Reliability: putting your product into the world

[TStehmann]

Hi

After 5 years of manufacturing specialized products at low volumes (between 10 and 50 per year), I have become more scared than ever to market my products. Most of the high-value items are shipped half way around the world and support is difficult. No matter how much testing is done before shipping a new device, there can always be problems. The funny thing is, that it is often not clear what the root causes of the problems are. Sometimes customers use old PCs (Pentiums with Windows XP) to interface with our devices. Other times, devices just stop working and need to be sent back. Ok, maybe it is not that bad. I am a perfectionist and if one device out of ten fails, the impact is detrimental. It also depends on the customer. With some customers I know there will just not be a problem.

One of the products is a multi-camera system. These are linear array CMOS sensor cameras. They use FPGAs, uPs and USB interface. Cameras need to be synchronised with each other and also with a photo-diode to sort measurements. This is a system of devices, that need to work together. All devices support firmware updates and the PC software also supports updates. So, at least there is the infrastructure to provide support via updates.

The one thing I have learned is that things can go wrong, but how to deal with it? It is maybe not a good thing to be the designer, sales person and support engineer at the same time. How does one gauge the quality of a product? Which integration problems or failures are acceptable and at what stage does one have to say, this is not working? Up to now, all problems could be solved and customers seem happy. However, the more products are sold, the more I will need to provide support.

How do other companies handle this situation? Are there start-ups going through the same phase?

 

[itsmoked]

I've done the same for several different products. Always a hard call.

Whether or not it makes sense depends on a lot of details. I had one product that was sold to the general public for about 100 bucks. I sold about 750 of them all over the planet. In the first 100 I found the issues I had to fix but unfortunately the product was unenclosed and so it was at the mercy of the customer's static. It became pretty clear that overseas had support travails when shipping cost a considerable fraction of the product.

You may want to restrict your sales to the USA if you're faced with a lot of support.

Quality? Does it work without frequent 'repairs'? If you often need to deal with computer issues as related to your product I'd consider you have robustness issues.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

Typically, large productions amortize expected warranty and customer service costs into the cost of the products. That might not be realistic for small production runs.

Post production testing can eat your lunch, so best design, manufacturing, and handling practices are a necessity, as you can never cost effectively test your way to robustness

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529

 

[MikeHalloran]

I used to work for a medium-large company that had an excellent reputation, worldwide, based on its service staff's availability and talent. Customers said "I love your products because your service person, <service person's name> is always here."

Eventually, after the patents ran out, the Japanese showed up in our market, and slowly took it away. Customers said, "I'm buying from the Japanese because your service person is _always_ here, but the Japanese stuff just doesn't break."

We responded by designing an internal product development and review system that was essentially a narrowband filter, designed to only allow surefire blockbuster products into production. Narrowband filters don't pass much unless the input is perfectly aligned with the passband, and we didn't know how to design surefire blockbusters; the company had grown to prominence basically by trying anything, and fixing the stuff that sold, and forgetting the stuff that didn't sell.

Since I was asked to leave, the company has merged, and then been sold, and may be available again...

Sorry, no happy ending there.


All I can suggest is limiting your market for the first few hundred units to a day's drive, or investing serious money into failure analysis and prevention, with internal design standards and regression testing and such...






Mike Halloran
Pembroke Pines, FL, USA

 

[zappedagain]

Sadly at my last company the RMA rate was over 75% for the first 50 products shipped around the world (4th generation design). They are better now (approaching 150 units shipped) but that division still isn't making any money. Nice to have a cash-cow in the company to support products like that until they can take off. When we moved from to the 3rd generation our field-service guy finally got to reconnect with his family.

Z