Origins of Constant Recoil: The Ultimax Mk3 (feat. Mae & a Yeti)

This is Lot 2052 in the upcoming October 2019 Morphy Extraordinary auction.

The Ultimax light machine gun was developed by James Sullivan for the Singaporean military, and it is the first mass-production machine gun to be designed with the “constant recoil” system in which the bolt never impacts the rear of the receiver and the recoil impulse is felt as a continuous push instead of series of rapid impacts. So, is it really as good as people say? Let’s find out – and let’s get more than one opinion on the question.

Thanks to Mae and Othais at C&Rsenal for appearing in this video!



  1. I know about KAC lmg thats beltfed, but I think this slow rate of fire constant recoil is more suited for mag/drum fed mags, because with belt you need additional energy to strip rounds and operate it,
    thats one of the reasons why older MGs have higher rate of fire, even Negev thats newer design than Ultimax, is plagued by it, making in heavier but less controllable.
    (its widely known but also strong point of the Ultimax is being also the lightest weight, they forgotten to mention that in the video as another wow thing).

    Second part of the conondrum that made Ultimax less desirable to widespread issue is the very feeding; drum works ok, but also you need modified ar15 mags with 2 punched holes meaning you can not get a feed from your fellow soldiers unless all their mags are modded (and mod is potentialy shortening the mag service and damages it). Later, there is special upgraded mk8 model or so, that could use regular ar15 mags but then you need special Beta C drum.

    I think in last iteration (1-2 years ago) they come up with model using regular drum, but you could add on adapter for regular mags, unfortunately that is still finicky in the combat, so for now(and last 35 years) there is no silver bullet solution.

    • Note that original intention for Ultimax drums was to use it in disposable manner
      Drum magazines are made mostly from plastic, with translucent rear wall which permits easy check of available ammunition. Original intent was to make these drums disposable but currently these are reusable
      This might influence design choices regarding weapon-magazine interface, as disposable magazine do not have to be as durable as reusable one.

      • Its true that there have been ideas of disposable magazines (that are to be preloaded and stored that way), in ar15 first iirc, but none of the world militaries have ever gone with that concept in the end, meaning they could not found a sweet spot between magazine reliability, durability and small cost that permits single use throwaway item.

  2. Yes, the marketing term “constant recoil” IMPLIES benefits only in full-auto.

    BUT physics tells us that letting spring(s) rather than the rear of the receiver (thus your shoulder) dissipate recoil is beneficial regardless.

    This is not mere classroom speculation but an effect Ian has noted himself in at least one semi (Kel-Tec RDB, which I bought after that review and can vouch for myself – Thanks!), but keeps saying otherwise for some reason.

    Try bolting some blocks to some solid, unyielding structural component of a truck frame (or just use short or under-strength springs), such that the axles will bottom halfway through suspension travel – not for a three-to-five-bump burst, not for a “mag dump” of 100 bumps, but just one good solid bump – and then tell me that spring dissipation of impact is only useful in full auto.

    • Granted that “constant recoil” dissipation system requires extra-long receiver, there is some inconvenience to it. If for instance a hydraulic buffer (with no action overtravel) or even low-tech elastomeric bumper is used, the effect may be similar. And as Storm said, this system is likely limited to small caliber, magazine fed weapon.

  3. Ian, recoil is not caused by the the bolt hitting the back of the receiver – in fact it only in threory, and is only a tiny small part of it if the bolt does hit the receiver, and not of any observeable or measurable quantum. In fact the action of the bolt against the spring is a counter-force to the recoil of every bullet fired.

    Recoil is the return force on the whole rifle of the impulse caused by the departing bullet. Impulse is momentum over time – meaning that the change in momentum of the departing bullet over the time (about 1 millisecond) of the bullet in the barrel.

    Recoil impulse is a force, and Force = Mass x Acceleration (so many metres per second per second of bullet acelleration from zero to say 2,700 ft/sec when the bullet leaves the muzzle) – and there again you have your “change in momentum over time”.

    • There’s no question as to what causes recoil. It’s just a matter of whether that recoil is dissipated via spring(s), or allowed to hammer rigid structural components of the weapon.

      There’s also nothing theoretical or immeasurable about whether or not a given gun’s recoiling mass strikes the receiver.

      -Some don’t (constant recoil).

      -Some do and have elastomer buffers (consumables that show visible wear and must be replaced at intervals).

      -Some do without buffers (or have worn-out buffers) and show frame wear or even damage as time goes by.

      -Some do or do not depending on the power and recoil of various ammunition.

  4. Since this gun being of a gas operated locked breech firearm, the impulse at instant of the initial discharge can not be cushioned by anything but the mass of the whole unit. The backward impact to the receiver should be made by the bolt with connected parts and they begin their rearward journey with initial speed of whole guns gained momentum through the locked mode and the plus speed caused by the excess gas pressure within the bore. Since the higly big amount of recoil impulse being collected by the whole gun mass within the initial locked breech mode, the cushionable part caused through simple blowback mode from unlocking stage to the backward stop should not be too much… lMHO…

    • That’s a good point, but the shooter’s (elastic) body does not absorb recoil momentum all at once, and the rearward motion of the recoiling mass is able to dissipate some of it gradually.

      I can’t cite any experiments with accelerometers as hard proof, but my own experience and the general consensus both indicate lower recoil for (properly designed) semiautos vs. manual actions of comparable weight. I would liken it to the recoilless launchers that rely on launching a countermass rather than simple backblast. Firing still creates a recoil, but something (in the semiauto, a recoiling mass whose momentum is naturally proportional to that of the ejecta) creates an offsetting “recoil” and then (hopefully) comes to rest against an attenuating spring.

    • Its not only springs, its also open bolt and jolting the gun forward when firing, thats all described in Sullivans patent.
      Closed bolt Ultimax would not be so controllable in auto.

      • I’ve heard that a lot, but it doesn’t really add up.

        An open bolt machine gun only differs from a closed-bolt on the first shot (with the possible exception of advanced primer ignition blowbacks).

        In a locked-breech automatic in cyclic, every shot after the first is essentially “open bolt” – ignited a predetermined, minimal safe interval after locking.

        • There is no API in blowback smg, API is in Becker, Oerlikon cannon etc. I’ve said it before, we need a new name.

          Unfortunately “closed bolt, open being the same after first shot” is not true, as in open bolt you have a fixed pin mounted on the bolt carrier that fires the very instant it lockes.
          In closed bolt firing is some time later as you have first lockup then a disconnector tripping a hammer,
          and it seems that in practice there is a noticable delay that negates jolt forward beneficial to recoil reduction. Thats why I said that german open bolt Mkb42 is actually even more controllable in full auto long burst than later MP 44.

          As patent vividly described “if you have unknown outside force that pushed the gun forward in the moment of firing, there would not be any recoil”

          • Never stated “smg” [sic].

            Different open bolt guns mount their firing pins differently. My M11/9, for instance, does not even contain a bolt carrier. Some open-bolt guns, with or without carriers, have spring-loaded or lever-actuated firing pins.

            Every automatic firearm, open- or closed-bolt, is timed to have the firing pin strike the primer after a (very short) delay. The fact that open-bolt MGs as a class are not noted for dramatically or even noticeably higher rates of fire than closed-bolt guns would seem to negate the statement that the delay is categorically shorter in these guns. It would seem intuitive from the number and inertia of FCG parts in closed-bolt (especially hammer-fired) autos, but also keep in mind that initiation may begin before full battery, thus still achieving near-simultaneity.

          • I think we got here mixed up with terms and operating systems which are similar but not the same.
            Its understandable and known that open bolt blowback smg have fixed firing pin, yet in locked breech action firing pin is mounted usually on the bolt carrier, as it would not be beneficial if gun fires before it is fully locked, these two systems are similar but not the same.
            Lever actuated open bolt firing pins like in thompson still fire faster than disconnector tied hammer or striker in closed bolt auto.

            There is ofc no effect on rate of fire in this locked breech open-closed firing method, as rate of fire is tied to bolt weight and travel, than firing method.
            In practice its proven fact of greater controllability for the shooter when you have parts that slam forward and fire instantly, compared to closed bolt slam forward and delay that distrupts in some way the shooters aim.

            Think of it simply as a complete reverse of open bolt single shot inaccuracy effect.

            However, there is a connection to the rate of fire in the end, as I believe, higher rate of fire (like in Mp5) with its short bursts negates by some or even great degree that closed bolt auto instability (but ONLY in short bursts).

          • This is a reply to your most recent response to me, but for some reason that post has no reply button.

            I still contend that – if there were some significant delay in closed-bolt firing after locking – there would be a corresponding, measurable, across-the-board tendency of closed-bolt weapons to fire at a slower rate. Time is the ultimate, unforgiving zero-sum game, and each shot is only tenths of seconds or less.

            I don’t see the “proven fact of greater controllability” in practice. There are other factors in play, obviously, like cartridge power / recoil impulse vs. weapon weight. Nevertheless, closed-bolt guns like the M16 and M2 Carbine are notably more controllable than the (both weaker and heavier) open-bolt Thompson. I have consistently seen the MP5 described as MORE controllable than slamfire SMGs, and never limited to short bursts. What other contrary examples would you cite, excluding cases like the Ultimax with longer bolt travel?

          • You are not listening or not (wanting to) understand – there is not a significant delay in firing that influences the rate of fire,its negligible, but still it does have an influence of mitigating the percieved recoil, from gun jumping forward which in open bolt subsequent auto shot, is used for a controllability benefit.

            Study this video and compare the groups in mag dump, especially pay attention to him saying about Uzi “up and down motion” – you do not get that in closed bolt mp5, it only climbs up :


            Also, invest time and effort in studying Colts M16 open bolt lmg prototype version (there on exist video) and how it is more controllable and precise in auto compared to regular service rifle.

            Lastly, there is a case of shotguns, Usas 12 and AA12, which are of same gauge, similar mass and dimensions, same bolt locking system, and similar rate of fire, yet it is known that AA12 is more controllable in auto due to being open bolt, for the same principles I’ve disclosed earlier.

          • Well, I saw a moderately better (~10%?) FA grouping with the Uzi. I also heard him explain that his controllability challenges with the MP5 were due to its higher rate of fire (Ian has commented numerous times that either <600RPM, or very fast, is ideal; and that 800 or so is choppy) and his inability to stay on target through a too-small aperture sight. It also sounds like he's more experienced with the Uzi.

            I also see (both) SMGs bouncing forward a bit as the bolt closes – the Uzi more so, sure – because its bolt is heavier.

            The Uzi also, with its telescoping bolt, has very little bolt behind the chamber and generous free-recoil distance compared to either the M11 or MP5.

            If there were really some significant disadvantage imposed by the timing of the MP5 FCG, all the relevant variables (auto-sear and trip-surface placement, spring constants, angles, etc.) are within the control of the fine engineers at H&K. The only true, inescapable limiting factor is to prevent firing before it's in battery – which would apply equally if it were open-bolt.

    • What you describe in first sentence is being overlooked and it is substantial part of felt recoil force. In fact this is typically lot greater in terms of transferred energy than the “tap-out” of action into receiver at the travel’s end.

      What I think is in play here is prevention of stacking of both inputs, which allows shooter’s body to collect first (bigger) input and recover without much of disturbance to shooter’s composure.

  5. Besides being late to the game, the issue could be that in society in general, and especially in military procurement, there is this idea that one has to quantify a bunch of data points then rank them to come up with the “scientific” best answer.

    In firearms that was certainly the case in past decades even in the civilian market. It seems when a gun magazine tested a gun, the whole point was to test the velocity, count the capacity, and assess the bench-rest accuracy. Then presumably one would buy this gun over that gun because of 10 more FPS velocity, or 0.1″ better accuracy from the bench. Basically, we were supposed to buy whatever came out on top of the margin of error that particular day. Meanwhile, some old timer with a short, handy, highly-ergonomic 30-30 lever action was hitting his bag limit year after year.

    Jeff Cooper noticed that and started talking and writing about how desirable it is for guns to be easy to carry, ergonomic to use, fast to get on target, and that being able to quickly hit a pie-plate at a hundred yards from off-hand was actually pretty good practical accuracy for nearly all shooters. I do not know if he started the trend or just documented it as is happened, and it does not really matter. The point was that even in civilian circles they had fallen in to the trap of latching onto whatever was quantifiable then ranking everything accordingly and ending up with guns that were not necessarily user-friendly or all that appropriate for the task. Constant recoil is a great thing, but how to quantify it?

  6. Looks like I had better watch my step out here in the woods of Washington state, we’re supposed to have Sasquatch here.

  7. Seeing is believing. So at this point all other LMGs are pretty well – useless. Or disadvantaged, to say at least. Just wonder, why is it that U.S. Marines did not buy it yet.

    • “(…)all other LMGs are pretty well – useless. Or disadvantaged, to say at least.(…)”
      I would not be so sure. Firstly, if you want to use LMG as crew-served weapon, rather than operated by single, bottom mounted magazine is inconvenient, secondly there were some doubts regarding 5,56×45 NATO cartridge reach (effective range), which is more important in LMG which are often to fire at greater distance, finally due to way LMG are deployed (with emphasis on volume of fire) minimizing of spread might give smaller advantage than in case of rifles.

      “(…)U.S. Marines did not buy it yet.(…)”
      Well, current development of plastic-cased cartridges might eventually lead to adoption of cartridge of greater performance.

  8. The 5.56 cartridge has a low impulse force and recoil is easily handled by most people. Having fired the MINIMI on numerous occasions, the recoil, due to the weight of the weapon was very controllable. Whether constant recoil control on a 5.56 machine gun is of any benefit is debatable. On a .308 machine gun it may be a different matter.

    • I think that MINIMI being heavier empty, than Ultimax loaded with 100 rounds is extremely debatable of benefits in terms of a regular soldier who wil lug it around all day, unlike the people who shot one of them, or both only at the range.

    • This is something I expected to read in this connection; albeit it came little late – still good. As far as I remember we had discussion on Russel’s guns in past. It gets more fun as we go on with it 🙂

  9. Given its advantages, I am surprised that the ‘lesser’ (no insult intended) countries and specialists e.g. elite forces etc. users haven’t bought some (I would have thought it ideal for less well trained troops such as Kurds, Ukrainians etc. and usefully sufficiently uncommon as to help control its supply / use) – is it too expensive or the supply in some other way restricted?
    p.s. it nice to see the increasing rapprochement between you & the sachwach & his more feminine side – being a great fan of both and you.

    • There is all kinds of small arms and support weapons on scene in Middle East – including some rarities. I saw Chinese grenade launchers and I am pretty sure Ultimax is there too. Probably on side of so called rebels.

      When comes to “training” (silly buzzword) you can trust that Kurds in particular are top notch fighters. Americans left the scene not to get into Turkish way, but Kurds stayed and will fight – you shall see soon.

      • Agreed,
        Neither ISIS nor the Syrian army were keen about getting into a fight with the Kurdish forces.

        The Turkish army may see itself as the defender of Kemal Attaturk’s legacy of a modern, secular Turkey (the opposite of what Erdoghan stands for).

        But it will be interesting to see how many Moslem Turkish soldiers will risk getting killed by Kurdish girls

        Got killed by a girl?
        Don’t go to paradise
        Don’t collect your 72 virgins

    • Kurds (I bet 90% of US citizens never even knew about that unfortunately unrecognized nation till few months, or years ago) do not have official state, so I dont see how they could officially buy Ultimax, other than that US buys it directly from Singapore and delivers to them (which Sinagapore probably would not want to), as there are not many, if any, of Ultimaxes “floating” on worlds black market.
      Highly doubt any of the Ultimax are on rebel side right now, there were none on any of the photos in last years of the Syrian conflict.
      Basicly, (regular everyday news from Syria here, stopped few years ago-talking about media credibility), from what it seems, rebels are done for, and the destroyed state will be split between Al-Assad and Kurds – unless Al Assad strikes some devilish deal directly with Turks and f***s the Kurds (with US watching from side, just like in 1991. Shia Iraqi uprising that they instigated and provoked).

      As for the Ukraine, being ex soviet republic long time tied to Russia, they do not have officially 5.56mm caliber in their stock.

  10. Rate of full auto fire,

    Without using rate reducing mechanisms,

    is more related to the weight of the moving parts, the preload and rate of the return springs, and the distance that the parts can move

    Than it is related to barrel length.

    Basically, heavy reciprocating parts, soft springs and long travel, all favour and low rate of fire.

    You can play tunes with clever buffering devices, for example the additional weight and stiffer spring, that go into the receiver tube of a sterling smg, behind the bolt and recoil spring.

    The weight can have several elastic impacts and cycles against the bolt, on one single cycle of the bolt.

    It is able to buffer the bolt, and speed the cycle, without anything whacking into the end cap of that long, skinny, receiver tube.

  11. Felt recoil;

    You are never going to escape from newton’s laws of motion.

    Muzzle brakes can be very effective at reducing recoil, but come at the cost of deafening muzzle blast directed at the firer and friends, and a big cloud of dust, mud and vegetation disturbance.

    The other tricks spread the recoil impulse over a longer period of time.

    And designing guns that have the impulse located closer horizontally and vertically to the centre of the firer’s shoulder, to minimise climb and disturbance.

    The ww2 era solothurn 20mm anti tank cannon fired from a closed breech, but the action moved forward on the chassis of the gun before it fired

    In that way it emulated the api Becker and oerlikon cannons, in delivering part of the recoil impulse before it fired. Spreading recoil over a longer period of time.

    • Yes, with a (not so small) difference that in Becker and Oerlikon api was primarily a method of operation, not a method of mitigation recoil.

      • It was indeed primarily a method of operation,
        But the reduced peak recoil stress imposed by that system of operation, made the guns that used it very attractive for some roles, for example aircraft cannon.
        I’ll also add in the Swiss Furrer designed aircraft cannons, which used a toggle locked breech, and which also fired during their forward overtravel, once the toggle was locked.


        I’m slightly perturbed by the factual inaccuracy in the video.

        Sullivan was probably responsible for the more recent interest in constant recoil, but, as the guys have already noted, Robinson did a lot of work on constant recoil, including smgs.

  12. One issue that you didn’t mention is the effect of the constant recoil mechanism on group sizes. Since it doesn’t jump as much group sizes should be smaller, but you didn’t address that.

Leave a Reply

Your email address will not be published.