Talk:Ring oscillator

Loop time matters
A single inverter does not provide enough phase difference for the ring oscillator to oscillate (Nyquist criterion of stability)


 * Does the physical size of the ring affect this? --64.232.164.46 19:18, 3 July 2006 (UTC)

Given a sufficiently long connection between the output and input, a single gate structure will oscillate. For a TTL logic gate the rise time is only a few nanoseconds so it will take a couple feet of wire to create enough phase difference between in and out. —The preceding unsigned comment was added by 192.128.134.68 (talk) 21:03, 19 January 2007 (UTC).


 * True, and if the wire wasn't long enough the output would sit at the Vmid of the p and n transistors based on the gate strength Guerberj (talk) 22:58, 20 January 2009 (UTC)

Applications?
Why would anyone go through the trouble of fabricating one? — Omegatron 01:51, 28 April 2007 (UTC)

They're commonly used in PLLs, which are an important part of many current high-speed chips. 153.96.171.70 (talk) 18:38, 16 January 2008 (UTC)


 * Good question. I've added an "applications" section to the article, which lists the 3 applications I know about. Are there any others? --68.0.124.33 (talk) 20:45, 7 October 2008 (UTC)


 * I find the statement "The voltage-controlled oscillator in most phase-locked loops is built from a ring oscillator." odd. Having seen a lot of PLLs, I think most of them do not use a classic ring oscillator (rather just a cross-coupled differential pair, and while one of the stages in the differential pair could be considered a inverter, it is not an odd amount). I would propose to modify the statement to "In some PLLs, a ring oscillator is used as voltage-controlled oscillator". Comments? TheUnnamedNewbie (talk) 05:59, 30 June 2018 (UTC)

SRAM
Possible error in the following sentence? : "A circular chain composed of an even number of inverters cannot be used as a ring oscillator; the last output in this case is the same as the input. However, this configuration of inverter feedback can be used as a storage element; it is the basic building block of static random access memory, or SRAM."

static random access memory, or SRAM should, maybe, be replaced by 'dynamic random-access memory or DRAM. — Preceding unsigned comment added by 137.138.196.171 (talk) 10:31, 29 November 2011 (UTC)


 * No, it shouldn't. DRAM is pretty much a single transistor with a high enough gate capacitance to store a bit, for a short while. SRAM, as you noticed, is a flip-flop, nothing like this, tho you can make one from 2 NAND gates feeding each other. So apart from, maybe, an unreliable delay-line, no sorts of memory are made from NOT gates, or inverters. 92.40.254.58 (talk) 10:46, 25 April 2013 (UTC)


 * How could we improve the article so people are not left with the above misconceptions?
 * Yes, SRAM almost invariably uses two inverters (two NOT gates) -- what Static random-access memory calls "two cross-coupled inverters".
 * I suppose one could make SRAM from 2 NAND gates feeding each other, or from 2 NOR gates feeding each other -- something like a SR latch.
 * And I suspect that such an arrangement was pretty common in the days of resistor–transistor logic and perhaps a few generations after that.
 * But my understanding is that the vast majority of all SRAM manufactured today stores a bit in two cross-coupled NOT gates built exactly as described the inverter (logic gate) article as "Static CMOS Inverter". I can't find a standard NOR gate or a standard NAND gate in the standard 6T SRAM cell. --DavidCary (talk) 03:09, 25 July 2013 (UTC)

Drawbacks?
Are there any drawbacks with using these? Are they used to generate clocks signals? Doesn't a ring oscillator consume a lot of power because of constantly switching the inverters? 79.136.62.165 (talk) 09:58, 10 February 2012 (UTC)

Ring oscillators have some characteristics that make them less suitable than other electronic oscillators in some applications. In particular, a crystal oscillator is a better frequency reference -- the crystal has far less sensitivity to temperature and voltage than a ring oscillator.

A ring oscillator is sometimes used to generate clock signals, although far less often than the Pierce oscillator.

Some ring oscillators use amazingly little power. A ring oscillator typically consumes *less* power than a Pierce oscillator or other square-wave oscillator at the same frequency. Any kind of electronic square-wave oscillator, in turn, typically consumes *less* power than any other kind of oscillator. For example, the Vratislav reference in this article describes several ring oscillators. Many people think the amount of power required to turn on one green LED ( 40 mW ) is pretty small. The *worst* power consumption of the ring oscillators described by Vratislav (0.65 mW) use better than an order of magnitude *less* power than the LED. Other ring oscillators described by Vratislav (0.003 mW) use better than three orders of magnitude less power than the LED.

How could we improve the article to make this clearer? --DavidCary (talk) 20:47, 28 July 2013 (UTC)

Comments on the article
1. "Period of ring oscillator vibrates in a random manner T=T+T' where T' is a random value."

This looks like a C++ statement. The expression needs to be revised and it doesn't work here. Possibly, t should be used because T is used for period.

2. "When the local silicon is cold ..."

This sentence seems to imply that a device is made of silicon. If that is the case, it should be clearly stated.

3. Has anybody been able to simulate one of the two provided schematics and obtained oscillation?

ICE77 (talk) 00:11, 1 August 2015 (UTC)

Proposed addition: chaotic signal generation
( Lminati (talk) 15:23, 12 March 2019 (UTC) )
 * Ring oscillators having lengths equal to the smallest odd prime numbers, i.e. 3, 5, 7 etc., can be cross-coupled (for example using diodes, or other similar elements) to collectively form a chaotic oscillator, whose principle of operation is linked to the fact that the primal lengths hinder the emergence of stable phase relationships between the rings. The resulting circuit is purely CMOS, namely, does not contain discrete resistive or reactive elements, hence is suitable for area-efficient realization of on-chip chaotic networks, with possible applications related to emergence in coupled oscillator systems, and more generally to chaotic signal generation.