How many design -> prototype -> test iterations?

[apelly]

So, you're a hobbyist. How many goes does it take to get a functional project?

Obviously that's a 'depends' sort of question, but I feel like I'm probably too worried about making mistakes. How many bucks worth of parts do you write off on prototypes?

A related question might be, how many prototypes work first time?

[nctnico]

In my case it usually goes like: breadboard prototype -> PCB prototype -> production ready circuit -> version 2 (due to changing customer requirements).

[georges80]

Obviously depends on the complexity of the design.

For simple boards I often get to production in the first iteration. Sometimes I'll do a proto run of boards to provide my customer some initial boards to test/evaluate and then any changes/refinements are rolled into the production version.

Other times a board in production gets some refinements and a new rev then becomes the production version.

Sometimes a board needs layout changes due to performance/operational issues in the layout - high power switching circuits that require tweaks for reliable operation over the full range of input/output voltages etc. Or thermal refinements...

Rarely a footprint is wrong for a complex part that escapes the human checks and the prototype is doa. This has happened a couple or so times in my lengthy career - quite embarrassing when it happens... Incentive to print 1:1 solder pad layers and placing suspect components onto the printout.

Cost, well I have a lot of parts that I've collected in various designs. I also try to stick to proven design blocks I've used before and reuse common parts. Everything I do is SMT when possible and use 0603 where possible and move up in size as needed for voltage/power ratings etc. Reusing design blocks also means that footprints are proven and already exist.

Anyhow, those are my experiences on boards I have designed.

cheers,
george.

[SL4P]

You should get to a first release within two prototypes, unless someone in the review & rework phases has dropped the ball.

• Customer Requirements & Specification
• Design
• Signoff
• POC hardware
• POC software
• Review Design
• Signoff
• Rework
• First Functional Prototype
• Review functionality & variation order$
• Signoff
• Rework & variation order$
• Develop BOM
• Reworked Functional Prototype
• Rework & variation order$
• Signoff
• Production Prototype
• Review / Error identification
• Rework / kludges
• Finalise BOM
• Signoff
• Production release v1.0

[Dago]

I agree with the previous posters. Based on own experience with work projects depending on the... Accuracy of the specifications and size of the project etc. one or two prototype rounds seems typical. Depending on the bugs or issues the second prototype USUALLY seems to happen but it varies if it is prototype number 2 or just a rev2 of the project (so some units made based on first prototype which worked well enough).

[SL4P]

The big deal in 'getting it right' is swirling around liability and professional indemnity...

If you make a $200 gadget, and put 100 units out in the field with an $80 profit - and at the very least, discover you have to perform a field update of software - (hardware is worse!), then factor in say a minimum of $50 to make a return trip to site... $50 on site time - no parts just reflash, and you'd pray the configuration, calibration and operation is not affected...

 $10,000 outlay to remove v1.0 bugs on top of code revision time - rapidly eliminates your $8000 profit margin!  Now do the math with 1K or 10K units deployed!

[dmills]

I have just had to go to a third prototype on something due to a faulty 5V regulator control loop (embarrassing!).

To be fair, we needed to get the second run out to a trade show, so had less then a week to test the first prototype before spinning the second one so these things happen, and I would not have been too shocked had it been something subtle and FPGA or analogue related, but to have it be the 5V regulator is just stupid.

The firmware is always the big risk from a deployed device perspective, and as a matter of policy we ship NOTHING that cannot have the firmware updated by sticking a USB key in.

Regards, Dan.

[T3sl4co1l]

I like to shoot for two (proto build, pilot production build).  But I'm a professional, and besides having quite a lot of experience at this, I also know to test anything I can't quite nail down.  This might entail development protos (usually deadbug of a fragment of the full circuit) or simulation.

Tim

[dom0]

So, you're a hobbyist. How many goes does it take to get a functional project?

Usually one. I don't make PCBs, so I almost always work on the same prototype until it is finished. That prototype is then also the "completed product". So that's very cost-effective, if you like.

Sometimes however you build a prototype and you'll measure and test stuff and you'll notice that it's just the completely wrong direction. Then the prototype goes into the bin and you start over.

Obviously that's a 'depends' sort of question, but I feel like I'm probably too worried about making mistakes. How many bucks worth of parts do you write off on prototypes?

Well, that depends. Small stuff (resistors, small caps etc.) goes into the bin when removed from a prototype. Sometimes I'll blow something up, but that's somewhat rare. The most expensive parts in a project are usually those that you won't blow up anyway. Stuff like cases, connectors, transformers, ...

OTOH integrated power ICs can be somewhat expensive (couple bucks to 10, 20, 30, 50 €) and you may blow them up, especially when supplied with high power with no preceding testing done. Rule of thumb: If that part costs more than 5 € or so and/or has a lot of pins, you wanna make sure you don't blow it up. Or you avoid them in the first place, because they often have few advantages for a hobbyist.

A related question might be, how many prototypes work first time?

Well, that a prototype just works fine out of the box happens, but not for more complicated circuits. Although that happens, too.

[Philbywhizz]

Sometimes however you build a prototype and you'll measure and test stuff and you'll notice that it's just the completely wrong direction. Then the prototype goes into the bin and you start over.

I don't throw my failed prototypes in the bin. I keep them and have them displayed on my 'fail shelf' with a note/tag on each one reminding me why it failed and what I did to fix it.

Some of my projects (my current one is an eeprom programmer) is at revision 3 (and I use the last 2 versions as reference for the bits that 'worked').

It's an great learning and motivation tool for when I do finish a project I can look back on how I got there.

[AMS]

I think I agree with most people here that say it depend son the complexity of the design. In my case its also depends on whether i have used a certain part before or not. If I haven't used it before and its a complex beast then I tend to buy the eval board if available of the chip and make a breadboard/veroboard prototype with it. If its not available I then tend to buy for example a ssop to dip cheap pcb converter and try the exact board that i have designed on breadboard/veroboard. So its usually breadboard/veroboard---->prototype PCB (With all the effort to use it as final)------>final production version (if the prototype PCB had problems). Some problems are not visible in breadboard/veroboard phase, such as cross talks or sometimes board outline or physical dimension incompatibilities (oops). DONE put too much stress on yourself as you will make more mistakes. Do your best and learn from the mistakes.

I don't throw my failed prototypes in the bin. I keep them and have them displayed on my 'fail shelf' with a note/tag on each one reminding me why it failed and what I did to fix it.

Some of my projects (my current one is an eeprom programmer) is at revision 3 (and I use the last 2 versions as reference for the bits that 'worked').

It's an great learning and motivation tool for when I do finish a project I can look back on how I got there.

Loved this idea and will start doing it I might call it wall of shame 

[apelly]

Thanks guys.

I think I was either worried about nothing, or being a tight arse.

My projects also suffer from scope creep. I should know better.