CS320A-FRA injection point in multilloop system

Hi,

I am using CS320A-FRA for feedback loop stability analysis of SMPS. Our power supply design consist two feedback path as shown in Image-1. As per control IC manufacturer application note(AN-57 Power Integration) they suggest to inject signal as per Image-2.

Here my questions are,
Will it shown correct responce with cleverscope ?
What will be R ref value as per Image - 2?
What will be connection point for channel A and B?

Hi, I am using CS320A-FRA for feedback loop stability analysis of SMPS. Our power supply design consist two feedback path as shown in Image-1. As per control IC manufacturer application note(AN-57 Power Integration) they suggest to inject signal as per Image-2. Here my questions are, Will it shown correct responce with cleverscope ? What will be R ref value as per Image - 2? What will be connection point for channel A and B?

Dear Mr Lalani,
Thanks for posting on the forum. I am assuming that the output is to the left of your diagram, and the part to the right of your diagram is further feedback related circuitry. If this is not the case, please let me know. I assumed but should point out, that the Scope Probe Chan A and Chan B reference points are ground. The Sign Gen is NOT grounded - it is across the reference resistor.

The goal here is to measure the overall Gain/Phase of the Response/Stimulus. To do that all we have to do is inject a Stimulus signal between the Output (which is assumed low impedance) and the Error measurement system (which is assumed to be high impedance). Then we measure the response on the output. You want the Fig1 Concept (see below). So we put a resistor between the Output and the Error System and inject the stimulus across it. As the output is low impedance, the stimulus will be fully presented to the Error system and ADD to the signal from the Output. Chan A measures the Stimulus. The Error system will therefore try and correct both the stimulus, and any errors in the Output signal. We measure the response (Chan B) directly from the Output. As we use a correlated narrow band measurement, we can isolate the stimulus, and the response to it, from any other error being generated (eg an external load or switching noise).

We already have an example using pretty much exactly this topology, and I attach it (made years ago when the FRA control panel was a bit different). We want the combined response to both feedback paths.

So for your application you should wire up your system as Fig 2. In this way you see the over-all response.

Remember with the Cleverscope FRA, the signal generator is isolated, and so it can be used directly without an isolation transformer across the the signal injection point. You will need to cut a track and add in the reference resistor. For best matching it should be 49.9 ohm, but anything from 10 - 100 ohms will work - the assumption is that the feedback system impedance is quite a lot higher than the reference resistor impedance, and that the output impedance is much lower than the parallel feedback system impedance.

If you want to measure the individual feedback paths, you can put the injection resistor in series with each of the two feedback paths.

Good luck with your work.

Bart

Dear Mr Lalani, Thanks for posting on the forum. I am assuming that the output is to the left of your diagram, and the part to the right of your diagram is further feedback related circuitry. If this is not the case, please let me know. I assumed but should point out, that the Scope Probe Chan A and Chan B reference points are ground. The Sign Gen is NOT grounded - it is across the reference resistor. The goal here is to measure the overall Gain/Phase of the Response/Stimulus. To do that all we have to do is inject a Stimulus signal between the Output (which is assumed low impedance) and the Error measurement system (which is assumed to be high impedance). Then we measure the response on the output. You want the Fig1 Concept (see below). So we put a resistor between the Output and the Error System and inject the stimulus across it. As the output is low impedance, the stimulus will be fully presented to the Error system and ADD to the signal from the Output. Chan A measures the Stimulus. The Error system will therefore try and correct both the stimulus, and any errors in the Output signal. We measure the response (Chan B) directly from the Output. As we use a correlated narrow band measurement, we can isolate the stimulus, and the response to it, from any other error being generated (eg an external load or switching noise). We already have an example using pretty much exactly this topology, and I attach it (made years ago when the FRA control panel was a bit different). We want the combined response to both feedback paths. So for your application you should wire up your system as Fig 2. In this way you see the over-all response. Remember with the Cleverscope FRA, the signal generator is isolated, and so it can be used directly without an isolation transformer across the the signal injection point. You will need to cut a track and add in the reference resistor. For best matching it should be 49.9 ohm, but anything from 10 - 100 ohms will work - the assumption is that the feedback system impedance is quite a lot higher than the reference resistor impedance, and that the output impedance is much lower than the parallel feedback system impedance. If you want to measure the individual feedback paths, you can put the injection resistor in series with each of the two feedback paths. Good luck with your work. Bart

Dear Sir,

I have attached complete schematic of circuit. Here suggested injection point is on primery side (on auxilary winding which replicates output voltage and provides control supply or feedback to PWM controller). This path represents combined effect of both secondary side feedback paths, but this combined feedback is in current form. The current in this path correlates output voltage error.

So, We need to inject current signal(stimulus error) in this path, which add to error current. I think we should add offset to injected signal, because supply for error current is DC and current is unidirectional toward PWM controller. Also, both 14 ohm resistors are used as current sense resistors. Here we have to measure voltage across sense resistors (to imagine current waveform) and do our mathematic on that measurments, because finally we are interested in ratio of amplitudes and phase difference between signals.

Now, if we see voltage drop across both resistors are in same polarity but probe connections should be in reverse polarity (otherwise channel A get shorted - both channel Ref should tie together). Channel A is connected reverse, so we get inverted image of current signal that we have to correct before going for phase measurement.

If is there any correction on my understanding, please let me know.

Dear Sir, I have attached complete schematic of circuit. Here suggested injection point is on primery side (on auxilary winding which replicates output voltage and provides control supply or feedback to PWM controller). This path represents combined effect of both secondary side feedback paths, but this combined feedback is in current form. The current in this path correlates output voltage error. So, We need to inject current signal(stimulus error) in this path, which add to error current. I think we should add offset to injected signal, because supply for error current is DC and current is unidirectional toward PWM controller. Also, both 14 ohm resistors are used as current sense resistors. Here we have to measure voltage across sense resistors (to imagine current waveform) and do our mathematic on that measurments, because finally we are interested in ratio of amplitudes and phase difference between signals. Now, if we see voltage drop across both resistors are in same polarity but probe connections should be in reverse polarity (otherwise channel A get shorted - both channel Ref should tie together). Channel A is connected reverse, so we get inverted image of current signal that we have to correct before going for phase measurement. If is there any correction on my understanding, please let me know.