Designing vs. Voicing and Tweaking circuits and crossovers to get that sublime sound?

There are design tweaks and then there are consumer tweaks. A design tweak might be changing an edge radius on a cabinet, which lo and behold makes a meaningful difference. A consumer tweak might be putting your cables on little telephone poles, which gets into the "all in your head" level of improvement, i.e. your spouse thinks you're nuts, and your friends can't tell the difference, and there's no solid evidence that there is any difference.

Depends on what you classify as "design" and "tweak".

If we take GaryProtein's description above - trying different parts is design..... then my work is 100% design.

However, by the time I finish computer modelling and settling on the values on a crossover, the design is nowhere near complete. If you also regard trying different configuration of parts (for example different brands of parts, mixing up the parts (parallel or serial) getting to the same value, the layout of the traces on the PCB, how components are oriented, etc) as tweaking, then my mix is closer to 10/90 design/tweak. Along the way, the values may change slightly, but the essential rigorous design work would already have been done.

There is never any solid evidence that there is any difference beyond what I can hear, my sister can hear the difference sometimes even sitting at her desk a couple of doors away from the lab, and my wife can sometimes come in and say "I liked what you did 2 weeks ago better."

buying and learn to use loudspeaker measuringequipment has brought the greatest gain it gives a lot of insight ,``tweaking `` is just trying what works or not ,after that ;
housing construction/material
ls units
cirquitry
x over components

i dont have a high opinion of tweaking x overs ,the x over cirquitry and value of the parts are for 100 % derived from measuring , no voicing what so ever, i dont use computer x- over software either ,all custom design

caesar said:

Does anyone have any insight on how much time high end designers spend designing vs tweaking? 60/40? 80/20? 90/10?

And what do they do when they voice? try different parts? try different wires? perform voodoo ceremonies?

Click to expand...

Knowing a few designers,I would say 90+ pct design and 10pct or less finalizing the product. Circuit design,computer modeling, PCB layout,enclosure design. sourcing parts,ect. Voicing takes some time and troubleshooting if needed. I have seen some grounding scheme tweaks afterwards,but the process has always been pretty thorough. For the ones I know they border on genius,not their first rodeo.

For me the ratio is around 95/5... The initial design is worked out mathematically, briefly proven on test bench, then circuit boards are laid out. Once the bare boards arrive they are assembled and tested at bench level. Next phase is to assemble a loose approximation of the product with the bare minimum required parts. Here is where the bulk of the work is done, grounding schemes may change, wiring harnesses created, board fitting, metal fitting, basic specification tests and listening tests. Once these look relatively sane then onto final assembly with full metal and all of the bells and whistles "whatever that might be" attached. At this level is where (for me) the final tweaking will be done. Final tweaking is usually comprised of changing resistor or cap values in different sections of the circuit to either maximize it's measured response or inevitably it's sonic performance. This is time consuming, but at this point I am over 95% satisfied with how the product performs and now I want to make the magic happen in the voicing... Specs and models are not reliable indicators for real world listening sessions. Later on I may play with different types of parts and wire etc.. but not until I am fully satisfied with the product as it stands.

I do "voicing" to compare and select different components, such as crystal oscillator or capacitors. This is an ongoing study. Occasionally, I will compare different designs for voltage regulators for analog and digital sections to make sure they are optimized for speed and temperature effects. Once a design is complete, I don't usually mess with it unless there are opportunities to improve it, and these are usually not discovered by voicing. I am however one of the few manufacturers that will change modules on an existing product to update it, and offer these as updates to existing product owners.

The design is optimized as I do it, including optimizations for:
1) handling transmission-line effects
2) minimizing cross-talk and ground-bounce
3) minimizing power current-sharing
4) optimizing speed to minimize jitter

Steve N.

caesar said:

Does anyone have any insight on how much time high end designers spend designing vs tweaking? 60/40? 80/20? 90/10?

And what do they do when they voice? try different parts? try different wires? perform voodoo ceremonies?

Click to expand...

As others have said, I would think and hope the amount of tweaking would be limited to a few percent. Conversely, I heard from a fairly reliable source that a small manufacturer (who shall remain nameless) of very expensive tube equipment voiced his phono stage purely by ear. Why, when the RIAA EQ curve is so well defined? It seems akin to throwing darts blindfolded. Alas, that manufacturer's tube gear proved to be so unreliable that they switched to making solid state only. Who knows if it's any more reliable, but I wouldn't touch it.