|1973||CDC 9760 (SMD)|
|The first significant departure from IBM standards for disk storage media.|
Why it’s important
The then newly emerging minicomputer manufacturers needed low cost removable drives and media. Prior to Control Data Corporation’s (CDC’s) SMD (Storage Module Drive) the industry generally provided the industry with media and disk drives very similar in design to IBM’s offerings.
In 1973 rather than follow the new IBM 3348 Data Module (Winchester), a complex and expensive removable head disk assembly (35/70 Mbyte), the industry responded with new non-IBM lower cost disk packs and associated drives. CDC's SMD won out over similar products, e.g., Century Data’s Trident and Ampex’s DM series, to become the dominant approach to disk drives for other than IBM computing systems through the middle of the1980’s.
Additional products using the SMD interface followed thereafter from CDC and its competitors As a consequence, by 1980 CDC was the largest independent disk drive supplier and the disk drive industry was following a product design path separate from IBM.
Until the 1970s the products and technologies of magnetic disk storage were dominated by the designs of IBM and the market principally consisted of removable disk packs, first introduced as the IBM 1311 in 1963. By the early 1970s it was clear to the technologists that the challenges created by media removability and interchangeability (e.g., misalignment between heads of different drives, different performance characteristics of heads, etc.) would ultimately require that removability be eliminated. However, it was equally clear that the market, particularly in the minicomputer class of systems, desired removable media.
In the 1970s CDC's development and manufacturing of storage peripherals was located principally in Minnesota and Oklahoma. In 1975 it was spun out as a subsidiary, Magnetics Peripherals Inc, (MPI). MPI later was renamed Imprimis and then acquired by Seagate.
In December of 1970, a group of engineers, led by Tom Murnan, started design of a fixed media disk drive, called the Memory Module, which could be rack mounted with up to 4 independent HDAs (head disk assemblies) sharing a common set of electronics with the entire drive mechanism as a field replaceable unit.[Ref. 1,2]
During the spring of 1972 the fixed drive Memory Module evolved into a removable disk pack, the SMD disk pack. The genesis of the idea to use a disk pack rather than copy the IBM 3348 data module design is unclear[ref 3]. The IBM 3348 although a year away from announcement was understood to be removable albeit in a complex data module form. Figures 1 and 2 to the right illustrate the relative simplicity of a conventional disk pack when compared to the data module (both have covers removed). Both have disks but the data module then adds heads, base casting, spindle carriage and electronics, none of which are in a disk pack. In addition removing the data module requires a relatively complex load/unload mechanism in the drive. Although much of the same materials reside in an SMD disk pack mounted in an SMD drive, in a removable media environment with many disk packs per drive the cost disadvantage of the data module is apparent.
By August1972 the decision had been made to develop a removable disk pack that would fit in a standard rack mount (10.5-inch height) as an alternative to the data module approach [Ref 4, 5, 6] Shortly thereafter the Normandale team, principally Larry Matthews for engineering and Phil Arenson for marketing, agreed upon a $4.5 million plan which was first approved by Normandale's VP Lloyd Thorndyke and then authorized by Kamp.
Tom Murnan was assigned as the program manager; releasing the first SMD specification on September 19, 1972 and held the first design review on September 30, 1972 [Ref 7]. The first product was the 40MB 9760 and the second product was the double track density (to 384 tpi) 80MB 9762. Subsequently the 150 MB 9764 and the 300MB 9766 drives and corresponding disk packs were announced.
Technical innovations included:
- the SMD unipad head [Ref. 8] with a spoiler hole to get the stability with the close flying height needed for the bit density and track width. The figure to the right depicts the SMD head versus a more conventional IBM 3330 head. The spoiler hole is visible as a tiny dot near the middle of the chevron
- a light-weight mechanics set that allowed a self-contained removable pack disk drive to fit into a 10.5 inch height,.
The disk pack was developed in Normandale and produced in Omaha. Gary Warmka was the lead engineer. The necessary production equipment such as servo track writer and disk pack tester were developed in Normandale with the effort led by Dick Yonke. Ultimately there were multiple independent sources for SMD disk packs.
Noel Allen designed the rack-mounted SMD mechanics set, winning a CDC Technical Excellence Award.
Bruce Johnson took the lead role in the creation of the SMD Interface architecture. The market success of SMD let to its interface becoming first a de facto interface standard and in 1982 it was adopted as ANSI standard X3.91M.
The 9760 "40 Megabyte tabletop disc drive" was announced in June 1973 [Ref 9] with volume production beginning in 1974.
The 9760 achieved its 40 Mbyte capacity by increasing the bit density by precisely 50% over conventional disk pack drives such as the IBM 3330 to approximately 6000 bits per inch, essentially the same as Winchester, but at a higher data rate due to the slower rotational speed of Winchester. This was a huge risk since it required major controller changes to handle more data per track at a higher data rate; traditionally the industry increased track density while maintaining track format to minimize controller changes . Other industry participants such as Calcomp and Information Storage Systems responded with traditional designs.
Nixdorf was the first customer receiving its unit in December 1973, followed by by many others, including DEC and most if not all of the other minicomputer manufacturers.
The 9762, an 80 megabyte version, was announced in June 1974. [CDC77] The capacity was achieved by doubling the track density and with the industry availability of controllers supporting the SMD interface the price per megabyte advantages of this higher capacity was a major contributor to the success of SMD over the more conventional alternatives. The risky unique interface turned into a competitive advantage.
At the 1975 NCC, CDC announced 150 and 300 MByte Storage Module Drives The increase in capacity was achieved by increasing the number of disks (and heads) without any technology change. The 300 MByte 9766 SMD drive was for many years the largest capacity removable disk drive on the market.
The SMD disk drive’s success from the mid-1970s into the 1980s was in meeting the market’s demand for removability by initially having a rack mountable removable disk pack and drive at a much lower cost than IBM's Data Module and associated drive and then providing a second generation at higher capacity and lower cost per megabyte. Collectively, it's volume exceeded far that of the IBM 3340 (Winchester) disk drive and associated 3348 Data Module. CDC produced its 100 thousandth SMD disk drive in July 1981.
Imitation is the sincerest form of flattery - the success of the SMD drives brought imitators - Ampex was an early supplier of SMD disk pack compatible disk drives and others including Calcomp then came out with pack and interface compatible disk drives.
CDC followed its SMD disk drives with families of SMD interface compatible disk drives, both removable and fixed that were the basis for its continued successes into the 1980s.
By 1983 at least 25 manufacturers were supplying disk drives with SMD compatible interfaces.
|December 1970||Memory Module Project Kickoff|
|January 1971||Memory Module Proposal Generated|
|Jan – Dec 1971||Engineering Feasibility Studies and Technology Development|
|April 1972||Fixed versus Removable Media Study Initiated|
|Summer 1972||SMD Program Plan & $4.5M funding request approved by Tom Kamp|
|September 1972||SMD Specification Created|
|June 1973||9760 SMD announced at NCC|
|August 1973||First SMD 9760 Demonstration Units Shipped to Customers|
|November 1973||SMD 9760 Manufacturing Plan Authorized @ 200 units/month|
|December 1973||First SMD Preproduction Unit Shipped to Nixdorf|
|March 1974||Manufacturing Plan Modified to Include the SMD 9762 [80 Mbytes]|
|April 1974||Initiated Development of the SMD 9764/66 Drives [150/ 300 Mbytes]|
|June 1974||Announced 9762 at NCC|
|May 1975||Announced the SMD 9764 and 9766 Products at NCC|
|May 1977||Delivered the 5,000th SMD to Systems Industries|
|December 1979||Delivered the 50,000th SMD to Philips|
|August 1981||Delivered the 100,000th SMD to Datapoint|
|1982||SMD interface approved by ANSI as industry standard X3.91M|
(copies generally available from author)
- Memory Module Memo #1, T.L Murnan, December 14, 1970
- Memory Module Technical Proposal, T.L. Murnan, D. Conway, O. Nicholson, January 26, 1971
- Some credit the impetus for a removable disk pack to Bill Morgan (ex- IBM) a consultant to Tom Kamp, President, CDC's Peripherals Products Company. During the period when the decision was made to change from fixed to removable, Morgan authored a memo, “OEM/End User Strategy "’Winchester’" dated April 26, 1972 which unfortunately is missing. Lloyd Thorndyke states it started with Tom Dugdale's observation that "...the IBM 3330 pack is too damn much data for the minicomputer market." (see Oral History of Lloyd Thorndyke, p.21)
- 30 Megabyte File, L Matthews, July 5, 1972
- Memory Module Specifications, T.L,Murnan, August 7, 1972
- Storage Module Design Review, August 30, 1972
- Preliminary Storage Module Design Specs, T.L.Murnan, September 19, 1972
- 1971 Annual Report Of 6000 BPI Head Development, December 31, 1971
- Datamation, August 1973, p. 101
The following documents have been given to the CHM as donation "CDC/IMP SMD documents (3), X3831.2007 (X3834.2007), Gift of Tom Murnan":
1979 SMD (9760/9762) Product Specification (flat cable)
- November 1972 development plan extract listing the properties of what became the SMD 9760.
- [CDC77] 1977 CDC newsletter including an article, "Year of SMD"
- Dec 14, 1979 internal Magnetics Peripherals (CDC subsidiary) publication, "Normandale Builds 50,000th Storage Module Device." Twenty-eight pages apparently consist of face shots of the persons within the various SMD organizations at related CDC locations in late 1979.
- Aug 10, 1981 internal Magnetics Peripherals (CDC subsidiary) publication, "Normandale Celebrates Shipment of 100,000th SMD."
History of Thomas G. Kamp, included in CDC History at Charles Babbage Institute
Berreth (Richard) and Murnan (Tom) Oral History, Computer History Museum, June 29, 2009, Accession Number 102702131
Kamp (Tom) Oral History, Computer History Museum, June 12, 2010, Accession Number [to be provided]
Lloyd Thorndyke Oral History Interview 303, February 18, 1980, Charles Babbage Institute
Oral History of Lloyd Thorndyke,Computer History Museum, August 13, 2010,CHM Reference number: X5969.2011
Tom Murnan, “The Emergence of the CDC Storage Module (SMD) 9760 and 9762 Drives”, January 14, 2008, updated August 1, 2011, available from author
Phil Arneson email on The History of the SMD, April 26, 2009, updated July 29, 2011, available from author
This article was authored by Tom Gardner with extensive input from Tom Burniece and Tom Murnan.
It was reviewed at the Storage SIG meeting of August 17 with changes suggested, a few of which are incorporated herein.
Revision 25 is essentially the version reviewed by the SIG.