1. Home
  2. Electronics Design
  3. Precise voltages from imprecise parts..

Precise voltages from imprecise parts..

Take a random LM285Z 1.24V 1% part. Select a 124K 1% resistor and select a 2.37M 1% resistor from nominal population by measurement, each one to 5 digits. Goal is to get 25V as close to 0.1% as possible.

ASS-u-ME one measures the LM285Z reference voltage in the 20uA to

500uA region,to 5 digits. What is the best strategy to determine that resistor selection? Pick the 124K first or the 2.37Meg first? I calculate 24.703V if all are at the low end of spec, 25.177V if all are at the high end of the spec, and 24.940V if all are at nominal value. I have found that these daze, that the film 1% 2Meg resistors are within 0.3% and nominally high by about 0.2%. I assume other values behave the same way; this makes a target of 25V 0.1% harder to achieve.

Suggestions?

Reply to
Robert Baer
Loading thread data ...

Use three resistors, two large and one small, and you don't have to do the selection, just pick the right standard value.

Getting 0.1% also requires attention to the tempco and aging, of course. (For a 25V supply, I commend the LM329 to your attention.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net
Reply to
Phil Hobbs

I suppose you could write a program to just enumerate all the combinations and find the best. I do that when I need to construct a divider from 2 or 3 standard values.

I did another one for the LT5400 that does all networks of four resistors.

--

John Devereux
Reply to
John Devereux

That's clearly not going to work. Robert is trying to get

0.1% precision using 1% components. Necessarily, at least one needs to be selected --case by case-- after a measurement of the actual error.

Jeroen Belleman

Reply to
Jeroen Belleman

Hi Jeroen

I meant he measures all his resistors, then sees which pairing gets closest. But Phils answer is best ini general though.

--

John Devereux
Reply to
John Devereux

On a sunny day (Fri, 17 Apr 2015 01:32:57 -0700) it happened Robert Baer wrote in :

Use a trimpot. I got myself some cheap Bourns 25 turns a while back. Solves all yer problems... Well almost all.

Reply to
Jan Panteltje

.

Trimpots aren't graduate-student proof. They can always persuade themselves that they can set it better that the guys back at the factory - the factor y technicians may have had better test tools, but the graduate students thi nk they have a better understanding of what they are doing ...

With soldered-in select on test resistors they'd need even more chutzpah.

While I was a post-doc I did modify a photomultiplier power supply to work across a wider voltage range - same upper limit, but I persuaded it to go l ower - but that took more than fiddling a trim pot, and I'm fairly sure tha t I actually improved on the original circuit.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

I've got a current source that uses mostly 0.1% R's. But 10M ohm 0.1% are spendy. I measured a bunch of (much) cheaper 1% metal films. (Stackpole, 1/4 W, through hole) I got this pile that were good to 0.1%. And then discovered that when I heated them to solder them into place the resistance went up by ~0.1% or so.. apparently permanently. You might want to test the 2 meg resistors.

George H.

Reply to
George Herold

Another squaddy-resistant approach is to have binary-weighted resistors, all shorted out with top-layer traces or 0-ohm resistors. You measure the initial voltage and look up in a table which jumpers to remove.

Cheers

Phil Hobbs

Reply to
Phil Hobbs

Thanks; not useful for my app. Need 20uA to 500uA operation.

Reply to
Robert Baer

Err..did i not say "from nominal population by measurement, each one to 5 digits"?

Reply to
Robert Baer

Phil Hobbs' answer is excellent for most uses, but i need this "zener" to operate in the 20uA to 00uA region.

Reply to
Robert Baer

Define "cheap" then add in board space then multiply by 12.

Reply to
Robert Baer

...AND heatsink the leads as well. Prolly same story for the 124K resistors.

Reply to
Robert Baer

Now THAT is something i did in another application; needed so many iterations of different voltages that i wrote an Excel macro to do that and "draw" in the jumpers needed (PCB layout in the sheet), for a resulting PDF appropriately labelled.

Reply to
Robert Baer

On a sunny day (Fri, 17 Apr 2015 20:10:07 -0700) it happened Robert Baer wrote in :

formatting link

Mount vertically :-)

Hell if you need to live from the difference in price of 12 pots adjustment time versus typing yer problem on usenet and doing all the complicatiated math.. Profesional equipment (for that sort of accuracy) is FULL of trimpots and other adjustments. Or in some cases digital pots and ADCs are used, or go digital all the way.

If it is a mass product trimpots are easer and faster than soldering in selected resistors to make a match.

Reply to
Jan Panteltje

It does not change the principle - there are the same 2 resistors as far as the reference is concerned it is just that one of them is a "composite" of 2 in series. The low-value one tweaks the overall value of the composite to get you the division ratio you want.

--

John Devereux
Reply to
John Devereux

I think the OP means the LM329 suggestion rather than the 3-resistor suggestion.

Cheers

Phil

Reply to
Phil Hobbs

BTW, use ONE turn trimpots (learn from Mr. Pease).

Reply to
Robert Baer

Check.

Reply to
Robert Baer

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.