Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Crystal oscillator start-up times

Status
Not open for further replies.

zirissjk

New Member
Hi,

I'm new to electronics forums and only a beginner circuit builder. I'm looking to find the start-up times of various crystal oscillators, primarily OCXO stratum class. I know it is difficult information to find due to the random start nature of crystal oscillators but still need to find min/max start times and was wondering if anyone here can point me in the right direction. Thanks.
 
One of the IC houses (TI, National, Motorola, HP, Analog?) had a great App Note about startup of crystal oscillators.

Not the one I remember, but... Freescale
 
Last edited:
Welcome, zirissjk!

An OCXO stratum3 class oscillator startup time is one issue (probably ranges from 10-500ms) but a stable frequency (from ambient temp) is altogether another story, depending on what level of stabilization you're looking for. These can range from 5min to over an hour.

As such, I'm wondering why the interest in startup times, as that class oscillator is of little use until its output frequency has settled.
 
Hey cowboybob,
I'm looking at the stratum 3 class as a good pairing for ToF measurement circuits. More then the initial start-up time I was looking to find the time it takes for the clock to fully settle and restart to a stable oscillation after the oven has already achieved a stable temperature, if that makes sense?

The big part of crystal oscillator start-up that throws me off is the random AC signal that starts the piezoelectric effect off.
How large a window does that part of the start-up sequence get. when I read that "X" XTAL's start-up time is between 134ms-187ms what are the governing factors of that differential?
I know that does not apply only to OCXO's but it is the main question I am challenged with.
 
Depends on what you are considering startup time. Oven control oscillators may define start up time as time to temperature stablize oven and crystal within it. This could be 10 to 30 minutes before the frequency settles down to overall frequency stabilty spec for OCXO is reached.

If you are talking about just time to steady state output oscillation it depends on crystal type and oscillator circuit design. Most OCXO are AT cut crystals held at high temp inflexioin point around 45-50 degs C. AT cut oscillators run in range of 200us to 1.5 msec time from power applied to steady state oscillation output.

But just as reference, low power oscillators running 32.768 KHz crystals in watches or cellphones can take 3 or 4 seconds to reach steady state. In cellphones there is often a low frequency RC oscillator that activates when you first switch on phone to get startup sequences going then switches clock over to more stable 32KHz crystal when it stablizes.

It is an issue when trying to do SPICE simulation of a crystal oscillator circuit. 1 msec simulation is a very long time when you are running 100 to 1000 calculation points per AC cycle of a 10 MHz oscillator. Not to mention the computer memory storage requirements for all the data points.
 
Last edited:
Hey cowboybob,
I'm looking at the stratum 3 class as a good pairing for ToF measurement circuits. More then the initial start-up time I was looking to find the time it takes for the clock to fully settle and restart to a stable oscillation after the oven has already achieved a stable temperature, if that makes sense?

The big part of crystal oscillator start-up that throws me off is the random AC signal that starts the piezoelectric effect off.
How large a window does that part of the start-up sequence get. when I read that "X" XTAL's start-up time is between 134ms-187ms what are the governing factors of that differential?
I know that does not apply only to OCXO's but it is the main question I am challenged with.
(My emphasis)

I don't believe there's a definitive answer to your basic question in that it is, in my mind, an anomalous event (i.e., some manner of an electro/mechanical interaction) that "kicks" the xtal into oscillations and is utterly random (as you noted). Hence the predicted range is made up of statistically significant startup times.

The total range, if ALL startup times were included, would be a lot wider.

And, of course, in most applications the delay is of no consequence or there are work-arounds that cover for (but do not eliminate) the anomaly.
 
If oscillator start/stop time is a problem...
... dont stop and start the oscillator!

As a general rule, if you need a very stable oscillator, you dont stop and start it.

If you need to effectively turn the oscillator on/off, use a gate of some kind to turn the output on/off, but keep the oscillator running.

JimB
 
when I read that "X" XTAL's start-up time is between 134ms-187ms what are the governing factors of that differential?

I have never seen a startup spec so precise as "134 ms to 187 msec. First this is very long for a startup of oscillator (exclusive of oven temp stabilization).

Startup is based on total oscillator circuit loop gain which not only involves the circuit of the oscillator but also includes crystal parameters, most significant being the effective Rs (indirectly Q) of the crystal.

Sometimes ultra low noise oscillator (oscillators that have good sideband noise and low harmonic content) have an ALC (or AGC if you prefer) on the oscillator circuit to adjust the gain of the active stage. You need a loop gain >1 to startup. The lower the loop gain the longer it takes to startup. You can have a startup circuit that increases loop gain at initial startup to reduce its startup time then backs gain down to improve spectral purity of the output.

The actual startup process is an impulse of energy due to power on and broadband noise of the active gain stage. Because the crystal is very high Q, only a small narrow band of this startup noise passes to the amplifier input which in turn is amplifed and put back to excite the crystal further. Loop gain greater then 1 means it will continue the feedback process building higher and higher output from the oscillator over time. Evenual something has to limit the build up in amplitude which is usually the amplifier going into compression making the loop gain exactly one, maintaining a constant steady state output.

As you started talking about oven controlled oscillator is it a bit strange to be concerned about starting and stopping the oscillator. The power consumed for the heating of the oven container is usually dominate and must be maintained to keep the temp at the high side inflexion point so why shut the oscillator circuit on and off. It is not going to save much overall power. If you need to turn the output off and on for some reason then leave oscillator on and just gate the output.
 
Hey, I can not thank you enough for your help all of you, I have scoured the net looking for a good explanation of the xtals start-up sequence and you if not all ready teachers you would make some great ones. RCinFLA you really helped me picture in my head what is going on.
I think I might have been misleading about what Im looking for, more then the speed of start-up I'm looking for stability of start-up, from powering on of the oscillator to wave form positive edge. I believe I can get that from a simple LRC circuit but want stable oscillation as well. I don't have any issues with the slow speed of the crystals start-up but more that it has such random start times. That's why I was hoping when reading "random noise" was used to begin the oscillation It only accounted for a few picoseconds of variability.
 
There will probably be an edge very soon (think microseconds or less) after you power up the amplifier. The question is, do you ask for "start up" as in when the oscillator sort-of does something i.e. toggles the output occasionally, or as in where the output frequency is within some limits?
 
An OCXO stratum3 class oscillator startup time is one issue (probably ranges from 10-500ms) but a stable frequency (from ambient temp) is altogether another story, depending on what level of stabilization you're looking for. These can range from 5min to over an hour.

I had an AN/URQ-10 frequency standard in the cal lab once and had to replace the NiCd battery pack because when sent to the "big" cal lab in San Juan, it couldn't make the trip (one problem of keeping NiCd on constant charge). Anyway, when it came back to us it was dead, so I left it cold and ordered the new battery pack. When the pack was finally installed, I turned it on and the frequency was off by 3 Hz or so according to our H-P5245L which should have had nothing but a 1 followed by all zeros on it's Nixies. That standard sat on the shelf running for about two weeks 3 Hz off and suddenly one day "BIND!" it hopped right to its correct output, counter showing exactly 10,000,000 Hz. The standard, when compared against other standards showed virtually no drift. So, sometimes it can take quite a while to operate on-frequency when brought up from a cold start.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top