Wiring a voltage divider with coarse and fine adjustment

[icon]

Hi

I know DJ favours 10-turn pots, but I haven't got one - I do have 1k and 10k pots to hand though. But now I come to put pen to paper I'm not sure how best to wire up a voltage divider consisting of a coarse and fine adjustment. My best guess is:

Code: [Select]

+V o¬
     |
     Z
 1k  Z<-¬
     Z  |
     Z  |
        |
        |
        |
        Z
        Z<--- off to op-amp input
  10k   Z
        Z
        |
       gnd


Is that hopelessly wrong? Is there a better way?

Cheers (and sorry about the ascii schematic!)

John

[amspire]

A 10:1 ratio of the pots is probably too small as a single pot can easily be adjusted to 1%. I do not know your application, but if you want maximum resolution from the two pots, I would probably use a 100:1 ratio. You can still use your pots, but put a 100 to 120 ohm resistor across the 1K pot, and connect the lower end of the 1K pot to the wiper of the 1K pot.

Richard.

[robrenz]

Leslie greens analog seekrets  https://www.eevblog.com/forum/general-chat/leslie-greens-analog-seekrets-re-upload/msg86752/#msg86752  has a very enlightening section on proper potentiometer use. 

Richard, I just happend to be reading about this last night.  Page 56 in analog seekrets shows that paralleling a pot with a resistor of 1/10th of the pot value makes it severely non linear and can actually decrease the rotational sensitivity.  What do you think?

robrenz

[icon]

Page 56 in analog seekrets shows that paralleling a pot with a resistor of 1/10th of the pot value makes it severely non linear and can actually decrease the rotational sensitivity.

Yes, I wondered that - I just bunged it in a spreadsheet now and checked it. Most of the variation is compressed to one end of the travel. If you replace the fixed resistor with progressively higher values it becomes more linear, of course - but you lose the ratio-increase. You can never have something for nothing in electronics!

John

[amspire]

Leslie greens analog seekrets  https://www.eevblog.com/forum/general-chat/leslie-greens-analog-seekrets-re-upload/msg86752/#msg86752  has a very enlightening section on proper potentiometer use. 

Richard, I just happend to be reading about this last night.  Page 56 in analog seekrets shows that paralleling a pot with a resistor of 1/10th of the pot value makes it severely non linear and can actually decrease the rotational sensitivity.  What do you think?

robrenz

It is slightly non-linear but not too bad with a 10K load. Do you need a high linearity for this application? If it is just something like a course/fine adjustment for a power supply, it should be fine. If you think of it when the 1K pot is at 10%, and 50%, there will be something like a 4% divergence from linearity (an estimation - didn't bother calculating it). Remember, this is the fine pot, so in most cases, I doubt if you would notice the non-linearity. It will not get in the way of your ability to adjust the final voltage.

Richard.

[robrenz]

Richard,  I was not concerned about the linearity but setability.  If Leslie's graph is reasonably accurate, the problem is the delta Ohm per delta pot rotation slope (setability) gets extremely steep (poor) at one end of the range. It is the same at mid travel and gets extremely shallow (much better) than the bare pot at the other end.  I am not disputing your assertion that the circuit will work fine.  I just find this shunted pot thing very interesting.

robrenz

[amspire]

You are right. I was partly thinking of something else. Better get a 100 ohm pot.

My suggestion will be a lot more non-linear then I suggested.

Richard.

[mariush]

You could get  one of those digital potentiometers with  up/down pins ... so you don't need a microcontroller to change it, just have to use 2 buttons which send pulses to those pins.
The only downside is you need to power it separately with 5v or something like that.

http://eu.mouser.com/search/ProductDetail.aspx?qs=1THa7WoU59EFipBrQDYtaw%3d%3d&cm_mmc=findchips-_-na-_-na-_-na

You should also be able to find similar ones with 100 steps. 64 step ones are definitely easy to find and in DIP formats but 64 steps may be too little for you.

This one above... 128 steps , 22k , 2.7-5.25v, remembers the position set last time  ... kind of a bitch to solder with a simple soldering gun but no problem with large pads or a heat gun.

[Mechatrommer]

Leslie greens analog seekrets  https://www.eevblog.com/forum/general-chat/leslie-greens-analog-seekrets-re-upload/msg86752/#msg86752  has a very enlightening section on proper potentiometer use. 

Richard, I just happend to be reading about this last night.  Page 56 in analog seekrets shows that paralleling a pot with a resistor of 1/10th of the pot value makes it severely non linear and can actually decrease the rotational sensitivity.  What do you think?

robrenz

dude!

first: i'm reading the seekret rite now (on noise section) and you directed me to page 56

second: i just did this shunting trimpot yesterday for adjustable LM317/LM337. 10Kohm trimpot shunted by 1.8Kohm resistor, thats N=5 1/5th of trimpots value because i need a trim from somewhere 1.5K to 0 ohm (1.25-16V adj volt) and i dont have that value trimpot for the model. my finding is its not so that bad, i still can get some resolution even tweaking for 3V, thats at the low resistance side of the trim and according to seekret, its in nonlinear region. but i guess not so bad, considering thats a 1 turn small trimmer and is a badass by itself.

third: as for the OP, to get better resolution, why dont put 3 stage of trim? 10K, 1K and 100-200 ohm? (to get quickly where you want) and also consider what will happen if one trim is damaged (overload etc) and the leg goes opened circuit. a shunt fixed resistor might help. or connect the other leg with the middle leg, in case middle leg goes open, the trim will go to its max ohm instead of opened.

edit: and fourth... looking closely the schema OP provided. if there's enough power from the psu, when you move those trimmer to their very low value, they can easily go poof! i'd rather put a fixed series resistor with calculated value to avoid exceeding trims rating.

[A Hellene]

John, I think that this is what you might be searching for:

Code: [Select]

                                       
        x               x             
        |               |             
        \ .             \             
        /       P1      /             
    P2a \<---/\/\/\/--->\ P2b         
        /       ^  .    /             
        \       |     . \             
        |       |       |             
        |       |       |             
        o       o       o             
                                       
 P1  Potentiometer #1 (Linear, single)
 P2  Potentiometer #2 (Linear, dual)   
 .   Wiper max. setting position       
 o   Connection terminals             
 x   Not connected                     
                                       
Using linear potentiometers you will get a perfectly linear result.


-George

[icon]

Leslie greens analog seekrets  https://www.eevblog.com/forum/general-chat/leslie-greens-analog-seekrets-re-upload/msg86752/#msg86752  has a very enlightening section on proper potentiometer use. 

What a fantastic book! Thanks for the pointer.

John

[icon]

Since I've played with this a bit further, I just thought I'd share my findings. I was interested to know whether in practice I could live with the failings of this circuit:


Where RB is the lower value pot, shunted by RA, and RC is the higher value pot. Taking 'pB' and 'pC' to be the variable proportion of RB and RC (ie between 0 and 1), I think the voltage divider has the two 'legs':

R1= RA||pB.RB+pC.RC and
R2= (1-pC)RC

Sticking the resulting formula in a 2-dimensional spreadsheet table (with some representative values for the various 'R's) shows what's wrong with the circuit:


The fine control works well when output is near full voltage, but as the voltage approaches 0v, the fine control ceases to do anything at all. Obvious really, when you look at the circuit. I've soldered up the pots to check my calculations, and the Real World agrees.

If I actually had a dual-gang pot to hand, I'd try George's idea, but I don't.

Leslie Green is also dead-set against using pots as variable resistors, saying they should be wired as pots 'whenever possible', but I'm stuck for how you'd wire two in series without making one of them a variable resistor.

Regards
John

[grumpydoc]

If you have a negative rail handy you could do this (with apologies for the ASCII art)

Code: [Select]


+VE -------+----
           |
           \
       10R /
           \<--+
           /   |
           |   |
           +---+
           |
           \
        1R /
           \<--+
           /   |
           |   |
           +---+
           |
           +---------> Vout
           |
           \
       11R /
           \
           /
           |
-VE -------+------

Assuming equal +VE and -VE rails Vout goes from 0V to +VE, Choose R according to your requirements, say 10K, 1K and 11k for the fixed resistor.

[thilo]

A few months ago I've seen a blog post showing hire to get a linear coarse and fine adjust by connecting a stereo and normal pot. I can't find it anymore though

[lk]

I found this earlier, chrisw957 made his version of Daves, constant current load, in his schematics he uses 2 x 50K Potentiometers both in "potentiometer" mode, with one of them having a 56k resistor in series on the wiper.

http://www.sleepyrobot.com/wp-content/uploads/2011/05/currentload.pdf


I have simulated it and it works pretty well, but i still prefer the 10 turn pots.

I don't know if this helps you.

-lk

[A Hellene]

A few months ago I've seen a blog post showing hire to get a linear coarse and fine adjust by connecting a stereo and normal pot. I can't find it anymore though

Do you mean something like this?


-George

[icon]

After a few more moments staring at the hand-drawn schematic I posted earlier, it became obvious that if I turned it upside down (making Vin into gnd and vice-versa) that I would get fine control at lower voltages and coarser control when Vout nears Vin. I could even call it a feature.

John