Results 1 to 11 of 11

Thread: Fine Tuning DOS Calibration

  1. #1
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33

    Fine Tuning DOS Calibration

    Hey All,

    I finally bought a DOS pump with the intention of setting up an AWC system. I went through and calibrated with a lab grade glass 50 mL graduated cylinder. I'm set up to remove 1250 mL from my tank from 7:00 to 7:30 and then add 1250 mL from 7:31 to 8:00. I have a Tunze ATO that is disabled during this timeframe.

    I watched the pump the first day and there is a fairly significant deviation in the water column level in the before and after tank quantity (note: I manually disabled the ATO for the entire timeframe so it isn't skewing the outcome). So I decided to try calibration on a larger scale, thinking that any error is amplified as total dosed volume increases.

    I reset the calibration to 40.0 and moved up to a 250 mL graduated cylinder. I programmed the Apex to dose 240 mL of new salt water over a 10 minute period. I get 243.5 mL. I ran this three times for consistency and each time was within about 0.50 mL by line of site.

    So next, I looked at a ratio of what my calibration numbers should be. Here's my math:

    X mL / 40 mL = 243.5 mL / 240 mL, with X being the number that should be entered in the calibration screen. This gives me X = 40.6 mL. Next, I ran through the calibration program again, discarded whatever 40 mL it tried to dose and entered 40.6 as the actual dosed amount. So I run my 240 mL test again and get a wide variety of numbers, within about 5% of target.

    I reset the calibration back to 40.0 and ran to the original 240 mL test a second time (test that consistently yielded 243.5 before). Now I'm consistently getting 230 mL from a 240 mL dose. This is a 5% error from what I was getting previously.

    I've always heard I could expect accuracy within 1%. 5% seems incredibly high. On a 1250 mL daily water change that will put me +/- 4 oz, which could affect my salinity on relatively short order.

    Is there a way for me to see the calibration multipliers that the Apex is using? Every time I go through the calibrate menu it sets back to 40 mL for the test, making it difficult to make micro adjustments.

    Thanks for any thoughts all,

    CS

    20180811_095420.jpg

  2. #2
    Frequent Contributor zombie's Avatar
    Join Date
    Dec 2013
    Location
    Denver, CO
    Posts
    13,176
    What do you mean when you say reset to 40mL? That isn't an option in the calibration.

    Sent from my SM-G965U using Tapatalk

  3. #3
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33
    All I meant by that is it defaults any bias back to zero when I go into the calibrate menu again. So for example if I set 41.5 previously and it was a tad too much, the multiplier resets back to the default during calibration. In this case I might enter 41.2 and see where that gets me. Lots of trial and error and I'm really struggling to find any linear response to my calibrations. Once the calibration factor has been set the pump seems to be relatively consistent.

  4. #4
    Frequent Contributor zombie's Avatar
    Join Date
    Dec 2013
    Location
    Denver, CO
    Posts
    13,176
    The value in the calibration does not reset. The multiplier stays intact after subsequent calibrations. Also setting a larger number in the calibration should reduce the actual amount pumped since calibration is expecting the amount you see.

    Sent from my SM-G965U using Tapatalk

  5. #5
    Frequent Contributor zombie's Avatar
    Join Date
    Dec 2013
    Location
    Denver, CO
    Posts
    13,176
    Also of note, the +-1% accuracy spec on the DOS is incorrectly defined. It is actually precision not accuracy. The actual accuracy error is the RMS of precision error, measurement error, and calibration error, which comes out closer to +-2% and comes out to about +-3% if you assume compounding error instead of RMS.

    Sent from my SM-G965U using Tapatalk

  6. #6
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33
    That's exactly what I'm finding. The amount seems fairly consistent, but the calibration isn't very linear. Just a second ago I was consistently getting 235 mL from a calibration factor of 41.8. I wanted a little more so I dropped the calibration factor to 41.7. Now it pumps 218 mL. Shouldn't a reduction in calibration factor result in more fluid pumped?

  7. #7
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33
    Quote Originally Posted by zombie View Post
    The value in the calibration does not reset. The multiplier stays intact after subsequent calibrations. Also setting a larger number in the calibration should reduce the actual amount pumped since calibration is expecting the amount you see.

    Sent from my SM-G965U using Tapatalk
    Zombie -

    After re-reading your response, what you said right here might be why I'm struggling with this so much. Let me try and rephrase this and please tell me if this is how it really works...

    I assume the calibration is something like [40 mL dose] * [correction factor]. Initially the correction factor is 1.

    So the first time around calibrating, say for example, it spits out 44 mL, which means it is pumping too much for what it thinks is 1 mL. I enter 44.0 in the display and the correction factor changes to 0.91. Next time it attempts to dose 40 mL, it should spit out 44 * 0.91 = 40 mL.

    Next is what I'm hoping to clarify with you... When I run a subsequent calibration, does it spit out 40 mL at the current correction factor (in this case 0.91), or do we start from scratch? I had been trying to calibrate the unit assuming it sets it back to 1. If it's adjusting from the existing point, I've been going about this entirely in the wrong way.

    I sure wish there was a way for me to see and edit the number that the Apex is currently holding as a correction factor. Compounding variables really makes it difficult to dial in. Especially when we're forced to use such a small quantity for calibration. If Neptune is really documenting this as a solution for AWC, we should be able to calibrate it at or near the volumes it is going to be dosing. Even if I'm only doing 1 L per day, that'll be a 25X error factor for any deviation from a 40 mL calibration.

    Thanks again for all the correspondence Zombie,

    Carl

  8. #8
    Frequent Contributor zombie's Avatar
    Join Date
    Dec 2013
    Location
    Denver, CO
    Posts
    13,176
    Heres an example

    Say you calibrate and it doses 44ml so you enter that. The internal correction factor becomes 0.909. Say you do it again and this time it delivers 41ml and you enter that. The correction factor is now 0.887. The internal correction factor compounds, which is why I always run my calibration at least 3 times to try and fine tune and I will keep doing even more until the meniscus is exactly on the 40 mark. I also dry the graduated cylinder in between calibrations to eliminate error from leftover drops.

    Sent from my SM-G965U using Tapatalk

  9. #9
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33
    Excellent! Thank you Zombie! That was my issue, I was compounding the error by assuming it went back to a zero bias for each calibration. I'm now within 1 or 2 mL between pump 1 and pump 2 over a 240 mL dose.

    I'm a little bit surprised that more people aren't talking about the small sample size in the stock calibration and how any small deviations from 40 mL magnifies greatly at volumes used for AWC systems. For anyone else who might be reading this thread, I used the largest graduated cylinder I had (250 mL) to make my calibrations. I used the following formula to make my calculations:

    (amount desired) / (amount dispensed) = 40 / X

    Then just solve for X.

    So in my situation, I was dosing 240 mL. If I measured an actual dispensed volume of 242 mL, I would solve for X being 40.333. I'd enter 40.3 into the calibration menu (regardless of how much of the 40 mL it pumped). All this does is magnify any error so that it is easier to measure. You could do it with 1000 mL if you had a graduated cylinder large enough to make an accurate reading.

    Anyhow, thanks again Zombie.

  10. #10
    Frequent Contributor zombie's Avatar
    Join Date
    Dec 2013
    Location
    Denver, CO
    Posts
    13,176
    Using a larger cylinder does help as long as the resolution of the cylinder is the same. For example a 50ml cylinder with resolution to 1 ml vs a 250ml cylinder with resolution to 1ml helps a ton. But if you say have a 500ml cylinder with a 10ml resolution, you actually dont gain anything vs the provided 50ml cylinder. The ideal scenario would be a Burette or a low graduation graduated cylinder that is tall and narrow (like a 500ml with 1ml graduations), but these are often expensive and hard to obtain for a hobbyist.

    It is also important to remember that hobbyist grade graduates cylinders have a "graduation range", so you cannot fully trust the level being 40ml just because the meniscus is at 40ml. The only way to be certain is to fill the miniscus to exactly 5ml, and then read the difference between the starting and ending range. Without doing this, there can be up to 2% error introduced simply from that alone because the allowable error from that is one graduation whereas the allowable error between graduation marks is limited to 25% of a graduation (if I remember my chemistry from 8 years ago correctly).

    Sent from my SM-G965U using Tapatalk

  11. #11
    Regular Vistor
    Join Date
    Oct 2016
    Location
    Atlanta, GA
    Posts
    33
    In my case I used a glass 250 mL graduated cylinder with 2 mL graduations and +/- 2% error.

    However, for an automatic water change, I was more concerned with negating any differences between pump 1 and 2 than I was absolute accuracy. I was able to get the two close enough to be within the precision of each pump (I was unable to obtain reproducable results with less variation).

    It's close enough that even if I had a 500 mL cylinder, I don't think I could get much better than I did with the 250. It's a far cry from what the 40 would have been though.

Similar Threads

  1. DOS Calibration
    By ubiq in forum APEX Fusion
    Replies: 1
    Last Post: 01-02-2021, 15:13
  2. Need help tuning MRC Ca reactor
    By kris2001 in forum Apex Programming for Dosing, ATO, and AWC
    Replies: 0
    Last Post: 12-24-2020, 20:19
  3. Question: Tuning a calcium reactor to reduce the day/night alkalinity fluctuation of your reef tank
    By justhes in forum Apex Programming for Dosing, ATO, and AWC
    Replies: 4
    Last Post: 04-25-2019, 09:56
  4. Dos calibration
    By madstyle1 in forum DŌS & DDR – Dosing and Fluid Metering System
    Replies: 4
    Last Post: 01-17-2017, 07:47
  5. Kessil color temps and intensity tuning
    By PhoticZoner in forum Apex Programming for Lighting
    Replies: 3
    Last Post: 01-16-2014, 13:50

Bookmarks

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •