Talk:Depletion-load NMOS logic
|WikiProject Electronics||(Rated C-class)|
The article's first heading has three "=" and later there is a heading with two "=":
Sources (Aug. 2008)
I have looked-up for references according to HMOS. All I've found is that it means the technologic level (see de.WP article) but no reference with the circuit topology cu --Biezl (talk) 17:36, 25 August 2008 (UTC)
- Good point. I really should've been better about that. Unfortunately the single de wp ref is for a late-generation version and hence not a good into. The topology involved is implied by the term "depletion load": there is a load device (ie pseudo-NMOS) is a depletion-mode transistor. Google for depletion load for more info. Potatoswatter (talk) 20:42, 25 August 2008 (UTC)
- Now I've visted a technical library and looked-up some books. I've found non who mentions HMOS, but always NMOS-logic with depletion load. One was written in english. Maybe you can find this in your library. Something I've recognized is that Intel_8080 had NMOS mit enhancement load and 6 µm size the orginal Intel_8086 was build using NMOS with depletion load and 3 µm size (compare HMOS I). So intel started with HMOS as depletion load was introduced.
- Finally I can't definitly say what's correct. I think the article can be left as it is and hopefully someone finds more definite sources in future.
- VLSI Design Techniques for Analog and Digital Circuits: Geiger, Allen, Strader ISBN:007100728-8 page 535
- BTW: To clearify what we're talking about look at Image:Nmos depletion and.svg which shows circuit diagram. (actually NAND)
- --Biezl (talk) 16:20, 16 September 2008 (UTC)
- HMOS was used more as a marketing term than in technical journals, but lives on in history. Therefore I think it deserves an article. Neither term is uniquely "correct." Thanks for the image! As depletion mode transistors are no longer used in logic, it would be nice to clarify the symbols though. Potatoswatter (talk) 17:35, 16 September 2008 (UTC)
high performance nmos?
Hi! I don't mean to be rude or aggressive here, but, what (significant) stylistic errors did I introduce? And, shouldn't pure facts have somewhat higher priority than such details? Also, I was planning to add information later (and let others do the same), I didn't have the time to fix everything at once.
- You misspelled "depletion," bolded words for no reason, etc.
- What "pure facts" did you add?
Only very early nMOS designs, such as the 8080 (the 8008 used pMOS), used enhancement-load, demanding +/-5V and 12V, which depletion-load circuits does not. It is also fairly clearly stated/implied in datasheets (for 8086 etc) as well as in other documents (for instance, by 8086 designer Stephen Morse et al) that Intel's HMOS, HMOS-II and HMOS-III processes are (geometrically scaled) variants of depletion-load nMOS. I don't know exactly what it is that Motorola calls HMOS, but probably something similar.
- OK, so we agree that depletion-NMOS and HMOS are the same thing.
- No, not really, Intel's HMOS is a special case of depletion-mode NMOS, they call it scaled NMOS (see Intel microprocessors - 8008 to 8086 by Stephen Morse et al, for instance). Motorola seems to call it scaled or high-density NMOS depending on the particular 6800/6809 data sheet. HenkeB (talk) 03:59, 13 January 2009 (UTC)
It's therefore quite odd (to say the least) to try to contrast HMOS against nMOS like the article now does (again). Not to be harsh, but it is nothing but a plain confusion with the distinction between depletion-load and enhacement-load nMOS. I belive depletion-load nMOS is a by far more important encyclopedia entry than is HMOS, as the term is technically well defined and actually mean something in itself (also, why call depletion load nMOS "tuned" transistors etc?). HenkeB (talk) 20:52, 12 January 2009 (UTC)
- NMOS refers excusively to enhancement-load. HMOS refers excusively to depletion-load. As for which term is more widespread, see Google. Any more questions? Potatoswatter (talk) 02:08, 13 January 2009 (UTC)
- Yes I have more questions: What did you get that from (superficial google searches?), and what do you then like to call the processes or technologies used to implement the 6800, 6502, 6510, 6809, Z80, etc (as well as early 8085 and 8086 chips)? These were all marketed as NMOS or N-channel MOS while at the same time using depletion-load (and silicon gate) technology, according to data sheets. Do you say we should retroactively rename these to "HMOS", because there is no such thing as depletion-load NMOS? That's weird man!
- BTW, the following excerpt is taken from the paper Recollections of Early Chip Development at Intel by Andrew M. Volk, Peter A. Stoll and Paul Metrovich. "The 8085, 8086, and SRAMs used the same NMOS processes. In the mid-70s, the SRAM business was seen as a larger revenue source than the microprocessors. Tweaks were made to the process to improve SRAM performance without worrying about the impact on the microprocessors. Today, it would be strange to think that an SRAM process requirement was more important than a microprocessor design. A bit later, Intel developed its dual implant NMOS process called “HMOS” for high-speed SRAMs.".
- HenkeB (talk) 03:58, 13 January 2009 (UTC)
<- Dual-implant is the (expensive) development which allowed depletion-mode transistors, and is not a "tweak." Being able to accurately align various implants is also the requisite for CMOS, so I'd expect HMOS to have been short-lived. I doubt the original 6800 had any depletion-mode xistors. Incidentally, microprocessor technologies are still optimized for and prototyped with SRAMs, as a sizable part of any MPU is SRAM cache.
- By the way, I don't want to take the silly position that depletion-load NMOS isn't a kind of NMOS… the force at play here is that adding the second implantation step was expensive, and no marketing department would miss out on the chance to attach the appropriate buzzword to their product. And, even if some products weren't marketed with the HMOS "brand," it was ubiquitous enough to become synonymous with the technology, and as Google shows, more widespread than the technical terms. So it is reasonable to call any such circuits HMOS. Potatoswatter (talk) 03:11, 14 January 2009 (UTC)
- I agree with HenkeB. The whole article is totally misleading, at best. And of course that the original 6800 had depletion-load. What references do you want Potatoswatter? You are not going to find authoritative references because as you are saying, HMOS is mostly a marketing term. It is not a term that you would find in text books. You didn't provide any references either. Where you got the idea that HMOS is a synonymous (or become one) of depletion-load NMOS? That is almost ridiculous.
- As I said to him, point out a distinction between the two terms or a source that says the 6800 generation had depletion load. I looked around for a bit and downloaded technical and marketing lit for that chip, and found no evidence at all for depletion load. And it's something that was trumpeted on the next generation, so there you go. The bare assertion that "the terms are different" is incredibly vague, you know. Potatoswatter (talk) 17:46, 23 March 2009 (UTC)
- No, that's not how Wikipedia works, and you know it. It is you the one that must provide references for your assertion. It is not us the ones that must provide references that you are wrong. This rule might sometimes be a bit too drastic, but in this case where you are the only one making the claim, and the rest of us challenging you, it makes all the sense.
There is plenty of evidence that the 6800 has depletion load. If you don't see it, it is possibly and with all due respect to you, is because you don't know enough about the topic. For starters, look at the pinout, it has a single voltage pin. Now look at the pinout of any know enhancement-only part, such as the 8080, how many voltage pins and levels do you see?
Anyway, the debate about the 6800 is not the main point. I'm sure you agree that all the rest of the devices that HenkeB lists (such as the 6502 or the Z80), do have depletion load. Some of them actually specify so in the datasheet. And nobody in the world (except you) would call the 6502 HMOS. Check, e.g., the 65C02 datasheet on the WDC site, which were written at the very least at the CMOS era (hence, after HMOS). They name the "old" (not CMOS) 6502 part as NMOS, not as HMOS.
At the very least, if you want to be coherent and you are so sure you are right, then go and modify all the WIKI articles about NMOS depletion-load devices. Modify the 6502 article, the Z80 article, the articles about Atari and Commodore ASIC, etc, and state that they are HMOS. Let's see which kind of reaction you would get.Ijor (talk) 22:16, 23 March 2009 (UTC)
- Recalling what I read during the last debate, the 6502 was designed as a depletion-load part at the insistence of the engineers but manufactured without the depletion-mode mask as the execs wanted to avoid expense and yield loss. Enhancement-mode transistors don't need high voltage to function as resistors, they can alternatively be made bigger and/or run the part slower.
- Anyway, I'm not saying that any particular chips weren't depletion load. All chips labeled HMOS were depletion load: many manufacturers chose to use this as a buzzword. Some chips not labeled HMOS were also depletion load. Some internet museums/databases use HMOS as a synonym for depletion load architecture in those chips. It is a widely recognized acronym by people who have not heard of "depletion load," or who could otherwise not connect these two different-sounding concepts.
- If you can give an example of a physical distinction between HMOS and depletion-load NMOS, then I will concede they are different. But the argument that "the term doesn't apply because they weren't marketed as such" I find to be irrelevant. HMOS is a generic term which was used by various manufacturers to denote nothing except heavily-doped load transistors, and it may also be applied to other manufacturers who didn't advertise their technology as such. Therefore I don't think it's something that needs to be referenced. Potatoswatter (talk) 01:28, 24 March 2009 (UTC)
- Also, there are many other instances within wikipedia where contention results from the choice of different synonymous terminologies. The megabyte/mebibyte debate comes to mind. In specific articles, original manufacturer terminology and "local language" must be followed. However an article covering one concept by two different names can and should introduce the terms as interchangeable. Potatoswatter (talk) 01:31, 24 March 2009 (UTC)
- LOL. So now you are claming that the 6502 is (or might be) enhancement only as well. FYI, the 6502 has been decapsulated and high magnification microphotographs of the die were taken and published. The depletion-load transistors can clearly be seen on the images.
- Potatoswatter, I think your obstination is starting to be legendary. I'm not going to beat this to dead. I repeat once again. Are you so sure you are right? Then be coherent and correct/modify all articles that mention depletion-load. Modify the 8086 article that claims (wrongly according to you) that the first version was NMOS depletion-load and then replaced with an HMOS version. Modify the 6502 article that claims it's depletion-load (and according to you it should be either HMOS or enhancement-only). Modify the VIC and SID articles (Commodore C64 ASICs) that also "claim" that first versions were depletion-load NMOS, then later versions were HMOS. etc, etc. I'm sure you would find there other people that would beat this to dead against you, I don't have the time.Ijor (talk) 06:00, 24 March 2009 (UTC)
- You're being obstinate in insisting I'm making a claim that I am not... but this is not uncommon for Wikipedia. Let me make this simple:
- HMOS and depletion-mode NMOS refer to the same technology. If a chip is advertised as HMOS, that only means it is depletion load NMOS.
- I make no other claims and no other claims are relevant to this article. I don't know for sure and don't really care about particular chips. This article isn't about particular chips. I only know Intel and Motorola advertised their parts as HMOS. Other manufacturers made depletion-load chips and advertised the process as "depletion-load" or simply "NMOS." That is irrelevant to this article.
- And I found the link: http://silicongenesis.stanford.edu/transcripts/mensch.htm . Bill Mensch talks about how the 6800 was originally not depletion load, tho the 6502 is. That's what I remembered wrong. Anyway, irrelevant. Potatoswatter (talk) 17:00, 24 March 2009 (UTC)
- Yeah, I am obstinate, HenkeB is obstinate, those that wrote the 8086 article, and the VIC and the SID one, they are all obstinate. We are all wrong, and you are the only one that are right.
No, you are not just claiming that HMOS chips are depletion-load. We all agree with that. You are also claiming that HMOS is a synonym to NMOS depletion-load. And you are also claiming that any NMOS chip that is not HMOS, is not depletion-load. You are now changing the wording slightly (you said before that "NMOS refers excusively to enhancement-load"), now you are saying that it depends on the manufacturer.
HMOS is just the term used for the later generations of NMOS depletion-load. Not for all depletion-load technologies. It is not just a matter of some manufacturers using one term and other manufacturers using the other. HMOS was always used for the later depletion-load devices, and it was never used for the earlier ones.
This is from the Motorola M6805 user manual:
"HMOS Features The current state of the continual shrinking of NMOS is called HMOS (High-Density NMOS)."
You see, there is no mention to depletion-load as being the *new* characterstic of HMOS. Depletion-load is not mentioned at all because it was implied. What is new and different in HMOS is the transistor and feature *size only*, based on improvements on the lithographic and fabrication process. There is no new topology, that is what we are all saying from the beginning.
From the Motorola MC8621 datasheet:
"NMOS (N-CHANNEL SILICON-GATE DEPLETION-LOAD)"
You see. Motorola used both NMOS depletion-load and HMOS terms depending on the part. It is not that Motorola always used HMOS for all his depletion-load devices.
From the MOS 8561 (SID, HMOS version): "HMOS"
Once again MOS using both NMOS depletion-load and HMOS depending on the generation and shrink size technlogy.
The Intel 8085A datasheet doesn't mention HMOS, only the 8085AH and later versions mention the term HMOS (and btw, why do you think they added the "H" to the part code?).
So *all the evidence* shows that HMOS is the term used for the later generations of NMOS depletion-load chips, and only for the later ones. That's why the "H" that stands for High Density or High Performance.
You were right about the 6800 though. At least according to that patent mentioned by Mensch, it is enhancement only. However it does use multiple voltages, only that they are generated internally inside the device. Again, at least that's what that patent describes.Ijor (talk) —Preceding undated comment added 20:01, 24 March 2009 (UTC).
- Okay, so we all agree that we are all obstinate. I admit that I furthermore oversimplified things a bit with a slapdash "excusively" comment. Will you also admit that you oversimplified a bit with "The plain truth is that HMOS in just one variant of depletion-load NMOS"? Now you seem to be saying that HMOS refers to different variants without a unifying characteristic besides being "later generation." If HMOS were simply one variant, then it would be easy to write an encyclopedic article about it.
- Thank you for providing these examples. I finally now see what you're trying to say, and that I was mis-educated on this.
- This still leaves the question, however, of what an encyclopedia should say about HMOS, if anything, besides that depletion-mode pullups are used. Certainly it looks like the article needs to be moved.
- In the course of trying to figure this out, I ran across such oddities as 10 µm process implying that the 4004 was CMOS and that the "next shrink" was 3 µm process, and Intel 80286 simply stating the minimum feature size as 1.5 µm. So I kinda wonder if this is top priority. But anyway. Potatoswatter (talk) 03:23, 26 March 2009 (UTC)
- Speaking of the 286, did Intel manage to be so far ahead of the game in 1982? Do you happen to have any old magazines lying around? I found one scanned book on Google claiming 2.5 µm and various other sources claiming 1.5 but 1.5 sounds anachronistic. Perhaps that was the nominal gate length, not the half pitch? Potatoswatter (talk) 04:02, 26 March 2009 (UTC)
There's not a single bloody picture on the entire page
- Good point; I added three; well, not bloody, but feel free to work on it. Dicklyon (talk) 03:24, 29 March 2013 (UTC)
As someone not yet very well-versed in electronics, I agree that the article seems overly technical. Nothing wrong with being comprehensive, but at least the introductory parts could probably convey the essence of the subject without thrusting terms like "depletion-mode n-type", "load transistors", and "pure enhancement-load" on the reader in the first sentence. -184.108.40.206 (talk) 02:26, 24 April 2013 (UTC)
- Some things are inherently obscure and technical, though. I can't imagine a Popular Mechanics-style explanation that doesn't very quickly have to explain what a "transistor" is, etc. - the whole signifiance of the topic is based on very narrow difference from other types of circuits. It's like explaining wine varieties, or almost as bad. --Wtshymanski (talk) 18:25, 24 April 2013 (UTC)