You are wandering around in the dark and the illumination you have provided does not allow us to see any except the broadest details. More illumination is needed. It's 'rather hard' to zero in on your problem(s) with no links to module used or circuit diagram of what you started with. You have an informally defined feedback phase response as you vary voltage and presumably are getting positive feedback across the areas where output varies widely. It's not obvious without knowing your circuit (both topology and component values) why this causes full rail output at one end of the range - your error amplifier apparently "lacks authority [tm] over the lower intended Vout range but this could be for simple and obvious reasons or something arcane. You are liable to be better off limiting the gain of the error amplifier. The LM339 can be compelled to believe it is a linear amplieifer but an eg LM324 is easier to convince. Also, the LM339 has open collector output so is at the mercy of the passive pullup response if it ever wishes to take V_FB high. As this is what turns the output off you MAY have asymmetric on/off drive and depending on what you are using to pull FB up (and we have no idea as [[repeat large hint follows ]] we do not know your circuit or component values) this could be a major factor in your problems, or unrelated to them. One of the many available software oscilloscopes - some of which were recommended in prior discussions on this list - would probably help. PWM input should produce a DC reference that the error amplifier compares to the divided Vout so response rate of the DC ref does not affect stability - but ripple on it may. Your filter is probably OK. It probably is not a major factor in your loop instability, but adding a small capacitor (maybe 1 nF or less) across the upper resistor in your feedback divider from Vout will (may) aid step response. XL6009 datasheet here: http://www.xlsemi.com/datasheet/XL6009%20datasheet.pdf Apparently shameless 95% copy here with irrelevant and interesting occasional stylistic variations http://www.haoyuelectronics.com/Attachment/XL6009/XL6009-DC-DC-Converter-Da= tasheet.pdf Note that Vin max is shown as 32V and abs max is 36V. What the spec is for your IC may or may not match. _________ Your XL6009 circuit **MAY** have looked like the one shown here http://www.electroschematics.com/10625/12v-universal-laptop-notebook-charge= r/ Or some of the zillion here https://www.google.co.nz/search?q=3Dxl6009&espv=3D2&biw=3D1280&bih=3D841&tb= m=3Disch&tbo=3Du&source=3Duniv&sa=3DX&ei=3DyWZmVbGNN6iymAX0g4Bg&ved=3D0CC4Q= sAQ _________ Here is someone using one as a CC LED driver wit an LM358 as the current sensing error amplifier, driving the FB pin. They use an amplifier gain of 13 (12k feedback with 1k to ground on inverting input, and 1 uF (!) negative feedback. http://uphotos.eepw.com.cn/l19900099/pics/e80845a690de29716e533b7cf8713516.= png >From here https://earthelectro.wordpress.com/2014/10/30/the-new-dc-to-dc-circuit/ What they are doing translates almost exactly into what you want if you replace their CC sense voltage input with a resistive divider from Vout. Russell --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .