Reducing digital scale drift

[RickKr]

I found this thread to be extremely interesting and helpful. I have one of those "higher resolution" scales (Ohaus Scout SPX123, 100 gram, ±0.001 gram) and have been very happy with its apparent stability, while not being aware of the auto-zero function, but wondered how it always seemed to remain at 0.00 grains. After playing with the unit's calibration weight (100 grams) and some Lyman check weights, I made a 100 grain "fake zero" weight in the hope of using it when weight charges and getting away from the auto-zero zone. Please let me know if I've understood this thread and if what I've done is appropriate.

Scale was on overnight. My shop is fairly consistent 69-71ºF year-round.

1) Zeroed the scale (0.00 grn).

2) Calibrated with the 100 gram calibration weight (1543.24 grn).

3) Placed a 50 grain check weight to get the scale away from its auto-zero (50.10 grn).

4) Added my new 100 grn fake-zero weight (150.10 grn).

5) Removed the 50 grn check weight, leaving only the new fake-zero weight (100.00 grn).

Limit is 6 photos, but this is a good stopping point and will continue in another post.

Rick

 

[RickKr]

Continuing with a trial of how the fake-zero weight would actually be used.

6) Placed the film canister I use when measuring charges (61.46 grn).

7) Tared it to 0.00 grn.

8) Added the new fake-zero weight (100.00 grn).

9) Added a 20 grn check weight to simulate a powder charge (120.20 grn).

10) Weighed the 20 grn check weight by itself. This has been consistent from the beginning. The other 20 grn weight in the set actually weighs 20.00 grn.


I have the scale set up with the new fake-zero weight to do a time trial of drift. It has only been going for an hour with three readings at 100.00 and one at 99.98 grn.

Rick

 

[Rocketvapor]

I have a couple sets of Lyman check weights and find them to be marginal for precision weighing.
Notice your 50.10 reading.
But, if you keep the 100 grain fake zero, even if slightly off, you can monitor your process.
Sameness has value.
Keep going and see if nothing else, your confidence improves.

I zero with pan, add Fake Zero, and NEVER let the scale go back to zero.
With pan and no charge the scale should always indicate the fake zero.

(I see you are aware of the check weight errors)

 

[RickKr]

Yes, I've noticed that about several of the supposed check weights, off by 0.01-0.04 grns. The 50 grn weight is actually from a Hornardy scale kit, not the Lyman. My intent in using them is for reference and consistency, not accuracy. That the fake zero is not exactly 100.00 grn is as you say, for monitoring the process. I could tell I overshot it slightly when making it, as it consistently weighed between 99.98-100.00. I figure if I keep it on the scale when measuring powder charges and disregard the 100 digit, I'll be more consistent than when relying on the auto zero function of the tared canister. And as you say, my confidence is already higher

Rick

 

[Rocketvapor]

Thanks for demonstrating this in grains.
I normally use grams, but it works with either.

 

[RickKr]

...snip...
I zero with pan, add Fake Zero, and NEVER let the scale go back to zero.
With pan and no charge the scale should always indicate the fake zero.

(I see you are aware of the check weight errors)

Yes, I plan on keeping the fake zero on the pan and never letting the scale go back to zero.

And what could one expect for $50 on the Lyman weights... I looked at some precision check weights, but passed as they were pretty pricey.

Rick

 

[Rocketvapor]

Auto zero isn't the only cause of zero drift.
It will accumulate with time. The Fake Zero will tell you when that happens.
The only place I have an issue is with a battery only, 2 minute auto off scale.

 

[RickKr]

Results of my time trial,
1) to monitor drift with no changes in weights place, and
2) to monitor drift with multiple changes, however, more extreme than when measuring charges.

Readings were very consistent for a few hours, with the Fake Zero (FZ) scale in place after initial calibration. Then I started changing weights between to FZ and adding and removing the 20 and 50 grn weights, in an attempt to simulate changes that would occur during charge weighing, although the magnitudes different a lot more than charge weights would. Displayed weights showed differences on the order of ±0.02 grns. Late in the day, the scale began displaying larger differences, up to 0.06 grn, and what seemed like erratic behavior, which may or may not have been helped with repeated allowing it to go to zero and recalibrations.

Time trial to determine drift of an Ohaus Scout digital scale

Constructed a "fake-zero" weight from stainless steel 1/2" dia. rod, target weight 100.00 grains.
Scale had been on for about 24 hours, powered by AC adapter, no batteries installed.
Scale calibrated using the Ohaus provide 100 gram check weight.
100 grams = 1543.2358 grains, rounded for this trial to 1543.24 grains
Check weight from a Lyman weight set and one from a Hornady scale were used.
Nominal Measured
50 grn 50.10 grn (Hornady)
20 grn 20.20 grn (Lyman)
Room temperature 69-71ºF
Date Time Result (grn) Action/Notes (no entry means no action take, simply read the display)
24-Jan 9:00 AM 0.00 Balance zeroed
9:02 AM 1543.24 Balance calibrated with 100 gram check weight
9:08 AM 100.00 Fake Zero weight place on the pan (referred to below as "FZ")
9:20 AM 99.98
9:30 AM 100.00
10:20 AM 100.00
10:44 AM 100.00
12:00 AM 100.00 Cycles to 99.98 now and then
12:54 PM 100.00 Cycles to 99.98 occasionally
1:03 PM 99.98 Remained continuous for a little while
1:50 PM 100.00
3:10 PM 100.00
3:11 PM 100.00 Lifted and replace FZ without letting scale reach zero
4:13 AM 100.00
4:14 PM 100.00 Lifted and replace FZ without letting scale reach zero
4:15 PM 1543.24 Put 100 gram weight on in place of FZ
4:15 PM 100.00 100 gram removed, FZ returned to pan
5:00 PM 99.98 Flipping between 100.00 and 99.98 frequently
--- 100.00
--- 20.06 FZ removed, replaced with 20 grn check weight
--- 100.00 20 grn removed, FZ returned
--- 150.10 50 grn check weight place, FZ left in place
5:03 PM 100.00 50 removed, FZ retained
5:08 PM 99.98 Alternating between 100.00 and 99.98 a bit
--- 150.10 50 grn check weight place, FZ left in place
--- 100.00 50 grn removed
--- 120.22 20 grn weight placed
--- 100.00 20 grn removed
--- 100.00
--- 120.22 20 grn place
5:12 PM 100.00 20 grain removed
--- 100.00
--- 150.12 50 grn placed
--- 100.02 50 grn removed
--- 120.20 20 grm placed
6:08 PM 100.02 20 grn removed
--- 0.00 Allowed to zero
--- 1543.32 Calibrated with 100 grm
--- 0.06 Allowed to zero
--- 0.00 Scale zeroed
--- 1543.24 Scale Calibrated with 100 grm weight
--- 0.06 Allowed to zero
--- 0.00 Scale zeroed
--- 1543.24 Reweighed 100 gram weight - not recalibrate
6:10 PM 100.00 FZ placed
--- 150.10 50 grn added
--- 99.98 50 grn removed
--- 120.18 20 grn place
--- 100.00 20 grn removed
6:14 PM 99.98
--- 0.02 Allowed to zero
--- 0.00 Scale rezeroed
--- 1543.24 Calibrated with 100 grm
--- 100.00 FZ placed
--- 150.08 50 grn added
--- 99.98 50 grn removed
--- 120.18 20 grn place
--- 99.98 20 grn removed
7:23 PM 99.98
7:25 PM 1543.24 100 gram weight reweighed
7:26:30 PM 100.00 100 gram removed, FZ placed
--- 99.98
--- 0.00 Allowed to zero
--- 1543.24 100 gram reweighed
--- 100.02 100 gram removed, FZ placed
--- 150.10 50 grn placed
--- 120.20 50 grn removed, 20 grn placed

 

[kiwijohn]

So when you have the fake zero drift off, you would re zero with the pan on and then add back on the fake zero weight? Or would you go througf the recalibration process?

 

[RickKr]

So when you have the fake zero drift off, you would re zero with the pan on and then add back on the fake zero weight? Or would you go througf the recalibration process?

More the latter.

After the initial calibration at the start, I didn't rezero or recalibrate for 9 hours, just changed weights without letting the scale return to its "auto zero" state. During that time, except for once at around 5pm, the scale only varied/drifted ±0.02 grns. That one time, it had drifted up 0.06 grns with the 20 grn weight (5:00).

At 6:08 pm, I decided to try letting the scale return to zero and then saw the need to recalibrate, several times. In doing the zeroing and recalibration, it was zero it first, then recalibrate and when it read high by 0.06 grn after being allowed to go to auto-zero following recalibration, it was rezeroed (6:08 and 6:14) prior to further FZ or check weight readings. It got confusing there toward the end. I have never worked anywhere near that long with the scale during any reloading, so hopefully, its much more consistent performance in the first 9 hrs is enough to feel it is reliable in normal use times.

I've been very pleased with this scale, far more so than with the Hornady before (±0.1 grn resolution) or the "antique" Redding beam balances I've tried. I'm also very pleased with having come across this thread to help me tune the weighing process with the Fake Zero. And it seems I've correctly understood the OP's process and found that it works for me as well.

Rick

 

[kiwijohn]

Thanks Rick I am in the process of doing a simple test myself now. I am using a 200 grain projectile as the FZ. The scale has been warming up and I have done a recalibration, still reading 200 grains after an hour, I will check it in the morning.

 

[RickKr]

Thanks Rick I am in the process of doing a simple test myself now. I am using a 200 grain projectile as the FZ. The scale has been warming up and I have done a recalibration, still reading 200 grains after an hour, I will check it in the morning.

Projectile is a good idea, except it can roll. Is it an even 200.0 grns? Not having to deal with stray decimals was one thing I was going for. I never got to the point of trying to hit it by filing a few specs of SS off as it was that close after the last touch of the turning tool on my lathe.

Rick

 

[Rocketvapor]

Ain't this fun?

A heavy load, like 100 grams on a 100 gram scale, left on the scale can cause the load cell to Creep.
Returning to zero for a couple of minutes would let the load cell relax. Leaving the scale powered with the platen, which weighs about 47 grams on mine, and the powder pan, and the Fake Zero should not be an issue.
Even with a Force Balancing scale, the force coils and driver circuit may warm up with a long term heavy load.

The fake zero should be a convenient value with 0000 where your charges would normally display.
The temperature test I did with 1 gram, 11 grams, 111 grams shows a low value and a high value that proves zero is OK, a mid value, and a large value (calibration factor) were OK. This was a 300 gram scale (built with a 500 gram load cell) loaded to 30% of full scale. I did zero before each 3 point check at each temperature. Some temperatures did not have sufficient time to really stabilize though. Scale factor should be very stable or there may be a problem with the scale. I did not have to recalibrate. Also note that I tried to measure scale temperature, not air temperature.

Milligram scales get more bad press than they deserve. 0.02 grains or a kernal of Varget is the accepted precision, not one milligram. Many use grains, even with a scale that displays milligrams as the native unit.
Playing with various techniques to evaluate any new scale should be done every so often especially if the scale was bumped, dropped or run over.

I find it hard to believe that this cheap scale performs as well as it does. 300,000 counts, rechargeable battery or AC powered, and now a way to monitor zero drift, not hide it with software. No ground wires or antistatic mats required.
Those with really expensive scales don't have to worry about zero drift. They can always check the receipt or their bank statement for stability

I had to put the big 300 gram scale away as the wife needed the stove.
I have another cheap scale, 20 gram/milligram coming from ebay and an A&D EJ-54D2 (the Japanese model) have tracking for both. More drift tests on the horizon

 

[JFrank]

I’m confused on the fake zero stuff.

 

[RegionRat]

I’m confused on the fake zero stuff.

Auto-zero or drift-cancellation is a feature designed into some scales.

This type of feature can be intertwined with temperature compensation on a strain gage load cell, or it may be separate.

They are bypassing it by working with an intentional bias load to keep the scale offset with a load that prevents the zero drift software from kicking in.

 

[JFrank]

Got it, thx

 

[Rocketvapor]

Did you zero with the pan and add a 20 grain check weight?

 

[kiwijohn]

@RickKr the project is bang on 200 so it will live with the digital scale now. Crept overnight....no biggie at least I can see the drift now.

 

[JFrank]

Did you zero with the pan and add a 20 grain check weight?

Yes.
Is that bad ?

 

[RickKr]

Yes.
Is that bad ?

I think you have to do that, that is, the container that carries the powder charge must be tared (zeroed using the Tare function if available) and the "Fake Zero", in your case the 20 grn weight, is added to move the scale away from it's auto-zero state. That is what I take as the major point of all of this, or at least one of the major points.

Rick