and power level.  Planning on conditioning the input with some filtering and agc, then clamping it with diodes, run it through a Schmitt trigger and finally a counter.  Still working figuring out whether I want the time-base itself to automatically drive both the counting then interrupting the mcu for a quick read phase.  Hopefully at 10Mhz, even someone like me can make it work.

Someone mentioned earlier about making their time base input 1Mohm impedance, would it be better to use 50ohm to match most sources, or at least the ability to switch in a 50ohm impedance like an oscilloscope front-end with that option?

[codeboy2k]

Someone mentioned earlier about making their time base input 1Mohm impedance, would it be better to use 50ohm to match most sources, or at least the ability to switch in a 50ohm impedance like an oscilloscope front-end with that option?

Yes a 50 ohm switch is useful especially if you plan to measure VHF (30 to 300 Mhz) and UHF (300Mhz to 3Ghz).

Also, since you're building it, you can put in an attenuation pad on the 50 ohm input path, so you can switch in 5db, 10db, 20db, 40db attenuation, something like that, whatever you think you might need.  Or build an external pad if it's more useful to you.

You don't need AGC on your input, you only need to gain it up alot and catch the rising edge, which you will do no matter what, if you gain it up enough.
You don't care that it clips. An AGC will try to keep the signal looking nice, but you don't need to, you just need to count the rising edges.

[madshaman]

Someone mentioned earlier about making their time base input 1Mohm impedance, would it be better to use 50ohm to match most sources, or at least the ability to switch in a 50ohm impedance like an oscilloscope front-end with that option?

Yes a 50 ohm switch is useful especially if you plan to measure VHF (30 to 300 Mhz) and UHF (300Mhz to 3Ghz).

Also, since you're building it, you can put in an attenuation pad on the 50 ohm input path, so you can switch in 5db, 10db, 20db, 40db attenuation, something like that, whatever you think you might need.  Or build an external pad if it's more useful to you.

You don't need AGC on your input, you only need to gain it up alot and catch the rising edge, which you will do no matter what, if you gain it up enough.
You don't care that it clips. An AGC will try to keep the signal looking nice, but you don't need to, you just need to count the rising edges.

Thanks codeboy2k, I guess agc isn't needed, I'm new in this realm of electronics and my intuition told me I should bring each stage into a known operating realm.  Was planning to over gain the signal and clip it deliberately with the diodes anyway but was worried there'd be some effect I wouldn't foresee (due to my lack of experience) if the signal power wasn't constrained at this point, my knowledge at this point in this realm, what little there is, comes from a massive crunch of book learning with no experience (which is why I've set a number of projects for myself which will generate useful bench equipment

I can't wait for all the screw ups and hallucinations I'll encounter bringing this project to completion; maybe the worst that could happen is that it just works ^^'

[olsenn]

I can't wait for all the screw ups and hallucinations I'll encounter bringing this project to completion; maybe the worst that could happen is that it just works ^^'

Why would it working be the worst thing, and why are you hallucinating... don't design on salvia...  it doesn't turn out well!

[madshaman]

I can't wait for all the screw ups and hallucinations I'll encounter bringing this project to completion; maybe the worst that could happen is that it just works ^^'

Why would it working be the worst thing, and why are you hallucinating... don't design on salvia...  it doesn't turn out well!

Lol, my lab is off limits to myself unless I'm well rested (well, having slept in at least the last 48 hours..) and sober :-)

If everything works, you learn less, it might even work for the wrong reason and you'll never know.

Hallucination is a term I picked up playing go, it's my way of describing that situation where everything looks good to you at a given moment in time, but things don't work, the reason being that what you think you're seeing isn't what's actually there, then later you slap your forehead saying "oh my god, why couldn't I see that!".  I find this happens *all* the time with every type of engineering.  To put it simply: you don't notice what you don't notice.  The same thing happens with math proofs, computer programs, wood working, practically everything.  Sure, as you gain more experience you make less mistakes, but you generally learn from your screw-ups caused by perceptual blindness.

Some people may feel they are very solid and accurate and sober and clear thinking, and maybe they are, I tend to find the world is *full* of constant misconceptions, and they can all be a fun source of learning.

[Rufus]

Also, since you're building it, you can put in an attenuation pad on the 50 ohm input path, so you can switch in 5db, 10db, 20db, 40db attenuation, something like that, whatever you think you might need.

If your 50 ohm input could handle lets say a modest +20dbm to have any point a 40db attenuator in front of it would need a 1kW power rating.

[kfitch42]

Speaking of DIY frequency counters:
http://dangerousprototypes.com/2013/03/27/1hz-to-10mhz-frequency-meter-using-attiny2313/

[codeboy2k]

Also, since you're building it, you can put in an attenuation pad on the 50 ohm input path, so you can switch in 5db, 10db, 20db, 40db attenuation, something like that, whatever you think you might need.

If your 50 ohm input could handle lets say a modest +20dbm to have any point a 40db attenuator in front of it would need a 1kW power rating.

I worked with large transmitters in the past, from 5KW AM up to 250KW FM so large attenuations are always something I think about (and dummy loads that fill half a room!)

Probably not needed at the hobbyist level, yeah. 

[madshaman]

Lol, not quite sure how soon I'll want to jump into 250KW face-sublimating power levels.  In the perhaps distant future I *could* see myself getting near 1KW, but that's quite a ways off.