Before activating any of the stimulus outputs, you need
to set them to your particular requirements. You do this
by entering a special Setup submenu for each output you
intend to use: DAC 0, DAC 1, or DigOut. The appropriate
submenu will pop up over the main Stimulus Generator menu.
Each submenu has multiple pages. For DAC 0 and DAC 1, there
are four main pages each, labeled A through D, which control
four different components of the stimulus. You change pages
via CTRL-PgUp or CTRL-PgDn from the DAC 0 or DAC 1 submenu.
For DigOut there are typically 8 pages, labeled A0 through A7,
for the 8 individual bits of the output port. As above, you
change pages with CTRL-PgUp or CTRL-PgDn. However, you may
elect to have four separate SETS of pages (A0-A7, B0-B7,
C0-C7, and D0-D7 corresponding to the different Pg Mode
settings) via the Dig Pg control on the
Sync menu page.
In that case, CTRL-ALT-Pg keys move between sets.
NOTE: The stimulus parameters can only be changed when the
Stimulus Generator menu is active to rebuild the stimulus
buffer. If you go to the X-axis menu and double the
sample rate, for example, the same samples will come out of
the buffer at twice the rate... doubling the stimulus wave
frequency and halving its duration. As soon as you return to
the Stimulus Generator menu the signal will be updated to
match the parameters you had previously set... in this case,
creating twice as many samples to compensate for the doubled
sample rate. The rule is thus to set sample rate FIRST,
then set the stimulus.
Also note that during DDisk recording
or signal averaging you
will be prevented from changing the stimulus. The cursor
won't move to any output control that actually changes the
signal. You can bring up the DAC 0, DAC 1, and DigOut Setup
submenus in order to review the settings, but the cursor will
be blanked or will not move to certain items. It will allow
you to bring up any lower submenus as needed for review only.
When Master is On, all selected stimulus outputs are active,
at the effective sample rate determined by the normal X-axis
(acquisition) rate times the Factor setting.
In Sequential mode, the trigger source is forced to
so each trace or acquisition run will start in sync with the
This is the only trigger source available for
operation (and probably the only one that makes sense most of
the time), but for RTime operation you can also select the
Internal source to trigger on features of the input signal.
This might be needed if the response you are observing is not
highly correlated with the stimulus, such as an increase in
"spontaneous" responses to a long stimulus tone.
In Dynamic RTime mode the Stim trigger
sync can be set to
any repeating component of the signal, either main wave or
If you set Master to On but there are no outputs selected,
there is no change... no outputs are active.
When Master is Off, all analog and digital stimulus outputs
are off and the input trigger source reverts to Internal
instead of Stim. Trigger mode reverts to Auto.
Static signal generation means that the entire signal waveform
is pre-computed and stored in a buffer, and signal output
consists of reading values from the buffer and sending them to
the DACs. This is very efficient if the signal is a tone
burst that repeats exactly, or a continuous wave with a
frequency such that an exact integer number of cycles can be
stored in the buffer.
Dynamic mode allows continuously-changing signals with
arbitrary frequencies and unlimited durations. The signal
buffer is updated in the background such that it always keeps
ahead of the values needed by the DACs. Since it is not
necessary to store the entire signal duration in the buffer
ahead of time, this mode effectively removes the constraint of
buffer size. For example, a noise source generates a sequence
of random values that takes millions of years to repeat.
Static and Dynamic generation modes differ in regard to
Sequential acquisition modes. In Sequential mode, either
Static or Dynamic, the stimulus is turned off between traces
or acquisition runs, so it is appropriate for tone bursts but
not for continuous waves.
In Static Sequential mode, the entire stimulus must fit into
the buffer. The stimulus output begins at the start of each
acquisition pass or trace and continues to the end of the
specified number of acquisition samples N or the end of the
specified stimulus, whichever is greater. Processing and
display are then performed, and the cycle is repeated. If a
stimulus Cycle time has been set via
the Trigger control menu,
the stimulus and data acquisition start will be delayed until
the proper time for the start of the cycle.
If the stimulus is divided into two alternating page pairs or
four alternating pages according to the Page Mode, then all of
these pages must fit into the buffer.
In contrast, Dynamic Sequential mode recomputes the entire
buffer for each trace. Each alternating page or page pair may
thus be as long as the entire buffer. Furthermore, the
stimulus may change between traces. For example, suppose you
set up a single tone burst. In Static Sequential mode this
would be the exact same burst on each presentation, whereas in
Dynamic Sequential mode the tone starting phase would change
between presentations to continue where it left off on the
Whenever any Sequential stimulus output is active, the
Trigger Source is forced to Stim,
making the start of the
stimulus the effective trigger event. (A special
option allows the same effect as the normal
Trigger Source Pulse option,
if you choose not to use the full Digital Output capabilities.)
In Static RTime mode, the stimulus output just repeats
continually. If the Trigger Source is set to Stim, then the
N samples displayed will
be synchronized with the first sample
of the stimulus. (In RTime mode you can optionally set the
trigger source to Intern.) Continuous background waves may
be generated by proper setup of the stimulus parameters, but
the frequencies must be such that an integer number of cycles
fits into the buffer. The StepN frequency control option
makes this easy to adjust, but the frequency resolution will
be governed by the sample rate and buffer size, and will often
be several Hz... rarely an integer number. Since the buffer
contents are static, noise-type outputs are "frozen" and
repeat exactly, giving a "machine"-like quality.
In Dynamic RTime mode, the stimulus samples are precomputed on
a semi-continuous basis, such that they are always at least
1024 samples ahead of the corresponding acquisition samples.
So even though this mode allows signals of indefinite
duration, it only requires a 16 Kbyte
buffer (22 Kbytes with
DigOut). You may to set continuous waves of any frequency
down to the resolution of the system (typically 0.0001 Hz),
and continuous noise has no apparent repeat pattern.
In any mode, you can look at events that take place before
or after the start of a stimulus cycle by setting the
Trigger Delay to negative or positive values,
Activates or disables the corresponding DAC channel output.
The item will be highlighted when that channel is active.
To activate any DAC output, at least one of its wave component
pages must be active or you will get an alert and warning:
'Must have at least one active page per output.'
The default here is the maximum resolution of the DAC, but you
may reduce that to simulate operation on another system, or to
experiment with special effects like dither.
For a dramatic demonstration of dither, start with Bits at
maximum and set up a continuous Dynamic RTime
sine wave of
about 440 Hz on page A with 75%
Level. Next, set up a
continuous White noise
source on page B with 25% Level, and
turn that page Off.
Listen to the output, and you should
hear a clean sine wave.
Now set Bits down to 2. The waveform becomes a staircase of
4 levels (counting 0), and you hear very strong harmonic
distortion. Check the spectrum, and you can see all the
harmonics. But toggle page B on and you again hear a lot of
background hiss, but the sine wave is miraculously
undistorted. You can see from the spectrum that the
distortion components have indeed gone away, not just been
buried in noise. Their energy has been redistributed from a
series of large peaks, down into the lower background noise
You can improve upon this further by using noise with a
different amplitude distribution, which concentrates more of
its levels near zero. This can be done with a
source, but the most common distribution used in digital
audio is triangular. This not only does a good job, but is
also easy to create.
To create a triangular noise distribution, you simply add
together two uniform sources. Cut the page B Level down to
12.5%, and set up an identical continuous White source on
page C, also at 12.5%. With both of these on, the average
noise level is lower by 3 dB, but you still get rid of the
You can verify that the distribution is triangular by using
the Histogram (Hist) option in the
averager control menu.
(Be sure to turn the sine wave off first.)
Invokes the Component Page submenu system for each DAC
channel. Each channel allows adjustment of up to four wave
component pages, each page with controls for
CTRL-PgUp or CTRL-PgDn to move between pages, or ESCape to
return to this Setup item.
The only label on this control is the tiny 'dB' to the right.
This controls the overall level of the DAC output, in
a resolution of 0.01 dB. Maximum output is 0.00 dB, and all
other levels are negative by default. However, you never need
to enter a minus sign for direct entry. Scrolling up gives
greater output, (less-negative dB values). You can change
this default direction with the
This control works by taking direct control of the SB16
attenuator, if present, and making coordinated small changes
to the effective Level
settings of each active component
The SB16 attenuator has steps of 1.5 or 2.0 dB (depending upon
model). This resolution is too coarse for some applications,
but it does provide attenuation down to -84 or -120 dB (-90
or -138 dB via ATN-SB16 with a separate lab-type ADC board).
On the other hand, the individual page Level controls provide
exceptionally fine resolution, but only at high levels. For
example, a setting of 99.88% is equivalent to -0.01 dB. The
resolution of this system is considerably better than 0.001 dB
down to settings of 50% (-6.02 dB).
The problem is that reducing the Level reduces the effective
number of bits. At a setting of just under 50% the highest
bit is never used, so a 16-bit DAC becomes effectively 15.
Basically, a bit is lost for every 6 dB reduction obtained via
Level instead of a true attenuator. By -48 dB the loss would
be 8 bits. This causes increased quantization distortion
since the wave is created with fewer, coarser steps.
The STIM3A Output dB control solves both problems by combining
the best features of each system: The fine resolution of the
Level system is used only to reach values between the steps of
the SB16 true attenuator. For example, to set an overall
attenuation of -0.01 dB the SB16 is set to 0 dB and the
effective Level is set to 99.88%. As the attenuation is
increased, the effective Level is reduced further... but only
until the setting reaches the SB16 step size. At that point,
Level is returned to 100% and the next step of the SB16
attenuator is set. On a model with 2.00 dB steps, the
effective Level never goes below 79.43% at any setting.
All of this is done by the Output dB control transparently, so
you can treat it like a simple attenuator with 0.01 dB
resolution over the entire attenuation range. Note, however,
that this system depends upon the SB16 steps being exactly
1.50 or 2.00 dB. If they are a little bit bigger or smaller
than specified, then the control response won't be smooth down
to the 0.01 dB resolution of the Output dB control. A setting
change of 0.01 dB thus might not give a true attenuation
change of that amount for values that fall near multiples of
the SB16 step size.
The changes that the Output dB control makes to the effective
Levels are also done transparently, so you won't see any
differences in the Level control settings.
If you attempt to adjust Output dB below the most negative
value supported by your SB16, it switches to 'Mute'. In this
state the SB16 attenuator is muted (better than -120 dB in all
models) and the effective Level is set to minimum (0.003%,
effectively -90 dB). You can simply scroll up from there to
get to the quietest active output setting.
When STIM3A starts up, or when a new
setup is loaded, the
Output dB control is set to 'Mute' by default. This is
intended as a safety feature, to prevent an unexpected blast
of sound. You can use the
A:1 parameter in DQA.CFG to change
this behavior so that Output dB jumps to the setup value upon
If you are using a separate non-SB16 attenuator, the Output dB
control works independently over a range of 0 to -6 dB. This
allows you to get fine resolution even with a simple manual
stand-alone attenuator, although it does require a more
cumbersome 2-step adjustment process.
Since STIM3A takes over the SB16 attenuators, the controls in
the SB16 (or ATN-SB16) menu are locked out: They show the
current settings of the SB16 attenuators (only), but you can't
change them there. If you want independent controls, you can
use the A:2 parameter
to tell STIM3A to ignore the SB16 and
provide only a separate 6 dB range.
The overall performance of the Output dB control depends upon
the exact SB16 model, and whether you use the Line or Spkr
output. (In general, you should probably use Line where
possible.) The table below is reproduced from the ATN-SB16
Model: CT1740 CT3600 CT4170
SB16 SB32 PnP ViBRA 16X
Output: Line Spkr Line Spkr Line Spkr
Max Input, Vpp: 2.7 3.3 3.0 3.0 2.3 3.8
Unity Gain dB: -18 -32 -18 -32 -12 -24
Output Noise (broadband) at unity gain, RMS mV:
0.18 0.27 0.12 0.18 0.25 0.50
-1 dB 60 12 29 12 90 75
-3 dB 110 22 53 22 >140 125
-6 dB >140 35 90 35 >140 >140
Left-Right Leakage, dB:
1 kHz >-73 -53 >-70 -54 >-73 -54
20 kHz -62 -49 -67 -58 >-73 -54
40 kHz -57 -50 -66 -63 -70 -56
60 kHz -55 -51 -66 -69 -69 -56
80 kHz -50 -52 -67 >-73 -71 -56
100 kHz -50 -53 -67 >-73 >-73 -58
- Leakage measured as Right output re: 1 Vpp Left input,
with both Left and Right attenuators set to unity gain.
With Right attenuator set to Off, leakage typically
decreases by an additional 20 dB.
- Standard 6-foot mini-plug to RCA paired cables were used
for input and output connections.
- Broadband noise includes high-frequency components from
video and other system sources. No attempt was made to
shield the cards or try other ISA slots.
- CT1740 thumbwheel set to maximum (fully DOWN) for all
- Actual attenuation versus setting was very accurate for
This reverses the effective output connections, such that
the normal DAC 0 signal goes to the DAC 1 output, and the
normal DAC 1 signal goes to the DAC 0 output. This control
and the following Dual controls are provided especially for
psychoacoustic experiments using headphones.
When this is active, the DAC 0 signal goes to both outputs,
and the DAC 1 signal is ignored. This requires both DACs to
be active; if you want to toggle between DAC 0 going to one or
both outputs, set Dual 0 on and toggle DAC 1.
If Dual 0 and both DACs are active, and you toggle DAC 0 off,
the output toggles from both outputs getting the DAC 0 signal
to only the DAC 1 output active with its own original signal.
The Dual 0 control will retain its active status, even though
there is no DAC 0 output, and when you toggle DAC 0 back on it
will again go to both outputs.
When this is active, the DAC 1 signal goes to both outputs,
and the DAC 0 signal is ignored. Otherwise, it operates like
the above Dual 0 control.
Activates the 8 bit digital output stream. To do this, at
least one bit menu page must be active or you will get an
alert and warning:
'Must have at least one active page per output.'
Invokes the Digital Output Setup submenu system, with a
separate menu page for each of 8 output bits. Use CTRL-PgUp
or CTRL-PgDn to move between bit pages, or ESCape to return to
this Adjust item.