Franc Zabkar wrote:
I just repaired an old Cirrus Logic VLB graphics card. The problem was
an open RF inductor in one of the RGB outputs. This caused the card to
believe that it was connected to a monochrome monitor. Anyway, I
replaced all three inductors with ferrite beads.
My question concerns the original part which was a 3.3uH
resistor-shaped inductor that looked something like this:
http://www.vishay.com/docs/34056/irf24.pdf
According to the Vishay datasheet, the parts are tested at 8MHz.
It seems to me that something is wrong with the design. Here's my
reasoning.
The monitor's resolution is 1024x768 at 60Hz which equates to a
maximum signal frequency of at least 1024x768x60/2 = 24Mhz. The
inductor's impedance at this frequency would be 2*pi*24*3.3 = 490
ohms. Yet the input impedance of the monitor is only 75 ohms.
Shouldn't this circuit behave like a low pass filter with a 3dB point
at about 4Mhz? FWIW, I didn't notice any improvement in the image
after the repair.
- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.
Franc:
Without seeing the actual schematic, it's hard to be sure, but I bet
the input impedance of 75 ohms is not purely resistive, but also has
considerable shunt capacitance, so the output of the graphics card
impedance is going up with frequency due to the inductor while the
monitor input impedance is going down at the same time. As the monitor
frequency goes much below 75 ohms at some point, the RGB output is
going up, so you are right about the low pass filter effect. There may
be other circuit components that tend to give a higher output with
frequency that overcome some of the low-pass filter effects. I
remember such circuits from tv video design courses I took many, many
years ago.
Also, I don't think that your calculation of the frequency response of
the monitor is right, however, as I don't believe the monitor specs
consist of both an output and no output per pixel, but rather an output
and no output per pair of adjacent pixels. This would reduce the max
frequency you calculated to just 12 MHz. Others may have more to add.
I enjoy your reasoned posts, btw. Whenever I see your name on a post I
always read it.
H. R.(Bob) Hofmann