# Troubleshooting a faulty filter circuit in a PAL CRT display

I have an old CRT monitor (Phillips CM8833) which I am attempting to repair. When displaying a composite PAL signal, it only shows a black and white image, but it is still able to display RGB signals correctly.

The composite signal is filtered into chrominance and luminance portions, with the chrominance signal sent into a TDA4510 color decoder. The U and V output are fed with the luminance (Y) into a TDA3505 control chip (which can also switch between the composite/YUV and RGB inputs, hence the RGB is unaffected). Here's an overview of the circuit:

I see the indicated signals at test points 9 (input) and 8 (luminance) but not at test point (chrominance), which suggests a fault with the filter circuit between points 9 and 7:

Probing between R529 and R554 shows an AC signal of 150mV peak-to-peak, one-tenth of the amplitude at testing point 9. Measuring between C554 and C556 the AC component is down to less than 30mV peak-to-peak, and after C556 (testing point 7) it's indistinguishable from noise.

I've desoldered and tested all of these components individually, and all the resistors and capacitors appear to be within specification. (I've also replaced the TDA4510, as a precaution.) The inductor, S555, doesn't have a value listed on the circuit diagram or the parts list, but it matches the value of the color bands on its side (red, violet, black, gold, making 27uH).

I've been reading up on the operation of PAL systems, and I think that this monitor is using a bandpass & notch filter to isolate the chrominance and luminance signals. (The CM8833 service manual mentions adjusting S533 to calibrate the "4.43MHz blanking circuit in the luminance circuit", which suggests it's using a notch filter.)

C555 and S555 look to be forming a parallel-tuned bandpass filter, but using $$\f= \frac{1}{2\pi \sqrt{LC}}\$$ with 12pF and 27uH I get a resonant frequency of 8.84MHz, twice what I'd expect for a chrominance signal.

I'm left with some questions:

• Is my analysis of C555 and S555 correct?
• Are C554 and C556 serving as decoupling capacitors, or do they have some other function?
• What are resistors R529, R553, and R554 doing?
• It is possible that the input of the oscilloscope reduces the signal. It is better to check the shape of the signal after the buffer cascade( pin 6 of the TDA4510). At the input of the monitor, it is necessary to apply a test signal - colored bars. Apr 14 '19 at 18:47

## 1 Answer

This parallel resonant circuit includes both caps in parallel to tune at 6.0 MHz. This is the sub-carrier channel in PAL for audio.

Both C554 and C556 add in a shunt equivalent cct, (27pF) to resonate with (27uH), S555 in parallel, which is the ferrite tuning slug

The resistors form a 'T' pad attenuator to reduce reactive loading effects on previous source from shifting phase, while attenuating 18dB then raising the load impedance on LC tank to reduce the resonant Q and increase bandwidth, BW of the tank circuit to about 1MHz.

Q = Xc(f)/Rp

The R pad is necessary to prevent reactive loading effects ( phase shift) on the previous stage shared with video yet raises the source impedance to parallel resonance to reduce the Q and sensitivity to component tolerances or drift but resonance still provides a gain of 6dB with a -3dB BW of about 1MHz,

For it work as the Chroma Burst resonator, L555 must be around 52uH

Falstad Simulation in browser

• Thank you for the answer, I'm currently digesting it. But I'm surprised that the filter should be tuned to the 6.0MHz audio carrier as opposed to the 4.43MHz chroma carrier, as it definitely feeds a colour decoder chip. (The monitor in question has a separate connector for audio input, so it shouldn't be mixed with the PAL video at any time.)
– Kaz
Apr 12 '19 at 15:53
• Pin 9 adds another 4pf, and perhaps there is additional stray capacitance. It is also possible that the 27uH inductor may be significantly close to its self-resonant-frequency, raising its effective inductance in the 4-5 MHz region. Apr 13 '19 at 0:03
• @glen_geek yes all input pF or gnd effects must be added to any calculation or simulation Apr 17 '19 at 15:10