Gantt Charts

[jimjim]

Hi there EEVblog fans - hey Dave!

I'm looking at proposing my first full scale electronics project, as a part of the business plan associated with the product I need to make a Gantt chart that shows the time scale involved in researching and developing the electronics.

I've got a basic idea of the circuit and what components I'd like to use, however - all the software and PCB will need to be designed from scratch. It's the first time I've had to do a Gantt chat for this kind of work and I'm not too sure on what's the convention, what kinda time is normally quoted to do x, y and z?

To give y'all a basic idea of the project, it'll include - wireless communication (wifi / bluetooth), instrumentation amps, a USB programmable MCU and multi channel ADC, I also plan to run the whole thing with batteries.

I need to get the whole project designed, tested and ready to go to the PCB manufacturer for volume production (batches in excess of 1000 units)


So - my questions are.....

* What processes will need to be performed in order to get a product ready for commercial sale? (For example - Would you recommend an initial prototype, proof of concept - whack it together scenario, followed by a production prototype - using the proposed production components?)

* What processes are commonly under estimated - from a time perspective? ( Is there an area where "young players" often underestimate the time required)

* What's the expected time period for developing the whole product? (Roughly speaking, within industry - what's the norm? , what's too long or not nearly enough?)

* Any other good pointers for "young players"?

Cheers

Jim.

www.sonodrome.co.uk

[Mechatrommer]

i dont have any real experience developing and implementing gannt chart. but i think:
1) you'll have the trouble following it once its done. unless you are disciplined and nothing else to do
2) if the time is critical for you, then only gannt chart is necessary (planning).
3) the duration will depend on how much work force do you have, experience needed to develop certain project and what kind of project itself and how its functioning, and how much pressure you have to complete the project in a very short time.
4) the less the time you spend, the less the quality of the product.
5) you only will be able to create an excellent gannt chart that suitable/effective for your workflow, if you are mastered enough in the workflow itself.
6) the best feature of gannt chart IMO, is that you can visualize clearly and plan for the shortest time completion where simultaneous tasks (tasks that can be done at the same time) is maximized. but then again, you need another fren to do the other task for you while you are doing another task.
7) the other feature is slack time, where you can modify your gantt chart to minimize your idling pals.
8th) i stand to be corrected

http://en.wikipedia.org/wiki/Gantt_chart

[Zad]

What it boils down to is a) "how long is a piece of string" and b) "On time, on spec, on budget. Pick 2". This is another part of the art of electronics (and engineering in general) that isn't taught at Uni.

I'm afraid the only helpful tool is experience really. You need to know as much as possible about the project, clients have this habit of "knowing" in their head what they want and not telling you. Small things like "oh, did you not design the enclosure?" when you were clearly never contracted to do so, and it was never mentioned at any stage. I'm afraid timelines can be rather nominal at best. As a rule of thumb, double whatever seems about right. Only you can know how much effort is likely to be required and what resources you have.

If it is a small business and comparatively flexible, the chance of you hitting your date is higher than if you are in a 9-5 corporate entity, where work stops at 5pm. Of course the big corporates can bring in resources at any time, should the accountant see fit. But they rarely do, a late product just results in shouting and finger wagging. It seems obvious, but ask the client when they would like the product. Chances are that they have some date in mind which will help drive your milestones.

Oh yes, if the design is anything but simple, definitely build prototypes. Even the humblest embedded micro has silicon bugs which can be a show stopper. Not great if you have just ordered 10,000 PCBs. In a complex system, the prototype can go through quite a few evolutions between start and finish. There is nothing quite like having physical hardware in your hands to satisfy you and the client that the product exists and actually works.

Many management books have been written about Gantt charts and, so far as I can tell, they exist mainly to impress managers and marketers. Critical path / PERT charts are more useful, not because they make the dates any easier to hit, but because they can show up bottlenecks before they happen. They need quite a bit more detail than Gantt, but this detail allows you to make closer estimations.

[jimjim]

Thanks for the great responses - And Yeah, I know, things aren't likely to follow the course of the Gantt chart, to a certain extent I'm mostly doing it to satisfy investors.

It's good to get some outside opinions on this tho, cuz you never know what people expect in the way of time scale - I'd love to be able to respond with "well......how long is a piece of string?" but for some reason it's too vague for them

In regard to my experience level - I've never studied electronics in an official capacity, however, I've worked professionally as an AV engineer and I also sell simple electronics kits on my website (of my own design), this is my first proper step into the big bad world of electronics design.

I guess I've mostly been messing with audio circuits and MCUs.  
 
When it comes to it - we'll be hiring in some extra EEs to help with the design process, but I'd imagine it will be a very small team, probably just consisting of me and someone else to be honest.  

Oh - and the project is not for a client - it's for my business, so I'm kinda the client, ultimately I get to decide the time scales, but I want them to be comparable to the norm (don't wanna scare off the investors just yet!) - but because we're getting investors involved much of the planning is needed to satisfy them. 

cheers

Jim.

[Mechatrommer]

what do u mean by "comparable to the norm"? gantt chart can expand from few days to few years? it also depends on your resources, tasks and type of project, requirement etc, as mentioned above... so whats the norm?

[jimjim]

Hey Shafri,

I'm not after precise figures, just a general idea of the time scales involved, I understand that's it's not possible to give completely accurate figures with such vague information.

I'm kinda just sniffing about for any related advice, I'm also looking out for any common underestimations, school boy errors etc.

Sorry to be so vague, but I've still gotta get some legal advice before I can start sharing the project details, to give you a rough idea - it will be more extensive than designing an Arduino shield (lets say the wave shield for arguments sake) but should be less complicated than designing say - the Beagle board.

I'd put it on a par with the complexity of Dave's mini calc watch as it will require a fair bit of programming, but that's just a rough comparison.

Cheers for the responses

J.
 

[Zad]

I'm not sure it would help a lot if we knew what the project was, because we don't know what resources you have and what sort of work rate you can get out of them, only you can know what your capabilities are. The difference between a theoretician trying to develop something, and an experienced engineer doing the same job can easily be over 10:1. If you don't have the right test equipment then it can be doubled. If parts are single source and on back-order, then that can add 3 months.

You could probably hack a working model together in a week or two, with basic software. But then it may take 6 months after that just to get the pcb and software production ready, and a couple more months to get the enclosure done. I don't like to quote under 9 months from start to fully finished. Not because I'm slow, but because "stuff" gets in the way. Compliance testing, pcb turnaround times, error correction, debugging, people going on holiday, client changes-of-mind, all non- electronics stuff that just slows things down. With the best will in the world, if I need something from Digikey, that's half a week gone.

[mikeselectricstuff]

In your position, before even thinking about talking to investors and messing with charts etc., I'd develop it up to a working prototype stage, with enough of the software functionality done to a) demonstrate it, and b) give you a good idea of how much more time it will take to finish.
Once you're at that stage, it will be a lot easier to predict the remaining time and cost factors to get it into a production state, and although this WILL still take longer than you expect, at least the variables will be more quantifiable.
 

[jimjim]

Thanks muchly for your frank answers Zad, that's just the sort of stuff I needed to know, never made that many of anything before, don't wanna stuff it up!

Cheers

[EEVblog]

Oh - and the project is not for a client - it's for my business, so I'm kinda the client, ultimately I get to decide the time scales, but I want them to be comparable to the norm (don't wanna scare off the investors just yet!) - but because we're getting investors involved much of the planning is needed to satisfy them.  

Oh boy, I get the heebie-jeebies every time I hear the word "investors" associated with projects like this, it almost always ends badly. I hope you know what you are getting yourself in for.

At this stage if you have no idea of the exact design or  who is going to be working on it etc then it really is pie-in-the-sky stuff, Zad's proverbial piece of string.

Generally for something like this you would start out with a demonstrator project. You'd generally produce that first before fishing for investors and talking gantt charts and project schedules etc. A complex project might have a project schedule for the demonstrator/prototype, but your design sounds like it hardly fits that category.

My advice would be to just build a prototype on your own and get something up and running before even start talking to investors of any kind. Once that's done you'll have a good idea how much work is left.

A sub Beagle-Board type design is easily doable by one person, depending upon skills.

A typical design like that might run at least say 2 or 3 prototypes before it would be in a state to take to production. A typical design cycle might be 9 months for such a project. Software and tweaking takes time.

Dave.

[Mechatrommer]

And Yeah, I know, things aren't likely to follow the course of the Gantt chart, to a certain extent I'm mostly doing it to satisfy investors.

whats the point of satisfying the investors if you cannot perform? empty promise? otherwise you are going to struggle to gather resources, and hence cost thereafter. dont you want to impress the investors with costing? you asked about the things thats usually underestimated. these are what i can think of:

1) silicon bugs
2) software bugs (time needed to fix will depend on your skills)
3) higher part cost than expected
4) out of part stock
5) change in specification (too bad)
6) investors requests (can you quicken? can you lower the cost? can you produce massively? can you change this and change that etc)
7) your accomplice? run out of you
8th) dead project = next project = new gantt chart.

my advice, if you wanna to buid a gantt chart, do it frankly of what you think you are capable of, if timeframe is not specified. by the way, underestimated things isnt suppose to reflect in the gantt chart. it is something that we "underestimate". the only thing that implicitly reflected in you gantt chart is your experience, and nobody can help you about that. IMHO. Cheers

[jimjim]

Cheers Dave & Shafri,

It's nice to have some input from an outside source, I've started out my business as a small partnership and I'm now looking at expanding to the next level, I'm just gathering as much info as possible so I can go into it with my eyes open.

We have just got some outside help to assist in applying for a mixture of funds and investment, they we're asking about gantt charts for technical development, hence this question thread.

I've made similar circuits to this before, but this is the first time I'll be turning one like this into a product, plus I've got the added problem of using a new microchip that's not yet available on the market - so I'm sure there will be plenty of unforeseen problems to tackle, including the dreaded silicon bugs!

Thanks for all the advice and comments