Talk:CMOS

Simple English please?

IF somebody would just explain the job a CMOS does, that'd help http://simple.wikipedia.org/wiki/CMOS

First of all, there's no such thing as "a CMOS". CMOS chips do almost ALL electronic jobs these days; every watch, phone, computer, television, etc. is full of them, doing all those jobs. Dicklyon 22:36, 6 November 2006 (UTC)

Digital electronic jobs, at least. We can't completely discount analog systems :) Do we have an article on BiCMOS? (yes, seems we do) Edit: The task of adequately explaining CMOS in "simple" terms is an unenviable one. (though not as unenviable as simple:Quantum mechanics) -- mattb @ 2006-11-06T22:54Z

Most analog electronic jobs these days are also done by CMOS. Image sensors, ADCs and DACs, op amps, and systems made of such things, all CMOS. BiCMOS plays a minority role. Dicklyon 03:56, 7 November 2006 (UTC)

Maybe I'm missing something, but to me CMOS implies complimentary logic gates in the design; something that's not particularly useful in analog design. Sure the fabrication process for Si CMOS and Si analog ICs is similar, but I can't understand how one could justify calling the analog design "CMOS". -- mattb @ 2006-11-07T04:31Z

I don't know if I am right, but I have no idea what a CMOS is, besides an electrical component that's used in many ways, even after reading the discussion. If one can't explain what is this? what it IS, maybe one can say what it DOES. Since I know I'm not dumb and I do not get what a CMOS is, and want to know it, the information given here seems insufficient to me.--Rudolf Saathof (talk) 13:49, 25 June 2009 (UTC)

After that I reacted I talked to an electrical engineer. He told me that the point of CMOS is that it is not so much a functional description. CMOS only indicates the fact that both NMOS and PMOS transistors are used. You could put it as follows: a couch, two chairs and a table could make a nice sitting room, but we can also describe it as furniture. That is actually what CMOS is; Furniture. If we would want to classify any circuit to it's function (sitting room), then just stating it's CMOS won't suffice. We'd have to call it an inverter, amplifier, switch or whatever the specific function of that particular CMOS circuit is. Maybe one can write some better description on the page?.. Rudolf Saathof (talk) 14:44, 28 June 2009 (UTC)

CMOS misnomer

(pasted from an edit in the article that should have been here in the first place) "anyone got any sources for this? i'm sure modern motherboards still store settings in battery backed ram (i know from having to pull the battery from them when people forget the bios password) the bios itself is in flash though maybe whoever wrote this peice was confusing the bios itself with the bios settings" -- Plugwash

The actual technology used to create the NVRAM varies. Generally flash memory is cheaper these days and is thus the most obvious choice for storing BIOS program settings. The battery backup is mostly for the hardware RTC in these cases. Removing the battery to clear flash memory isn't necessary (indeed, it wasn't really necessary with old CMOS RAM if the motherboard properly erased the RAM when the appropriate jumper was set) -- uberpenguin 01:54, 2005 Apr 17 (UTC)

sure you can use the jumper if you have the motherboard manual handy but often you don't. The battery on the other hand is easilly spotted and in my experiance it does reset the settings. do you know of any reliable sources saying that settings are stored in flash? Plugwash 02:32, 17 Apr 2005 (UTC)

Look up the numbers printed on some NVRAM chips. I can't give you a good reference because the best reference would probably be motherboard manuals, and those are usually written in some long lost dialect of Engrish. -- uberpenguin 03:15, 2005 Apr 17 (UTC)

thanks

THANKS FOR THE INFORMATION PROVIDED ON THE CMOS DEVICES AND THE TECHNOLOGY... I FOUND IT VERY MUCH USEFUL ... ACTUALLY I WAS LOOKING FOR SOME SIMPLE LANGUAGE FORMAT OF THE MATERIAL ON THE INTERNET FOR THE STUDENTS......... IT IS REALLY VERY NICE...........

==5 Nov 2003 (UTC)

both expansions seems to be used pretty interchangablly. Truth is its very hard to make an igfet (the key component of cmos) in anything other than silicon (silicon dioxide gives a very robust and simple way to make an extremely thin insulating layer).Plugwash 08:55, 7 Mar 2005 (UTC)

Thank you for this information. I am a high tech professional and was looking for a definistion for this word, CMOS. After I did google and glance through a few links, I found this site very useful. Thanks Again!!!

thank you for this information.I am a comp. hardware student and i was seeing this history or definition for this word.CMOS after i did get this term through a few links. i found it now in your site thank you and god bless......

To whoever created this section: please sign yourself and try to add to previous posts instead of creating new ones such as this one ("thanks" does not say anything about the topic being discussed).

ICE77 (talk) 00:51, 22 February 2011 (UTC)

Image sensor

Here's a link i'll bookmark here about CMOS imaging.

• PDF Format - Why CMOS Image Sensors are Poised to Surpass CCDs

I'm no wikipedian (yet...) but it might be a good idea to create a CMOS(image_sensor) thingie? /Erik

yeah, you can be linked here hoping to find information about the image sensor, and you find a completely unrelated article, if nobody else does it i'll give it a try when i have some time. -- Joolz 16:29, 6 Mar 2005 (UTC)

Um, this article isn't completely unrelated. A CMOS image sensor is a sensor that uses the P/N junction available in a standard CMOS process as the photodetector. Thus the image sensor can be fabbed on a normal CMOS production line, which means they get to take advantage of either (a) older 0.25um lines which have been completely amortized at this point, and so produce fairly cheap silicon or (b) newer, tighter lines which get smaller pixels on the sensor.Iain McClatchie 08:26, 7 Mar 2005 (UTC)

Static power consumption

I added some info on the recent problems with static power consumption. I wonder if is is too much or too little. There is, of course a lot to be said on the subject. For instance, it is not only sub-threshold leakage, but also effects like gate tunneling and several others. The current way of handling this is to use MTCMOS, Multiple Threshold CMOS. This makes it possible to have transistors with different threshold voltages on the same chip. So we can have both high speed, high leakage gates and low speed, low leakage on the same chip. This has been used all the way back to 0.18 micrometre cell libraries.

I have heard reports of some designs that had a static power consumption of up to 50% of total power.

What do you guys think? How much shuld we write about this?

Where has your text gone? The basics should be explained. There's a badly rated article on this, see my merger proposal from multithreshold-CMOS. —Preceding unsigned comment added by 153.96.104.2 (talk) 11:08, 17 August 2009 (UTC)

Hyphens?

Where are the Missing HYPENS

CMOS stands for Complementary METAL-OXIDE-SEMICONDUCTOR.

This is NOT the same as Complementary Metal Oxide Semiconductor. The Construction method is A base layer of N-type or P-type Semiconductor material (usually SILICON). An insulating layer of Silicon Dioxide is then deposited on the semiconductor layer. A layer of a metal usually Aluminum or tungsten is then deposited on the metal layer . Regions of both P-Type and N-type semiconductor is then made on the the surface. This creates the necessary complementary transistor junctions needed for the basic GATE package. The use of the three layers of METAL,OXIDE,SEMICONDUCTOR is the derivation of the ACRONYM MOS NOT a mythical not existent Metal Oxide Based Semiconductor material. — Preceding unsigned comment added by 203.87.74.39 (talk) 23:20, 23 July 2004 (UTC)

Currently, the article is using the phrase "Complementary metal–oxide–semiconductor". Why is this using dashes instead of hyphens? Why is the last term joined together with the others? This seems contrary to MOS:DASH and MOS:HYPHEN. —BarrelProof (talk) 13:19, 26 April 2019 (UTC)

Now starts WW3. EEng 18:23, 26 April 2019 (UTC)

Apparently not. It's been more than 24 hours since I made that comment and edited the article to use "Complementary metal-oxide semiconductor", and there is no sign of trouble so far. —BarrelProof (talk) 22:04, 27 April 2019 (UTC)

Your version would suggest that there's a metal oxide involved, and that the relationship of the semiconductor to the metal and the oxide is not symmetric. That's just wrong. It's about a stack of materials: metal–oxide–semiconductor. The oxide is usually silicon dioxide, and the metal is often not a metal, but that's what it refers to and that's why dashes are used in sources that want to help the reader parse the meaning. Dicklyon (talk) 22:08, 27 April 2019 (UTC)

It's good to see that someone is paying attention. Your explanation is much more clear than that of the IP in 2004. —BarrelProof (talk) 22:51, 27 April 2019 (UTC)

I'll drink to that. Dicklyon (talk) 02:18, 28 April 2019 (UTC)

Cleanup

This article is too technical, and needs to be better organized. Andros 1337 21:32, 8 Jan 2005 (UTC)

Pardon me for saying, but this article has little to no context unless you DO use highly technical terms. Trying to explain it in laymens terms would be like trying to explain vector calculus to someone who knows only basic algebra or microkernel theory to someone who has only programmed in Python; there is no context. You can't expect the article on, say, integrals to be simplified for someone with little exposure to formal mathematics, likewise you can't expect an article about an electrical engineering concept to be simplified for someone with no EE experience or knowledge... I can see a few places where clarity (and certainly organization) can be improved, but it is pretty difficult to explain CMOS in terms everyone can understand. -- uberpenguin 23:00, 2005 Jan 14 (UTC)

• It strikes me that the article would at least seem less technical if the "history" section preceded the more technical discussion. I think it helps the layman reader if the invention, manufacturers, and applications of CMOS are given priority, with the more difficult technical details occurring later, for readers who have the necessary context and want to read further. -- Wapcaplet 23:09, 14 Jan 2005 (UTC)

• I agree with you on that point, better organization would help a lot. -- uberpenguin 20:46, 2005 Jan 19 (UTC)

The article was not too technical, IMO, but it did need to have additional introductory material at the beginning for those readers wanting just the gist. -R. S. Shaw 04:02, 2005 Jan 21 (UTC)

This topic has plenty general interest and should give lay readers several facts:

• CMOS is built using MOSFET transistors. See MOSFET for more info.
• Complementary static CMOS gates have a p-network and an n-network. This is how 1's and 0's are generated and used.
• Other ways of connected the transistors are possible. See dynamic logic and cascode for more info.

That's should keep any lay reader happily learning for hours. Which is what they came here to do. Potatoswatter (talk) 08:09, 7 May 2008 (UTC)

2005.02.02 Edits

I am reverting this article to the version prior to Akidd dublin's edits on 2 February. The reason for this is that the material added in the introduction is already covered in the introduction and elsewhere and doesn't really clarify anything in my opinion. Furthermore it really breaks the flow of the article and doesn't fit stylistically with the rest of the content or the general Wikipedia standards. If the user who added these edits still feels they belong, could they please try to reformat them to fit better and post here explaining why they feel these changes are necessary before editing the article again. Thanks! -- uberpenguin 13:13, 2005 Feb 2 (UTC)

being stylistically: digital logic can be tricky. the article needs to be rewritten. "wire" "burn"

my NOTICE shows how the CMOS NOT gate works, and covers the power consumption. capitalization: within programming, listings look very much better, thats all.Akidd_dublin 200502061007 1049

My main problems with what you included were stylistic; they broke the flow of the article. Including an example of one basic CMOS logic implementation is a good idea, but it probably should be put in its own section, not in the introduction, and should be formatted a little bit more clearly. I don't believe including a description of a specific gate clarifies or simplifies CMOS any, but it seems ill at place in the introduction section, which is why I removed it. If you still would like to include an example, try rephrasing it to fit better with the article, post it here, and we'll talk about where would be an appropriate place to put it in the article. -- uberpenguin 13:55, 2005 Feb 6 (UTC)

the term MOS is already explained in the MOSFET article. my include would describe the term complementary and also the static nature, power is consumed by electric capacities within the structures. the include would make many lines obsolete: means more than a minor edit. indeed there are actual cpu's which are described as being CMOS, but i would guess their technology is derived from CMOS. sometimes i have read the term static design: the technology is in use for low-power applications.Akidd_dublin 200502070852

The `complementary and static nature' of CMOS is already adequately described elsewhere in the article, I don't see why you want to edit it to be more simplistic rather than more complete. Furthermore, nearly every CPU and microcontroller design within the last decade has been CMOS and not some derivative (the manufacturing process of the FETs has changed, but CMOS is a design technique not a manufacturing technique). NMOS and TTL are far too dated for the high density applications needed today. CMOS is also used in low-power applications, there are just some additional considerations for creating low-powered microprocessor. No offense, but is English your first language? I'm having a hard time understanding some of what you are saying. -- uberpenguin 14:17, 2005 Feb 7 (UTC)

let's parse the sentence:CMOS circuits burn power by charging and discharging the various load capacitances (mostly gate and wire capacitance, but also drain and some source capacitances) whenever they are switched

the circuits burn power: no charges with different electric potential are equalized.*
charging and discharging: charging contains discharging logically or would charge up forever.
various load capacitances: hhm most people can make the assumption of "various"
gate and wire capacitance: basically these are structures of integrated circuits, but i am not happy with the term wires as such.
(mostly..): electrical capacities within the silicion structure? mostly, but also.., some source..: this can vary due to the layout.
whenever they are switched: logic does not have to be clocked but usually it is.

most circuits are CMOS: as far as i know, there is also LOCMOS, SOS and if you have a mobile phone: it's GaAs technology. i am not an expert, have read PIC references and have build circuits with around twenty chip cases.

about my languages: people have never understood me.

however, i could make up examples. the article should orientate to intel, microchip, motorola documentation style, without words like "some".

probably you know yourself from programming references the feeling of instant understanding. but language does not always model the exact physics model: and its only a model. Akidd_dublin 200502071756

• add a colon after no: have seen it like this in physic books
  
  

the SRAM article is of poor quality. the statement chips would consume around one watt is objection. by the way i know the CMOS cookbook but made the experience that circuits have become obsolete because of PAL/GAL PIC. however, i have found a 4060 in a toaster two years ago, used for timing. this would mean the 4000 CMOS logic product line. phillips is providing detailed information about what ACT and so on means.

perhaps the article should outline the timeline, and the excact terms for current circuits, if you say nearly everything nowadays would be CMOSAkidd_dublin 200502071820

In the above I see objections to the terms "burn" and "wire". I think there's some merit in changing the uses of these terms in the article:

• "burn" - This is a sort of slang usage: I think "use" would be better.
• "wire" - Seems a little misleading, since 'wire' seems to connotate round, flexible things. I don't know what term might fit better though. Any suggestions?

-R. S. Shaw 22:59, 7 Feb 2005 (UTC)

"burn" is indeed colloquial usage, maybe we should change that to "dissipate", which is what circuit designers use when they're being formal.

"wire" is correct. They aren't round or very flexible, but they are referred to as wires in all the english technical literature. (Sometimes they're referred to as interconnect.)

Overall, though, I agree that the article is a mess and should be rewritten. Iain McClatchie 23:06, 7 Feb 2005 (UTC)

Yes, it really is in rough shape... I'd love to do some major re-writing, but right now I'm pressed for time... I think it would also be a nice idea to create a linked article describing several CMOS logic implementations (also something in the category of `Things I Want To Do') -- uberpenguin 04:43, 2005 Feb 8 (UTC)

MIS

Is CMOS a special case of MIS (metal insulator semiconductor)?

--Abdull 17:48, 18 Jun 2005 (UTC)

CMOS is a design methodology (complimentary gate networks), not really a type of gate manufacture technology. CMOS uses MOSFET gates. -- uberpenguin 20:50, 2005 Jun 18 (UTC)

I find it confusing when the same word means 2 very different things. In this case, we have:

• the "static logic" CMOS design methodology logic family that involves complementary gate networks, both active pull-up and active pull-down, such that no matter whether the clock or any other input is high or low, the output pin is either pulled-up or pulled-down, but not both. Such a design technique is a kind of static logic, since it holds its state properly when the clock is stopped, and also uses almost no power when the clock is stopped. Such designs could theoretically be implemented with several fabrication technologies such as discrete nFET and pFET transistors, BiCMOS, silicon on insulator, etc., although most implementations use the CMOS chip fabrication technology. Many modern CPUs do not use this CMOS design methodology -- instead they use dynamic logic (digital logic).
• the CMOS chip fabrication technology, which puts both nFETs and pFETs on the same chip. Practically all transistors produced in the world today are produced using this fabrication technology, including all modern CPUs. This fabrication technology can be used to produce both digital static logic ("CMOS design methodology") and digital dynamic logic chips, as well as analog chips (in particular, CMOS sensor) and mixed-signal chips.

What can we do to reduce the confusion? Is there 2 other terms (other than "CMOS") that would be more appropriate for these 2 things? --68.0.124.33 (talk) 04:02, 23 August 2008 (UTC)

How-to tips should be moved

Is there a better article for all the how-to tips on configuring PCs? It doesn't feel to me as if they belong here; I wouldn't put a set of instructions for my CMOS wristwatch here. Where to put the how-to information? --Wtshymanski 17:18, 20 Jun 2005 (UTC)

The BIOS CMOS deserves a separate article of its own, linked from both CMOS and BIOS. The howto roadmap (it would have to be more detailed to warrant calling it tips) could then fit in better. --Shaddack 21:02, 20 Jun 2005 (UTC)

Done. Iain McClatchie 07:53, 26 Jun 2005 (UTC)

Now at CMOS RAM, aka Nonvolatile BIOS memory. --68.0.124.33 (talk) 16:31, 22 August 2008 (UTC)

WTH happened to Image:CMOS_NAND.png?

It appears this image has been deleted for some reason. Can someone tell me why? I created that image, and if I need to tag it with some rights release that's no problem.

Iain McClatchie 21:41, 27 July 2005 (UTC)

Deleted because: "Listed at WP:PUI for 30 days and not tagged by the uploder."

Yeah, it was deleted because you didn't have a license tag. I can undelete the image description but the image is deleted forever so you'll have to re-upload it. Make sure you add tags to all your images. They should have notified you on your talk page that it was being deleted, but if you had the image on your watchlist you would have seen it, too. - Omegatron 21:52, July 27, 2005 (UTC)

I found the image very helpful. I hope you can find the time to reupload it. -R. S. Shaw 23:41, 27 July 2005 (UTC)

All right, I'll try to get to it this weekend. This time I'll put in a GFDL tag. Grrr. I definitely didn't see this on my home page or watchlist. Iain McClatchie 22:42, 28 July 2005 (UTC)

The image uses unconventional symbols for the MOSFET transistors. Usually you indicate N and P channel with an arrow - the bubble on the top two transistors is non-standard. I don't think it's quite complete, either, since multiple-input gates would have increasing thresholds. And what does "&h8203" mean? It doesn't reproduce as anything comprehensible on my little-known and obscure Web browser. --Wtshymanski 20:05, 26 August 2005 (UTC)

I removed the #8203 unicode char I put in; Firefox interprets it correctly but IE doesn't. I don't know nothin' 'bout thresholds or symbols, but I do note that MOSFET shows alternative symbols using bubbles and arrows. -R. S. Shaw 21:14, 26 August 2005 (UTC)

My digital VLSI teacher always used the bubble for pFET, and never drew a circle around the transistor. I suspect that's because he drew hundreds of the pFETs and nFETs, and the little bubble thing is just as unambiguous and much quicker to draw than the little arrow thing.

And he always drew the same 2 symbols, no matter how many inputs there were to a gate, or whether or not there were "increasing thresholds".

I haven't yet learned analog VLSI, so I wouldn't be surprised if there were slightly different symbols to indicate variations in channel "length", thresholds, etc.

I would be interested in learning about other ways of drawing transistors, although I suspect MOSFET#Circuit symbols or transistor#Types would be a more appropriate place for such alternative drawings. --68.0.124.33 (talk) 22:50, 22 August 2008 (UTC)

Stuff I'd like to see here

These are notes for myself or anyone else who feels like contributing. It seems like a lot of artwork, I guess because hardware design, especially process and layout, is quite visual.

• Better explanation of the difference between a CMOS circuit and a CMOS process

• Cross section of a PMOS gate, and another cross section of an NMOS gate, in a double-well process (substrate and a simple NWELL implant).

• Same diagram, pointing out the unwanted PNP transistors and the resulting SCR. Description of latchup (forward bias the source/drain diode to well/diffusion, turn on the PNP, turn on the SCR, kablooey). Description of how latchup is prevented.

Diagrams of various major process options:

• Cross section of PMOS and NMOS gates, triple well process

• Cross section of PMOS and NMOS gates in a SOI process.

For each of those process options, a series of pictures showing how they get built up would be nice. This is a lot of artwork, but I think it's the best way to convey the information.

I'd also like to see a representation of the scale of modern chips. Perhaps artwork of a transistor (100 nm indicated), then an inverter (1 um indicated), that inverter amidst rows of standard cells (10 um indicated), rows of standard cells in a block (100 um indicated), and a micrograph of a chip or full chip layout (1 mm indicated) with a block labelled. Each pic could have a scale and highlight the component in the previous pic.

And, um, if anyone is going to comment that this material has to be structured, broken into many pages, yes, of course, but first it must be written and drawn, which is a lot harder.

Iain McClatchie 21:30, 30 October 2005 (UTC)

Latchup already has its own article. --Shaddack 05:24, 21 January 2006 (UTC)

About CMOS

Hi,, Can you please explain if their is a chip call "CMOS RAM"? Or where we can find CMOS chip? is with in ROM chip? pat..

See CMOS RAM. It used to be a SRAM with a back-up battery, but today's it's mostly a serial EEPROM. --Shaddack 05:20, 21 January 2006 (UTC)

THANK YOU SHADDACK FOR HELP....... pat

Comparison

Could it be said, that

• TTL is a series circuit of a
  • common base
  • common emitter
• and ECL is
  • common emitter
  • common collector
• and CMOS is a single stage
  • common emitter

? -- Arnero 17:33, 27 January 2006 (UTC)

Please answere at TTL

I can't understand what you are trying to get to or what your logic is. Common emitter, common base or common collector are BJT configurations. CMOS does not even use BJTs.

ICE77 (talk) 01:00, 22 February 2011 (UTC)

CMOS vs CCD

So how do CMOS sensors compare to CCDs? I remember reading somewhere that although the CCDs are harder to manufacter, they are better for image quality; but then, I've also read that CMOS sensors are better in that way. And how do 3CCDs compare to a single CMOS? —The preceding unsigned comment was added by 24.23.6.222 (talk • contribs) .

A CMOS sensor uses a CMOS inverter per pixel as a preamplifier. This reduces the space available for the photodiode, and in turn the sensitivity and dynamic range.--Arnero 14:55, 28 January 2006 (UTC)

but presumablly it also means they can be fabricated in a standard cmos fab and i believe there is a LOT of spare capacity in older (Larger feature) CMOS fabs that people are looking to find other uses for. Plugwash 17:51, 28 January 2006 (UTC)

So what is the difference then? A CCD only needs nMOSFETs and under-utilizes a CMOS-Fab. But presumablly it needs higher quality silicon, to not loose any electrons in the very long chain to the first amplifier.--Arnero 13:34, 29 January 2006 (UTC)

gate leakage reduction with a silicon superlattice

In addition to high-k dielectics, another approach to gate leakage reduction is to use a silicon superlattice to enhance drive current whilst siultaneously reducing leakage. This has advantages that it can be incorporated into the production process without costly re-tooling and also superadded onto other techniques such as silicon-on-insulator (SOI), giving additive benefits. See "Addressing Gate leakage with Rerengineered Silicon" (www.reed-electronics.com/semiconductor/article/CA6418540?pubdate=3%2F1%2F2007Lesterruss 10:55, 25 March 2007 (UTC)

CMOS' invention

Who invented CMOS technology? --Abdull 22:00, 30 November 2007 (UTC)

According to this page, Frank Wanlass of Fairchild Semi in 1963. Others say Wanlass and Sah. Dicklyon 05:58, 1 December 2007 (UTC)

NAND Layout

I think that the layout is bad. There are no connection to the polysilicon gates, and the desing won't work. See [1] for a good nand layout. —Preceding unsigned comment added by 200.126.158.167 (talk) 06:53, 18 September 2008 (UTC)

It seems alright to me. The picture you linked to doesn't have anything more that resembles a connection to the polysilicon gate. I think the overlapping of the polysilicon onto the NMOS and PMOS is clear enough.Liuyuan Chen 21:01, 19 September 2008 (UTC)

The one he linked does show contacts to metal lines that come in from the edge of the cell. It would be nice to see that on the layout we have, but some cells like this one are designed so that you can lay down contacts later. The fact that they're missing doesn't make it wrong. Dicklyon (talk) 02:12, 20 September 2008 (UTC)

For CMOS setting - For cange or edit in CMOS features in Computer system.we used two methods 1. Standard settings 2. Advance setting —Preceding unsigned comment added by 122.168.14.35 (talk) 09:26, 4 January 2009 (UTC)

Enhancement or depletion?

do they use enhancement or depletion mode mosfets? or both? - Omegatron 02:19, July 11, 2005 (UTC)

CMOS always uses all enhancement-mode MOSFETs (in other words, a zero gate-to-source voltage turns the transistor off).

(Should we specifically mention that in this article, or delegate it to the MOSFET article?)

Depletion-mode MOSFETs are on (conducting) when the gate-to-source voltage is zero.

It's physically possible to wire up some depletion-mode MOSFETs in a typical CMOS configuration. Such a hypothetical depletion-mode gate, by itself, might seem to "work" if you applied an input voltage far enough *above* the VCC rail to turn off the pFETs, or far enough *below* the GND rail to turn off the nFETs.

But an input voltage anywhere between the power rails (such as the output voltage from a similar gate) would turn on all the transistors in the gate, both depletion-mode pFETs and depletion-mode nFETs, shorting the power rails together.

As far as I know, the only depletion-mode mosfets ever used in digital logic were in the long-obsolete HMOS logic family.

For more information on enhancement mode vs. depletion mode, see MOSFET#Depletion-mode MOSFETs and Field effect transistor#FET operation.

--68.0.124.33 (talk) 22:11, 22 August 2008 (UTC) (Edited) --68.0.124.33 (talk) 12:09, 28 June 2011 (UTC)

I'll mention it briefly in Technical details. - Rod57 (talk) 08:33, 4 January 2016 (UTC)

Depletion CMOS?

I have a little problem w/the lack of explanation for the NAND layout. If you look at the polies, below them is diffusion. This makes them depletion-type MOSFETs. Now I realize that by convention when a CMOS layout is drawn a single monolithic diffusion is drawn for both the drain and the source, making channel under the gate by default, and that it's up to the reader to intuitively know that below the poly is really substrate or well. Unfortunately, I didn't know that last week, and had to find my prof before I could understand. I think this should be explained in the article. AngusCA (talk) 05:42, 15 February 2010 (UTC)

You could be bold and put something in yourself if you understand it, and think it will improve the article. Jwoodger (talk) 05:50, 15 February 2010 (UTC)

With the kind of trash-talk I've taken for making a perfectly reasonable but infeasible suggestion on a talk page, calling attention to a point of confusion about something I only learned last week is the boldest I get. AngusCA (talk) 04:20, 16 February 2010 (UTC)

Source/Drain mis-identification?

In the image Cmos_impurity_profile.PNG, which shows a cross-section of a chip with a CMOS inverter, aren't the Source and Drain mis-identified in the PMOS transistor? Thank you 69.27.178.6 (talk) 18:15, 6 February 2009 (UTC)

Most likely, that transistor would be hooked up the other way around, based on the proximity of the transistors to each other and to the power rails. But as drawn, there's nothing to actually determine which would be source and which drain. It would be better to leave off the S and D, or maybe label them each as S/D. Or draw some connections to make an inverter. Dicklyon (talk) 15:20, 14 October 2009 (UTC)

Merger proposal : Multi-threshold CMOS

That badly rated page nicely explains some details of state-of-the-art measures to bring down static power consumption of CMOS circuits. It should be merged into the section on static power on this page.

Additionally, the section on static power here needs a brush up as well. -- Joerg Schreiter

Perhaps the multi-threshold article should be changed to a power-saving page, and other techniques of power saving could be added, including a link to this CMOS page? Brews ohare (talk) 14:22, 14 October 2009 (UTC)

Years later...since this seems to be a specialized topic that may be too technical and specific for a general overview article, I've taken off the merge tags and added a see-also here. --Wtshymanski (talk) 22:03, 3 October 2012 (UTC)

Baker reference

Baker is an excellent source for the entire subject of this page. It is cited in connection with its discussion of the original patent, but has a lto more to say (about 1000 pages). Brews ohare (talk) 14:20, 14 October 2009 (UTC)

Yes, Jake did a good book; but it's not the source used to write the article. And I notice that he or someone spammed his book and web site over a bunch of pages already, including this one: [2]. There's no point adding "further reading" junk like that, nor citing a ref in irrelevant places, esp. in the lead paragraph. Dicklyon (talk) 15:10, 14 October 2009 (UTC)

It is not pertinent whether Jake's book was used to write the article; it is a pivotal work in this area and should be cited. The reference to the patent number is fine, but it is not linked, and the link to Jake's book provides not only the historical figures from that paper, but a full intro to the CMOS technology. Eliminating this source as an in-line reference with a link is a backward step. Brews ohare (talk) 15:40, 14 October 2009 (UTC)

First paragraph and topic repetition

I'm not sure, but it seems that first paragraph of article contains sentences it should not contain. It references how wide is CMOS in use. It does not define CMOS by itself.
Article contains repetitive topics, also.
User:Vanished user 8ij3r8jwefi 13:59, 28 November 2009 (UTC)

Temperature range error?

A sentence in the "Temperature range" section reads "There were theoretical indications as early as August 2008 that silicon CMOS will work down to 40 kelvin, or -243 °C.[2]"

Isn't 40 kelvin -233 °C, not -243 °C? Conversely, wouldn't -243 °C be 30 kelvin?

If anyone has read the cited paper (Edwards C, "Temperature control", Engineering & Technology Magazine 26 July - 8 August 2008, IET), please correct one or the other to make them consistent.

Based on the paragraph alone, 40 kelvin is probably correct and -243 °C is probably erroneous. . . —Preceding unsigned comment added by Needlesslystilted (talk • contribs) 00:44, 11 January 2010 (UTC)

Output is inversion of input

this section is misleading, as it is describing the characteristics of a CMOS Inverter. A CMOS inverter is one example of a CMOS circuit, but not all CMOS circuits are inverters. For example if you were to say "Output is inversion of input" when talking about a CMOS NAND gate (or indeed anything else besides an inverter) you would be wrong.

The other problem with this section are the first two sentences: "CMOS circuits are constructed so that all PMOS transistors must have either an input from the voltage source or from another PMOS transistor. Similarly, all NMOS transistors must have either an input from ground or from another NMOS transistor." The word input should be replaced by source. Usually (and definitely in this case) one thinks of the gate as the input, not the source. Obviously the author is considering the input to the inverter to be separate from the inputs to the individual transistors, which is confusing and unhelpful in my opinion. —Preceding unsigned comment added by Denki23 (talk • contribs) 05:56, 11 April 2010 (UTC)

I agree with you that this section is misleading and it has to be reconstructed. I noticed this and I reconstructed the section six months ago (see the result after my final edit). But then the page was reverted without any explanations and my edits were totally removed (see the result after removing). Since then I have been waiting for some response and some moral support from wikipedians responsible to this page...

Of course, inverting is the most inferior CMOS property as even the simple MOS circuit inverts. The complement trick is the great idea behind CMOS circuits (and not only behind them). But it is too primitive to present this configuration only as two series connected switches. This notion works well only at the ends (logical one and logical zero); it can't show what happens during the transition. To show the operation during this short time, we have to present CMOS stage as an analog amplifying stage. CMOS and other logic circuits are actually analog circuits; they are overdriven amplifiers having no biasing. And what is more interesting, they are no simple amplifying stages... they are amplifying stages with dynamic (active) load... and this is not simple active load; this is a controlled active load. What does it mean? I have explained this great phenomenon in my edits:

...In the transition region, where the input voltage is half the supply voltage, both the transistors conduct; they operate in active mode and the whole circuit behaves as an inverting amplifier with very high gain. The two transistors may be thought as two voltage-driven current sources (current-stable resistors) that are connected in series and are driven by opposite-varying complementary input voltages. They try to set the desired current magnitudes by changing significantly their present resistances in opposite directions. As a result of this interaction, the output voltage changes significantly; the gain is high and the transition zone is narrow. From this viewpoint, each of the two transistors serves as an active load for the other one...

Now about the two sentences: "CMOS circuits are constructed so that all PMOS transistors must have either an input from the voltage source or from another PMOS transistor. Similarly, all NMOS transistors must have either an input from ground or from another NMOS transistor." I removed them then as they were vague and wrong. What does "to take an input from the voltage source or from the ground" mean? Or what does "to take an input from another PMOS/NMOS transistor" mean? There is no sense in these sentences. I suppose that their author would like to describe the basic CMOS NOR and NAND configurations. Maybe, I'm not sure, he/she wants to say that PMOS transistors may be connected directly to the positive rail or through (in series with) another PMOS transistor that is connected to the positive rail; also, NMOS transistors may be connected directly to the ground or through (in series with) another NMOS transistor that is connected to the ground. From this viewpoint (that looks very, very strange), the sources of the transistors are "inputs" instead the gates...

I will start discussing and reconstructing this section soon and you may also contribute the page. Circuit dreamer (talk) 12:22, 11 April 2010 (UTC)

I find it interesting that denki23 would start this discussion as his/her first and so far only contribution to wikipedia. There is another examples of this unusual pattern Talk:Voltage-to-current_converter, also involving edits or desired edits by Circuit dreamer. Is denki23 a sock puppet? The user name doesn't even come up normally and this user may in fact not exist. It sure looks like an amateurish attempt at sock puppetry. Maybe its time for someone to investigate. Zen-in (talk) 16:04, 11 April 2010 (UTC)

Were denki23 and Circuit Dreamer's suggested changes made ? (they sound OK to me) - Rod57 (talk) 13:51, 28 May 2018 (UTC)

Article Rewrite Required

This statement from the "Inversion" section is completely nonsensical:

"CMOS circuits are constructed so that all PMOS transistors must have either an input from the voltage source or from another PMOS transistor."

The original author is trying to make the point that, with pFETs in a static CMOS digital circuit, the drain/source terminals are either connected to a positive voltage (greater than the threshold voltage, V_T), or be tied to a node which includes the gate of another MOSFET.

This article is hectic and not well-written. The entire article requires a complete overhaul. For example, the "NAND gate in physical layout" section shows a cross section of the silicon stack itself, but there is no gate. There's just two, disjoined transistors. The diagram of the NAND gate's top-level layout doesn't include the layer for a p-well or n-well. Further, I believe that the article should be split in to two: CMOS-as-a-process-technology, and CMOS-as-a-circuit-technology.

I know that Wikipedia has a tag to indicate that an article requires a re-write, but I can find no policy on when a re-write is justified or allowed. I'd really like to fix this article, but I realize that it will require a lot more than some strategically-placed bandages. Can anyone tell me where I should be directing my efforts in requesting an article re-write? —Preceding unsigned comment added by Cinnanom (talk • contribs) 14:48, 28 August 2010 (UTC)

Corrections and comments

I don't think this is a very nice article on CMOS. At a minimum, there should be a reference to the 5400 family and a schematic of a NOR so it can be visually compared to a NAND. The inverter's image is repeated twice in the article. This article does not show good substance. Here are some additional comments.

1. The sentence "There are small reverse leakage currents which is formed due to formation of reverse biased between diffusion regions and wells." needs fixing.

2. The sentence "In modern process diode leakage is very small compared to sub threshold and tunnelling currents, so these may be neglected during power calculations." needs corrections.

3. "1.d. Contention current in ratioed circuit" is an incomplete section and it needs expanding.

4. I made several corrections in the "Power: switching and leakage". For some reason, lots of words were capitalized. There were also plenty of misspellings so I assume the section is relatively new.

ICE77 (talk) 01:08, 22 February 2011 (UTC)

WP:SOFIXIT. Dicklyon (talk) 03:40, 22 February 2011 (UTC)

I am already fixing lots of spelling, grammar and punctuation errors for many articles. I have no intention to assume ideas a writer was intending to convey or expand a section that should have been completed by somebody in the first place. I list the things I see that need work and if somebody has more time than me to add, he/she is welcome to add.

ICE77 (talk) 04:01, 26 February 2011 (UTC)

Dynamic power consumption

The most of data signals usually have very low activity factors (0-0.2), except maybe for domino logic (which is used rarely today). I fixed "0.5" used in the article by 0.1 and added link to the paper in which there is a distribution of % of signals vs. their activity factor. In general, I think the "activity factor" is not explained well in the article, in my opinion it should be added as an independent article in Wikipedia (working on it...) Mkostya (talk) 18:23, 18 August 2012 (UTC)

unambiguous terms, two uses of CMOS

Some people correctly point out that pass transistor logic and dynamic logic (digital electronics) don't seem to match the description at CMOS#Duality. So when they see circuits in those and other logic family#On-chip design styles, they ask "why is this ... considered a CMOS circuit?" (Talk:XOR gate#Why is that CMOS?)

My understanding is that there is a very specific logic family F that does match the description at CMOS#Duality. My understanding is that there is a more general category C of circuits constructed from both nFET and pFET transistors that includes many logic families, including F, pass transistor logic, and dynamic logic. The ambiguity arises because most people talking about F call it "CMOS", and most people talking about C call that "CMOS".

There are many other cases where people use the same word to refer to some general category C in one sentence, and the same word to refer to some specific case F in another sentence -- genericized trademark, metonymy, etc. Is there some more specific terms we could at least mention in this article, one term that unambiguously refers to F and a different term that unambiguously refers to C? --DavidCary (talk) 01:57, 16 October 2015 (UTC)

Cinnanom (above: Rewrite required) suggests CMOS as a logic, and CMOS as a technology - is that the same distinction you are making ? - Rod57 (talk) 13:38, 28 May 2018 (UTC)

high Vth vs low Vth wording

"high Vth transistors are used when switching speed is very important, while low Vth transistors are used in speed sensitive paths" doesn't read properly for me. Missing the word "less", perhaps? Sorry if I'm using the wrong approach to flag the issue -- I'm NOT an experienced contributer! Cabbageears (talk) 01:37, 17 October 2015 (UTC)

Someone has fixed with "not" for high vth. - Rod57 (talk) 14:35, 28 May 2018 (UTC)

Should we mention FinFET here

The denser VLSI processors 14 nm, 10 nm, Intel's 22 nm etc seem to all use FinFET to achieve the high densities (and low power). Could mention here. Maybe need a clearer separation of CMOS logic from CMOS technology processes ? - Rod57 (talk) 10:23, 23 June 2017 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified one external link on CMOS. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20100915140806/http://blogs.amd.com/home/2009/01/15/breaking-records-with-dragons-and-helium-in-the-las-vegas-desert/ to http://blogs.amd.com/home/2009/01/15/breaking-records-with-dragons-and-helium-in-the-las-vegas-desert/

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 18 January 2022).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 00:51, 27 July 2017 (UTC)

Proposed talk page edit

Per ongoing discussion at Wikipedia:Village_pump_(proposals)#RfC:_Delete_IABot_talk_page_posts? and Template_talk:Sourcecheck#Can_we_change_the_standard_message_to_says_its_OK_to_delete_the_entire_talk_page_section I'd like to delete the above External links modified section. Any objections ? - Rod57 (talk) 18:57, 30 May 2018 (UTC)

What can be used instead of silicon

Under Technical details is it ok to say that other materials can be used apart from silicon. eg Germanium [3] for research. Does GaN enhancement mode transistors [4] mean GaN could be used for CMOS? If not, what else is needed ? - Rod57 (talk) 13:32, 28 May 2018 (UTC)

OK, HRL claims CMOS using GaN (and AlGaN) (just 2 transistors on one chip!) : HRL Laboratories claims first gallium nitride CMOS transistor fabrication. Feb 2016 "gate widths of the PMOS and NMOS transistors were 500μm and 50μm, respectively." "With 5V pulses with steps of 10ns rise/fall times, the output swung between 0V and ~5V with 90ns fall and 670ns rise. "Slower rise time is due to higher on-resistance of the PMOS. Although the performance is yet to be improved, the functional inverter IC proves the feasibility of the GaN CMOS technology."" - Rod57 (talk) 18:41, 30 May 2018 (UTC)

NMOS vs CMOS

"CMOS technology was initially overlooked by the American semiconductor industry in favour of NMOS" ... is the reason given correct? (NMOS was more powerful) Wasn't CMOS more complicated to fabricate vs NMOS? More layers? NMOS allowed more transistors per die?