When to look at custom hardware?

So I have generally shied away from developing custom hardware / paying someone to develop custom hardware. But I haven't actually looked into the economics of it.

Does anyone have an anecdote of when they looked into this? Particularly interested in sensing technology.

For example, I can get a particular 4-20mA sensor for £500. Add £50 for the input card.

I notice that the IC with an I²C output costs £8 in quantities of 5+.

I could probably justify greater quantities of sensors and hence a more intelligent machine if there was say, a high development/maintaining the design price and low cost per part.

If you are building multiple identical machines then it might be worth looking at, but certainly not for any one of a kind machine. In addition to the initial cost you also have to consider the maintenance costs.

How many is multiple? 100?

What type of company do I want to enlist to help me with such a task?

AustralIan said:

How many is multiple? 100?

What type of company do I want to enlist to help me with such a task?

Click to expand...

Those are the two questions that are the most difficult to answer. Both of those answers are going to depend entirely on the specific equipment involved and how complex the components that you need to develop are.

Awesome! I'll ask around and see what I come up with. Will try to report back my findings.

But I haven't actually looked into the economics of it.

Click to expand...

The economics is pretty simple.


Let X be the number of sensors you need.

500X is the cost of continuing to use your current solution.

The break-even point is when the unit cost of the custom solution plus development cost equals the cost of the current solution or:

X * 500 = (X * Custom Unit Cost) + Development Cost

X = Development cost / (500 - Custom Unit Cost)

If you need to use fewer than X, you are better off continuing with your current solution. More than X and it makes sense to pursue the custom option.

It's up to you to figure out development cost and unit cost of the custom solution

I dont know about Australia, but in the EU and in the US you cant just jumple various bits and pieces together and use it in a machine or plant. It doesnt matter that it works OK and is of sound design etc.
You have to state that it follows all relevant regulations. For the EU, there is the "Declaration of EC Conformity" that all equipment in must have. It is a legal document.
I would be surprised if there isnt something similar in Australia.

Just something to be aware of if you consider to manufacture your own homemade devices and use them in machinery.

Your math makes sense Steve. I was hoping to get some anecdotal examples of someone doing this before, but I think I will contact a few electronics manufacturers and see what they say.

Also Jesper, I have moved to the UK, so I would need to get someone to build my sensor with the EC conformity.

AustralIan said:

For example, I can get a particular 4-20mA sensor for £500. Add £50 for the input card..

Click to expand...

What is the sensor in question ?

Custom hardware is fine until the little outfit you used to design it and build it fold and your supply suddenly dries up, or they decide that the next guy is more important to them than you are as you are only buying x amount of them a year.

I can make many sensors for cheap price, Like a Pt100 for around 20 euro with a 4-20 mA output.
i have made current converters from 4-20 mA to a valve running on DC with 2 Amps.
If you need intelligence have a look at Arduino, a very cheap(free) development

tragically1969 said:

What is the sensor in question ?

Custom hardware is fine until the little outfit you used to design it and build it fold and your supply suddenly dries up, or they decide that the next guy is more important to them than you are as you are only buying x amount of them a year.

Click to expand...

The little outfit is often not the problem, its the big outfit (Texas Instruments, Marvell etc) that makes the ICs that stops making IC number 1234 and launches the new version 1235 that has all these new bells and whistles (which you don't need) and now comes with 3 extra pins, which means the PCB you had no longer works (even though you just have 1000 of them made for economy of scale) oh and the new IC also draws 0.1amps more which means that your cooling system is not big enough.....

I could go on but believe me, where I work we used to make custom PLCs to control one of our machines, it causes all sorts of problems including the biggest issue of them all: What happens when one of them goes bad?

If its an off the shelf component from lets say Sick or AB, simply send it back to them and they´ll replace it at no cost to you (other than a little inconvenience). Now when its custom hardware the cost of that replacement is coming out of your pocket! Either you throw it out and put a new one in its place or someone has to repair it.

Also your quality will never come anywhere close to the quality you get from one of the big name manufacturers. They have machines running making the same part 24 hours a day 7 days a week. They don't have to do changeovers, they have Test-Molds made to test out each and every circuit board as it comes off the line, they have teams focused purely on quality control etc. etc. etc. (And even still sometimes we receive dead parts).

As for production scale, we would have been producing about 150-180 plcs per year. When an off the shelf solution became available, we jumped at the opportunity to change!

It may seem like its cheaper but in the long run its probably not!