Bode plot

Pixelx said:

Is it worth hiding the transformer in an aluminum housing connected to PE?

Click to expand...

I'd be very cautious about shields and/or filters to the injection transformer.

One of the fundamental assumptions of this method of measurement is that the injection transformer does not affect the loop transfer function at all. The more stuff you add, the more likely you are to violate this assumption.

It's unclear what purpose putting the transformer in a shield would serve in the first place.

This is an old waveform, now I don't have such problems. I took the loop characteristic as you saw above, taking the information about the amplitude.

Can I connect this Y capacitor from the secondary side of the injection transformer to PE?


It's mainly about common mode interference, which adds noise. In my opinion, if I don't affect the disturbing signal, it doesn't interfere at all. In converters, add CAP Y between the converter grounds to reduce common mode currents.

Pixelx said:

. Do you know a better method of determining the phase? And not susceptible to noise?

Click to expand...

Synchronous detection. Also known as Lock-in amplifier. With this method measurements below noise floor are possible.

A notch filter on input sine is possible but captured signals have lots of regulator interference.

Is there 33 kHz sin on blue output? That's the difference.

I think it's time to do some step load tests. Do you have any specs for your error budget?

D.A.(Tony)Stewart said:

Is there 33 kHz sin on blue output? That's the difference.

Click to expand...

Just expectable to get little gain at switching frequency. As mfntioned early in this thread, there's no useful modulator transfer function above 0.5 fsw. Problem of simplified measurement method is response to components not related to excitation signal.

kd502 said:

Synchronous detection. Also known as Lock-in amplifier. With this method measurements below noise floor are possible.

Click to expand...

Yes, as mentioned in post #9, commercial loop gain analyzers are using this method.

D.A.(Tony)Stewart said:

I think it's time to do some step load tests.

Click to expand...

Pro and cons of loop gain measurement versus step response have been already addressed. Most practioneers are happy with step response method for stability check and feedback loop tuning and rarely measure loop gain in frequency domain. Nevertheless loop gain measurement is a valid method, opions about unique advantages are obviously different.
Feedback loop can be stable at small signal and still expose large signal instability, in so far I'd rate step response or similar large signal methods as necessary for stability assessment.

Pixelx said:

Can I connect this Y capacitor from the secondary side of the injection transformer to PE?

Click to expand...

I certainly would not try this unless all the other circuitry on the secondary side is also referenced to PE. Basically that means I would only connect this capacitor to the local GND (which might also be PE). But even that is something I would be very cautious about.

Pixelx said:

It's mainly about common mode interference, which adds noise.

Click to expand...

Fundamental question when discussing EMI issues: What is the source of the noise, and what is the victim you're trying to protect?

Pixelx said:

In my opinion, if I don't affect the disturbing signal, it doesn't interfere at all. In converters, add CAP Y between the converter grounds to reduce common mode currents.

Click to expand...

There are sometimes valid reasons to add capacitance between "grounds," and this is based on identifying the source and victim in the system. In this case, I highly doubt your signal generator on the primary of the injection transformer is either a source or a victim of common mode noise. Thus I don't see any benefit to adding a Y capacitor.