# Atomic Clock disciplined with Rubidium

#### truetime320

Joined Aug 5, 2023
6

I was able to get my hand on a FEI-5650A Rubidium Frequency Standard and I was wondering how I could use it to make a really overpowered wall clock. I was able to power it using some information I found on the internet from a defunct website. Using an oscilloscope, I was able to get a constant 10mhz sine wave. I guess what I'm asking for is help devising a wall clock that is disciplined with this rubidium standard. Do I need to flip flop the 10mhz down to 1? How do I make the sine wave to a square wave? I hope to hear everyone's thoughts!

#### Attachments

• 199.1 KB Views: 8
• 186.9 KB Views: 7

#### crutschow

Joined Mar 14, 2008
34,452
I assume you mean 10MHz (m is milli).

Yes you would reduce the 10MHz down to whatever frequency you need for the clock using high-speed counters (such as the 74HC4020 14-bit which will divide the frequency by a factor of 16,384).
To convert the sinewave to digital, you can use a high-speed analog converter (which has a digital output).

#### truetime320

Joined Aug 5, 2023
6
I assume you mean 10MHz (m is milli).

Yes you would reduce the 10MHz down to whatever frequency you need for the clock using high-speed counters (such as the 74HC4020 14-bit which will divide the frequency by a factor of 16,384).
To convert the sinewave to digital, you can use a high-speed analog converter (which has a digital output).
Oh my apologies, I meant megahertz.
Thank you for the tip!
So my current plan is to power the Rubidium > convert the sine waves to digital using a High-speed analog converter > reduce the 10Mhz using the High-speed counter to 1Hz. From there the 1Hz should tick per second and then connect it to a display.
Do you think I’m missing anything?

#### MrChips

Joined Oct 2, 2009
30,810
You don't need a high-speed analog converter.
What is the amplitude of the output signal?
Amplify the signal to 0-5V and feed it into the input of any 74HCxxx divider circuit, for example, 74HC4060

#### BobTPH

Joined Jun 5, 2013
8,967
You don’t need to do that. A Schmitt trigger will turn the sine wave into a square wave.

#### crutschow

Joined Mar 14, 2008
34,452
Before we can determine how best to convert the 10MHz sinewave to a square-wave, we need to know what the amplitude of the 10MHz sinewave is (state whether the answer is peak, peak-to-peak, or RMS).

Last edited:

#### Ian0

Joined Aug 7, 2020
9,819
I thought that the 74HC4040 etc had Schmitt trigger inputs, but I was mistaken - the CD4040 does, but the 74HC4040 doesn't.
CD4040 would only manage 10MHz on a 12V power supply.

#### ericgibbs

Joined Jan 29, 2010
18,849
Hi,
The HEF4040B is a fast counter, has slow rise tolerant inputs, also available in DIL package.
E

#### Attachments

• 207.2 KB Views: 2

#### Ian0

Joined Aug 7, 2020
9,819
Hi,
The HEF4040B is a fast counter, has slow rise tolerant inputs, also available in DIL package.
E
That's interesting, because the Texas version is only guaranteed for 5MHz @5V.

#### ericgibbs

Joined Jan 29, 2010
18,849
I have run the HEF at 10V, 10MHz, with no problems

#### crutschow

Joined Mar 14, 2008
34,452
I have run the HEF at 10V, 10MHz, with no problems
But it would appear to be marginal at 5V.

Anyway, after some further thought, I don't think the 4020 or the 4040 would be a good choice here, since the output frequency from a 10MHz clock would be fractional cycles, which would not lend itself well to obtaining a 1Hz signal.
Likely better to use binary counters configured to divide by factors of ten.

#### Ian0

Joined Aug 7, 2020
9,819
But it would appear to be marginal at 5V.

Anyway, after some further thought, I don't think the 4020 or the 4040 would be a good choice here, since the output frequency from a 10MHz clock would be fractional cycles, which would not lend itself well to obtaining a 1Hz signal.
Likely better to use binary counters configured to divide by factors of ten.
unless you want to decode 100110001001011010000000. A 4024 and two 4040s and a 8-input and gate, or seven 4017s.

#### crutschow

Joined Mar 14, 2008
34,452
unless you want to decode 100110001001011010000000. A 4024 and two 4040s and a 8-input and gate, or seven 4017s.
Or how about four 74HC390 dual-decade ripple counters, configured to give a total division of 10 million.

#### Ian0

Joined Aug 7, 2020
9,819
Or how about four 74HC390 dual-decade ripple counters, configured to give a total division of 10 million.
I'd forgotten about that one. Now there are two four-chip solutions!
I think that decoding an 18-stage ripple counter clocked at 10MHz might be a recipe for race-hazards.
I thought I'd found a three-chip solution based around a pair of 40103s but it needs to divide by 78125 (max is 65536)

#### Ian0

Joined Aug 7, 2020
9,819
It needs to divide by 10^7 which is (2^7)*(5^7).
74hc4024 dividing by 128, (74HC)4059 dividing by 625, (74hc)40103 (or another 4059 but the 40103 is smaller and cheaper) dividing by 125.
Having the 4024 at the end instead of the beginning does give a squarewave at 1Hz which can be used to flash the colon.

#### Ian0

Joined Aug 7, 2020
9,819
But don't you need extra logic chips to generate those divisor values, making the total chips more than four?
A 40103 can manage it on its own - not sure about the 4059 as I've never used it - I assumed it worked like a 40103.

#### crutschow

Joined Mar 14, 2008
34,452
Sorry, I removed my thread when I realized you didn't need any extra gates.

#### AnalogKid

Joined Aug 1, 2013
11,048
The SMA output is 0.5 Vrms with a 50 ohm source impedance. That's a bit small to drive a Schmitt Trigger CMOS input reliably. The output voltage should be higher into a high impedance load, but there is not guarantee of a specific value. Still, I would start with a simple 2-tranistor comparator. It could produce a 10 MHz square wave with a duty cycle of about 25%. That should be enough to drive a decade divider string.

ak

Last edited:

#### AnalogKid

Joined Aug 1, 2013
11,048
Amplify the signal to 0-5V
That's the high-speed converter part.

ak

#### AnalogKid

Joined Aug 1, 2013
11,048
You don’t need to do that. A Schmitt trigger will turn the sine wave into a square wave.
At 10 MHz, with ships that can barely keep up with that clock speed, you don't need a Schmitt Trigger input. No standard 4000 series CMOS part is going to interject false or spurious edges.

ak