Rossi: About 95% of the Way to Sigma 5

Thanks to Jean Yves Metivier for a pointed question on the Journal of Nuclear Physics today. Andrea Rossi has been regularly saying in response to requests for updates something like “we’re on the way to Sigma 5”. Jean Yves was not satisfied with that reply and wrote this:

Jean Yves Metivier
October 5, 2017 at 7:10 AM
Dr Andrea Rossi,
Your sigma 5 quest is interesting, but your answer to JPR is pointless.
What is your final goal, is it related to how much cycles you do, how many days without failures, or how many Watts without interruptions?
How far are you from the goal in percentage?
The answers to these questions will give the Readers a feel of your progress.
Keep it up!

Rossi replied:

Andrea Rossi
October 5, 2017 at 1:25 PM
Jean Yves Motivier:
The final goal is related to how many hours without failures.
We are very close to the target, I’d say we are around 95%.
Thank you for your attention to our work,
Warm Regards,

So there’s a little bit more detail regarding the whole Sigma 5 thing. I’m not sure exactly when Rossi started this Sigma 5 campaign, but I think it has been close to a year. If a reactor running reliably for that long with no refueling at 2000+ degrees C with tiny amounts of electrical input can be verified, then I guess any partner he would be very impressed. I feel pretty confident now that the November demo is going to happen, and so I hope we get a much clearer picture about what the E-Cat QX is capable of quite soon.

  • Jag Kaurah

    Great, we have some clarity.

    I am in R&D and fully empathize with Rossi and feel for him for all the delays he has faced. R&D is not a preset path and one never knows the twists and turns it takes or how long it is.

    • Rene

      But it wasn’t just R&D. His process was fake commercialization, fake market territories, fake products. We now know and proven the products were fake, though we have yet to know if his effect is real, and if so how reliable. It appears it is, but it is not independently verified. His Now November meeting will be under NDA, so again we’ll not really know.

      So, yes, he did R&D on the back of investors who were told any day there would be products to sell. That was 2011-2013. People were asked to sign up for domestic e-cat nonbinding purchase intent, which basically was building potential customer list to entice investors.

      He said he paid back his previous investors, and not seeing any howls of protest, it suggests he was good on his word.

      Then he made (by his word) several improvements and he found he could never sell a domestic unit without disclosing his IP. In actuality, by his words, he was having problems with stability. So what did he do? He made bigger ones, and develop huge units for large scale commercial heating. He claimed he had the reaction under control. He marketed a 1MW shipping container sized unit. Claimed to have sold it to the military and other undisclosed partners. There were runaway reaction problems.

      He realized the market for low temperature heat was limited. He wanted to get into electrical production which, as he rightly pointed out, needs high temps to generate steam efficiently. Herein was the era of the hot-cat with cool photos of red to nearly white hot cores. There were measurement issues brought on by the use of photo-determination of temperature and derived heat. The hot cats also it didn’t last long, lots of material failures. That was probably one of his more R&D parts of the journey.

      Then came more partners, IH, a performance demo of the warm cats a demo of actual use of e-cat heat, a fake customer, dubious claims on performance, a lawsuit, a shyster partner, san undisclosed settlement, and during all that mess more R&D on a shiny new kind of e-cat: the quark.

      So, quark, blurry pictures, lots of popping units (was R&D after all), the quest for low MTBF, striving for sigma5, a new partner (undisclosed), more claims of robots and factories, Yet another performance meeting, more NDA cloaking.

      This is not just R&D, it’s really bad attempts at commercializing an effect that needs a lot of theoretical and applied engineering work. There’s your clarity.

      • Jag Kaurah

        Thanks Rene,

        Well stated. It is not so much about his personality, more about a useful product.

        Underlying all that is my gut feel that the effect is real, which you agree with, it is unstable and individual units tend to occasionally come on and go off again. It seems that the units that go off seem to come back on later.

        So Rossi has personal highs and lows which he reflects in his actions.

        My gut feel is that by making his units small, bundling a lot together and allowing for some to go off and come on, he gets an more stable average output.

        That is the reason for my more positive feel that we will see a usable product soon.

      • Peter Wolstenholme

        Rene – Howl of protest, or of anguish follows.
        No, he has not repaid all the investors. He sold licences, with a three year expiry clause if sales were below a certain limit, and then refused to deliver anything to sell. Some were repaid, others not.

  • Rene

    The problem with 95% (0.05) and what that means about the time to achieve 99.999+:

    • DrD

      That’s very true but my interpretation of Andreas answers are that they are rarely precise i.e. not to be taken too literaly. He probably meant something like “I’ve been working on it for 300(?) days and it looks like I’ll get there in round about another 15 (5%) days.
      Atleast we now know (possibly) that his measure is “mean time to failure” (MTTF). I had rather suspected he was measuring something parametric, like core temperature.
      Also, the apparent lack of progress (from 4.? to 5) meant he thought that simply collecting more data improves sigma which of course it won’t. He needs to make changes to improve sigma.
      I’d like to see a chart of sigma against time but that’s asking a lot.

  • Ophelia Rump

    Is he talking about an current test in process, or the entire development process?

  • MH

    Consider you have a system containing thousands of quarks then to calculate the system reliability you have to multiply each quarks reliability to get the system reliability. It could be the case to achieve 5 sigma. We dont understand the consequence if one quark fails.

    • Ophelia Rump

      We certainly do, the core stops producing energy. If it takes out a string of other quarks like cheap Christmas tree lights, the circuit is designed incorrectly.

  • Anon2012_2014

    What is Sigma 5 to Rossi? Is this some kind of proof the device works.

    • Stephen Taylor

      Five sigma or five standard deviations is a probability of one in 3.5 million. For Rossi, it seems to be the tolerable failure rate he claims to be striving for.

    • Buck

      My best guess is that Rossi is looking to control COP, and the two factors defining COP: energy in (electricity) and energy out (heat at this point). Further, I suggest that a customer is more interest in temperature control.

      My evidence of this is the information coming out of the 1-year pilot plant test. The chart below shows how the standard deviation, a measure of variability, of the COP is tightened as the year progressed. Major improvements show after day 175 in the chart. You can see the tightening after each of 3 major changes after day 175.

      Here is additional background:

      • DrD

        That’s pretty much what I expected but his reply says it’s not so.

        • Buck

          “The final goal is related to how many hours without failures.”

          Rossi did not define his term “failures”. I am guessing that he has improved the physical makeup of the E-Cat QX (the materials used, the type of connections made, etc.) from many years of experience. So I exclude a physical failure. In my simplified thought process, that leaves his control software and its ability to maintain a target COP.

  • Axil Axil

    From first hand reports, the quark is a small tube that is transparent. Light comes out of that tube and is very intense. That plasma containment tube holds a plasma at 2700 °C. The material that that plasma containment tube is made of must hold together for 1 year without leaking. This containment tube material is where a failure of the quark will come from.

    My guess is that this tube material is Boron nitride. This isolator begins to sublimate at 2,973 °C. My guess is that Rossi has figured at a way to keep the plasma temperature lower than 2700 °C so that the plasma is not so hard on the tube material.

    It also come to mind that the COP measurement was calculated with the assumption that the light coming from the quark was produced by blackbody radiation. This light from the Quake may not be produced by black body radiation.

    The light may be coming from a polariton bose condensation

    Marrying superconductors, lasers, and Bose-Einstein condensates
    June 17, 2016 by Sheri Ledbetter

    also see

    High-energy side-peak emission of exciton-polariton condensates in high density regime

    • Rene

      This will be the second time in e-cat history that photo-temp measurement and derivation of energy produced will have been done erroneously. And without independent analysis of the measurement techniques and independent total heat measurements, we’ll again not be able to verify claims.

      • AdrianAshfield

        Don’t jump to conclusions. The hints so far suggest the Quark X tube is surrounded by oil that then transmits the heat to water. Heat output being measured by the water temperature increase.

      • Axil Axil

        Suppose you has two sources of light, an incandescent light bulb and an LED. And you wanted to measure the temperature of these two sources using a spectrometer. If you assume black body radiation in measuring the temperature of the incandescent light filament, you will get the right answer, but if you make the same assumption about the light coming from the LED, then you would be off by thousands of degrees.

        • Obvious

          The bremmstrahlung light emitted by electrons in a plasma almost never reaches blackbody levels. Bremmstrahlung is responsible for the broadband light output. Light emitted by electrons entering and leaving orbitals from excited ions in a plasma often does reach blackbody levels, but only in the respective characteristic wavelength bands. Plasmas are often semi-transparent in the bremmstrahlung range, and reflective in the characteristic emission bands, so using a blackbody to characterize output is foolhardy unless the spectra has actually been appropriately measured to determine that it does approach a blackbody, and the optical thickness of the plasma is sufficient to prevent transparency.

  • Anon2012_2014

    Is that what he means, or is it quality control of his robot factory. He supposedly is making Quark-X devices that have overall COP >3, i.e. 100 watts in => > 300 watts out. That’s useful, so what does he need to observe phenomenon for if he is making so much obviously useful energy.

    Rossi is the Enigma Machine.

  • Wholewitt

    I think if his goal for 5 sigma is about failures he should use MTBF (mean time between failure) which is usually rated in hours ( like 1 million hrs.) Sigma is normally a process variation rating in industry.

  • sam

    October 6, 2017 at 3:56 AM
    Andrea Rossi
    October 6, 2017 at 7:29 AM
    On the homestretch.
    Warm Regards,

  • Rossi – Using random scientific sounding words with very own interpretation since 2011 – quality feature.

    • Gerrit

      Sigma 5 of what ? The 2011 ecat, the hot cat, the gas cat, Curie, the mouse and cat, the home depot cat, the 2017 demonstration cat, the 2020 coming now very soon cat, the cat just before the even better cat ?

      • Ophelia Rump

        It is not a relative measure.

  • Just for your information – Disqus service was hacked and account data leaked!

  • DrD

    In fact, it has a much wider use than that. For example, it’s much used as a precise well defined measure of manufacturability. Individual manufacturing processes are commonly expected to meet 3-sigma whilst the component designer commonly targets 6-sigma. The latter might be applied to many parameters related to a single product, typical semiconductors each may have dozens of such parameters. In this example it might mean that a product has a capacitance that is consistent from device to device to within 6-sigma.
    Rossi is doing neither of these, he just claimed he’s monitoring MTTF / MTBF but whats puzzling is exactly how he’s defining it because a high sigma would result from all failures occurring after the same period (infinite in that case) and that period could be very short indeed, like a minute. So clearly it’s not that or he’s kidding himself.

    • DrD

      One way he could do this, at least in theory, is to set a “target”, say 300 days to failure and then monitor a very large number of units to establish a distribution of time to failure plus the MTTF from which he can then calculate sigma. Obviously impossible though.

      • Chapman

        To be honest, the entire Sigma 5 issue has me scratching my head!

        If a QX is expected to operate for 1 year, then the only realistic way to achieve a TRUE Sigma 5 reliability is to MANUFACTURE, and RUN, well in excess of 3.5 million units for more than 1 year. If you have a demonstrable anticipated performance success, with failures lower than 1 in 3.5 million units, well then – there’s your Sigma 5. But anything less is just wordplay and marketing…

        Not even Ford, Maytag or Microsoft (hardware division) can boast a Sigma 5 reliability.

        But I am REALLY enjoying reading your ideas as you try to figure out what he is talking about! Keep at it! I think you are on the right track to pinning down what HE is thinking, even if it is not a real Sigma 5 as the rest of us know it.

  • wizkid

    Rossi explains the “5 sigma thing” to two people on jonp

    Andrea Rossi
    September 18, 2017 at 6:54 AM
    Frank Acland:
    Sigma 5 means high reliability, because the statistics of malfunctions caused by many issues make unlikely important shortcomings. Sigma 5 is a probabilistic formula normally used in science.
    Google “CERN Sigma 5”
    Warm Regards,

    September 13, 2014 at 11:10 AM
    Dr Rossi:
    Can you explain better the “five sigma” thing?

    Andrea Rossi
    September 13, 2014 at 3:04 PM
    The classic example is the case of flipping a coin: you have a 50% chance that it will be face 1 and 50% that it will be face 2. Obviously if you flip 100 times the coin, it is unlikely you get 50 F1 and 50 F2, most likely you will get 45-55 times either one face or the other: this interval 45-55 is in this example “sigma 1”; means that if you find from 45 through 55 times a face flipping 100 times, there is no event at all, because it is normal. If a stretch of 10 is sigma 1, we will have a stretch of 20 = sigma 2, a stretch of 30= sigma 3, a stretch of 40= sigma 4, a stretch of 50= sigma 5; this means that the higher the value of sigma, the higher the possibility that there is something, that you are looking for, that makes a work: in the case of the coin, can be a trick that makes the coin fraudolent, in case of Physics can be a force that breaks the symmetry or something else. If you get a sigma 5 the probability that there is a break of the normality is very close to be certain ( in Physics nothing is certain, everything has a higher or lower grade of probability).
    Warm Regards,

  • wizkid

    What he hasn’t explained is why it is taking more time than he ever expected to achieve 5 sigma. Maybe he keeps running into a brick wall at sigma 4.93333333333333333333 … Or maybe his “partners” keep changing the requirements to keep him on the treadmill. Or maybe he just has trouble with doing the tests correctly. It could be any of these or several other choices.

    • Obvious

      The 5 sigma thing is just a (barely) plausible device for buying some more time, and for separating the most gullible from the more critical audience. It seems to have worked.

      • wizkid

        I have to agree. 😳

  • artefact

    On JONP:

    “Nils Fryklund October 7, 2017 at 8:44 AM
    Dear Andrea!
    Has the invited persons already got the date for the demonstration?
    How many persons will be present at the demonstration?
    Will someone from Sweden be there?
    How will you show the demonstration to your fans?
    Best regards Nils Fryklund

    Andrea Rossi October 7, 2017 at 2:41 PM
    Nils Fryklund:
    1- Not all
    2- 40
    3- Yes
    4- It will be in direct streaming
    Warm Regards A.R.”

    • Toussaint françois

      I hope the MPFM Team will be invited

      • Toussaint françois


        • Buck

          You have repeatedly posted a YouTube link.

          However, it leads to the following message: “this video is unavailable”

          Could you clarify?


          • Toussaint françois

            No it is a bug from the JONP it keeps poping up, this link is 3/4 weeks old and I see that the link does not work anymore, it was a quote about Rossi not doing “open source” by Steven Greer.

          • Omega Z

            Would this be the video, 2hr 35 min

            Dr Steven Greer – NEW Documentary UNACKNOWLEDGED

          • Toussaint françois

            Yes, this is the video !

          • LION

            Hi Omega Z,
            I thought that you particularly would enjoy this from the BBC.


        • Rene

          Last time Bob Greenyer responded to this was his concern about an NDA. MFMP is about open science after all.

      • Axil Axil

        MFMP is a competitor and therefore will not be invited, IMHO.

    • Jerry Soloman

      Frank have you made plans to attend the November Demo in Miami,
      and what about Mats Lewan?

      • Frank Acland

        I am not able to confirm anything at this time, and can’t speak for Mats.

    • LION

      I will be signing up for 4 for sure.

  • Jas

    Toussaint françois
    October 8, 2017 at 1:51 AM
    Dear Andrea Rossi,

    I wish you a great success for your demonstration,I have the feeling that Sigma 5 will be announced before the demo.

    Warm regards,

    Toussaint françois

    Andrea Rossi
    October 8, 2017 at 6:55 AM
    Toussaint Francois:
    Your feeling is correct.
    Warm Regards,

  • Steve D
  • Wholewitt

    To add some sigma clarity (hopefully not confusion): The standard in industry is 6 sigma for process control, which is +/- 3 s. A good example would be the 12V output of a power supply which may be rated at +/- 5%. The target value is 12.00V and the Upper spec is 12.6V while the lower spec is 11.4V, this would be the 6 sigma range. If you use fixed resistor values you might get a variation of +/- 0.52V which is ok if it is centered on 12.00V but not ok if it has a center of 12.30V. So you have two problems, variation and centering. Like some have said, Rossi is probably talking about a temperature spec (which must be measured and controlled). There is variation in the measurement device itself (maybe a thermocouple) and any process depending on variation must have continuous measurement (such as production measurements at final test) or batch sampling.

    Software will just take the reading and try to control the temp, so no tolerance is involved but it may not take action until the reading reaches a trigger point. I don’t see much relationship of sigma to device failure unless some runaway condition occurs close to the spec limits. I think Mr. Rossi either is using flawed language or is being a bit cryptic.

    In the case of a power supply, it can be corrected by an adjustment pot or a select at test resistor with the second way being more long term stable. A problem happens with controlling multiple output voltages in high efficiency supplies using the same transformer but I won’t go deeper into that.
    I still hope for the best with the Quark X, we need to have some game changers.

    • Axil Axil

      Regarding: ” I don’t see much relationship of sigma to device failure unless some runaway condition occurs close to the spec limits”

      Could the out of spec condition be the generation of radiation above some set limit?