Talk:Storage Module Device

Disk Geometry

I've added a section on the disk geometry of the 80Mb and 300Mb disks with which I was very familiar due to writing software to interact with these devices. I believe the 40Mb and 150Mb versions were identical but with 411 cylinders instead of 823 cylinders, and twice the cylinder spacing. However, I didn't use these lower density versions, so I'm less certain of their precise geometry. I also have a vague recollection that the 40Mb disk packs could be read but not written on 80Mb drives (possibly with driver support to double the cylinder address?)
81.187.162.109 18:21, 1 October 2007 (UTC) (Andrew Gabriel)

Added some detail of the physical attributes of SMD drives relating to the necessity of drive servo alignment for removeable media drives. MK McCole —Preceding unsigned comment added by 65.197.19.244 (talk) 18:33, 31 July 2009 (UTC)

Dubious: Removable Media Addition

The added material is generic to removable disk media devices and as such belongs elsewhere, perhaps in its own article or as a section in the hard disk drive article. More important, the addition is unsourced and IMO, several ways incorrect:

• All interchangeable drives with multiple heads require head alignment regardless of the positioning technology (open loop, servo surface, and embedded servo). In the case of embedded servo the requirement is just much less stringent, but if they are very misaligned problems can be encountered.
• IMO CE pack stands for Customer Engineer pack.
• I cannot recall any drive vendor who required periodic realignment - they may have required periodic inspection, but even then I doubt it. I suspect most operated on the "if it ain't broke, don't fix it mode," deferring even inspection until a symptom occured, such as data errors.

Given the unsourced nature of the addition and its dubious nature, I would suggest we allow the author several weeks to move and correct the addition. If no action is taken, then I suggest we delete the addition. Tom94022 (talk) 18:11, 1 August 2009 (UTC)

The CDC Lark drives mentioned have no way to align the heads mechanically—the heads have no mechanical adjustment. The original single density drives rely on the manufacturing tolerances. For the triple density drives (the tracks are packed in 3 times more densely), this wasn't good enough, and at initial head load, the LMU (Lark Micro Unit) spends about 30 seconds doing a set of seeks over the 4 surfaces to work out the misalignment between the heads and thereafter to apply automatic head wobble to compensate when it receives the surface number for an i/o operation. (During this calibration run, it also calibrates the voice coil accelerometer, and works out what the fastest seek velocity possible is without missing the target track.)

The earlier CDC 14" drives with separate servo surfaces don't do anything automatic like this, which is why head alignment across the (6 or 20) surfaces is so important. It was possible for the system software to tell the drive to wobble the head in either direction—for systems which made use of this (very few, as far as I know, but the one I worked on did), it was done after getting a CRC error, but as the system couldn't know which side the head was off (or even if that was the cause of the problem at all), it just had to trial and error by telling the drive to wobble the head one way, and if that didn't work, then the other way. There was also the option of telling the drive to do an early or late access to the sector as another means to try and recover from a CRC error. A disk driver which made use of all these techniques to try and recover from an erronous sector would have often have as many as 27 goes at reading a sector before finally giving up.

The alignment of the 14" disk heads is supposed to be tested each time the drive is serviced. CDC kept changing their minds on recommended service interval (partly because in the early 1980's they produced a batch of really bad disk packs). Regular servicing required replacement of the filter pack, cleaning of the airways and bowl, and inspection, cleaning, and alignment checking the heads (and many other things, like the voice coil roller bearings). Over time, the leading edge of the heads (in the airflow) builds up a layer of dirt. On drives which were never routinely serviced (not uncommon!), this is clearly visible, and eventually bits get shed, causing a head touch, or in bad cases a head crash. Heads would typically need realigning every couple of years (unless a head crash happened first). You could go longer without doing so, but if you wait until you are getting CRC errors, you're in a real mess because after realignment, you will find lots of your packs which were written with misaligned heads can't be reliably read anymore. This was not uncommon with customers who scrimped on the regular maintenance schedule. If it ain't broke don't fix it really didn't work with the 14" drives, as very many customers who tried that philosophy found out. When properly maintained, these drives were very reliable.

I still have working CDC 8" Lark drives and manuals. I last used the 14" drives in 1995, and no longer have the manuals for them. Wrote lots of software for both types though. The information in the main article is correct as far as it goes (could add lots more!). I doubt any online references exist though—this stuff was all dead long before there was an online.

81.187.74.206 (talk) 15:03, 2 August 2009 (UTC)

You really haven't addressed most of my points. Lark is not SMD so again this points to a separate article or section. While it maybe true that the first Lark did not have any means to align the heads that is a long way from stating that all embedded servo drives did not require alignment. And apparently the triple density Lark did a form of self alignment. Tom94022 (talk) 19:35, 3 August 2009 (UTC)

Moved material to Disk_pack#Alignment Tom94022 (talk) 22:23, 7 February 2014 (UTC)

Interface vs Drive Technology

I see SMD meaning at least 3 things in the article and discussion:

1. In the minicomputer world of the 80's SMD was used largely to talk about the drive interface.

2. Many of drives and later interface-compatible drives by non-CDC manufacturers shared common geometry or common geometry factors (9760, 9762, 9764, 9766). (Although at the very end they just gave up on trying to be backwards-geometry compatible.)

3. Some are using this as a jumping off point for discussion CDC products that were not actually SMD (e.g. the Lark).

Perhaps it would be wise to split it up:

A: a page on the SMD interface (both CDC and later non-CDC but SMD interface) and how it was a common denominator for many 70's and 80's era drives and minicomputers. The archetypal CDC and non-CDC (e.g.: Fujitsu Eagle, which already has its own Wikipedia page) drives would be hyperlinked.

and

B: a page for the CDC drives (some, but not all of which used the SMD interface) and their technologies (which is where the Lark discussion would go).

Even though CDC originated the SMD interface, the interface was important far beyond CDC's product line, and I don't think the current article has that pervading it as much as it should. Similarly the CDC SMD drives used and developed many technologies that were important beyond their particular SMD interface (this would be related to the removable pack and calibration subthread.) —Preceding unsigned comment added by N3QE (talk • contribs) 23:41, 27 December 2009 (UTC)

Actually I think the current article is your suggested Article "A." and a page for other CDC drives would be appropriate (your Article "B.". The challenge is finding someone to write it. 17:10, 15 May 2010 (UTC)

The CDC Lark was sold initially as an SMD device and then an ISI device module was also added, selection depending on the IO module fitted. Later, CDC started selling it without an IO module for OEMs who wanted a reduced cost option and to use the LMU's (Lark Micro Unit) proprietary LDI (Lark Device Interface) directly. The vast majority of installations were SMD-based. 81.187.1.83 (talk) 12:03, 7 February 2014 (UTC)

Disambiguation page linked from Interface

Following reversion of my edit removing a link to the Interface disambiguation page, I felt I should explain my reasoning. Since I believe the term interface is referring to a type of electrical connector in this context, I feel that a link to this page is more appropriate than linking to the disambiguation page, where anyone confused about what "interface" means (which is the only reason they're likely to click on the link) would simply become more confused when presented with several varying options. GiftigerWunsch [TALK] 16:26, 15 May 2010 (UTC)

Again, the dictionary definition does not provide nearly as much relevant information as the electrical connection page. I welcome different opinions, but please let's try to achieve a consensus here before modifying the link any further. GiftigerWunsch [TALK] 16:48, 15 May 2010 (UTC)

I was explaining my change when you reverted the page, this is now your second reversion - please do not violate WP:3RR!

Most of the information on the Electrical Connector article is irrelevant and/or misleading. The interface in a computer hardware sense is far more than the physical connector and in fact there were several different physical connectors that were "SMD" compatible, including from CDC the round one and the flat one. What is important to compatibility at a hardware interface is the number and timing of the various signals. The Electrical Connector article provides too much useless information and arguably, the Electrical Connector link misleads the reader by implying it is just the physical connector. The reader would be better off with no link or the dictionary link than with Electrical Connector. If and when someone revises Interface (computer science) to accurately reflect a computer hardware interface perhaps we could link there. In the meantime, I am going to remove the link while we talk this over. Tom94022 (talk) 17:05, 15 May 2010 (UTC)

I have restored the link since the procedure is to leave the change in place until it is challenged and removed by consensus. If and when that happens, by all means remove it but bear in mind that one more revert constitutes violation of WP:3RR. I would welcome any opinions from third parties to help establish a consensus here.

Since "interface" in this context could be taken to extend to both the physical hardware connection and the protocol used to interface with the device at a software level, and as you say, the page to which my link refers does not fully explain this, perhaps a middle ground here would be to update the article to clarify this itself, preferrably with reliable references. I believe my explanation of the concept here is fairly accurate, with some tweaking and citation that could replace the current link.

It is not my intention to perpetuate an edit war, but please remember that my opinion on the article is no less valid, and since I made the original edit, that should remain until a consensus is reached as to a proper course of action as per WP:Consensus.

Opinions welcome, but please do not revert my edit again until we can agree on this, at which time I will gladly make the necessary changes myself. GiftigerWunsch [TALK] 18:16, 15 May 2010 (UTC)

In fact given the title of the page, I would think explaining the SMD interface would be fundamentally important to the article; instead of bickering, let's consider ideas to integrate this into the page, and then there will be no need for a link to another page. Perhaps we could copy the article into one of our user spaces, create a new section dealing with the interface, and update the article once we are satisfied. Sound reasonable? GiftigerWunsch [TALK] 18:25, 15 May 2010 (UTC)

Actually you have reverted three times, I have only reverted once, my other two changes were alternatives to the misleading link. So if you revert you will be in violation. I have given you three alternatives, you have insisted on your edit - sounds like you are the unreasonable one. If you want to go ahead and update the Interface (computer science) article to reflect the components of a computer hardware interface and then link it back to this article that would be great. In the meantime keeping this misleading link is inappropriate and should be replaced, in the absence of consensus the reversion should be to the original link not to your edit. I can live with any of the three I have proposed, what is your alternative or are you going to continue to edit war? Tom94022 (talk) 00:31, 16 May 2010 (UTC)

You are clearly being un-cooperative, and if you read WP:3RR you'll see that you don't have to click the "undo" link for it to qualify as a revert; you reverted my edit and replaced it with something you decided was better, three times without achieving consensus any of those times. I have no interest in continuing this discussion with you as it's clear you aren't going to make any attempt to compromise. GiftigerWunsch [TALK] 06:16, 16 May 2010 (UTC)

GW, you are mistaken and I have accordingly reverted your last change. You are arguing for the most misleading and confusing option, which is to have the word "interface" here link the user directly to a page called "electrical connector." Note that that page would appear without so much as a redirect notice at the top. Simply in terms of WP linking principles, this would be wrong as it would violate the principle of least astonishment.

It also engenders the notion that an "interface" specification is in this case is nothing more than a physical connector. This notion is, of course, completely wrong, and actively misleading; speaking as someone with considerable experience not only in this field, but also in instructional design, I will also tell you that would even be a hindrance to further learning.

I agree that linking to the DA page is not particularly helpful, but at least it is not actively misleading, and in the absence of an article titled e.g. Interface (electronics) or Interface (computer hardware) either this, or removing the WL completely, would be the best option.

Regarding your assertion that

the procedure is to leave the change in place until it is challenged and removed by consensus. 

Um, I'm going to have to call a big fat on that one. I've been editing here far longer than you have and I've never heard of this before. Indeed, please see WP:consensus:

consensus ... begins with an editor boldly changing an article or other page. In response, the viewers of the page have three options: accepting the change, trying to improve the change, or reverting the change.

There is nothing there that suggests a new edit has any special status before discussion occurs. See also WP:BRD.

btw, I note that your edit history does not suggest that you are a SME in this or any remotely related area, and you even described yourself as a WikiNoob just a month ago. I suggest you are not the right person to properly evaluate this particular point. Please accept Tom's and my views here. Jeh (talk) 06:30, 16 May 2010 (UTC)

SMD was originally the project name for the first 40Mb and 80Mb SMD drives from CDC. It later also became the name of the interface as later projects designed new drives to be compatible with the same interface. There were at least two standard types of SMD connectors (and maybe more I didn't come across). The original drives predate ribbon cable and use large Amphenol connectors for the multi-core cables. Later drives and controllers switched to using 0.05" pitch twisted pair ribbon cable for the data cable, and semi-screened ribbon cable (to get the impedance correct rather than for screening) for the control cable, both using standard 0.1" by 2 row insulation displacement connectors (60 way for data cable, 26 way for the control cable). Many of the later controllers and disks using the ribbon cable interface could drive the SMD interface faster than the original 10mBit/s rate (I vaguely recall interface rates of up to 35Mbit/s on the Eagle drives we used).

So, SMD started as a project name for the first two drives, became also a data interface name, and hence a generic name for any disk drive with an SMD interface.

82.21.185.32 (talk) 16:29, 14 January 2018 (UTC)