Tektronix 468 Adventures
Tek468 Adventures
While studying the digital internals of the 468 and experimenting with various settings I came across the following 2 issues:
Sometimes a small dot appears on the left side of the display
While showing two WFs on the screen while the Time cursor is active, the 468 shows the Time Cursor dots only on the primary WF. When freezing the display (SAVE) in that state, the Time cursor dots appears on both WFs.
At first, these looked like anomalies, but on further inspection, it turns out these are features of the 468!
The Operators Manual on page 62 describes a procedure to produce a "Ground Reference Dot" (GRD) on the screen, which is the small dot from issue #1. This dot can be used to make DC voltage references. Any change in the vertical mode will clear this GRD.
This GRD will not move when moving the WF with the position control knob, unless SAVE mode is selected. In that case both the WF and the GRD will both move in tandem.
The Operators Manual on page 71, item 14 mentions that the Time cursor dots will be present on both WFs if the SAVE button is pushed. This behavior make measuring phase differences on sinusoidal signals easier.
While looking into this, I found an barely documented feature of the 468: Line Cursor Snapping:
If a Ground Reference Dot was set previously AND
the cursor mode is "Coupled Volts Time" AND
one of the Time cursor dots is moved all the way to the left in fast cursor mode AND
the cursor dot stays there for over 512 mSecs, THEN
The cursor line associated with this cursor dot (coupled cursor mode) snaps to the GRD's vertical position.
This snapped cursor line cannot be moved by the Cursor Potentiometer unless the potentiometer is held to the right in fast cursor mode. In that case, the cursor dot and line will unsnap and move along the WF again. It is possible to select the other cursor dot (and its line) by pushing the Cursor Select button: the other cursor/line can now be freely moved along the WF.
The WF and the non-snapped dot/cursor can be moved using the position control know but the snapped cursor will stay in place, unless SAVE mode is selected.
This feature allows for easier DC voltage measurements than described in the Operators Manual pages 61 and 62.
This feature is documented in the Tektronix Service Manual Volume 1, page 2-12 but not in the Operators Manual (May 1981).
Minor problem discovered
Feed the scope a sine wave of 10 mV amplitude, any frequency. Setup for 1 channel, 5 mV/div. Turn on the coupled VT cursor and move the cursors to the top and bottom of the sine wave. The numeric window should read about 10 mV. Now turn on average mode (AVG). The window still reads 10 mV. Move the Volts knob for the selected channel to 2 mV/div. The numeric window still reads 10 mV. Move the volts knob one more position to 1 mV/Div. The numeric window now suddenly reads about 5 mV/div and, if you turn up the intensity, you'll see that the WF has lots of spikes going to the top of the display. The same effect can be achieved in 2 mV/div by positioning more than 50% of the WF below the bottom of the display.
The special case here is that the 468 only can acquire WFs at an amplitude of 5 mV/div or higher (internal code VOLTAGE4). It is able to display WFs at a lower mV/div setting by amplifying the WF in hardware (called VGAIN). The default value for VGAIN is 0xF0 (240) which means no amplification and applies to any setting >= 5 mV/div. If amplification is needed, a lower VGAIN value is used. This normally works great.
There is however lots of code dealing with this special case and AVG mode. The code wants to use a special VGAIN value of 0xC0 (192), which implies a amplification of 1.25. This is not enough. For example, going from 5 mV/div to 2 mV/div requires an amplification factor of 2.5, which equates to a VGAIN value of 0x60 (96). This value is used in the non-AVG case. Using a value of 0xC0 means that the individual samples need to be rescaled by a factor of 2 (doubled) to get the desired amplification of 2.5. The code actually does this by applying a correction factor when it is calculating the average value for each sample. The code is cryptic (to say the least):
First, the averaged value X is first calculated as 256 * X in 16-bit word (in a clever way), then the value 0xC000 is added, ignoring overflow and the result is doubled, taking care of overflow. The final byte sized averaged value is just the Most Significant Byte of this word (which is the same as divide by 256).
Unfortunately, this code doesn't always work, as per the desciption above. Why this code actually works most of the time is a mystery. Please look at the source code listing I published and search for label L0EDC.
For suggestions and corrections, please contact me at:
