Slope compensation for current mode SMPS 1

[marginal]

hello,

I am doing a current mode buck converter using a UC3842 PWM controller and a low side mosfet.

Input voltage is 160V.

Output voltage is to be 110V.

The load is a long series string of 150mA LEDs.

The duty cycle will end up being 110/160 = 0.69.

-However, this duty cycle is greater than 0.5, therefore i will suffer from subharmonic oscillations.

I believe i can mitigate these oscillations by using slope compensation. (?)

Do you know if this is right? also, i have an application note for a different PWM controller which says that the added slope should be equal to the downward slope of the buck inductor current. (equal in magnitude, not sign, as the slope must be rising)

I would be grateful if readers could possibly verify this.

 

[cabraham]

An added slope equal to the inductor current downward slope is excessive, which overdamps the system. With a duty factor of 0.69, the optimum added ramp slope is around 0.75 of the inductor current downslope. Visit the Ray Ridley site, "Switching Power Magazine" and download his app notes. He covers the slope comp issue in detail and shows the optimum amount needed.

Claude

 

[marginal]

Cabraham

Thankyou very much for your reference…..very good articles there

The oscillations from subharmonic oscillation will always (at lease I think so?) be of too small amplitude on the output current to bother me…

–However, these oscillations are said to possibly cause a high pitched whining in the transformer/inductor , and for that reason alone, I must eliminate subharmonic oscillation.

Unfortunately, I cannot simply reduce my output voltage to ensure getting rid of subharmonic oscillation, as Dr Ridley in his article (below) says that subharmonic oscillation can also happen at duty cycles below 50%

For slope compensation, I picked out the article
“Current Mode Control Modelling” By Dr Ray Ridley
(from logging in to

http://www.switchingpowermagazine.com

)

Though I am not so sure if this article applies to my current mode buck converter, because on page 3 of Dr Ridley’s article, he states that

“current mode control regulates the peak of the inductor current with a control signal Vc”

….But my buck doesn’t use a “control signal, Vc”, but just turns the mosfet off when the current sense resistor voltage reaches 1 Volt.

..Here is my basic schematic…

SCHEMATIC:

LT1241 DATASHEET.

http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1142,C1137,P1335,D1328

..-My circuit does present subharmonic oscillation on the LT Spice simulator (when duty cycle is above 50%).
-The subharmonic oscillation obviously exists because I can see the gate drive pulses being alternately long then short, etc, etc…

Another thing that makes me think normal slope compensation rules might not be applicable to my circuit is page 5 of the following Application Note concerning slope compensation.

LINK TO ONSEMI APP NOTE AND8029

http://www.datasheetcatalog.org/datasheet2/8/0uyr2uyq7gqst5k2pz42a89yuucy.pdf

..pgae 5 says that …

“injecting a compensation ramp diminishes the current loop
gain. This has the same effect as raising R(sense) on the
small–signal point of view.
-As a result, the controller grows its operating feedback voltage VFB (that sets Ip) to impose the same peak current.”

…….However, the above definitely is not my situation……my controller will not grow its “operating feedback voltage” –my circuit has no feedback voltage, -my circuit must not regulate voltage, -only LED current must be regulated.


In fact, an inspection of the electronics on my schematic reveals that adding slope compensation will in fact result in a reduction of my output current……
-since my trip voltage on the current comparator is always fixed at 1 Volt….

….The only way that I will then be able to raise my output current back up again is by reducing the current sense resistor…….
...but this will bring about a reduction in the current sense ramp….which means that I am reducing the effect of the ramp that I will have just injected.

-I will be “robbing Peter to pay Paul”.

The point about my circuit is that it’s purely current mode, there is no voltage feedback.
(-apart from open LED protection, should a LED fail, but this is inoperative under normal circumstances.)

Voltage regulation is not wanted in my circuit, only LED current is to be regulated.

In my buck circuit , the error amplifier in the LT1241 PWM controller is not ‘active’, -it is permanently ‘railed’ …
(…therefore, it would also be pointless for me to add feedback compensation components to it.)

I suspect that the power stage transfer functions presented in Dr Ridley’s article are for voltage, and thus not relevant to me.

I wonder if this article’s words on slope compensation are relevant to me ?

Dr Ridley’s article gives an equation for the amount of ramp to add on the middle of page 6 (Right hand side)

…it is …Se / Sf = 1 – 0.18/D

where:
D = duty cycle
Se = ramp to be injected
Sf = off_time ramp

…I am wondering if this is right for the amount of ramp that I should add ?

Interestingly, at the bottom (left hand side) of page 7, Dr Ridley discusses and illustrates a converter for which just the current feedback loop is closed....
...and states that the current bounces about a bit but is stable.

……interestingly , it then goes unstable with subharmonic oscillation when the voltage loop is then closed.

Apologies for the length of this post.

Do you believe I should inject a compensating ramp?
Also, how shall I deal with the subsequent drop in output current?

Thankyou for listening.

 

[cabraham]

It sounds like you're using the 28C43 family of current mode control IC. The 1.0 volt shut off is a maximum value intended for fault conditions. During normal operation, the PWM is turned off when the sensed current signal Is, attains the value of the control signal Vc. At the start of the clock cycle, Is is at its minimum, or "valley" value, the switch turns on and the inductor current ramps upwards towards the value of the control signal, then shuts off when the Is*Rs signal equals Vc.

The value of 1.0 volt applies only at full load, 100% output current. If the load is 40% of full rated output, then the PWM shut off occurs at the peak value of inductor current corresponding to an average value of 40% of the output.

The link below is a paper I authored for the 11 Oct 2007 issue of Electronic Design magazine. It details what I consider to be the optimium method to generate the slope comp ramp.

http://electronicdesign.com/Articles/ArticleID/16996/16996.html

Best regards.

Claude

 

[marginal]

Thankyou Cabraham,

You're right i will be going to use that family..the cheap UC38C42, the LT1242 as in the above (first post) schematic is very similar.

I am wondering what i am going to do about the control voltage, because my feedback pin is grounded meaning that the current sense trigger_off point is 1 Volt, always?

I am beginning to wonder if slope compensation is going to be possible here, and maybe should i just get the inductor's "dipped" in resin to stop any high pitched whining?

 

[Fluorescence]

You can use slope compensation in a current mode led driver

see

http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1094,C1766,P81080,D26364

it also has a Vc pin for loop compensation.