[SOLVED] Ripple on Variable power supply

[naeem.saleem.ns@gmail.com]

I am trying to implement a variable power supply in my circuitry built around OPAMP (LTC1637) negative feedback and NMOS (Farnell PartNo: 2114736) being a pass element. The idea is to supply the positive input (reference voltage) from a 12bit DAC and then vary the output voltage apprx 10mV steps from 2.8V to 18V DC.
When I started using this circuit, I faced few problems so I simplified the circuit top give a fixed reference voltage of apprx
3.3V (62K + 10K) (1/7.2 x 3.3 = 0.458),
Vout= 0.458 x (680+100)/100 = 3.5724V apprx
My input voltage for all the tests is 24V DC
I have connected a load that is variable resistor and it can take current from 16mA to 335mA approx. by rotating the POT
I am using R7 of 1R0 to sense the current. Circuit is designed so that current from 1mA to 350mA could be detected. This would be detected by using MAX9611 current sense IC that gives digital output (Max Full scale range of sense voltage as 440mV across 1R sense resistor is permissible as per datasheet)
The problem I have is the ripple in output voltage
Initially I didn’t have any resistor at C1 (output capacitor)
I didn’t not have C8 across opamp o/p & negative i/p
My C9 initial value was 22nF
I started changing the circuitry and by trying different values of components, the best results I have so far is with 1000uF output cap @ C1. C9 as 47pF & C8 as 12pF. This has given me comparatively better results so far BUT they are still far away from being ideal as they are not acceptable for the type of voltage resolution and current measurement accuracy I am looking for.
The output ripple’s amplitude and frequency increases with increase in load current
Clearly this is not acceptable for the sort of accuracy am after. I was expecting this circuit to work like a normal voltage follower but for some reason it is showing nasty ripple at the output voltage. Please help

I have attached my schematic & oscilloscope traces with this post

 

[dick_freebird]

First question is, is this really ripple (from your upstream power
source) or is it oscillation within the regulator itself? The upper
traces look unlike usual power supply ripple, which tends to be
more triangular than sawtooth. The duty cycle on a switcher
would have to be pretty narrow to get that kind of ripple. So I
am inclined to think that you have a relaxation oscillator on your
hands, under some confitions, and a generally unstable loop
since I see not one flat (or nothing-but-noise) yellow line in
the bunch.

Using a true DC source (battery stack) would put the question
of incoming ripple and its attenuation or gain, to bed.

A current limit too low on the incoming supply, is one way to
get a relaxation oscillation in power circuits.

 

[malli_1729]

I want to know few things about the circuit. What is this topology. No inductor....Do u have any reference for this....If u have tat much big capacitance 1000uf, for currents such as 17mA to 335mA it will any way will have less ripple....
Is there any PWM used.....? this topology may be generating required voltage for this current. But if any transient comes, i don't think it will respond sharply......I prefer to use any standard Buck topology using Voltage mode control or current mode control or any other control methods listed,,,,...

If i won't understand....please explain to me.....

........................................................................................................................................
for an input 24V and Output of let's say 3.3V at current 20mA --> Power dissipated in MOSFET = (24-3.3)*0.02 = 0.4W
for an input 24V and Output of let's say 3.3V at current 335mA --> Power dissipated in MOSFET = (24-3.3)*0.335 = 6.9W
MOSFET should handle this.....

 

[THINKER_KAM_MAN]

just a suggestion

is this a CUK PSU type circuit.
or is it a switcher design if so expect ripple, if not
why are you using a FET and not a transistor?
fets are voltage control devices and they are not all that linear

i think the current sense resistor should be within the feed back loop
and the best way to test the power supply output is to apply a step
change in load then you can see the settling times and over shoot.

 

[naeem.saleem.ns@gmail.com]

Guys thanks for the feedback. The circuit is linear voltage follower i.e., negative feedback of opamp is used to control the voltage at output of NMOS pretty much like it is used in LDO.

@ dick_freebird, yes you are correct the noise is actually coming from OPAMP as it is going to oscillation. I have done some more tests today, and replaced the OPAMP LT1637 by TI-LM7301IM5.
I have also removed C9 & C8. Output cap of 1000uF is still connected (I might reduce this 1000uF to check if it is still working). The circuit with this new OPAMP is appearing much more stable and is not oscillating so far for complete sweep of load.
I am not an expert on OPAMPs but I know some basics, looking at the datasheet both these opamps are Rail to Rail and that is what I am interested in so that my NMOS could be turned ON or OFF directly from OPAMP's output. What could cause these oscillations when using LT1763? As I mentioned before they were present even when no capacitors were present in feedback loop?

 

[naeem.saleem.ns@gmail.com]

Sorry there is type. Original OPM was LT1637 NOT LT1763

New one is: LM7301IM5

@ thinker_Kam_man: yes you are correct about the location of Rsense (current sense) but for this application, I didn't include flexibility of adjusting the current limit on this circuit which I could have done using the internal opamp of MAx9611. As one of the usage of this circuit is to determine if load has a Zener on it. (Different loads to be tested and one of them has Zener on it, so it is kind of automatic detection of type of load connected and then run the test specific to that load) And setting output voltage to 5V will cause 4.7V zener to conduct heavily with current of 300mA & 0.3V drop across R(sense) with existing setup. MOSFET is mounted on a heat sink so it should be able to handle that power dissipation