AC sweep (using PSIM)

Thread Starter

suzuki

Joined Aug 10, 2011
119
Hi,

Just wondering if anyone here is familiar using PSIM simulation software? I'm trying to do an AC sweep analysis/frequency response but having some trouble. If anybody is familiar with this, I'd like to send them my schematic file to look over.

Thanks.
 

Thread Starter

suzuki

Joined Aug 10, 2011
119
hmm, unfortunately I don't have access to this block. I am still using the old ac probe/small signal injection method.

Here is a diagram of the circuit.

 
Last edited:

Ron H

Joined Apr 14, 2005
7,063
Those two grounds (AC and DC) aren't actually the same ground are they?
If they are, then one of the diodes is shorted out.

I don't think an AC sweep (frequency response) is appropriate for a full wave rectifier. If it is like a Spice simulation, the nonlinearity and harmonic generation will be totally lost, because the DC operating point is established first, then the circuit is linearized around that operating point.

Suzuki, what are you really trying to do?
 

Thread Starter

suzuki

Joined Aug 10, 2011
119
Yes, that should be the same ground.

What I want to do is to get the s-domain transfer function of the full bridge diode. Essentially, I have some signal that I want to pass though the diode bridge and see how the gain and phase change due to the diode bridge. Surprisingly, I haven't found any literature on this, so I thought the next step would be to simulate it via ac sweep.

Not really sure if that makes sense, but it's one of those "why not" things I want to try out. If there's another way I can see the gain/phase, I'd love to hear how.
 

t_n_k

Joined Mar 6, 2009
5,455
If you are actually looking at the frequency response of the bridge rectifier plus RC load then that is a more uncertain problem.

Keep in mind that the diodes may operate in discontinuous mode with the presence of a load filter capacitor.

A simplistic view is to consider the main AC supply cycle period to be infinitely long in comparison with the sweep frequency cycle period range. One would then consider one opposing diode pair of the diode bridge to be in quasi-static forward bias. In that case one would have the equivalent dynamic diode resistances in series with the load and the sweep frequency source. So one might expect to see a typical single pole -20dB per decade roll off.

It eludes me as to why one would want to know this ....???
 
Last edited:

Ron H

Joined Apr 14, 2005
7,063
If you are actually looking at the frequency response of the bridge rectifier plus RC load then that is a more uncertain problem.

Keep in mind that the diodes operate in discontinuous mode with the presence of a load filter capacitor.

A simplistic view is to consider the main AC supply cycle period to be infinitely long in comparison with the sweep frequency cycle period range. One would then consider one opposing diode pair of the diode bridge to be in quasi-static forward bias. In that case one would have the equivalent dynamic diode resistances in series with the load and the sweep frequency source. So one might expect to see a typical single pole -20dB per decade roll off.

It eludes me as to why one would want to know this ....???
I'm with you.
If it were a lightly cap-loaded rectifier, the diodes would not cut off, but it REALLY eludes me why someone would want to know that. The DC bias is evidence that this capacitor really is for peak holding, as in a power supply.
 

Thread Starter

suzuki

Joined Aug 10, 2011
119
If you are actually looking at the frequency response of the bridge rectifier plus RC load then that is a more uncertain problem.

Keep in mind that the diodes operate in discontinuous mode with the presence of a load filter capacitor.

A simplistic view is to consider the main AC supply cycle period to be infinitely long in comparison with the sweep frequency cycle period range. One would then consider one opposing diode pair of the diode bridge to be in quasi-static forward bias. In that case one would have the equivalent dynamic diode resistances in series with the load and the sweep frequency source. So one might expect to see a typical single pole -20dB per decade roll off.

It eludes me as to why one would want to know this ....???
I actually really like this idea. So in each half cycle, you essentially just have 2 diode resistances, the load R and C. I think this should give a pretty good idea of the frequency response.

I guess to clarify, i had the dc source there as a test for the frequency response. It should be ac going into the rectifier, and the capacitor is mainly for filtering.
 

t_n_k

Joined Mar 6, 2009
5,455
I tried using PSIM for this 'concept' but obtained a better result using Simetrix.

The attachment shows the simulation model and two frequency response results for two different bridge AC input voltages - namely 2V p-p and 5V p-p. The result of differing diode dynamic resistances is readily observed with the lower cut-off frequency resulting from a lower AC source voltage.The dB ratio is obtained from the load capacitor back to the AC small signal source sensor / isolator.
 

Attachments

Thread Starter

suzuki

Joined Aug 10, 2011
119
Thought I would bump this thread and continue rather than start a new one.

That is a pretty interesting result, t_n_k. Sorry, I did not see it until now as I've been away. The fact that the response changes so much based on the input voltage seems to make it even more difficult to model or at least approximate the bridge rectifier as a single frequency response transfer function.

Since this time, what I have tried to do is the following, although I'm not sure if it is correct, so I'd love to have some input regarding my method.

If we take that circuit you provided, I called the input port \(V_i\) and then output of the diode bridge (before R1 and C1) \(V_d\), and lastly the port across C1 is \(V_o\)

As you suggested in a previous post, consider the path through the diode bridge as a series resistor \(R_d\). Therefore, the total output voltage is given by \(V_o =V_d*Z_o/(Z_o+R_d)\) where \(Z_o=R1/(1+sC1R1)\).

Would this voltage divider then give the correct transfer function? If you solve for \(V_o/V_in\), you get a single pole, but it doesn't appear that I would get the same DC gain values as you did in your SIMetrix model.

Thanks for your help.
 
Top