Talk:Transistor–transistor logic/Archive 1

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1 Archive 2

Relation to RS232?

This page doesn't explain what's the relation of TTL to RS232 in modern the modern OEM market. It just made me more confused, rather than help. It's like a history lesson for people who are already 'in-the-know'. If you were to carefully read it, you would notice it doesn't explain at all, in layman's terms, WHAT TTL is.

RS-232 is a serial communcation standard, based on signals where -12V is low (or binary 0), and 12V is high (or binary 1). RS-232 also defines timing, the way bits are formed into bytes, and other things such as handshaking. TTL, on the other hand, only defines voltage levels -- 0V for low, 5V for high. There are also standard ways of making the 0V and 5V using bipolar transistors, and standard currents which flow in various cases. There is no timing standard. So in a sense you are comparing apples and oranges. TTL is an electronic design standard, RS-232 is a communication standard.
Sometimes when discussing RS-232, people talk about "TTL levels". This just means that the circuitry is made so that it accepts 0V/5V, as well as the standard -12V/12V. Otherwise it is just normal RS-232. Most RS-232 circuitry is like this. -- Tim Starling 07:41, 25 Aug 2003 (UTC)
with relation to RS/232 TTL levels can also mean that its only 0V/5V and it needs an external tranciever chip to actually connect to a RS/232 line (PICs are like this for example) Plugwash 20:47, 1 June 2006 (UTC)

Picture

Does anyone else think a different picture would show the chips a little better? The breadboard in the present picture is hard to follow and doesn't show, in my opinion, a typical application. I'll have to shoot some pics of TTL-based things I have around the house. --Wtshymanski 20:05, 8 Apr 2005 (UTC)


Here's a picture of a TTL digital clock/calendar. Today this would be one chip powered by a coin cell. Things were different in 1979.

--Wtshymanski 20:12, 10 Apr 2005 (UTC)

Not sure where I should put this but TTL also stands for Through-the-lense which is a way that a camera can adjust the lighting/flash when used in conjuction with a mounted flash. 212.161.59.210 10:23, 26 January 2007 (UTC)John McGinty212.161.59.210 10:23, 26 January 2007 (UTC)

See the disambiguation page TTL and the article Through-the-lens. --Wtshymanski 19:57, 26 January 2007 (UTC)

Comparison

Could it be said, that

?--Arnero 17:30, 27 January 2006 (UTC)

well CMOS uses mosfets rather than bipolars but its essentially two common emmiter units built with opposite polarities of transistor wired in parallell. not sure about TTL and ECL. Plugwash 01:10, 8 February 2006 (UTC)
Actually CMOS is common Source. FETs have Sources, Gates, and Drains, Bipolars have Emitters, Bases, and Collectors. -- RTC 00:23, 18 February 2006 (UTC)
RTL uses a resistor network input stage and a common emitter output stage. DTL uses a diode/resistor network input stage and a common emitter output stage. TTL uses a common base multiple emitter input stage and a common emitter output stage (The classic TTL "totempole" output stage is actually both a common emitter and common collector output combined). ECL uses a common emitter differential amplifier input stage and a common collector output stage.
In all of these bipolar technologies the logic function is implemented in the input stage and amplification is implemented in the output stage to get adequate drive for the inputs of other gates. CMOS however can perform both the logic and amplification functions in one stage.
RTC 00:34, 18 February 2006 (UTC)

10,000 transistors is VLSI ?

"Each "chip" contains the equivalent of a few dozen to a few hundred transistors, contrasting with early very-large-scale integration (VLSI) devices that had the equivalent of up to 10,000 transistors"

10,000 transistors does not qualify as "Early VLSI" it is not Very-Large at all. LSI means at least a thousand gates, the term VLSI implies many tens of thousands.

"TTL Trigger"

I've used devices with a "TTL Trigger" or "TTL Line". Being a software engineer, I wondered why the type of logic used mattered with respect to interface. An electrical engineer explained that this just meant it spoke in TTL-levels and that it might be 5V CMOS inside the box. Is this correct? If so, perhaps something to this effect should be in the opening paragraph? —Ben FrantzDale 19:20, 1 June 2006 (UTC)

Different logic families have different rules for levels, in particular TTL and tradtional CMOS aren't really very compatible. TTL inputs tend to overload traditional cmos (especially the old 4000 series) outputs and cmos inputs tend to have the thresholds in the wrong place for detecting a high from a TTL output. There are however logic families (for example 74HCT) that will happilly interface with both. Plugwash 20:46, 1 June 2006 (UTC)

RTL vs. TTL

I'm afraid the schematic shows an RTL, not a TTL gate circuit. Suggestion: Use the image from the German site de:Transistor-Transistor-Logik. (Please help, I don't know how to transfer it here.) --84.150.154.189 19:20, 30 July 2006 (UTC)

Nope, that's a TTL-style circuit, though with fewer transistors than a real gate would have - the multiple-emitter transistor input stage is classic. Worth checking the German illustration, of course, but often there's text stuck in the illustrations which makes them non-portable between languages. --Wtshymanski 00:51, 31 July 2006 (UTC)
Oh I see, it's only the output stage that's different, which may be irrelevant. --84.150.183.132 05:27, 31 July 2006 (UTC) (was 84.150.154.189 yesterday)
The "totem pole" output stage is nearly as classic as the multiple-emitter input, so you are correct, a better illustration would show this more typical configuration. But I've always found drawing with a mouse about as handy as drawing with a bar of soap...someone with better skills is invited to come forth. Could use the German picture, I suppose...it is better in that sense. --Wtshymanski 01:09, 1 August 2006 (UTC)

Popularity of various logic alternatives

Tho it is hinted at that there has been an evolution in technologies, I think it would be interesting to see a more explicit discussion of the changes is choices available over time maybe with some numbers attached to the trends. Production or sales numbers vs time say. There was a time that one could buy TTL chips at Radio Shack. I dont think that is the case now. Choices of technology for a given application not only include technical considerations but also availability and cost of alternatives. I presume that "programable" technologies are more and more likely choice rather than discreet technolgies such as TTL (or whatever). A discussion of this would also be helpful. --Fholson 13:50, 7 August 2006 (UTC)

Well tandy (uk branch of intertan the owners of radioshak) no longer exists but maplin afaict stock both 74LS TTL and 4000 series cmos in thier stores (thier mail order also stocks some of the 74HC and 74HCT series chips) but anyone who is remotely serious ends up turning to mail order anyway a lot of the time.
The main problem with TTL chips for glue logic functions is that all the big modern integrated circuits are CMOS based and on many of them at least some of the inputs are not TTL compatible, power consumption is also a problem.
Where performance doesn't matter a lot of people (both hobbyist and pro) are turning to software on pics, they are cheap versatile, availible in dil packages to allow easy prototyping and easy to program
Programmable logic is another option but it tends to be fairly inaccesible to hobbyists except in the form of ready built demo boards due to programming issues (on older families) and nasty packages (on newer families). Plugwash 14:14, 7 August 2006 (UTC)

Logic levels?

The text says TTL high logic level is 2 volts. I vaguely remember that value to be 2.4 volts, but I don't seem to have any of my old TTL data books around ;-( I thought this was the reason for 3.3 volt CMOS logic as opposed to 3.0 volts, as it could directly interface with TTL. Can anyone confirm this? Madhu 03:03, 30 August 2006 (UTC)

I can confirm: for 5V TTL logic levels were declared as 0 to 0.4V as "0" and 2.4V and up as "1". At least my old books saying so.
As for 3.3V CMOS to TTL interoperability: they're INCOMPATIBLE! The only wide use of mixed CMOS + TTL circuits I know is 5V CMOS intermixed with 5V TTL. Reason is very simple. In standard CMOS design, input level should never exceed positive power supply rail level + 0.4...0.6 V. Failure to met this condition will cause protective diodes to open on CMOS IC input pin. These diodes are not part of sample illustrative CMOS schematics but they are present in almost all real-world CMOS ICs on all input pins to protect CMOS inputs from being quickly damaged by static discharges. Without these diodes CMOS logic too easily killed by static discharges. Earliest CMOS ICs suffered greatly from static discharges before protective diodes were added as part of inputs design.
In real CMOS IC, usually 2 diodes attached to input pin to GND and Vdd in way they're closed during normal IC operation. Diode attached to positive power rail opens when level on input pin exceeds level of power supply rail by approx 0.5 ... 0.6V so diode opened since voltage polarity is getting "correct". This causes static discharge to flow into power rail without causing any harm to input. However this circuit will not tolerate levels above power supply + 0.4 ... 0.6V. Let's admit that 5V TTL output loaded to CMOS input will output almost 5V as it's gated to power supply rail via upper transistor and CMOS input normally does not consumes current so there is almost whole power rail level. For 5V TTL this means 5V on CMOS input. A way too much for 3.3V CMOS logic! It's diodes will open, extra current will flow from 5V supply rail to 3.3V supply rail and since this is not static discharge, this condition will not disappear quickly. If this condition persists for a while, CMOS IC could get damaged. And it's correct operation is not warranted when input level exceeds Vdd + 0.6V (read any data sheet for any CMOS IC, take a look on absolute maximum ratings). So, 5V TTL output can't be interfaced to 3.3V CMOS input. Those new "5V-tolerant" 3.3V (and lower) CMOS ICs are using non-standard input design to withstand 5V while being powered by 3.3V or lower voltage. However, surely, 3.3 V CMOS output can drive 5V TTL input. And even 3.0V CMOS can do it as well. The only issues here are increased switching times and less noise tolerance. So, 3.3V CMOS output can drive 5V TTL input with some restrictions. —Preceding unsigned comment added by 195.210.145.118 (talk) 16:59, 1 October 2008 (UTC)

Title

Is it more proper to capitalize the initial letters of each word or not? Despite the title being "Transistor-transistor logic", the introduction uses "Transistor-Transistor Logic". I'm just hoping this inconsistency can be resolved by people who know how it should be referred. --Kamasutra 18:41, 14 September 2006 (UTC)

part numbering scheme

The Part Numbering Scheme section is useful, but I think it has a more appropriate home at the 7400 Series article. Does anyone object to my moving it there? -- Mikeblas 13:52, 21 October 2006 (UTC)

Please do! The numbering scheme was out of place in logic families and fits only a little better here (in my project-building days it seems to me there was always at least one non-74XX series TTL chip in everything). Watch for case significance in article titles - you meant of course 7400 series - darn C programmers have infected everything with spurious case significance, I bet 15% of all Wikipedia maintenance activity is due to this alone. --Wtshymanski 16:26, 21 October 2006 (UTC)

Historical question

From the History section:

The Kenbak-1, possibly the first personal computer, used TTL for its CPU instead of a microprocessor chip, which was not available in 1971. According to the inventor, the most expensive component of such a computer was memory, not the processor.

Maybe I'm reading this wrong, but I was under the assumption TTL describes a method for logic development, which could easily be used in a microprocessor. 19:31, 9 November 2006 (UTC)

TTL was mostly used for small-scale and medium-scale integrated circuits, but there may have been a few TTL microprocessors, too. I can't think of any, though, just the "slices" such as AMD's Am2900 series parts (actually, I'm not even sure those were TTL internally, but some kind of bipolar logic). TTL was generally too power hungry to make LSI chips with, which is why everything went to pMOS, then depletion-load nMOS, and then CMOS. Dicklyon 07:07, 24 June 2007 (UTC)
I think there may have been a misunderstanding.
I agree with anonymous that, using the TTL design style, a single-chip TTL microprocessor can theoretically be fabbed.
I agree with Dicklyon that it is highly unlikely such a chip was ever built -- around 1971, the biggest possible chip that could be fabbed at a reasonable cost was far to small to contain an entire microprocessor (no matter which logic family design style they tried to use); by the time chip fabs could make such a large chip, other logic design styles had made TTL design styles obsolete.
However, *lots* of computers used TTL for their CPU -- not a microprocessor (a single chip), but one or more boards full of MSI TTL chips.
I'm assuming that this Kenbak-1 was one of those computers.
Other computers with a CPU built from lots of TTL chips: PDP-15, HP 9830 (?), the original VAX, the original TV Typewriter (?), ...
(I've been told that) the PDP-8 was the most widely-sold computer in the world (of its time), and (I've been told that) most PDP-8 CPUs were built from TTL chips. ...
Even today, students in some college classes and hobbyists continue to build CPUs out of a pile of MSI chips (Wikibooks:Microprocessor_Design/Wire_Wrap) -- although recently, it seems that functionally-equivalent CMOS chips are often substituted.
Lots of other computer were built even before TTL chips were invented, such as the Apollo guidance computer built out of "4,100 ICs, each containing a single 3-input NOR logic gate ... using resistor-transistor logic (RTL)." ... and I think some computers built out of ECL chips.
Should CPUs notable to have their own Wikipedia article be listed in this article, simply because they were built from TTL?
--68.0.124.33 (talk) 06:51, 26 April 2008 (UTC)
I think it wouldn't hurt to mention which notable computers were built from TTL, as long as the info is reliably sourced. I don't think all that you've been told is correct (PDP-8, for instance, was TTL only after third generation or so). Dicklyon (talk) 16:02, 26 April 2008 (UTC)
A couple of months ago there was an IEEE article that said the limit to TTL circuit complexity was related to the heat-dissipation capacity of the DIP package; you couldn't build a plastic DIP and keep the die temperature under control while dissipating more than 1 or 2 watts, which set an upper bound on how many gates you could have on a die. I will try to find this and add it in a relevant fashion. This explains why there were no TTL microprocessors. Getting rid of heat is still a problem in processor design. --Wtshymanski (talk) 15:18, 17 September 2008 (UTC)
DIP packages are no less effective than square ones. I think that ceramic been in use even longer than plastic. FETs are smaller than BJTs in the same lithographic process, explaining the jump in gate count with MOS. TTL implementations of a few chips competed with single chip MOS processors for a while. Later, BiCMOS was perfected to practically put both styles on the same die. Since then, power issues have obsoleted BiCMOS and subsequently almost all dynamic logic. But that's not to say you couldn't integrate the ~10K transistors necessary for an MPU using an early or mid 80's technology. Potatoswatter (talk) 15:34, 17 September 2008 (UTC)

Packaging

The terms "flat-pack" and "beam-lead chips" are not commonly known outside the industry. Therse terms should carry a link or a further explanation for the sake of the uninitiated. —The preceding unsigned comment was added by 128.6.83.118 (talk) 16:11, 11 April 2007 (UTC).

Proper punctuation: en dash

Would anyone mind very much if I were to move the page and change the text to properly punctuate the subject with an en dash? For people who know English, the hyphen here is a bit nonsensical, but transistor–transistor logic makes good sense; same with diode–transistor logic resistor–transistor logic, which are better than the nonsensical resistor-transistor logic and diode-transistor logic. What is a diode transistor, anyway? Most refs don't do it right, either, but a few do, like this one. Dicklyon 06:47, 24 June 2007 (UTC)

Here is another ref that gets it right; and another, and another, and another, and another. Other things you sometimes see are a slash, or a space. Or multiple hyphens as transistor-transistor-logic. But none of these make as much sense as using the en dash, to people used to English punctuation and publishing style. Dicklyon 06:59, 24 June 2007 (UTC)

Hearing no objection, I went ahead and did it. Dicklyon 21:02, 26 June 2007 (UTC)

Expert editing needed

I was one of the few engineers in IBM that developed most of the discreet transistor switching circuits in the late 1950's and early 1960's. I believe I am about as much of an expert as there still is. Many Wikipedia articles on digital circuits appear to have been written by second hand observers and nearly every article I viewed on the subject have major and minor errors. I am forced to go by mostly memory and I am getting old. I do believe I know where my memory fails me. I fear much of the truth is about to be lost.

If there are others that lived this history or a historian who interviewed people like me twenty years ago, or more, then they may be able to help. If anyone has access to the actual documentation of that work that would be helpful?

I hope to write something but I am not sure how to present it. The subject seems to be scattered everywhere. I am thinking I should write one article in one place and link it to the many articles in Wikipedia.

Any thoughts? Any offers of help?UPCMaker 22:27, 25 August 2007 (UTC)

Wikipedia articles look like they're written by outside observers because there is no original research at Wikipedia; everything must be referenced. If you can write something that's referenced by third-party publications and not simply a collection of your memoirs, I'd be happy to help you get it formatted and posted. -- Mikeblas 04:09, 26 August 2007 (UTC)

The main problem with this article is that it misses the point. Transistor transistor logic refers to logic gates where the inputs and outputs are both controlled by transistors. That should be the entire premise of the article. Then the original operating regions of such circuits were carried over for many decades to other circuit technologies and called the "TTL logic levels" of "TTL-compatible" standard logic families of chips such as the 7400 series. Confused editors have added an insane amount of inappropriate information to this article. Potatoswatter (talk) 00:39, 7 May 2008 (UTC)

I'm somewhat expert on this; but I'm not sure what you're getting at. Can you be more explicit about what you think is "inappropriate" information? If you'd like to add some stuff about "TTL levels" or "TTL compatible", feel free. Dicklyon (talk) 04:24, 7 May 2008 (UTC)

Well... thinking how to rewrite/reorganize, "Functions" doesn't belong at all and "Applications" probably needs more context. "Theory" should be renamed and should be the first section, since that's what defines TTL and lets the reader know what it is. "History" should be after theory but should describe the development before getting to the first few applications, and should contain no unreferenced information. Right now the article gives the false impression that TTL was invented along with the IC, which it wasn't (as is revealed by the first sentence of History, then glossed over). "Comparison" should be merged into both Theory and History and should mainly serve to explain how and when TTL gained and lost its popularity.

Unfortunately, I'm really not familiar with the specifics of the subject, so I really need to study quite a bit to know how quickly TTL supplanted RTL and DTL. I know they coexisted for a long time... and the theory section kinda hints at why it's better than DTL but again writing the article would be much easier and less risky for someone who actually used both technologies at some point. Potatoswatter (talk) 08:02, 7 May 2008 (UTC)

Doesn't sound to me like the edits you want to do require an expert, so I'm going to remove that tag. Dicklyon (talk) 14:40, 7 May 2008 (UTC)
The part about what makes TTL faster than DTL is quite technical, involving parasitic capacitances and impedences. Are you sure? Potatoswatter (talk) 04:48, 8 May 2008 (UTC)
Sure of what? That it doesn't take an expert to understand or to write such a section? Pretty sure. It sounds right as it is, but it would be much better if someone would find a source to cite and make sure it agrees. I don't think that takes an expert, either; just someone with basic electronics understanding and ability to read books. Dicklyon (talk) 04:53, 8 May 2008 (UTC)
Well, both of us fit that bill, yet we're not actually doing it ;v) . Still easier for someone who knows what's what off the bat. In any case, what the expert tag does is flag it on the Wikipedia:WikiProject Electronics project page, which really might be appropriate. Potatoswatter (talk) 07:28, 8 May 2008 (UTC)
The thing is, I've never seen calling for experts do much useful. It just discourages normal editors. But I suppose you can try it for a while. Dicklyon (talk) 14:12, 8 May 2008 (UTC)
If that's your experience, mebbe it's not a good idea. Although "normal editors" seem to have simply increased the entropy in this article. Either way, we seem to be making progress for now. Potatoswatter (talk) 16:55, 8 May 2008 (UTC)

Dominance?

Potatoswatter [inserted] the statement " Despite this, for several years RTL and DTL remained the dominant technologies." In the edit summary he claims "(... and that when it was invented, it was done independantly. In any case, the basic concept is obvious enough...)"

I see nothing obvious about what logic technology was dominant in the 1960s. How can we know how long it took TTL to become dominant (that is, if it ever was dominant). --Gerry Ashton (talk) 19:23, 8 May 2008 (UTC)

Sorry. I meant dominant in the one market segment, general purpose chips. We know that TTL was not popular until the Sylvania introduction in 1963, and probably was still gaining steam as of the TI 7400 introduction in 1966. By obvious I only meant that an emitter follower followed by a common-emitter stage is obvious. Potatoswatter (talk) 19:29, 8 May 2008 (UTC)

*STTL families actually DTL?

Several places on Wikipedia it's stated that the difference between Schottky TTL and other TTL is the presence of diode clamps on the input transistor(s). However one primary source and another secondary source I found show that the input transistors are indeed removed and replaced with Schottky diodes, making STTL, LSTTL, ALSTTL and ASTTL actually DTL families. (The evaluation/inverter transistor and several others are clamped with diodes as described. Also, the DTL pull-up resistor is replaced with a complex secondary path which varies between the families.) Before I make a drastic change, is there any really good reason to class LSTTL as TTL? Potatoswatter (talk) 21:38, 10 May 2008 (UTC)

Don't change the article that much. Though I admit I was startled when I pulled out my TI TTL Logic Data Book (1988 edition) and found that the schematic for the LS TTL NAND gate does *not* have the distinctive multiple-emitter transistor that the regular Schottky does. If Texas Instruments thinks LS TTL is TTL, then I don't think we should come to the novel conclusion that LS TTL is really DTL. If it's good enough for TI, it should be good enough for Wikipedia. I would like to know why - up until a few minutes ago I'd always thought an LS gate was the same as an S but with higher resitor values. Oh, and other logic functions like NOR, etc. are much more similar in their LS and regular TTL schematics. Comment on this difference in LS TTL, but don't call it DTL. --Wtshymanski (talk) 22:00, 10 May 2008 (UTC)
Some of the diagrams in the reference you gave do have transistors on the input, although the input is connected to the base, rather than the emitter as in traditional TTL. Also, DTL refers to a specific circuit, and the circuits in the reference are not DTL. So STTL, LSTTL, and so on, should not be called DTL. --Gerry Ashton (talk) 00:55, 11 May 2008 (UTC)
The defining characteristic of DTL that the input directly drives the base of the NAND/NOR stage through a diode, not the simple presence of any transistor on the input. The transistors on the inputs are part of a power-saving active circuit in series with DTL's passive resistor, so it isn't a fundamental modification. Potatoswatter (talk) 01:05, 11 May 2008 (UTC)
I just found a reference that says multiple diode inputs to a DTL NAND can be implemented as a multiple-emitter transistor! So when is DTL "really" TTL ? --Wtshymanski (talk) 01:52, 11 May 2008 (UTC)
The diodes are fundamentally different from a transistor. The multiple-emitter transistor is pretty much equivalent to several transistors in parallel, which some other references show in place of the multiple emitters. Where's the reference? Potatoswatter (talk) 02:01, 11 May 2008 (UTC)
The ALS NAND of fig.12 of your primary source shows an active pullup transistor in place of the passive pullup resistor of DTL. So it's certainly not right to say it's really DTL-like. But more fundamentally, it not OK to make interpretations of sources, whether primary or secondary – just report what they say. Dicklyon (talk) 01:50, 11 May 2008 (UTC)
...But the secondary source I just added (which I actually found first) does have a lengthy explanation of how it's actually DTL. And the active transistor is in series with a passive resistor, so it looks to me more like an unessential power-saving measure. Potatoswatter (talk) 02:01, 11 May 2008 (UTC)

{{Citations}} template addition

User:Dicklyon added a {{Citations}} template in the middle of the reference list with the edit summary "need to cite these". I'm not sure what he means by this. If he is suggesting that no information may be used from a book unless the information is associated with a particular phrase, sentence, or paragraph in the article, then I disagree. For example, the work might have a good index and could be used to verify many of the statements that do not have inline citations. --Gerry Ashton (talk) 17:11, 16 September 2008 (UTC)

No, that's not what I'm suggesting. What I'm suggesting is pretty much what the tag says: "sources remain unclear" due to the lack of citations; in particular, if parts of the articles were written from these sources, it would be really good to put citations on those parts. I could sprikle of a bunch of "citation needed" tags in the article, but if whoever used these sources would be provoked by this to do what he should have done in the first place, that would be a big step forward. Dicklyon (talk) 21:51, 16 September 2008 (UTC)
Looks like they were mostly added by User:Wtshymanski in early 2005 when the article content was not much; so that gives us at least a clue what came from those sources. Dicklyon (talk) 21:56, 16 September 2008 (UTC)
My Wikipedia skills have improved since then, and the general frequency and quality of citations has also improved. --Wtshymanski (talk) 15:18, 17 September 2008 (UTC)
I have one of the sources listed (Horowitz & Hill) and two other relevant texts, so I'll take a look at the article. I would try to cite some of the less obvious statements line "the Sylvania parts were used in the controls of the Phoenix missile" but I don't see a need to cite obvious statements that can be verified on vertually any text about TTL, such as "Transistor–Transistor Logic (TTL) is a class of digital circuits built from bipolar junction transistors (BJT), and resistors."
My first step would be to separate the notes from the references, and create a complete list of references. This way, short footnotes could be used to give the author, date, and page, rather than giving full bibliographic details over and over again each time a different page in a given work is used. --Gerry Ashton (talk) 22:37, 16 September 2008 (UTC)
I've started the process, but Wikipedia is getting slow, so I'm going to wait a few hours. --Gerry Ashton (talk) 23:16, 16 September 2008 (UTC)
I've restructured the references and footnotes. Tomorrow I'll look for citations in my sources to support some of the less obvious statements in the article. --Gerry Ashton (talk) 05:34, 17 September 2008 (UTC)

I've only been able to find additional support for a few statements in my sources. My sources generally agree with most of what is in the article, but I can't verify things point-by-point. --Gerry Ashton (talk) 23:49, 17 September 2008 (UTC)

SSI MSI TTL? and early Chinese TTL manufacturing

I remember many years ago when I was taking EE lab courses that our professor used the terms SSI (Small Scale Integration) for gate-level TTL ICs, like 7400, 7402, etc. MSI TTL ICs were more complex. I think the transition between the two was when you went from logic gates to latches; possibly at 7474. Maybe this bit of information is unecessary, especially since there is no verification right now. Later on I attended VLSI seminars, including one where Robert Noyce was the speaker. Another interesting point that could be added to this page is the assertion that from a logic point perspective only NAND or NOR gates are really necessary. All other logic devices can be constructed from NAND and/or NOR gates. In one of my EE lectures the professor described his experience touring a Chinese TTL factory, circa late 60's or early 70's. They only made NAND gates. Everything else, including minicomputers, were made from them.Zen-in (talk) 20:37, 13 October 2009 (UTC)

IEEE Standard definitions for SSI, MSI, LSI, VLSI, ULSI

For what it's worth, IEEE Standard 100 gives the following definitions:

  • Small-scale integration - less than 100 transistors, less fewer than 10 elements
  • Medium scale - 100-500 transistors, 10-100 elements
  • Large scale - 500-20,000 transistors, 100-5000 elements
  • Very large scale - 20,000-1,000,000 transistors 5000-1,000,000 elements
  • Ultra large scale - more than 1,000,000 elements.

The standard doesn't say what it means by "elements" but "gates" would be my guess. --Wtshymanski (talk) 16:02, 27 October 2009 (UTC)

Does it say why they chose to say "less than" instead of the more appropriate "fewer than"? Dicklyon (talk) 16:08, 27 October 2009 (UTC)
The standard used both "less than" and "fewer than" in the definition of SSI (and I have amended my posting appropriately). Very democratic of them. Take it up with the standards committee, not me. This is obviously a critical grammatical point upon which the reputation of the entire Wikipedia depends. --Wtshymanski (talk) 16:19, 27 October 2009 (UTC)
And today I read in the Globe and Mail that the National Post had "less than 300 employees" - if it's good enough for our (surviving) national newspaper, it's good enough for me. --Wtshymanski (talk) 18:40, 30 October 2009 (UTC)

Theory

I disagree with some of the statements in the Theory part this article and since there are no citations to back up these statements they are fair game. It opens with a statement of the problems with DTL being related to diode inputs, and that TTL solves these problems. A few sentences later it looks like the theory is claiming these same PN junction inputs are why TTL is faster. It would be better if the 2 stages were described as common base and common emitter stages and the easily verifiable characteristics of these types of transistor stages be used to explain how TTL works. Also a consistent schematic convention for resistors is needed. Zen-in (talk) 03:07, 15 March 2010 (UTC)

Presenting the input stage as a common-base amplifier is a misleading concept here. The input multiple-emitter transistor would operate in a common-base configuration if the potential of its base was fixed by a voltage source ("stiff" base potential) and if its collector was connected to the positive supply rail. Instead, the input transistor is driven by a simple resistor-type current source ("soft" base potential) and its collector is connected to the ground through the base-emitter junction of the second transistor. Circuit dreamer (talk) 19:13, 15 March 2010 (UTC)
You are wrong about this. The input stage is a common base configuration and that should be used as a starting point for describing the circuit. A common base circuit doesn't have the collector tied to the positive supply rail - that is a common collector circuit. This talk of "soft" and "hard" "stiff" is POV and confusing. Can you produce a reference that uses this terminology in describing TTL circuits? Better to talk about impedance - such as looking into the emitter(s) of a CB, the impedance is very low. An earlier version of this article had this sentence in the start of the theory section " As shown in the top schematic at right, the fundamental concept of TTL is to isolate the inputs by using a common-base connection, and amplify the function using a common emitter connection." This (isolation of inputs) is an important aspect of logic circuits and it was expressed very well. Common base stages have very good isolation, or high return loss when considering RF applications, low input impedance, and voltage gain. Zen-in (talk) 19:24, 15 March 2010 (UTC)
I have used these figurative and colorful words only here, in the talk page, in order to make the discussion more interesting... Well, let's consider first the input part of the common-base stage. The main characteristic of this arrangement is that two voltages have to be applied - a reference voltage to the base and a varying (input) voltage to the emitter; thus the base-emitter junction of the transistor is driven by the difference between these voltages. So, in order to control the transistor (to change its base-emitter voltage) by the emitter voltage, we have to keep steady the base voltage. But in the input TTL stage there is no constant voltage source connected to the base. The power supply is connected through a resistor to the base (this is actually a current source). So, when we vary the emitter voltage, the base voltage will follow the emitter voltage and the base-emitter voltage will change slightly. The conclusion is: the input stage is not a common-base stage.
Now, some words about the output part of the common-base stage. I would like only to say that a positive voltage has to be applied to the collector, in order to set the transistor in an active regime. This means that the collector has to be connected in some way to the positive rail so that the collector current passes through the collector-emitter junction and the input voltage source toward the ground. In order to take an output voltage, we connect a (collector) resistor between the collector and the positive rail acting as a current-to-voltage converter and take the complementary (to the power supply) voltage drop between the collector and the ground as an output. But if the load is "flying", we may connect it in the place of the collector resistor.
Finally, some words about the input impedance. In a common-base stage it is low since the transistor passes all its collector (emitter) current through the input voltage source. Here it is high enough since there is no significant collector current flowing through the input source. Circuit dreamer (talk) 22:29, 15 March 2010 (UTC)
I've just looked at Millman "Microelectronics", Hamilton and Howard "Basic Integrated circuit engineering', and Dorf "Electrical engineering handbook" and none of them use the exact words "common base" to describe the input stage; on the other hand, a three terminal device with an input and output has to have one terminal common to both, and i don't see what else it could be. Since none of these folks felt compelled to call it a "common base amplifier", perhaps we need not use that phrase either. --Wtshymanski (talk) 00:00, 16 March 2010 (UTC)
Agreed, the input is a CB stage but most (possibly all) electronics text only consider linear applications of the CB stage. However the lead in to this section that referred to the input as a CB stage was much clearer than what we have now. TTL requires sink current to maintain a "0" input. The theory section of this article talks about voltages. " If one (or more) of the input voltages becomes zero, the corresponding base-emitter... " That isn't how TTL works. So even if it is easier for some people to understand it is still wrong. Then we have: " The input base-emitter junction deprives all the base current of the output transistor (the current is steered from the base to the input source); the transistor becomes "off" and the output voltage is high (logical "1"). " What is being said here? Which base-emitter junction? The first one is not connected to the second. How does it "deprive" current from the first? What does deprive mean? "the current is steered from the base to the input source" Which base? All the current that goes into a base comes out the emitter (NPN). How can a transistor deprive itself of its own current? I think this article should be reverted to an earlier version. Too much of it has become unreadable. Zen-in (talk) 00:30, 16 March 2010 (UTC)
Common base. There is no place here for the common base concept. The little reason is that the base is no common transistor terminal here. What does "common" mean? "Common" means "AC grounded", i.e. a transistor terminal that is connected (directly or through a perfect voltage source) to the ground. But the base is connected through a very bad (real) voltage source having 4 k internal resistance. The big reason is... that the transistor is not a transistor here:) It does not operate as a transistor in active mode (as an amplifying device) since its collector is not "pulled-up"; instead it is "pulled-down" by the base-emitter junction of the second transistor. Actually, the multiple-emitter transistor is just a set of diodes that are connected together by their anodes. The input transistor operates in active mode only for a while at the end of the high-to-low transition when it draws a current away from the base of the second transistor. At high input voltage, the input transistor operates in the odd (because ß < 1) and undesired (because of the small input current consuming) reverse-active mode. In this case, we don't need the amplifying abilities of the transistor (we don't need a transistor; we need just diodes).
Well you are out-voted 2 to 1 on that. There are 3 types of single transistor amplifiers: CB, CE, and CC. The input is a common base stage. Zen-in (talk) 02:05, 18 March 2010 (UTC)
Input resistance. The input resistance of a common-base stage is low since the transistor "attacks" the input source by passing the big emitter (collector) current through the source. In this arrangement, looking into the emitter of a CB transistor, the input source "sees" the very low output resistance of a good voltage source (emitter follower). But looking into the emitter of a TTL input transistor, the input source "sees" the very high output resistance of a good current source (at logical one) or the 4 k base resistance (at logical zero).
The input impedance of a circuit is not determined by figurative musings. You need to use a simple transistor model and perform an incremental analysis. This is how electronic circuits are analyzed. Zen-in (talk) 02:05, 18 March 2010 (UTC)
Input isolation. The sentence "...the fundamental concept of TTL is to isolate the inputs by using a common-base connection..." is a misconception. In a TTL input stage, the base-emitter junctions of the multiple-emitter transistor isolate (decouple) the input sources exactly as the input diodes of a DTL gate. These junctions actully act as diode switches implementing a logical AND function.
Again, this is determined using commonly accepted methods of analyzing electronic circuits. Zen-in (talk) 02:05, 18 March 2010 (UTC)
Input sources. Finally, I would like to consider the sentences "...TTL requires sink current to maintain a "0" input. The theory section of this article talks about voltages..." The logical OR finction is implemented by parallel connected electrically-controlled switches. In the 4-terminal switches (e.g., relays and optoelectronic devices) and 3-terminal switches (transistors), there is no connection between the input and the output circuit. But in the diode switches the input and the output circuit are the same. As a result, DL and the input stages of DTL and TTL gates, implementing logical AND, behave very strange at input logical zero in comparison with RTL, DCTL and NMOS gates as they pass their output current through the input sources (the input sources don't expect such an attack:) So, the input sources have to be perfect voltage sources with very low internal resistance; TTL gates are driven "by voltages". Circuit dreamer (talk) 19:16, 17 March 2010 (UTC)
If you consult a TTL data book you will see that for an output to provide a logic low signal to a TTL input, it has to sink current. For example 74LS TTL logic ICs can sink up to 16mA but only source 2mA. That is why TTL outputs all have a totempole stage. Zen-in (talk) 02:05, 18 March 2010 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:Transistor–transistor_logic/Archive_1&oldid=395056866"