With KiCad, and I suppose any design package, you need
to know the physical size of the components. So hopefully,
the following is not off-topic. Most component choice
results in a decision on component size, and that will
affect the layout and physical arrangement.
If I do a PCB design in KiCad for the Leach amp, and
post it, then would you like SMD or normal components?
KiCad is new to me, so I'd appreciate any pointers.
You will notice that there is a major load of doo doo
on the net (and in the shops) about speaker-lead quality,
oxygen free copper, "audio-gade" capacitors etc. The more
research I do on this, the more convinced I become that
we are being hoodwinked. For example, super-low loss
speaker cables can actually cause an amp to oscillate at
supersonic frequencies because the L/C characteristics
move the poles and basically mess with the amp design.
Some blind tests have revealed ordinary lamp-cord sounds
better! Similarly, snubber caps look like a waste of time.
None of the maths, physical measurements or simulations
show anything other than infinitesimal differences. Very high
bandwidth amps can be prone to oscillation. Speaker wires
are NOT transimission lines at ANY audio frequency.
self inductance of caps is also insignificant especially with
today's construction; 6nH of inductance per cm of lead
wire totally swamps the calculations.
Unfortunately the list goes on.
So I am pleased to see you tell me of some of the
tweeks that actually make a difference. But I'd
like to know why. :)
Do you know why to use metal film resistors? Would
SMDs be better these days for any of the components?
In no particular order, what I think is important in component
selection is:
1) Nominal value.
2) How its characteristics change with temperature?
3) What electrical / thermal stresses can it take?
4) Physical size.
5) How does it deviate from its ideal model?
6) Build quality and therefore reliability.
7) How the characteristics change over time.
8) Is it a very specialised component?
with 1) only some components need to be tight tolerance.
If you make everything 1%, then it just makes it expensive.
Although tight tolerance can sometimes imply better
quality and reliability.
with 2), if the value changes with temperature, and you go
for tight tolerance, then there had better be some temperature
compensation, or you may as well use a cheaper component.
Almost all practical calculations and component specifications
involve temperature. Hfe (Beta) of a bipolar transistor actually
goes up with T, some resistive devices fall with T, some increase,
Random noise increases with T.
with 3) These are the parameters that stop your equipment
blowing up when voltages and currents and heating takes
place. The biggest characteristic that is effected is probably
physical size.
with 4) See 3 ! Of course, this is also where component
construction technology comes in. The most notable being
capacitors. How big would a paper-oil 8000uF capacitor be?
with 5) It depends on the use as to the effect. As noted
above, at audio frequencies, self inductance of a cap
does not have a measureable effect in most (of all?)
circumstances. However, the characteristics of a mica
capacitor make it ideal for the Miller-Cap.
6) and 7) are self explanatory except that sometimes
smaller is better, and sometimes not!
with 8) specialised components might not be in the
market for a long time. So your design could be
rapidly uneconomical to repair.
Point 8) is one reason that I like the Leach design.
It's simple. Simple can often lead to better results.
Some off-PCB notes:
I am interested to know more about the issues you had
with the transformer. I don't know what is a
"constant voltage transformer". Why would the hum be
bad? Was this a physical hum? If so then this is because
of loose laminations, and unaccaptable eddy-current
losses in the iron. The iron could be poor quality for
a transformer. If so it was probably inefficient. Or it
might have been fixed by physical sheilding if the
hum was electromagnetic radiation from the transformer
picked up in the early stage of the amp.
I can't fathom how the transformer can affect base
response unless the losses were so great that the
supply voltage was affected. Any ideas on that?
Voltage is voltage, and current is current. If these
are available to the components upon demand
and at required levels, then how can one
supply sound better than another? I reckon an
amp driven by 15,000 mice in a wheel would sound
just the same as one supplied by electricity from
an oil-fired power station providing that the mice
were well fed, we have a few 'subs' and we don't run out of oil.
Toroids tend to be low-noise, compact but draw a large
surge current on turn on. They are very efficient. The
radiate less because of small or insignificant air-gaps
which otherwise radiate flux, and the construction gives
good sheilding.
Maybe the high efficiency of the toroid reduces its
regulation. (No load