Phase shift in X-Y mode

[yugu]

Hi guys. I have a 35MHz oscilloscope. When CH1 and CH2 were fed 1kHz sine waves (90 degree phase diff), it showed a circle in X-Y mode as expected. However, I noticed that when I increased the frequency to 1MHz, it no longer showed a circle. I have to change the phase diff to be roughly 120 degree to get a circle.

It seems to me when the frequency of the signals are increased, the two channels are out of phase. Is this normal? If calibration is needed, which part should I calibrate? (I have no service manual and need to look at the schematics to find the trimmers)

[kwass]

Hi guys. I have a 35MHz oscilloscope. When CH1 and CH2 were fed 1kHz sine waves (90 degree phase diff), it showed a circle in X-Y mode as expected. However, I noticed that when I increased the frequency to 1MHz, it no longer showed a circle. I have to change the phase diff to be roughly 120 degree to get a circle.

It seems to me when the frequency of the signals are increased, the two channels are out of phase. Is this normal? If calibration is needed, which part should I calibrate? (I have no service manual and need to look at the schematics to find the trimmers)

What cables are you using to between the generator and the o'scope?  If they are not exactly the same length and/or don't have the same velocity factor you'll see exactly what you're seeing -- the shorter the wavelength (higher the frequency) the more phase shift you'll see.

[nfmax]

I assume this is a CRT 'analogue' oscilloscope? The rated bandwidth applies only to the Y channels, the X bandwidth is always much less. This is because it only has to be enough to handle the fastest sweep speed's sawtooth scanning waveform, which is lower in frequency than the fastest signal component. The physical cause of the difference is within the CRT: the electron beam passes between the Y plates first, then through the X plates (usually there is another electrode between them which is used to control the astigmatism). Because the beam has been deflected up and down by the Y plates, the X plates have to be physically larger so the beam will always pass between them. Larger plate area means more capacitance which means less bandwidth.
The X channel phase shift you see is caused by the roll off of the X channel response

[yugu]

I assume this is a CRT 'analogue' oscilloscope? The rated bandwidth applies only to the Y channels, the X bandwidth is always much less. This is because it only has to be enough to handle the fastest sweep speed's sawtooth scanning waveform, which is lower in frequency than the fastest signal component. The physical cause of the difference is within the CRT: the electron beam passes between the Y plates first, then through the X plates (usually there is another electrode between them which is used to control the astigmatism). Because the beam has been deflected up and down by the Y plates, the X plates have to be physically larger so the beam will always pass between them. Larger plate area means more capacitance which means less bandwidth.
The X channel phase shift you see is caused by the roll off of the X channel response

Thanks nfmax! Exactly, it's an CRT analogue scope. Very interesting explanation!

What cables are you using to between the generator and the o'scope?  If they are not exactly the same length and/or don't have the same velocity factor you'll see exactly what you're seeing -- the shorter the wavelength (higher the frequency) the more phase shift you'll see.

I am using two oscilloscope probes of the same length. My other scope shows a perfect circle using the same cables at 1MHz. So it's not the reason in this case. However, it's a possible cause of this symptom. Thanks, kwass!

[yugu]

What cables are you using to between the generator and the o'scope?  If they are not exactly the same length and/or don't have the same velocity factor you'll see exactly what you're seeing -- the shorter the wavelength (higher the frequency) the more phase shift you'll see.

I am using two oscilloscope probes of the same length. My other scope shows a perfect circle using the same cables at 1MHz. So it's not the reason in this case. However, it's a possible cause of this symptom. Thanks, kwass!

Actually, giving it more thoughts, i think this could be a reason. Because when CH1 was translated into horizontal axis, the signal could travel longer/shorter distance than CH2, which may then cause the phase difference.

Any thoughts of how to verify it?

[Howardlong]

Swap the cables?

[nfmax]

Within the oscilloscope, the delays in each channel are carefully matched. Fancy DSO's allow the channel delays to be adjusted individually to compensate for different cable lengths - usually called 'probe skew adjustment'.
Velocity factor for most coax cables is around 2/3, so cable delay is about 5ns/m, so at 35MHz you get a 90° phase shift from a difference in cable lengths of about 1.4m.
Why not try it and see?

[yugu]

Within the oscilloscope, the delays in each channel are carefully matched. Fancy DSO's allow the channel delays to be adjusted individually to compensate for different cable lengths - usually called 'probe skew adjustment'.
Velocity factor for most coax cables is around 2/3, so cable delay is about 5ns/m, so at 35MHz you get a 90° phase shift from a difference in cable lengths of about 1.4m.
Why not try it and see?

Makes sense. At 1MHz, cable length is unlikely to be the source of the ~30 degree phase shift.

FYI I measured the X-bandwidth in X-Y mode. It's roughly 2.5MHz, much lower than the vertical channel's 35Mhz! It could explain the phase shift quite well.