Yeah , "art of the compromise".
Compromise is even more valuable in a DIY design where you have
to expect a wider sourcing of parts.
Start with an "ideal" simulated design (that is overkill) , expect performance
the halve with layout in the real world , and maybe halve again with sourcing.
OS
Compromise doesn't require any creativity, the art is in optimizing, not compromising a design.
OTOH, why compromise at all? Or compromise the least possible. Do the best you can and see what you get. See what happens when you do the best you can possibly do. Push yourself and your design further than ever before.
THx-RNMarsh
THx-RNMarsh
Last edited:
You first have to define best. Now of course compared to a satellite, or a comms network a domestic audio system is simple so you can roll requirements management and systems engineering into the one person who delivers the board. Of course without checks and balances that one person could end up thinking that they are an artist in silicon...
My 100W VMOS amp counts 22 transistor in total, 6 of it are CCS or Cap multipliers. I don't think its to complex or over engineered and still THD is below 10ppm at all frequencies and all output powers. That is simulation of course, but from built examples it quite accurate. The main obstacle to get that low distortion is the PCB layout and I can't say I am anything but expert for it, someone more experienced in audio layout could make it very close to the simulated result.
I recall that Mr Bob Stuart of Meridian once stated in an interview that there should be s few transistors in the signal opath as possible because no matter which transistor and no matter how one applies it, each transistor will introduce some distortion and change the sound even if just a little, However, "as few as possible" relative to the design goals of the circuit, the point being that there should be only as many as you need to meet your design goals.
I find this "art" thing weird. It's as if we have to apologise for being physicists, scientists, engineers, and try and make a rational process sound more exotic. Excellence does not need to be defined as art.
Oh yes, the "F" is well defined in this case! 🙂
I agree on the low semi count despite what others say. It's a nice design aim, to get the desired result in the simplest way. And, in general, simple is more predictable, less likely to do something unexpected!
I don't think anyone ever lobs in a few extra devices for 'lolz'. 'As few as possible' is silly. Much better to quote Antoine de Saint-Exupéry in this case. Now whilst some will argue the use of a CCS where a resistor will do I would expect a professional to understand the trade offs and know which one matters, even if the cost constraints limit him to the sub optimal solution.
Bob of course is a very good salesman as well as having produced some fine equipment over the years.
Don't know of any pro PCB designers that use autoroute these days and for the last 20 years or ever....
Autoplace is a joke, placing components is the key skill a PCB designer learns over the years.
Audio layout is no different from any other layout both analogue and digital aspects...
I am guessing Alteon (Protel to anyone over 35). That seems to be the standard for hairy chested layout software. And have the most vertical learning curve!
Looked at all the main packages (Cadence, Zuken, Mentor and Altium), they are no good, seriously the art of PCB design is all in the placement, get that right and the rest is easy.... And learning placement takes practice and more practice and feedback from your designs so you build up your knowledge base for the various circuitry you will have to place and route... Analogue with feedback and power supplies are two prime examples (some digital may only be a few BGAs but even here getting the correct pattern of connections is critical to getting a good routing results. A good design should flow and not look clustered with components placed as indicated by the circuit functionality.
Even if you spend weeks setting them up they are still a bag of *****, how many rules and decisions do you have to define to determine where to place a component... If you don't apply rules they just go into mental breakdown mode....
I never lay out a PCB without the schematic open in my second screen so I can cross probe between the 2 and I use the schematic as my guide to the layout, both (schematic and PCB) are always open during the layout cycle...
Anything that is autoplaced and routed is likely to be quite horrible.......
Protel was **** so are its descendants...
Zuken my first preference because PReditor is the best...and used it since 1985Visula, Redboard and Redlogic in them days now Cadstar. Cadence my second preference...
I have access to and avoid using all the main packages, one of the joys of bureau work, I am a believer in the phrase "Jack of all trades master of none" I am very proficient in one main tool, the others I leave alone, again all the main people I do layouts for use the same tool so I am a happy! PCB designer.
Kindhornman, I am in Poole and with travelling and work load I haven't had much chance to use my PC, having to use a company one down here so haven't looked at emails etc. for a week!!! (been here 3 weeks now and spent 32+ hours enjoying the UKs efficient motorway system....
I am not a professional PCB layout person, but I second this 100%. I've tried, and seen on other packages, a few 'auto route' attempts - laughable.
Everyone I know uses it, possibly because of the industries I have worked in. The use may have been driven by clients having selected it and therefore you could earn more by actually being able to use it! At least that would explain why nobody ever really praised it!
As per my favourite demotivator Consulting - Despair, Inc. 😀
To quote Dogbert, consulting is a what you get when you add "con" to "insulting"...
- Status
- Not open for further replies.