Monday, June 24, 2013

Good, Cheap, Fast–Pick two

This is broken down into a low volume startup company, and an even lower volume at home company.

Breakdown is that both end up with products around $750. More after the break

The Good – Fast option


Above is my current calculations on how much a powermeter needs to cost in order to run a company on this as a product. Some of the numbers are rough, but if you look through you see that total cost for parts is around 210 dollars, however each requires a minimum of 5 hours of labour. This is based on a low volume production method. Initially it'll be closer to 20 hours per unit by my estimate but could easily be brought down to 4 – 5 in batch production methods. I could see it might even drop to < 2 hours in larger batches. Unless I needed to churn out 20 – 30 a day this isn’t possible.

What is your time worth? What is my time worth? That’s the big question. Let’s look at the basics. A fresh out of University undergrad engineering graduate essentially expects 60k/year. Without health, dental, benefits, etc this works out to approximately $32/hr once vacation is accounted for. This totals to about 200 dollars of labour per crank. Lets add that on but lets examine skilled labour as an alternative.

If the engineer was replaced with skilled labour it could be lowered – however a high level of accountability and skill would be required. Based on my experiences this will end up being $24/hr. It may actually be lower, but they will require training, supervision, and…. ergonomic breaks. All of this is tedious work. Strain gauge installation can be very difficult and not something I could do every single day. A bad day can cost hundreds of dollars in gauges, but the real cost is lost time. If it takes 1 hour to install a gauge, it takes an additional 30 mins to remove a bad gauge, and then another hour to get it installed again. I always factor in 1 gauge in 2 to be a bad install, but in reality my ratio is 1/10 or better.


Back to the skilled labour, we arrive that there is no management or facility. Based on some preliminary rental research and rough estimates, approximately $1250/month minimum is needed to have even a basic facility. Based on the expected time frame to make each unit, this works out to 480 powermeters per year. An additional $31.25 is now tagged on per unit.


The next big expense is going to be all the testing required, this totals out to be about 25k. I’m going to amortize that over half a years maximum production capability. Realistically I’d guess that it would take a year to two years to hit that total demand. Though, that is a guess. This brings the price per unit to 533 for engineer assembly, 485 for skilled.

Back to management / R&D / Support / Etc this works out to another engineer employed, but this time it’s full time labour divided over the shipped units. This results in an additional $125/powermeter. I firmly believe that there is a need to continuously improve, design, expand, and debug as well as being there for consumers. A lot of company’s get flack for that last issue. I want to make sure that the time for me to reply to a problem is less than 24 hours at a maximum. It can also be amortized over a higher volume as I’d likely be the one in this role.

What does that work out to? Well, $741/$694 Eng/Skl.

I’m sure that this could elicit reactions such as “you just showed it could be cheaper, your estimates are bad”, or the other end of the spectrum I could get responses of “wow that’s cheap, you just proved everyone who makes these are greedy!”.

I’ve intentionally left out a lot of things. I’m also not to this point of operation and I don’t know when or if I will. I’ll be making prototypes at home, and later maybe in a rented garage or something similar. I don’t know if 500 units/year is a reasonable number or if it is too high or low. If I was to run a small facility with several employees the cost per unit skyrockets.

Cost per meter, $700 – $725.

The Good – Cheap option


Lets assume low volume, second “job” production. You’ll notice costs have gone up for items such as cranks, strain gauges, ancillaries, etc all due to reduced volumes. Electronic components and PCB haven’t changed significantly but what has is the labour. The labour goes up significantly per unit. To compensate, and with the assumption of myself as the only real employee, the labour rate goes down (I’ve halved it in fact). In order to sell production units (though development products are exempt) I need FCC and ANT+ approvals. Assuming free time construction, I could probably pull off 50 units in theory over 6 months if I needed to. Likely more, but it’s hard to say.

Cost per meter, $725.

In conclusion

What does this all mean. It means that moving from one type of production to another type means more overhead, but more efficient production. That being said it looks like I’ve hit a pricing number. Rounding up, approximately $550 for developer units, and $750 for production units. Without high volume manufacturing techniques that seems like the peak. Now, if I were to include “profit” it’d have to go higher. Some more research has to go into these numbers, but it’s a start.


  1. Hi Keith,

    I stumbled on your blog a couple of weeks ago and have been following it closely since. I like what your are doing.
    In terms of "profit" - remember that some of that needs to be taken in to keep you going through the rough patches - when you need to replace defective units, or deal with a difficult customer who is being vocal...
    Getting through that will either take your own money or a bit of profit on each to cover the potential issues. IF everything goes according to plan you have a bit of extra money to show for your good planning and execution. If not then the profit will mean your not totally sunk by some small misfortune.


    1. Mike,

      That is an excellent point and I'm glad you commented about it. It's hard to ever know what to expect so some buffer is certainly needed to be built in.

      I'm sure there are other things I haven't thought of too so I'll likely have to keep updating this as I go.

      Thanks again

    2. Keith,

      In reviewing your 2 scenarios I think you've missed a 3rd option - sub-contracting.
      You are effectively doing this for the circuit boards if you have them assembled for you, why not sub-contract the strain gauge installation? I realize this is a greatly simplified suggestion as the number of companies that can / will install SG's correctly is probably small - but it may be a viable option to get product out there without needing to rent a facility / hire people.
      Plus it frees up your time to work on more important things like sales and R&D.

    3. You are correct. This is actually more by design than anything. The short is it's possible to outsource the board but I need a lot of orders (at least 200 I estimated) to make it viable. I'm actually working now on getting my BOM and boards priced for automated manufactured.

      Strain gauge outsourcing actually increases cost. I've professionally worked with industry on this. The local charge out rates are 100 - 140 dollars CAD/hour with 1/hour per gauge estimates for easy installs and 2 - 3 for harder. That means a minimum of 400 dollars for cyanoacrylate installs + shipping them the cranks. In epoxy transducer rated installs this number increases due to more equipment, time, setup, etc. The only option is China then, but I need to have the financial muscle to source, test, fly over multiple times to check quality, etc. Again, more volume than I'd expect.

  2. This analysis is one of the most blunt-honest that I've read about start up costs. The membership fees in order to certify the equipment seem to me like the most unfortunate aspect of the whole spreadsheet. Governing bodies should have a means for companies to certify much cheaper when they are in a start up phase. If those fees were cut in half for the first two years you could sell at $750 and operate with more profit in the early years when it is more crucial for success.

    My gut says that you don't have enough profit margin at $750.

    Out of curiosity, do you think there is a market for a hybrid kit/DIY package? Are there enough people who could assemble and test this themselves (and who would pay for a kit) to run a business? If people built up there own unit for personal use the professional membership fees would vanish. I'm just not sure if there are enough people who can build these systems reliably.

    1. Thanks Duane, this comment means a lot to me. I debated hard about how much I should reveal going on in my head in terms of pricing.

      I've been debating a "developer kit". I'm not sure it's feasible to let people install strain gauges themselves. It'll end up with a lot of angry / sad people. However, that doesn't mean a complete powermeter couldn't be built sans the FCC and ANT+ approvals with source code available. The only thing I couldn't include would be the ANT+ network key, but that is free to get by signing up as an ANT+ adopter - it's generally closed source. However, this will be beyond a lot of people's skill set to compile and upload the code limiting the potential market to hardware hackers. It could still be a starting point. Early adopters are usually willing to jump through some hoops.

      The Pioneer Dura-Ace System was originally suppose to be installed by LBS on a crank. I have my doubts that it'll see the light of day in this form. It's too easy to mess up. I suspect the same might be said about mine.

    2. You are welcome. Just so you know, I meant the honesty more in terms of being honest with yourself than necessarily being transparent toward the marketplace. The cardinal sin of an entrepreneur is to overstate the market appetite for a product and to understate the operational costs to deliver. I'm really impressed at how clearly you have identified the costs for this, I just wish the numbers footed a bit better.

      You confirm my suspicious that there is a very small DIY market for this type of thing.

  3. One thing that isn't mentioned her is tax. Remember to think about things like shipping & tax, as this can add easily 10-15% onto the price of the unit (which makes a $700 power meter into an $800 one).

    This also doesn't include business setup overhead costs. Insurance, business registration, domain registration, website, etc.

    I think that China may not be as hard as you think. There are factories that are willing to work in smaller quantities, and they are willing to send your finished samples of units to your door directly. This allows you to test the quality without needing to travel. But as you say, this would not be what you do on your first production run of say 100-200 units (or first 6 months). After you have the documentation systems and procedures in place, China would definitely be a place to start looking into.

    China's labor rate can be as low as $10 an hour. This could drop the subtotal price by about $100 dollars per unit.

    The other one to keep in mind is that after the first 12 months the FCC testing money will be profit an increased profit margin, or alternatively an increased R&D budget, or money that you can spend on customer support.

    Good cost analysis though. I am glad you are thinking about these things.

    With regards to the ANT+ key and needing to compile the code, it may be possible as part of the kit to sell for early adopters to instead be provided with a TTL to USB converter (can be sourced for $2 USD) and they enter their ANT+ key by using a serial port on there PC that is then saved into the flash memory of the micro-controller.

    This may reduce the cost of the first 100/200 units and thus reduce the risk.

    1. I think I'll address some of this in a post coming this week. Things have been slow waiting on PCB's and I won't have much to say until I try and build them up, so commenting on a few of the things that you and others have mentioned in the comments seems like a valid to-do.

      The costs that are currently behind me should be accounted for like you mention, but I haven't as it's always been under the guise of a "hobby". I should fix this.

      Insurance is a subject that I sigh before touching on. I haven't looked into the specifics and cost, but I have a good idea of what I need and I am related to someone who has just about every insurance course under the sun so I'm covered in terms of advice / consulting at least.

      As for China... All I have is anecdotal evidence of others experience. Now that Kickstarter is sorting out the legal for Canada I am seriously looking at this for October time frame assuming beta's get back on schedule. I poorly estimated on the long shipping time from OSHpark. I'll discuss this as well in the forthcoming post after doing some more research.

  4. Keith:

    I just stumbled on this blog entering it via the Jan 16 2013 entry, which, in summary seems to explain in detail for the interested layman the physics and engineering of power meters, and suggests how the very interested layman can build his own, which seem really intimidating, but you have broken it down into easy bites.

    Next, I went to your last post June 30 and am amazed that you have got gone further from straight engineering and have generated some economic modelling and have ordered some circuit boards. I flipped through the posts in between, and I am trying to sort it out. As I understand it your position has morphed along the slippery slope (not a criticism, just trying to understand your process) from teaching the lay person that he can do it entirely himself if he has the tenacity, to acknowledging that the average joe really does not have the expertise to do some high tech steps well and economically, and now you are proposing to offer small lot manufacturing of some parts in your home setup, which will be a lean establishment, all construction and no promo, with sales dependant on those coming in through this blog and word of mouth. Do I have it in a nutshell?

    Are the circuit boards you have ordered destined for customers that have pre-ordered in a process that I have missed, or are these all on speculation for your own R&D, or build-a-few-prototypes-and-some-customers-might-show-up kind of thinking?

    I echo the questions and thoughts the other posters have raised; I won't add to them as I haven't thought sufficiently deeply about the situation. I certainly commend you for the effort you have put in and the clarity of your exposition -- it truly has the hallmarks of a labour of love -- although, as I have alluded, I am a little confused as to where you are going with your economic and sales model.

    Actually, I do have a question -- with truly separate Right and Left power measurement your PM would be in the extreme minority of those on the market, even before the availability if ever of the Garmin Vector and Brim Brothers PMs. While some question the utility of the L /R separation, if there is any merit to this metric, the interest in the minority of "experts" who claim it is useful seems to be on the shape of the individual power/torque curves of L and R foot strokes as they rotate through the 360 degree (or 2 radian) pedal cycle. Do you propose to have a sampling frequency high enough to be able to generate a L and R cyclic or polar power curve of reasonable precision?

    1. This is all a slippery slope. Right now it's still teaching things, but it's less about the nitty gritty. I've hammered out a lot of the technical issues over the months and there seems to be less interest in actual hacking and more in the question "why are power meters so expensive?" and "Can they be built more cost effectively". This post on cost analysis has gained more views in a few days than some of my other major posts in months.

      So all this is telling me that maybe their is a market, at least in some capacity, for a more "entry" level but "no compromise" power meter. Frankly I spend days wrestling with if I should even make a detailed post about force and thermal sensitivity on various designs. That research would cost a company days and thousands of dollars of time.

      I'd like to say that my project is like an Arduino. You tend to buy an Arduino and program and develop with it. While you can go build one yourself it requires a huge skill set and it's expensive to develop that skill set. The same can be said about strain gauging, however someone out there will want to, but most people will not.

      I want this to be clear -- there are no customers. Beta testers will receive a powermeter on loan only. The hardware (crank with pre-installed strain gauges and circuit board, debuggers, etc) will all remain my property until some time I deem them as acceptable development kits or a product. All successes and failures will be talked about openly. Should the beta test prove successful I'll look at dev kits, kickstarter, pre-orders, etc. I have a plan, it just isn't online right now. (I've been pressed to put up this plan, maybe in 2 - 3 weeks)

      Thing of this as following the insides (in detail) of a potential product, if the product fails all the details are there to help educate others.

      That last question is a good one. I've been asked about this feature. The ANT+ protocol now includes a metric I discussed previously called Torque Efficetivness and Pedal Smoothness. These are simple versions of a polar torque plot. It's an easy endeavour to provide this capability and I've been in talks with people from bicycle manufacturing and the scientific community about providing units for this very reason. I'll be making that open as well in time such that people can see those torque plots themselves, or record them for a coach, etc. I don't know how hard it will be to run both simultaneously so initially it'll be the generic profile OR this scientific power meter profile.

  5. OK, I think I get where you're at. You're just a terminally curious geek. In a good way. I was worried that you had some oversimplistic view that because you thought you could build a more precise PM you could take on Garmin etc., but I see now you're way smarter than that, on one hand, and not really necessarily driven by the profit motive, whose success would require, in my cynical world view, among other things, the ability to lie convincingly, or at least to finesse the truth, or at very least, the ability to gussy up some trivial capability to to attract the ignorant masses whose incomplete understanding of technology is influenced by some poorly understood presumed benefit that has been hyped by some media "expert" or athlete whose prowess has trumped his understanding of physics and physiology.

    My consumer's take on the question you posed -- why are powermeters so expensive -- is that to some extent they are difficult to design and build, at least in a form factor that is foolproof enough for the unwashed public to accept, and to this has to be added the cost of making it cosmetically acceptable, easy for civilians to use and understand, and then the cost of hyping and marketing. The first reason, while not necessarily the biggest cost factor, I would have guessed not to be trivial, and your cost analysis seems to have borne out my guess. Once you have got a stable and reliable design -- something that apparently has so far eluded Garmin, so it obviously must be reasonably difficult --, the other costs have to be added on, and for the big players to bring the costs down, I would imagine they have to sell in large numbers, which would mean hiring expensive marketers and marketing campaigns, which I would guess involves dumbing down the product some, and promoting a lot of frivolous useless frills. I understand your intent is to cut the process off way before it reaches that stage.

    Of course it is your exclusive right to publicize or not publicize your strategy as it evolves, or not to have a cohesive strategy yet, or even for ever; although, I'm sure you'll find that if you have a strategy that resonates with others, they will be more likely to commit to making the effort to contribute.

    I am very interested in your project, and would volunteer as a beta tester, except that I live in Calgary. I would have no problem with not owning the hardware, and having to do a lot of fiddling. I consider myself an early adapter, I am inquisitive, understand physics, a tinkerer, and someone who enjoys getting things right, but am enough of a non-geek, electronically, that I had to look up Arduino.

    I understand your viewpoint -- as a kindred spirit I myself have taken off on involved projects, usually of the home improvement type, that to family and friends have defied common sense and certainly economic (worth/cost of labour involved) balance. However, I have always been well rewarded in the effort and also, usually in the final outcome, you know what I mean. I admire your expertise and your enthusiasm, and if there is any way I can contribute from a distance, just ask.