1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
Between the development of Synchro-Mesh (launched for 1929) and Hydra-Matic (launched for 1940), General Motors also developed a four-speed semi-automatic transmission, sometimes called the Automatic Safety Transmission, which was available on 1937–1939 Oldsmobiles and some 1938 Buicks. Here’s a 1974 article from Special Interest Autos discussing this obscure, often misunderstood transmission, to which I’ve added an assortment of corrections and explanatory notes, plus pictures of some of the few surviving cars still equipped with it.
The article presented here originally appeared in Special Interest Autos (SIA) No. 20 (January–February 1974), the same issue as Karl Ludvigsen’s article on the Chevrolet Cadet. Like the Cadet article, it was never reprinted in any of the Hemmings collections of Special Interest Autos material, and the version that appeared on the Hemmings website for some years is now long gone.
I think this article is historically important enough to be worth keeping available in some form other than a handful of ancient print magazines. However, it also contains some errors and misconceptions that I’d like to clear up for the record. While I’ve never had a chance to get my hands on one of these transmissions (only a handful still survive), I have several of the factory service manuals, technical analyses from contemporary trade journals, and copies of the underlying patents, so I probably understand it about as well as anyone still living.
Unlike the various Chrysler “Fluid Drive” M-5 and M-6 semi-automatic transmissions, the GM semi-automatic transmission was available only on a limited number of models, and only for a relatively short time. Oldsmobile, which called this the Automatic Safety Transmission, introduced it as an option on late 1937 eight-cylinder cars, and then offered it on both sixes and eights in 1938 and 1939. Buick, which called it the Self-Shifting Transmission, offered it only on 1938 Series 40 Special cars. No other GM division ever offered this transmission, and production ended in September 1939.
1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
The GM semi-automatic transmission was developed by a GM Engineering Staff team led by Earl A. Thompson, the inventor of Synchro-Mesh, in collaboration with Oldsmobile. (All the production transmissions were actually built by Buick, although Buick’s interest in the project was low, and their participation was apparently very reluctant.)
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission / Darin Schnabel – RM Sotheby’s
This transmission had a conventional plate clutch, sliding gears for reverse and neutral, and two sets of planetary gears. There were four forward speeds, divided into two driving ranges: low, which gave 1st and 2nd, and high, which gave 3rd and 4th. Here’s a labeled cross-section from the 1938 Buick service manual, showing everything but the manual clutch, which would be to the left in this view:
From the 1938 Buick service manual; the 1937–1938 Oldsmobile version was identical
If you’ve heard of this transmission at all, you’ve probably heard that it was basically just a simpler version of the early Hydra-Matic Drive with a plate clutch instead of a fluid coupling. That’s not really true — in fact, the semi-automatic transmission was quite a bit different from Hydra-Matic, although they were both developed by most of the same people and were similar in concept.
The photo captions read, “GM semi looks a lot like Hydra-Matic, but sans torque converter. This is one of two discovered recently, still in their crates, by a Palo Alto Olds dealer. H-M Div. bought both for display. Here’s where a standard 1938 Buick would plant its shift stick. Now there’s room for three. GM called semi-automatic a ‘safety’ trans because the driver could keep both hands on the wheel.”
The first caption contains this article’s most embarrassing glitch: First- and second-generation Hydra-Matic transmissions did NOT use torque converters — they used fluid couplings: two-element fluid clutches that did not multiply torque. Also, only Oldsmobile called the semi-automatic a “Safety Transmission.” (Buick didn’t use that name.)
SIA began their report:
WE’RE TESTING this transmission in the wrong car. It ought to be in an Olds rather than a Buick. The semi-automatic was much more Oldsmobile’s baby than Buick’s. And while it’s true that Buick manufactured the Automatic Safety-Transmission for Olds from June 1937 through Sept. 1939, Oldsmobile actually stood behind its development and got it ready for production.
So few of these semi-automatics still exist, though, that we felt lucky to find one at all. Our test car, a lovely 1938 Buick Special business coupe, was kindly lent us by Harrah’s Automobile Collection in Reno, Nev. We’d been curious for some time to find out what the semi-automatic felt like—how it performed and how we performed behind the wheel—so we drove this Buick out into the hills behind Reno and gave it a thorough trial.
Depending on your age, you might recall hearing about Bill Harrah’s vast automobile collection in Reno, which was the automotive equivalent of the Library of Alexandria. At its peak, it included something between 1,100 and 1,800 cars, some quite rare. However, Harrah did not make provision for the collection in his will, so after he died, Holiday Inn ended up with the collection, and sold a big chunk of it at auctions in the 1980s. Some of those cars are now part of the National Automobile Museum collection, but the 1938 Buick business coupe Special Interest Autos drove apparently is not.
1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
I don’t know what became of that particular car, or whether it still exists, but I found RM Sotheby’s listings for two other survivors with the semi-automatic transmission: a 1938 Buick Model 40C Special Phaeton (the maroon car above) and a 1938 Oldsmobile L-38 convertible coupe (the blue car below). There are several other survivors, but not many. I’ve seen estimates that there may be as few as SIX surviving 1937–1939 cars that still retain the semi-automatic.
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission / Darin Schnabel – RM Sotheby’s
SIA continued:
In boning up about this transmission, we’d read that the factory recommended starting out in LOW range and then shifting by hand into HIGH range at about 20 mph. We tried it that way at first and then also tried starting off simply in HIGH range. We found very little difference in acceleration, and of course, in HIGH range we eliminated that one manual shift.
The transmission has a total of four forward gears. Low range starts out in first gear and shifts automatically to second, then holds it there. You powershift by hand, but without using the clutch and without letting off the gas pedal. At that point the transmission goes to third gear, and the final shift to fourth is again automatic. That’s how it performs if you do it by the book.
But we soon found, as did most owners, that it’s much simpler to start off in HIGH range. If you do that, it’s automatic all the way. You do lose second gear, though. In HIGH, the transmission moves off in first gear, then shifts automatically to third, and finally shifts to fourth, again automatically. We decided we didn’t miss second gear, and acceleration didn’t suffer very much.
In principle, starting in high range should have given a 3rd-gear start. However, because the oil pumps for this transmission were driven off the main shaft, the pumps stopped if the clutch was released. As the clutch engaged, it took a few beats for the pumps to build up enough pressure to operate the transmission’s clutches and servos. So, if you started from rest in high, the transmission would start in 1st and then abruptly shift to 3rd at about 7 mph.
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission / Darin Schnabel – RM Sotheby’s
Making a habit of that was hard on the transmission, so Oldsmobile very strongly recommended always starting in low. Buick wasn’t as emphatic, saying only that starting in low was “preferable,” and many owners probably decided that starting in high was easier.
Yes, the semi-automatic does still need the old clutch pedal. You use the pedal when starting, stopping, idling, or shifting past neutral on the column selector. We felt some chatter during rolling starts, and at first we thought this was clutch chatter. But we soon found it came from the transmission directly. What caused the chatter we don’t know, probably adjustment. We understand that the Automatic Safety-Transmission was very sensitive to engine tune, temperature, and internal adjustment.
We talk about this transmission being “semi-automatic,” because there’s that old devil clutch still on the floor. Actually it’s automatic except when starting or stopping. The only thing it lacks is the Hydra-Matic’s torque converter [sic], which allowed enough slippage to eliminate the friction clutch. There’s the same kickdown for passing, so if you’re driving along in fourth gear and want to accelerate quickly, tromping the gas at any speed between 23 and 55 mph immediately engages third. And upshifts respect that magic combination of car speed and accelerator pressure that automatically varies shift points on harder acceleration—the “brain” that distinguishes this transmission from rival semi’s and full automatics of its time.
Although the semi-automatic transmission was prone to gear noise, which gave Oldsmobile engineers fits, it should be understood that the way the SIA editors were driving the test car was also somewhat abusive. The designers of this transmission expected drivers to use both low and high ranges. You could shift from low to high at any time, and you could shift from high to low as long as the engine speed was low enough to do so safely; if it wasn’t, a lockout mechanism would prevent the lever from moving out of high.
1938 Buick Self-Shifting Transmission worked exactly the same as the Oldsmobile Automatic Safety Transmission (and was mechanically identical), but labeled the shift quadrant with two “F” ranges (for Forward) rather than “L” and “H” / Patrick Emzen — RM Auctions
Unlike Hydra-Matic, the semi-automatic transmission could not shift for itself through all four gears. Its actual automatic controls worked only on the front gearset. They were mostly mechanical, a complicated Rube Goldberg contraption involving rods and levers controlled by a flyball governor (turning at engine speed) and movement of the throttle linkage. Here’s the complete valve body when removed from the transmission — the automatic valve is at the right, the manual valve (which controlled the rear gearset) is at the top right. The vertical rod at the left is the relay shaft for the governor.
From the 1939 Oldsmobile service manual; earlier Buick and Oldsmobile versions were identical
The governor rod (highlighted in blue in the image below) acted on the shift rod through the equalizer bar (purple), whose fulcrum point moved forward or backwards depending on the position of the throttle linkage (green). There was also a hydraulic piston (labeled “C” and “D”) that engaged to further raise the shift points when the transmission was in high.
From the 1938 Buick service manual; 1937–1939 Oldsmobile versions were almost identical
This mechanical linkage was way too complicated for too little capability, so Hydra-Matic went a completely different direction, using metered hydraulic pressures to operate spring-loaded spool valves. However, the controls shown above DID allow the semi-automatic transmission to “power-shift” — it could change gears automatically without an interruption in power, even at full throttle — and it could vary the automatic shift point based on throttle position. SIA explained:
You can hold the accelerator to the floor, and shifts come at the high end of the rpm scale, just as they would if you were trying to get the best from a manual-transmissioned car. The final shift to fourth comes at around 55 mph under full throttle but at 28 mph in normal driving. The shifts themselves feel smooth but are noticeable. Yet the engine doesn’t rev unrestrainedly during shifts.
In rolling to a stop, we did find the downshift back to first gear rather too forceful, with a solid clunk around four mph. But we understand Olds remedied this in the 1939 version of the semi-automatic. To go into reverse, you have to push inward on a chrome button in the end of the column lever, and you also have to pull the lever toward you, same as shifting into reverse in a manual-transmissioned car with column shift.
The reason the downshift was clunky was the same reason the transmission would start in 1st even with the selector in high: When you disengaged the clutch when coming to a stop, transmission oil pressure would abruptly cut out, causing the rear band to reengage with a thump. 1939 transmissions used a hydraulic accumulator to cushion the band engagement, but even the earlier units were less thumpy if you selected low while slowing to a stop, before releasing the clutch.
Reverse in this transmission worked just like a sliding-gear manual gearbox. Like ’30s Synchro-Mesh transmissions, reverse gear wasn’t synchronized, so it was important to come to a complete stop before shifting into or out of reverse.
The photo captions read, “Shift points are too low for maximum acceleration. You have to use clutch when starting and stopping. Adjustment is critical. The weather, engine tune, and mileage affect the transmission. Earl A. Thompson, developer of synchromesh, pioneered work on semi and Hydra-Matic. Buick Special business coupe has an ample trunk, but it’s augmented by huge storage bin behind seat. Semi-automatic cost $80 in 1938. To shift into reverse, driver depresses clutch, pushes chromed button inward into column lever, then lifts lever up and toward himself.”
1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
Despite SIA‘s first comment, the semi-automatic transmission’s shift points were perfectly reasonable IF you actually used both low and high, as the manual recommended. If you did that, Oldsmobile claimed that the Automatic Safety Transmission cut 1 second off 5 to 25 mph acceleration times, and 1.5 second off 10 to 60 mph acceleration compared to a Synchro-Mesh car. The way SIA drove the test car amounted to always short-shifting in 1st and then using only 3rd and 4th gears!
Earl A. Thompson, circa early 1940s; the picture shown on the page above is from 1963
SIA then moved on to the most useful segment of the article: a discussion of the actual development of the semi-automatic transmission and its connection to Hydra-Matic. (Much of this information probably came from a pamphlet issued in early 1964 by the American Society of Mechanical Engineers when Thompson and his group received the Elmer A. Sperry Award.) SIA explained:
This semi-automatic was an ancestor and direct link to the Hydra-Matic transmission of 1940. Development of both was fraught with difficulties. R&D boiled down to years of cut-and-try.
The man initially in charge of the program was Earl A. Thompson, inventor of synchromesh. Thompson was one of those rare inventors who had knocked on GM’s front door from the outside, had been let in, was given a hearing, and he eventually sold GM his idea. Cadillac put Thompson’s synchromesh into production for 1928, and Thompson stayed on to become Cadillac’s assistant chief engineer (in 1929) under Ernest Seaholm. It was at Cadillac, then, that Thompson first began the research that eventually led to Hydra-Matic.
Just to get the chronology straight, Synchro-Mesh went into production in 1928 for 1929 Cadillac and LaSalle models. Seaholm hired Thompson as a Cadillac engineer (he’d previously been a consultant) in September 1929, but Thompson didn’t become assistant chief engineer until the following August.
The project started in 1932 with the purchase of several outside patents. It was Thompson’s habit to buy up potentially conflicting patents, and he had done this with synchromesh too.
This assertion has been repeated many times over the years, but as far as I’ve ever been able to find, GM only acknowledged buying one outside patent in connection with this project, US19845556, filed in 1932 by Dutch engineer Johan Machiel Vetter. That’s the only non-GM patent listed on early Hydra-Matic patent plates. (Also, as assistant chief engineer, I doubt Thompson would have had the authority to buy any patents — at most, he could probably have submitted a recommendation to the GM New-Devices Committee with his boss’s endorsement.)
Since all sorts of other semi-automatic and fully automatic transmissions were being developed during the early 1930s, it seemed only a matter of time before one or several would become commercial successes. Daimler had introduced the (non-automatic) fluid flywheel in 1930, Vickers-Coats had a true torque converter at that time, and there were transmissions in various states of automaticity called Spontan, Grade, Mono-Drive, Reo’s first Self-Shifter, plus several more. No-clutch shifting was being advertised widely, as were preselector gearboxes and, of course, overdrives and even freewheeling made people conscious of the need and potential for automatic transmissions. Henry Ford was trying to develop one, so if nothing else, GM was covering bases and accumulating patents through Thompson’s research.
In 1932, then, Earl Thompson was given a tiny lab and a 2-man staff to work on what was code-named the “Military Transmission.” This sign on the door was simply to disguise what was really going on. There was nothing military about it.
The first two members of Thompson’s group, photographed circa 1963
Earl’s first two teammates were Ralph F. Beck and Walter B. Herndon. Beck designed the gearbox for early versions; Herndon worked on hydraulic controls. (Herndon left the project in mid-1935 but rejoined Thompson in 1939 to refine the final Hydra-Matic for production [and then became the chief engineer of Detroit Transmission Division].)
A year or so after the Military project started, Earl added two more staffers: William L. Carnegie and Maurice S. (Rosy) Rosenberger. Carnegie worked on controls and general development, and Rosenberger ended up with full responsibility for testing the early prototypes for reliability. The transmissions, at that time, were pretty unreliable.
Carnegie and Rosenberger, circa 1963
Late in 1934, Cadillac decided it could no longer afford the Military project, so in Jan. 1935, Thompson’s entire operation was transferred as a product study group to GM’s Central Staff under corporate engineering v.p. O.E. Hunt. Thompson got two small offices in the GM Building plus a room in the research labs.
It was here that Charles L. McCuen, general manager of Oldsmobile, got wind of the Military Transmission. Olds was at that time GM’s most forward-looking division, thanks largely to McCuen, who was one of the corporation’s most forward-looking executive engineers. McCuen was named v.p, of engineering for GM in 1940 and then research chief in 1947, succeeding Boss Ket [Charles F. Kettering]. For the moment, though, McCuen saw great potential in developing an automatic transmission, so he decided to put up the money for it and hoped eventually to bring it out as an Oldsmobile option, which, of course, he did.
Oldsmobile general manager Charles L. McCuen, 1930s
There then began a period during which both Olds and the Thompson group worked side by side. Thompson’s staff kept doing basic research, while McCuen, along with his very capable chief engineer, Harold T. Youngren, got the transmission ready for production.
Says William L. Carnegie: “If you keep something like this in the development stage, you can keep developing it indefinitely. You come to a point where you ask, ‘Is this suitable for production?’ Once you release it for production, you get it out a lot faster.”
McCuen, Youngren, and Oldsmobile worked very quickly after deciding on production. There was no chance, though, to finish a suitable fluid coupling or torque converter at this stage, so Olds decided to go ahead with just the semi-automatic gearbox and use a conventional disc clutch and floor pedal.
The text doesn’t mention the other important member of Thompson’s group: Oliver K. Kelley, who was transferred from Yellow Coach to Engineering Staff in June 1936.
O.K. Kelley, circa 1963
It was Kelley who came up with the novel way Hydra-Matic obtained reverse gear (which was completely different than this transmission), and with how the fluid coupling was integrated with the planetary gears. Both were big departures from this semi-automatic transmission, and not just because they dispensed with the plate clutch.
The photo caption at the left reads, “Why Buick offered semi on lower series only isn’t known. Special’s engine delivers 107 bhp as against 141 for other 1938 Buicks. Olds refined semi for production, but Buick actually built the units and sold them to Oldsmobile Div.”
The 1938 Buick Special used a 248 cu. in. straight-eight, rather than the 320 cu. in. eight used in the senior Century, Roadmaster, and Limited / Patrick Emzen — RM Auctions
The reason Buick didn’t offer the self-shifting transmission on the senior series was almost certainly related to torque capacity: The 320 cu. in. eight used in other Buick models had 276 lb-ft of torque, whereas the Olds eight peaked at 200 lb-ft. Torque capacity limits may also be why Cadillac never offered the semi-automatic transmission, even though the project had started with them.
This 1938 Oldsmobile has the biggest Olds engine of the time, a 257.1 cu. in. straight-eight with 110 hp / Darin Schnabel – RM Sotheby’s
SIA continued:
When time came for actual production, GM’s central staff decided that Buick had greater plant capacity than Oldsmobile, so the gearboxes were built in Flint rather than in Lansing. Buick also built axles and other parts for Olds at that time.
At any rate, the first Automatic Safety-Transmission appeared on an Olds 8 in June 1937, and Buick waited several months before offering it. The Olds version cost only $80 extra, even though Buick charged Olds $180 to supply the gearboxes. Olds later made this transmission available on the 6 as well as the 8.
Buick actually built a completely new plant to manufacture the semi-automatic transmission: Plant 66, which cost them over $4 million to build and tool. This was why Buick charged Oldsmobile so much for each transmission — they had been obliged to incur a huge expense for a transmission Buick didn’t especially want.
The semi-automatic wouldn’t interchange with a standard transmission because of differences in size, so it had to be a factory installation. When something went wrong with one of these transmissions, GM instructed dealers not to try to repair them but simply to return them to the factory for an exchange unit. At the factory, gearboxes were torn down, inspected, repaired, and records kept on failures. This way, by the time the fully automatic Hydra-Matic came along, a good deal of field testing had already been done by owners.
Early on, Oldsmobile did ask dealers to exchange malfunctioning transmissions rather than attempt to repair them. However, Oldsmobile and Buick both subsequently issued factory service manuals for the transmission and made replacement parts and tools available for maintenance and repair. Disassembly, repair, and maintenance procedures were also described at length in the 1938 and 1939 Oldsmobile shop manuals, so dealer technicians COULD work on these transmissions, although some may have been reluctant to do so.
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission / Darin Schnabel – RM Sotheby’s
As for the second assertion, while Hydra-Matic may have benefited a little from Oldsmobile and Buick experience with the semi-automatic transmission, they were different in so many ways that I’m skeptical that the semi-automatic really provided much of a head start in field-testing Hydra-Matic prototypes, which Oldsmobile began conducting separately in 1938–1939.
Innards of the semi-automatic’s gearbox were very much like the Hydra-Matic’s. It had two planetary gearsets in series, one ahead of the other. Sun and ring gears were held by clutches and bands. The automatic selection of the most efficient gear was controlled by a centrifugal governor. Shifting from one gear to another was by hydraulic pistons that controlled the brake bands and clutches within the planetaries.
In first gear, the bands for both gearsets were held tightly against the drums, so power flowed through both sun gears. In second, the forward band released to duct power through the rear gearset. In third, the gear band let go and the rear clutch engaged. In fourth, both bands released and both clutches engaged for direct drive. Reverse used a separate set of convention[al] gears.
The semi-automatic transmission’s innards were NOT “very much like Hydra-Matic’s” except in certain broad ways. (Both did use planetary gears controlled by servo-operated brake bands and multi-disc clutches.) SIA also badly garbled their description of the planetary arrangement: Power flowed through both planetary gearsets at all times, but in 1st, power flowed through the front ring gear to the front planet carrier (the sun gear was held stationary), then to the rear sun gears (with the first ring gear held stationary) and the rear carrier, which was splined to the output shaft.
This section of Fig. 1a of Earl Thompson’s US2193524 patent (filed March 15, 1937, issued March 12, 1940) shows the semi-automatic transmission’s compound rear gearset — in this view, the output shaft (numbered 50) is at the left of the diagram, with the clutch and engine to the right
One unusual feature of the semi-automatic transmission was that the rear planetary gearset was actually an interconnected compound gearset, allowing a lower numerical gear ratio. Early Oldsmobile Hydra-Matic transmissions didn’t use this, but it later showed up on the early Cadillac and wartime Hydra-Matic units. Postwar units then went back to a simple planetary gearset.
Oldses and Buicks equipped with the semi had higher rear axle ratios than standard models, giving a 20% “overdrive” in fourth. This meant that at 2700 rpm, a semi-equipped Olds was cruising 65 mph while its conventional cousin was doing only 53 mph. Olds claimed 12% better acceleration when using all four gears. (Gear ratios were changed for 1939.) Other semi-automatic advantages included 2-handed steering at all times, thus the “safety” epithet; an unobstructed front center seat, with no stick on the floor; longer engine life; and less gas and oil consumption.
The semi-automatic transmission didn’t actually have overdrive gears, but Buick and Oldsmobile specified taller (lower numerical) axle ratios on cars that had this transmission. Buick normally specified a 3.615 axle on self-shifting cars (although a few had 3.90 gears for mountain driving), compared to 4.40 for the Synchro-Mesh Special, while Oldsmobile used a 3.64 axle, compared to 4.30 on Synchro-Mesh cars. The transmission’s lower gears made up for this in acceleration, but the taller final drive ratios reduced engine speeds for more economical cruising — all of the claims about lower fuel and oil consumption and reduced wear were due to the axle ratios rather than the transmission itself.
1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
Gear ratios for the Automatic Safety Transmission were NOT changed for 1939, and remained as listed below, still with the 3.64 axle:
Olds/Buick Automatic Safety Transmission Gear Ratios
- 1st: 3.16:1
- 2nd: 2.23:1
- 3rd: 1.42:1
- 4th: 1.00:1
- Reverse: 3.20:1
When the semi-automatic transmission debuted in 1937, column shift was rare on American cars. However, Buick and Oldsmobile adopted it across the line for 1939, and within a few years, it would be almost universal.
Buick no longer offered the self-shifting transmission for 1939, but all 1939 Buicks had “Handi-shift” column shifters for the standard three-speed Synchro-Mesh gearbox / Old Car Manual Project Brochure Collection
The caption of the cross-sectional illustrations reads, “Semi-automatic gearbox has four forward gears, with two sets of planetaries. Bands and clutches inside case are actuated hydraulically. The controls sense load and shift accordingly. Reverse comes by a set of non-planetary spur gears.” Here are bigger versions of those illustrations, which are from the May 29, 1937 Automotive Industries:
SIA concluded:
We haven’t been able to find out how many Automatic Safety-Transmission-equipped Oldses and Buicks were sold. Estimates put the figure at less than 7% of 1937-39 Olds/Buick production. So the Hydra-Matic’s grandaddy wasn’t a rampant bestseller by any means. But this reluctance on the public’s part is understandable, because the Reo Self-Shifter (see SIA #13, pp. 30-35) and other mid-1930s shifting aids promised a lot more than they delivered. So the public was a little gun-shy by that time. However, a less reputable automaker than Oldsmobile (and later Cadillac) wouldn’t have been able to put the semi-automatic and Hydra-Matic over at all.
I don’t know that the public was gun-shy so much as skeptical, and for good reason: Not only did this semi-automatic transmission still require manual clutching, it didn’t eliminate manual shifting so much as make it more complicated. To get the best out of it, you really needed to understand how the transmission worked, and even then, it didn’t allow either fully automatic OR fully manual control. I think once the novelty value had worn off, it would be more exasperating than useful. By contrast, Hydra-Matic Drive was riskier, but it offered a lot more.
1938 Buick Special Phaeton with Self-Shifting Transmission / Patrick Emzen — RM Auctions
Buick apparently only built and offered their Self-Shifting Transmission because they were ordered to (which they resented the whole time). So, the question becomes, why did Oldsmobile bother with this semi-automatic transmission at all? It had less in common with Hydra-Matic than SIA assumed; the Automatic Safety Transmission represented an array of dead ends that were obsolete almost as soon as they went on sale. (The related patents also include an interesting missing link that was never offered in production.)
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission — I don’t know what happened to this car’s speedometer, and the listing doesn’t say / Darin Schnabel – RM Sotheby’s
I suspect that the main reason for offering the Automatic Safety Transmission to the public was that treating the whole project as a production program (even one that clearly lost a lot of money) made it easier to justify the expenditure of resources involved. It was still the midst of the Great Depression, after all, and there were limits to how much money even General Motors was prepared to throw around.
1938 Oldsmobile L-38 convertible coupe with Automatic Safety Transmission / Darin Schnabel – RM Sotheby’s
People often (rightly) criticize manufacturers who treat their paying customers like beta testers for new technology, but I think the Automatic Safety Transmission wasn’t even that: It was basically a loss leader for a related R&D project, a rough draft of the significantly different, more ambitious, and ultimately far more successful Hydra-Matic.
Related Reading
1929 Cadillac/LaSalle Synchro-Mesh Transmission – “Shifting Is Made Simple And Clashing Of Gears Is Ended” (by me)
The World’s First Automatic Transmission – The 1904-1907 Sturtevant Automatic Automobile (by me)
Planetary (Not Interstellar) Overdrive (1934-72) (by Paul N)
From NDLR To PRNDL – How One Inventor’s Grudge Against GM Helped To Standardize Transmission Shift Patterns (by me)
Early Hydra-Matic Users: Many Non-GM Automakers Bought This Pioneering Automatic Transmission (by me)
Hydra-Matic History: GM’s First Automatic Transmission (at Ate Up With Motor)
THANKS! I’m always fascinated by semi-autos but frustrated by a lack of info. Most auto writers are great with engines but inadequate with transmissions. Most misunderstand freewheeling and Fluid Drive, some don’t even grasp synchromesh.
You’ve added tremendously to the stock of accessible info. Your discussion of the personalities and economic motives is also interesting!
I remember reading the SIA article when it came out, but had long forgotten the details.
The semi-automatic was a hot thing for awhile in the late pre-war years, but it seems that Chrysler was the only company that made one that was easy to use and worked well. It probably helped that it might have been the simplest design of all of them.
Ive driven a few 3 pedal automatics, all operated differently but this is different again, it predates the others by 80 years and its in a car not a truck, wow a learning day.
Somebody somewhere is always the Beta tester, my sister and BIL were for the widebody Camry, thats why they refined it for world consumption, the original was too fast too thirsty and cost to much to trim like that, less used cow better MPG and the world loved them, this semi auto or semi auto shift didnt make the cut, considering you can drive a full crash trans only using the clutch to start off and stop this semi auto was a complicated way of achieving very little.
The transmission seems like it would have been a pain to use, much less service. With standard transmissions, these cars were so flexible in high gear – note the Buick’s 4.40 rear axle ratio – that a semi automatic that still required declutching at every stop wouldn’t be very appealing.
The lower rear axle ratio would have been nice for economy, but opportunities for driving at sustained higher speeds, say over 50, were much fewer then, and B-W overdrive was extremely good for those situations, though GM treated it with the not invented here hauteur in this period. Even Ford’s two speed rear axle was more straightforward.
Thanks for the detailed overview, Aaron.
An interesting feature that Thompson and his team came up with for this transmission that wasn’t actually used in production was an additional “3” position on the quadrant, which would keep the transmission in 3rd gear until the engine reached the maximum safe upshift point (which would have been around 60 mph with these gears). This would have mitigated the flexibility issue by allowing the driver to choose whether to stay in 3rd or 4th in part-throttle cruising.
Later, the same concept was applied to Hydra-Matic, creating the “Dual Range Hydra-Matic” of the ’50s. It worked differently, since the controls weren’t the same, but it was the same idea.
Interestingly, the first self-shifting transmission Thompson created in 1932–1933 was an automatic overdrive, so GM was not unfamiliar with the concept even though they didn’t adopt it until much later.
The transmission is interesting. The dashboard of the ‘38 Olds as seen in the second-to-last photo – now that’s stunning! A Bakelite and steel work of art.
Yes. Writers tend to overuse “ahead of its time”, but the 38 Olds dash was WAY ahead. Olds dropped it the next year. The overall shape didn’t come back until ’49, and the side-facing controls didn’t come again until the ’80s.
Excellent column Aaron…like everything you write.
I was given for my 15th birthday the great “Buick” by the late Beverly Rae animes. That was in 1980, the book was new, and it had a chapter “The Buick shiifts for itself” with lots of descriptions and data that as far as I remember match much more your comments than SIA’s. Keep these columns fllowing, and thanks a lot!
Thanks for a long-overdue education on the history and operation of this unit. As always with your posts, I am personally grateful for the education and very proud to offer them here at CC. There’s just nothing remotely like it anywhere.
Extremely detailed article, much appreciated!
Regardless of any Hydramatic commonality – or lack thereof, the Automatic Safety Transmission at least pointed GM away from further development of mechanical transmission control systems and towards hydraulic control systems. Until electronic controls developed enough to become more practical transmission controls, hydraulic control was used by every automatic transmission that followed the Hydramatic. One transmission mystery that still eludes me is why GM never coupled a torque converter with any of the Hydramatic design iterations.
Aaron may add more detail, but in essence the HM with its four gears didn’t really have any use for a torque converter. As it is, it had one more gear than the usual 3-speed manual, which explains the improved acceleration. The first gera on the HM was quite low; a TC would have been overkill.
Back then, it was either one or the other, meaning torque converters were seen as a complete automatic transmission, with no actual mechanical gear shifts. This is how the Dynaflow and Power glide worked. Their Low range was intended to be used only for ascending or descending very steep hills.
Even the early 3-speed torque converter automatics (Fordomatic, etc.) only used two gears in automatic operation, with Low having to be engaged annually. It wasn’t really until the Chrysler Torqueflite came along in 1957 and showed the benefits of a proper 3-speed automatic with TC. That eventually became the template for all subsequent automatics.
They did, sort of, with the third-generation (1961–1964 three-speed) Hydra-Matic, which also speaks to the thought process involved.
A torque converter involves some significant tradeoffs: Torque multiplication requires slippage, which reduces efficiency and produces additional waste heat. You can mitigate that with a mechanical lockup clutch, but then you have to work out some strategy for deciding when to engage and disengage the clutch, which entails a whole additional set of tradeoffs.
Except at very low road speeds and when starting from rest, fluid couplings are usually arranged to get to their coupling stage as quickly as possible, so their efficiency is higher, which also means they run cooler. (Hydra-Matic got by with air cooling for a surprisingly long time.) Hydra-Matic transmissions also used torque-splitting in the higher gears, so at cruising speeds, only part of the input torque even goes through the fluid coupling, cutting way down on slippage. This meant it needed a higher numerical step-off gear (as did the Mercedes fluid coupling four-speed automatics), but overall efficiency was very high except at parking-lot-crawl speeds.
For the third-generation Hydra-Matic, which was intended to reduce cost and weight, Detroit Transmission thought, “Okay, we really only use first gear for a short period at low speeds, and having an extra automatic shift is complicated. What if we used some fluid torque multiplication, but ONLY at launch?” So, they added a torque multiplier to the fluid coupling, but instead of using a conventional stator, they splined it to the output shaft, which meant it only multiplied torque when starting from rest. Its blading was still like a regular fluid coupling, so its torque ratio was low (1.2 in first, about 1.3 in reverse), but once the car was moving, it acted as a “tight” fluid coupling, with a high efficiency. They did consider making it a conventional torque converter — there’s a patent for a version like that, although it wasn’t produced — but I think they decided the loss of efficiency wasn’t worth it.
Ultimately, that was kind of a dead end, and by the early ’60s, maximum efficiency wasn’t a high priority, so GM decided two- and three-speed torque converter automatics were a better all-around compromise.
Also, it’s worth noting that O.K. Kelley was the lead designer of GM’s later torque converter transmissions (Dynaflow, early Powerglide, Turbine Drive, and the triple-turbine Turboglide/Flight Pitch Dynaflow). He went from Hydra-Matic to focusing primarily on torque converters.
Great essay! I enjoyed reading about this interim transmission and its limitations due to torque. Attached is an ad for the 1939 Oldsmobile featuring Hyra-Matic with no clutch pedal.
Fascinating! I have heard the Safety Transmission mentioned but knew nothing of it. Glad Aaron saw fit to write it up in his usual thorough style. I wonder if the Safety Transmission was something of an ‘insurance policy’ in the event the Hydra-Matic had failed.
I’m not surprised this went nowhere. I’d be confused driving it – when do I need to shift? When do I not need to shift? When do I need to use the clutch? I like my automatic transmissions to be automatic – I don’t even like all the “sport modes” and upshift/downshift tabs on modern automatics; the whole point of an automatic transmission is not to need fiddling with what gear you’re in. Otherwise I’d rather stick with the simplicity of a regular manual transmission.