Poor adjustability of voltage booster based on Vout=Vref(1+R1/R2)

[Back2Volts]

This is my first time playing with one of those $2 Chinese voltage boosters.  This one in particular uses what I think is an SBD628, which has a Vref=600mV.   The actual circuit has the output divider consists of a 100K trim pot from Vout, with a 2K2 fix resistor to groung on the other end.   

I set it up on a breadboard, with a 235R load, trying to get 18V on the output.   I started turning the trim pot while measuring voltage.   After quite some turns, the voltage started to increase, with a reasonable sensitivity, but as voltage increased, the sensitivity grew exponentially, making it hard to set a value.    I soldered a few wires for voltage monitoring on the trim pot, and took a few measurement points.   

Eventually I looked for the IC datasheet and found the voltage setting formula.    With it I went to excel and created a spreadsheet with 1% adjustment steps of the trim pot.    It is incredible how badly it behaves as the pot goes up.   When the Vout at 6V, the Vout sensitivity is around 11% for a 1% trim increment.   At Vout=10, it is 20%; at Vout=12V, 25%; at Vout=14.7V, 31%.   A one% change takes it from 14.65V to 19.23V, making it very hard to adjust.

Are there any tricks to deal with this?    I may hack it replacing the 100K trim pot with two fix resistors and a 5K or 10K one.

Any ideas will be appreciated.       

               

[JPortici]

well R1 could be a resistor + a trimmer so the actual influence of the trimmer is reduced

[Back2Volts]

well R1 could be a resistor + a trimmer so the actual influence of the trimmer is reduced

That is basically the same as I expressed above, replacing the current 100K trimmer (which right now is R1 plus part of R2), with two resistors and a smaller trimmer.    I will probably do that, but that kills the use of advertised range for other projects.     

[Yansi]

The actual circuit has the output divider consists of a 100K trim pot from Vout, with a 2K2 fix resistor to groung on the other end.               

With such setup (supposed the reference voltage is reference ground) the voltage goes LINEAR with the ohm setting on the trimpot. If it does not, you're doing something BAD or it is not wired as you describe.

[Back2Volts]

The actual circuit has the output divider consists of a 100K trim pot from Vout, with a 2K2 fix resistor to groung on the other end.               

With such setup (supposed the reference voltage is reference ground) the voltage goes LINEAR with the ohm setting on the trimpot. If it does not, you're doing something BAD or it is not wired as you describe.

I may be wrong -- newbie here.    I see what you mean.   If I wire one end of a 100K trimmer to Vcc and the other to GND through 2K2 resistor, then the voltage in the trimpot sweeper pin will be linear with respect to the trimpot angle, with some small constant offset due to the 2K2 resistor.     That seems to be as close to Ohms law as it can be.    So if the booster "amplifies" this voltage and sends it to Vout, then Vout has to be linear.

I think the problem comes when Vcc is actually Vout and the sweeper voltage is compared with the 600mV Vref by the booster, resulting in a booster action by which  Vout= .600 * (1 + R1/R2)   where R1 and R2 are the paths from sweeper to V+ and to GND.  That causes some kind of a non linear mess.

Do I make any sense ?    I wish I was wrong because it would make my project easier.

This is the breakout board  http://www.ebay.com/itm/DC-DC-Booster-Conveter-Module-1A-2-24V-Input-/381212404573?hash=item58c207ab5d:g:yJgAAOSwNSxVG-Ft

And I believe this is the datasheet for the IC  http://hfdx.at.ua/_fr/7/SDB628.pdf

By the way, there seem to be a number of booster ICs that follow the same formula, like XL6009...

     

[Yansi]

Your formula is correct. The voltage is proportional to the trimpot value.  But make SURE the feedback sense pin is connected to the top of the 2K2 resistor and the trimpot is connected as a variable resistor (sweeper together with one of the ends).

But please note this arrangement, as linear it is, is a dangerous design practice. Think what will happen if the wiper looses its connection. (Correct would be to use another arrangement, that won't fire the full output, when the trimpot fails).

[Back2Volts]

Your formula is correct. The voltage is proportional to the trimpot value.  But make SURE the feedback sense pin is connected to the top of the 2K2 resistor and the trimpot is connected as a variable resistor (sweeper together with one of the ends).

...

Unfortunately, this is not my design, it is a booster made in China.    The resistors are connected the way they are connected, which is:  The  feedback is pulled from the sweeper of the trimmer as a pot, not a VR.

Code: [Select]

----------  +Vout
            |
            z
 FB     --> z   100K
            z
            z
            |
            z
            z 2K2
            z
            |
----------------   GND

Today I received another version with an XL6009 which appears to have the divider configured as you suggest.   I hope to be able to play with it tomorrow.   

[JPortici]

well R1 could be a resistor + a trimmer so the actual influence of the trimmer is reduced

That is basically the same as I expressed above, replacing the current 100K trimmer (which right now is R1 plus part of R2), with two resistors and a smaller trimmer.    I will probably do that, but that kills the use of advertised range for other projects.     

of course. i missed the last line. point is, trimmer resistor are terrible. period. multiturn are less terrible as they can be adjusted with more precision but their temperature coefficient is terrible compared to all the rest of the surrounding circuit, which is why you want them to do a fine trim and at the working temperature (so it may start not well adjusted but end up adjusted after some minutes

[Back2Volts]

I have been testing the XL6009 based unit, which has the trimmer as a VR instead of a POT.    There is a big difference in behavior for the better, and while adjust-ability it is a bit on the coarse side, it is acceptable for some applications.   

After testing this one, I have gone back to the other one, re configuring it as VR, and this time it behaves much better.   

You can notice in the pictures that I soldered some pin headers to facilitate meter hookup.   The yellow jumper converts the trimmer to VR.   

Any one using this module may want to short part of the trimmer to improve adjust-ability.   Obviously, as JPortici suggests, it would be better to use resistor plus small value trimmer calculated for fine adjustment  to the desired output.