How to get an LED to light when a signal is at 0.0mv?

[Kevin Mitchell]

I'm looking for a circuit that will shine an LED only when a signal is at 0.0mv. I'm no electrical engineer and am trying to figure out what circuit configuration would be the most ideal for this application.

The objective is to use a wheatstone bridge to find matching resistors and have the measurements between two points trigger an LED if it's at 0.0mv
Here's the bridge in a simulator with all resistors at true value - giving 0mv between the two test points.


Original site of idea;
https://www.head-fi.org/threads/tutorial-how-to-match-resistors-up-to-0-01-using-a-cheap-dvm.295983/

Any guidance would be appreciated!
-KM

[GibsonGM]

How about a comparator?  Interesting design question, I'm curious what the sharper minds will suggest!

[anotherjim]

I think it might be easier with balanced supplies and hang the bridge circuit in between.
Send the bridge test terminals to a differential amp. The output will go positive or negative according to the error.
Send the output to a window comparator. Arrange for the "go" LED to light when the diff amp output is in the comparator window.

[Kevin Mitchell]

I think it might be easier with balanced supplies and hang the bridge circuit in between.
Send the bridge test terminals to a differential amp. The output will go positive or negative according to the error.
Send the output to a window comparator. Arrange for the "go" LED to light when the diff amp output is in the comparator window.

Pretty much put the puzzle together there  A window comparator is exactly what I need.

I've drawn up a similar approach to the MFOS transistor matcher since it's really the only comparator design I've sunk my teeth into.


It's a simple design. I know I can tighten things up for better accuracy. Right now it seems to shine when at about +/- .16% and changing the 330 ohm to 100 brings it to about .06%. The goal is to get it to work at .01%.

-KM

[mth5044]

You've probably thought of this, but why not a pre-packed digital voltage display? This one is relatively cheap and appears to have a pretty good resolution on voltage, +/-0.0001V max.

Long link is long.

https://www.amazon.com/DROK-100115-Voltmeter-0-4-3000V-33-000V-Detector/dp/B00DRA5UPU

Edit: long link is shorter, gracias PRR.

[Rob Strand]

It's a simple design. I know I can tighten things up for better accuracy. Right now it seems to shine when at about +/- .16% and changing the 330 ohm to 100 brings it to about .06%. The goal is to get it to work at .01%.

When you start getting into things less than say 0.1%  you have to start considering the opamp offset voltage and input bias current and input offset current.     Also the test window needs to be very small and noise will make the LEDs flicker so you need to add filtering to your circuit.

You would probably be better off measuring with a multimeter then sorting the values to find parts with a suitable match.    It's also a good idea to measure on the same day as some meters will produce different readings in different ambient temperatures.  The resistors themselves vary with temperature.  You might see 10ppm to 300ppm per degree C.    100ppm per deg C is 0.01% for 1 degC change so even holding the resistors too long will cause the resistance to drift.

[ElectricDruid]

The goal is to get it to work at .01%.

Excuse me for being so bold, but what on earth do you need such accuracy for? Is it *really* required, or could you relax the requirement if you were able to improve some other part of the circuit? I mean, if this circuit includes any capacitors, transistors or even op-amps (unless perhaps you choose very carefully) your 0.01% is out of the window...so why fuss so about the resistor values?

(basically I'm fishing here, to see if there isn't another possible solution )

T.

[PRR]

> Long link is long. https://www.amazon.com/DROK-100115-Voltmeter-0-4-3000V-33-000V-Detector/dp/B00DRA5UPU/ref=asc_df_B00DRA5UPU/?tag=hyprod-20&linkCode=df0&hvadid=309722091285&hvpos=1o12&hvnetw=g&hvrand=6369312256812670258&hvpone=&hvptwo=&hvqmt=&hvdev=m&hvdvcmdl=&hvlocint=&hvlocphy=1013962&hvtargid=aud-801381245258:pla-599334115364&psc=1&tag=&ref=&adgrpid=60439548223&hvpone=&hvptwo=&hvadid=309722091285&hvpos=1o12&hvnetw=g&hvrand=6369312256812670258&hvqmt=&hvdev=m&hvdvcmdl=&hvlocint=&hvlocphy=1013962&hvtargid=aud-801381245258:pla-599334115364

Tip: when taking an Amazon sales-page URL, strip out "/ref=...." to the end.
https://www.amazon.com/DROK-100115-Voltmeter-0-4-3000V-33-000V-Detector/dp/B00DRA5UPU

Site, item name, "dp", Amazon number.

All the rest of that stuff is "tracking". Where you have been on Amazon, how you got to that page (so they can figure something else to sell you). Maybe even who you are. (If all of us hit "your" link, Amazon may think you really-really want it.)

I have been stripping at "/ref=..." since the 1990s (for giving links to professors' books and recordings).

If YOU earn a *commission* from Amazon links, some part of that over-long URL links to your kick-back account. You would want to suss that out. I never troubled to try to make pennies from linking to professors' books.

[mth5044]

Many thanks, I will get rid of that garbage now!

[Kevin Mitchell]

Excuse me for being so bold, but what on earth do you need such accuracy for?
T.

I had a feeling you'd chime in with this comment 
It's for the summing resistors in a Sequential Pro One (six per each VCO) I'm not sure why the build doc request such tight specs - but it does.

-KM

[ElectricDruid]

Excuse me for being so bold, but what on earth do you need such accuracy for?
T.

I had a feeling you'd chime in with this comment 
It's for the summing resistors in a Sequential Pro One (six per each VCO) I'm not sure why the build doc request such tight specs - but it does.

-KM

Aha! Well, this I have experience with! Keeping a Pro-One in tune isn't entirely easy, even with only two oscillators! Getting the 1V source that is used for the octave switching spot-on is key, otherwise you can tune it all up perfectly, and then you change octave and MEH! It all goes out again. So make sure that's right.

It's true, the original schematics do say 1% resistors with a little note (A) and (B) which at the foot of the page seems to say "matched to 0.1%", but (at least on my copy) it's pretty hard to read.

So they're looking for 100K resistors matched to within 100R. Not impossible to find, and presumably cheaper at the time for them to select them rather than buy them. I doubt that would still be the case for commercial quantities.

[duck_arse]

why not just use bridge with a centre-zero needle type meter? then you will have some idea of how far off and in which direction.

[Kevin Mitchell]

So they're looking for 100K resistors matched to within 100R.

First time I needed .1% resistors I just went and bought them. But with this circuit and me being a through-hole snob I've decided to get used to hand matching - especially when it comes to matching to even closer specs. You are right that the original schematic specifies 1% resistors there. But the clone document specifies .01% - likely in efforts to keeping it in tune!

why not just use bridge with a centre-zero needle type meter? then you will have some idea of how far off and in which direction.

Where's the fun in that? 
I could get by no problem just using my meter to match these but I'm a sucker for status LEDs.

Here's my MFOS transistor matcher



Ultimately this idea will come together as an all-in-one circuit for matching parts for synth builds - depending how often I'll be needing it.

-KM

[ElectricDruid]

You are right that the original schematic specifies 1% resistors there. But the clone document specifies .01% - likely in efforts to keeping it in tune!

That's not just the clone document - the original schematic does have a little note saying that the 1% parts were matched at the bottom of the sheet.

Full schematics here:

https://www.cyborgstudio.com/sequential-pro-one

(if anyone finds a better link to the Pro-one schematics...)

[PRR]

0.06% might be musically excellent?

Pre-baked resistors at Mouser are $8 for 0.01%, $3 for 0.05%, $0.50 for 0.1%. So there is some room for hand-sorting.

[skyled]

I know this is a DIY forum, but Behringer is going to release their Pro-1 clone soon for $299...

[Kevin Mitchell]

I know this is a DIY forum, but Behringer is going to release their Pro-1 clone soon for $299...

Yeah they seem to tail whatever synth projects I'm fixin' for 
I admire their efforts but I'm in it for a real-deal clone. Even though their teaser looks very promising it may fall short just as their model D had.

Pre-baked resistors at Mouser are $8 for 0.01%, $3 for 0.05%, $0.50 for 0.1%. So there is some room for hand-sorting.

Tryin' to take the fun out of it I see lol. I'd rather save $100 and sort though my inventory for matches. Lord knows I have enough resistors to find my candidates.

(if anyone finds a better link to the Pro-one schematics...)

Original;
http://www.synthfool.com/docs/SequentialCircuits/Pro_One/
Clone;
https://www.westonaudio.com/page1.html

-KM