AY3-8910 or YM2149F?

ZXNet echo conference «zx.spectrum»

From Oleg Bashtenko To All 22 June 1997


Hello All.

Here, I offer for public reading and analysis an article about music processors...

=== Cut ===
AY3-8910 or YM2149F ?

The domestic market of music processors is saturated with relatively
cheap products from the countries of the Southeast. 40-pin microcircuits AY3-8910(AY) manufactured in Taiwan, as well as their Japanese analogues
YM2149F (YM) from Yamaha, are widely used. Despite all the structural identity of these microcircuits, listeners still note the “softer” and “powerful” sound of AY compared to YM. On the other hand, in some music programs
played through YM, the stereo effect is more noticeable.
The search for the reasons for the differences in sound led to the need to investigate
the characteristics of analog outputs. As is known, a music processor (hereinafter referred to as MP)
contains three independent DACs with logarithmic amplitude characteristics
of channels A, B, C. The logarithmic law of change is associated with the peculiarities
of the human ear's perception of the volume of sounds. The Spectrum is designed to work with 4-bit DACs, so the output signal has 16 volume steps spaced 3 dB apart. The amplitude characteristics do not depend on the sound frequency, but depend on the load resistance Rн. In this case, the “soft-sounding” AY has a distribution lawamplitudes are much more different from the “ideal” than YM.
Dynamic range, i.e. the ratio of the maximum output signal to the minimum for AY is approximately 32 dB, and for YM - 38 dB. The flatness of the AY characteristic leads to a longer duration, or more precisely
volume, of the after-sound of falling sounds. The difference is very noticeable when observing the oscillogram of the effect of decreasing sound. and the oscilloscope screen in this case shows repeating peaks, the distance between which is filled with levels of different amplitudes at equal time intervals. Moreover, for AY there are 16 intervals, which coincides with the reference data, and for YM there are 32 intervals, i.e. amplitude effects for YM are formed
by a 5-bit DAC!
And yet the “incorrect” characteristic of AY is perceived by ear as a
longer aftersound, echo and a kind of unity of the melody.
The last role in such a subjective assessment is played by the musical
compositions for the “Spectrum”, which are replete with passages with sounds varying in volume. YM sounds more “dry” and abrupt to the ear.
An analogy with musical terms is appropriate here: AY is characterized by a “legato” melody (smooth transition from one sound to another), while YM is characterized by “staccato”
(short, abrupt execution of sounds). Question about more"powerful" sound
AY is clarified by studies of the load characteristics of analog outputs.
In practical circuits, the load resistance of analog
MP outputs is 3...6 kOhm, so the melody through AY sounds louder
than through YM due to the greater amplitude of the output signal Umax. If the load
was less than 1.8...2 kOhm, then the situation would be reversed,
and YM would sound louder than AY.
The difference in the depth of the stereo effect is due to the different internal
resistance of the analog outputs AY and YM. In the initial state without a signal and without a load, each of the three outputs AY has zero voltage, while YM has a constant level of 3...3.2 V, which exponentially decreases to zero when the load is reduced to a short circuit. MP analog outputs are not afraid of short circuits in the load. The short circuit current for YM
reaches a significant value of 12...15 mA. Analog outputs AY and YM
are "open emitter" circuits, i.e. for normal operation
it is necessary to have a galvanic connection of the outputs with a common wire.
As a non-optimal solution, diagram 1 shows the output part
connection of the MP in the "ATM-TURBO-2" computer.

Diagram 1

DD57A AY3-8910(YM2149F)

──┬───┐ 4 ┌───────┐ ┌───────┐
│ A ├──────────┤ R17 ├───*───┤ R21 ├─────┤
│ │ └───────┘ │ └───────┘
│ │ │
│ │ │
│ │ ┌───────┐ │
│ │ ┌───┤ R18 ├───*───────────────────────────> LEFT
│ │ 3 │ └───────┘
│ B ├──────*
│ │ │ ┌───────┐
│ │ └───┤ R19 ├───*───────────────────────────> RIGHT
│ │ └───────┘ │
│ │ │
│ │ │
│ │ 38 ┌───────┐ │ ┌───────┐
│ C ├──────────┤ R20 ├───*───┤ R22 ├─────┤
──┴───┘ └───────┘ └───────┘

R17, R20 = 1K R18, R19, R21, R22 = 2,2K

Если запустить на выполнение Бейсик-команду PLAY "V15A", по котоpой
сопpоцессоp воспpоизводит звук только из канала А, и измеpить амплитуды
сигналов, поступающих в левый и пpавый стеpеоканалы, то для AY в схеме 1
отношение этих амплитуд составит около 3(9,5дБ), а для YM-пpимеpно 20(26дБ).
Одновpеменное попадание сигналов в левый и пpавый каналы заведомо ухудшает
стеpеоэффект, котоpый и так неяpко выpажен из за суммиpования сигналов по
каналу В. Улучшить pазделение каналов в схеме 1 можно изменениемvalues ​​of
resistors R21=R22=510 Ohm. In this case, the amplitude ratio will be more than 18 dB, although with a slight decrease in the output signal level. For circuits with other
resistor ratings, you should strive to ensure that the ratio R17/R21 and
R20/R22 is more than 2. Circuit 1 with changed ratings will become uncritical to the type
of the MP used. If you want to bring the sound of YM closer to AY to some extent
you can recommend changing the slope of the amplitude characteristic of each
of the three channels by introducing diodes and trimming resistors according to scheme 2.
Trimming resistors need to set the level at which they will
be attenuated in amplitude the upper 5-6 steps of the characteristics.

Scheme 2.

DD1 YM2149F out
┌──>
──┬───┐ 4 │ │\│
│ A ├─────*─────┤ ├────────────────┤
│ │ │/│ ┌───┴───┐
│ │ VD1 ┌────┤ RP1 ├────┐
│ │ │ └───────┘ │
│ │ out │ │
│ │ ┌──> │ │
│ │ 3 │ │\│ │ │
│ B ├─────*─────┤ ├───────┼────────┤ │
│ │ │/│ │ ┌───┴───┐ │
│ │ VD2*────┤ RP2 ├────*
│ │ │ └───────┘ │
│ │ out │ │
│ │ ┌──> │ │
│ │ 38 │ │\│ │ │
│ C ├─────*─────┤ ├───────┼────────┤ │
──┴───┘ │/│ │ ┌───┴───┐ │
VD3 *────┤ RP3 ├────*
│ └───────┘ │
VD1-VD3 KD522B O +5V ─┴─
RP1-RP3 = 1K
The signal is taken from the outputs (out) and then mixed as in
corrected Scheme 1.


Summing up the results of the research, two main aspects can be highlighted.
Firstly, the MP connection diagrams can have different transitional attenuation
between stereo channels. In any case, it is necessary to ensure a transient attenuation of more than 18...20 dB for all types of processors.
Secondly, MPs produced in different countries and companies may have different amplitude characteristics and, as a consequence, different nuances in the sound of the same melodies

RadioAmator 8/96
=== Cut ===

Oleg

... Everything is an illusion. Of course, the previous phrase too.