AVR Freaks

I usually found that the batch run is the main driver behind costs. Runs above 1K can be priced in a similar fashion as you've posted: PI times the manufacturing cost. In my previous live, er, job, that was the usual pricing method for rough price estimation.

For short runs, there are other factors to take into account, starting with the R+D effort. This is now my current business: short runs that can go up to... 70 - 80, not more. The good thing here is that the 'customer' is the very same company. But costs are damn, prohibitively, high. Lots of time devoted to development, as few prototypes as possible, everything prepared for multiple possibilities, oversizes, etc. For that case, 'the usual fee plus expenses' would be the quotation system. And the PI factor is involved when estimating the time required to do the job. Partial payments at pre-stablished milestones are the best option.

BTW, I was involved in a project that, by involuntarily misquotation and totally wrong planning, things developed in your depicted 'pretty immoral way'. That was not good for lighthearted people. Tension, stress, and many personal issues were involved due the bad planning. Probably this is the main reason why in Spain there is so much pain with technology and R+D companies, that left us as a place where engineering is not well considered.

Dren,

I'll tell you exactly how it works in consumer electronics and the way Amstrad used to price things in the high street when we sold retail product.

To decide if it would be cost effective to make something we actually started with a target retail price and then worked back from there to find out what was available for the BOM cost and whether it was possible to make it for that price. If it later turned out that the BOM could actually be lower we'd either absorb the difference in profit or use it to have a knock on effect in the retail (often dropping the price 6-12 months after release to regenerate sales).

So say we wanted to make a hi-fi, telephone, fax machine, TV, computer or whatever to retail for £299.

First we'd have to consider the Chancellor of the Exchequer. He'd want (at present) 17.5% so the retailer will actually receive £299-17.5% = £246.68. A typical retailer like Dixons wants to make 30-40% profit to bother stocking an item on their shelves so let's say that they want to take 35% of that £246.68 (that is £86.34) That means we sell to the retailer for £246.68 - £86.34 = £160.34.

(luckily for us there's no distributor between Amstrad and Dixons, if there had been (selling through smaller retailers) they'd have taken 10%)

To justify our existence we'd like to make 30% ourselves so of the £160.34 we take 30% (typically) which means we take £48.10 (yes we make just over half what Dixons do). Admittedly some of that "profit" will be our marketing costs (and equally some of Dixons £86 will be their marketing cost - sometimes we'd work together for marketing).

Having taken our £48 we're now left with £160.34-£48.10 = £112.24. That is the "landed" cost of the product.

To land it in the UK from our Far Eastern manufacturer there's usually some form of duty to be paid - the level depends on the product but is typically 4% (lucky old chancellor again). When combined with the shipping cost from the Far East (usually boat, though we used planes for fast delivery at product launch) we'd generally allow 10% for shipping+duty which means the £112.24 is reduced by £11.22 to £101.01. This is known as the FOB (Full On Board) price and is what we pay the factory to procure the components and build the device.

The factory themselves are not philanthropists so they want to make some profit for their efforts. This is known as Labour Overhead and Profit and is typically 8%-10% (there used to be a VHS factory that just made $1 per unit for LOP!). In this case the £101.01 is therefore reduced further to £101.01 - £10.10 = £90.91 and finally this is the Bill Of Materials (BOM) cost. So for a £299 retail unit the BOM is a little under 1/3.

At this point we'd convert the £90.91 to $ as almost all components are bought in the Far East for $US. This is where we can often win/lose according to the current rate. Today the £ is strengthening against the $ and is 1.591 so in this case we'd have $144.64 to spend on components, casing, packaging, manuals, software royalties, licence fees etc. etc.

One thing I have not included in the above is our NRE. The software development (and hardware engineering) are a one-off cost and can range from £50K to £3-4M depending on the product. The tooling for the plastics typically ranges from £250K..£1M. These costs will be amortised across the number of units built and either included in the BOM costs or taken from out proit/marketing budget. If you have £1M of s/w dev costs and £2M of plastic tooling costs then the £3M could be amortised at £30/unit if you built 100,000 though you are in deep weeds if you can only make/sell 20,000 in which case the amortisation becomes £150/unit!!

I realise this is not directly related to your question but some aspects of this may prove useful.

Bottom line is that on the tightest of margins take 30% profit and expect retail to be 3X BOM.

Cliff

Well if you are just acting as a wholesaler buying something in for $10 and making a reasonably healthy 30% then that would suggest $13 (though obviously covering P&P as an additional cost). But to a certain extent the dictates of Adam Smith come into play and the market will be driven by supply and demand (in a free market economy). If you had some TQFP mega328's right now you could probably sell them for $10 each (for example) and make 200-300% profit simply because that's what supply/demand in the market will bear right now.

I believe we all understood that... :D

Who let him out??? or did he escape? :roll:
https://www.avrfreaks.net/index.p...

Damien, I bet this is the main reason driving ARM's in the low end, with OS's and low prices for such low volumes. Since I'm in an even lower volume business, we do it in a even more radical fashion: no uC at all, everything PC/Windoze based (huh, this sounds like a new terror film...), and the little work I'm doing on uC is only because we need some safety around certain part of the program/data/calibration that we can't leave on the PC.

On later times, our interest, like other engineers in the factory and other factories, had turned into something like Beagleboard and such.

Dren, in our case the PC is there for other reasons, so we only add some small task to it.

A long time ago, in a galaxy far, far away, I've developed one 'simple' product with a M128 (in a cheap 20€ module), a cheap 2x16 LCD, and some HW. In that case, I charged 'the usual fee, plus expenses'. I mean 30€ per labour hour, plus all HW paid by the customer. The bill for one off was about 600€, after some re-negotiation. I should be about 1000€, but that is another history.

If we made a production run, just the PCB's, components and LCD's would cost about 2000€ for 100 units (using M16 instead, though). You have to add any assembly labour, something like another 2000€, and the developing costs, 1000€. This would represent that to have a net income, the 100 units must be sold above 50€ plus taxes to the distributor/company/customer/whatever. The salesmen priced them at 300€. The boss considered the presumably net income (it would be smaller for sure) of 25000€ not worth.

Over here, R+D in electronics is considered an expensive extravaganza. An EE working in this area hardly can expect more than 33K€ brute per year (usually below 30K€, being a 24-26K€ a regular salary for such job). Would you work for such salary, given the current costs of living here (a regular flat is over 250K€)? Management consider this an overpaid engineer. No need for them, so they can increase their own salary in 33K€ brute per year.

With this, I mean that developing costs rise too fast, and they should be charged to the total amount of products sold. Usually, when I was in another busines with regular sales in the range of 1K to 6K units per year, there were always a sales estimation, usually done by some marketing guy and some knowledge of the market and the distributors. This simple (but really hard to determine) figure can indicate too much how to proceed.

Thus you have two estimations here to do: your developing costs in money and time (remember the Pi factor ;)), and the amount of units you think you will sell in two years. Sum the costs of production of those units, and add it up to your developing costs, then divide the result by the amount of units. You can find that easily, for small runs, the developing costs will be bigger than the manufacturing costs. Then you have to add your planned benefit, and all expenses that can go there (taxes, shipment, stock, distributors, discounts). The math and numbers are simple if you use common sense, but you need a 'solid' (if a sales estimation can be anything like 'solid') starting point. The developing costs are not easy to calculate, though.