Author Topic: RT9025 - Determining value of feedback capacitor  (Read 1947 times)

0 Members and 1 Guest are viewing this topic.

Offline ArahoTopic starter

  • Regular Contributor
  • *
  • Posts: 74
  • Country: no
RT9025 - Determining value of feedback capacitor
« on: February 28, 2015, 05:24:53 pm »
Hi!

I'm using the Richtek RT9025 for my final year project as a 3.7V 2A (peak) regulator. In the datasheet's Typical Application Circuit for the adjustable Vreg, there is a CF not mentioned anywhere else in the datasheet (see picture 1).

Since there is absolutely no mention of a feedback capacitor elsewhere in the datasheet, I'm somewhat struggling to figure out what it's there for and what value cap would be a good idea. The output cap selection is documented very well, but the feedback capacitor isn't even mentioned outside of the drawing.

A quick analysis of what it seems to do is shunting high frequency components from Vout back to the adjust pin, as far as I can see. This would either cause the regulator to try to adjust for these (maybe overcompensate?), or make it ignore them due to it skipping the adjustment voltage divider, I'm not sure.

If anyone could shine a light on this, it would be really appreciated  :)
(Oh, and also how to attach an uploaded photo attachment inline in posts, that would be nice)
« Last Edit: February 28, 2015, 05:35:28 pm by Araho »
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5226
  • Country: us
Re: RT9025 - Determining value of feedback capacitor
« Reply #1 on: February 28, 2015, 05:52:02 pm »
If you draw the block diagram from the data sheet into the typical application circuit, you will see that Cf basically sets the high frequency rolloff of the gain.  Set too small you will maintain gain until phase shift from other lags in the system will drive the circuit into oscillation.  The larger you set this the poorer your high frequency regulation will be. 

To select a specific value try to establish what frequencies are important for your regulator to control (obvious things like mains frequencies, but also operating frequencies in your equipment and environment, and so on) and pick the capacitor to make a corner (with the voltage setting resistor R1) to satisfy that goal. 

To do this really right you need to understand the frequency response, but a hammer and tongs approach is:  If it doesn't oscillate with your selected cap (over the range of operating conditions (temp, voltage, current etc) that you care about and meets your performance requirements, try again with a quarter the capacitance.  If it still doesn't oscillate you are good to go.  If so, you have some soul searching to do about the tradeoff between stability and performance.
 

Offline ArahoTopic starter

  • Regular Contributor
  • *
  • Posts: 74
  • Country: no
Re: RT9025 - Determining value of feedback capacitor
« Reply #2 on: March 01, 2015, 02:35:18 pm »
Ah, of course! Thanks for a great answer!

So in my application, I won't have any mains frequencies, as the product will be run on batteries fairly away from a lot of other stuff. The pre-regulator feeding this is a switching regulator running at 600kHz, so that will probably be one of these frequencies i should control. The subcircuit the RT9025 is feeding I don't really know the frequency of, but it's a HL6528G, a GSM / GPS-modem, so it will have some high-frequency bursts and processing, as well as some lower clock domains like an RTC.

Other places in my application, I'll have some 12MHz stuff feeding into a MCU, and some lowfrequency stuff, but nothing with much current. As such, I don't think this will radiate a lot and be necessary to consider here? What other typical stuff do I need to look at?

So, in theory, I should set the knee frequency of this somewhat below the lowest frequency I want to filter out, but with as small a capacitor as I can get away with and still maintain stability, to maintain good high frequency regulation, right? And the knee frequency is determined by finding 1/(R1*Cf), right?

For the moment, I've put a 10nF capacitor in the schematic, which sets the cutoff frequency at 1kHz. I'm using only 0603 capacitors, so I will be experimentally checking the stability of the regulator with different values if I have to when I get the boards back from the fab.  :)

(Sorry for being a noob about this!)
 

Offline CatalinaWOW

  • Super Contributor
  • ***
  • Posts: 5226
  • Country: us
Re: RT9025 - Determining value of feedback capacitor
« Reply #3 on: March 01, 2015, 05:12:16 pm »
Actually you want to set the knee above the frequencies you want to control.  This lets the gain of the amplifier help with control.  I haven't really looked into the gain-bandwidth product of the part you are using, but older parts of this type tended to have products around 1MHz, newer ones maybe a factor of 10 higher.  This says you don't really have much capability at your 600 kHz frequency.   I would probably set the knee at around 100kHz as a place to start.

The good news is that higher frequencies are easier to suppress passively.  That is the purpose of the output filter capacitor.  As you get into this you will find that capacitors are not ideal components, and that high value capacitors tend to have more parasitic inductance and resistance, which means they don't filter as well as expected at high frequencies.  For this reason you will see most power supply app notes showing two output capacitors in parallel with widely differing values.  The larger one provides a low corner frequency and helps from there up to a frequency set by the capacitors internal construction.  The second, much smaller one takes over at the highest frequencies.

In principal it is not too difficult to put numbers on this general discussion, but you do need to know the impedance of your load, and the parameters for the capacitors you use as well as the actual frequency characteristic of the regulator.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf