LM5575: LM5575 Vout Regulation

Hi Saul ,

I dont have the background on this issue . Can you please send all the details .

---Ambreesh

Hi Ambreesh,

we are using the LM5575 voltage regulator in our UPS reference design. We notice
that during testing (loading the voltage regulator), by sweeping the duty cycle for
more on time (and thus more current/power drawn from the regulator), the device
goes out of regulation as the data that I provided for a particular sweep shows. Now,
this is an existing design that I am supporting and keep running into this issue. I have
tried different compensator values as calculated per the LM5575 datasheet just to
see if I get different (better) regulation … to no avail. However, I did notice that the
original designers installed an ES3B diode (which is an ultra-fast type diode) for D1
in the datasheet (page 14) as opposed to a non ultra-fast type diode. Would this type
of diode affect regulation.

Can you please advise.

Thank you.

Regards,

Saul

Saul ,

You have opened multiple threads with information scattered in them .Please close all the other threads and share all the relevant information in one thread.

I have seen in other threads that you have shared information about the discrete components .What is the switching frequency of the design (share the RT value) . Further Share the complete Schematic . Its much easier that way .

On the debug side , Can you please share the Switch node waveform ? Is the switching stable ? Also your input capacitance seems to be very small . Can you add a bulk capacitance at the input (22uF or so ) .

---Ambreesh

Also i hope you are measuring the output voltage at the output capacitance only and not at the end of wire or so .

---Ambreesh

Hi Ambreesh,

I do not see a 'delete' button to remove the posts as you requested. On another note, I believe that I have solved the issue as confirmed by testing. Apparently, when the LM5575 voltage regulator was designed, they used an ultra-fast diode. The datasheet specifically states to use a non ultra-fast type schottky diode. I changed the diode and now it is working ... no longer losing regulation for the entire load sweep. I also added the 22 uF to the input per your recommendation.

For the R4/C5 selection, is it generally a good idea to select the error amp zero to be a few Hz above the modulator pole?It appears this way from the design example.  In the design example, Fm-p = 240 Hz, so they chose fea-p = 320 Hz. For our design, based on the load, it will generally be with a load of ~10 to 12 Ohms, with Cout = 205 uF.  This give a zero at ~60 Hz.  Should we choose the R4/C5 values to have a zero at ~200 Hz?  This would give us a crossover frequency of ~12 kHz (after cancellation and still having the error amp zero at DC)?

On another note, Figure 19 in the datasheet shows a gain of ~42 dB at 100 Hz. This is supposed to be the overall loop gain. By definition (as stated in the datasheet): Overall loop gain = Mod (dB) + ErAmp (dB) = 14 dB + 20 dB = 34 dB. Am I reading this wrong? Why does it not show 34 dB instead?

Regards,

Saul