Tropic Marin Precision Hydrometrer Corrections - Redux

[BeanAnimal]

Hi Randy - I am going a bit nuts with this and am sorry to drag you back into this (well kind of ). I know this is not chem or your exact field. We can remove the thread and take to DM if needed.

Put simply, I don't like uncertainty, even when it doesn't matter. Kind of an OCD data thing with me. I was building a new calculator for TM temperature conversion and the chart built by Johan Thelander is making me twitch, as I don't know how he built it or verified it. The BRS chart as well. They don't match each other. BRS is more linear than Johan's and appears to be nabbed from some other hydrometer (my guess).

I do know that neither follow the standard/accepted formulas for seawater (UNESCO, PSS-78, Equations of State, etc.) that Simon Huntington, Hamza, me and several other people have built seawater calculators around.

I understand that due to the TM Hydrometer being hand made, that each will have slightly different nonlinearities with regard to density at temperature. So there is that too..


Seeing that you wrote the Reefkeeping.com article so long ago, I was wonder if you are willing to revisit or recall any of the information form Johan, Boomer, etc. regarding this topic. All of the suspects are long MIA in this space.


Just trying to scratch a long time itch here, but understand if you prefer not to participate.

 

[Randy Holmes-Farley]

I’ll look into this and get back here. I’m assuming initially that the expansion of the hydrometer itself can be ignored as being much smaller than water expansion with temp, but we will want to verify that as well, and hopefully it won’t depend on the exact type of glass used since we will not know that.

I recall that TM also had a correction table, but I presume you mean to go beyond its limits or to not assume it is correct?

 

[BeanAnimal]

I’ll look into this and get back here. I’m assuming initially that the expansion of the hydrometer itself can be ignored as being much smaller than water expansion with temp, but we will want to verify that as well, and hopefully it won’t depend on the exact type of glass used since we will not know that.

I recall that TM also had a correction table, but I presume you mean to go beyond its limits or to not assume it is correct?

Hydrometer (non linear, based on materials) expansion, meniscus behavior, etc. Yeah I guess it all matters but doesn't in the scheme of things.

I think my sticking point is simply that I have at least (3) sets of data that don't really match in absolute value or in deviation magnitude as compared to each other. None of it likely matters at all but it is (very) bothersome nonetheless.

My fools errand of a goal is to determine which method is actually realistically closest and if that method is Johan's table, then interpolating it (for no good cause) a few rows and columns in each direction.

I am aware that NONE OF THIS is needed for any reasonable case or accuracy and it is a rabbit hole with no return value other than putting sense to the whole thing and acting as a vehicle for an interesting conversation.

 

[braaap]

Ohhhh this is exciting. Mine is being delivered today because I’m not trusting my BRS refrac or my Hanna. Or my salinity probe. They are all different even when calibrated.

 

[BeanAnimal]

Food for thought.
Here is Johan's table (I think I captured the data points correctly)

Below is chart showing each row of the table, nice and linear, right!

If we use the same data and flip the axis we get a rather odd set of data points, but they all follow the same pattern.

The device has a card in the neck. At 25C, the readings can't be anything other than linear, though conversion to PSU is not linear, it does follow a curve, not the crazy slope variations above.

Here is a quick compare at 25C for the Johan's (Empirical), New (BRS) and UNESCO

So my takeaway is this so far.

1 - I have no idea what Johan did, but it looks like he varied temperature at a few SGs 1.0202, 1.0242, 1.0264 and then interpolated everything in between, with some unknown method, then copied that to each temperature.

2 - it appears BRS just chose some random "well that says hydrometer correction, so let's use it" chart. Supported by the fact that they unwittingly appear to copy/paste everything else they can find and misinterpret it, including my overflow design and operational explanation. The BRS chart is useless and the wrong correction chart.


I absolutely do not want to sit down for hours on end and build my own data, but I am very unsettled with all of this.

 

[Randy Holmes-Farley]

This question appears to be one of those many aspects of science that we just accept, but when one digs down to first principles, is way more complicated than it initially seems. Given how complicated it is, I have no confidence that any company that sells into the hobby gets this exactly right both for temp and salinity effects (and completely ignoring the chemistry variations reefers experience at a fixed salinity). Good enough? Maybe or maybe not, depending on what they actually did.

To generate such a table perfectly, one needs to know the density of seawater as a function of temperature at a bunch of different salinities, and the thermal expansion coefficient of the hydrometer at different temperatures, and other effects, such as accounting for the meniscus (not just knowing where to read, but the added weight of it on the hydrometer if curving up or the subtracted weight if it curves down.

Here's a NIST document with some crazy complicated equations. See Appendix B especially. Equation B7 is pertinent. Appendix A is all about the surface tension effects. Have fun. lol

Loading…

www.govinfo.gov

"from which it is clear that the expansion of the liquid as well as the expansion of the hydrometer glass must be taken into account. "

"When aqueous solutions contact a clean, glass hydrometer, the rising meniscus can alter the hydrometer’s equilibrium position by the equivalent of up to 0.2 % reduction of the liquid density."

Here's a literally worked out example:

Example A3:
Given: A 1.0000 to 1.0500 sp gr 60/60 °F hydrometer was calibrated by the comparison method and at a reading of 1.0350 its correction is -0.0008. The surface tension of the liquid used during calibration was 68 mN/m. The hydrometer mass is 50 g and its stem diameter is 0.5 cm. The hydrometer is being used to measure the density of a liquid with a surface tension of 33 mN/m and the reading of the hydrometer in the liquid under test is 1.0350. What is the specific gravity of the liquid under test?

 

[BeanAnimal]

What is the specific gravity of the liquid under test?



This is discouraging. It indicates a deeper rabbit hole than anticipated. I don't do well with sunk cost fallacy decisions, as they leave the itch. I may trudge forward. However, the easy way out is not lost upon me... returning both the empirical table result and the UNESCO result with a disclaimer and expanding the columns exactly following Johan's predictable plotted values.

Thank you for the NIST document, I have not seen that yet.

I guess my only real burning question is what the heck Johan exactly did to derive his table.

 

[Randy Holmes-Farley]

What is the specific gravity of the liquid under test?



This is discouraging. It indicates a deeper rabbit hole than anticipated. I don't do well with sunk cost fallacy decisions, as they leave the itch. I may trudge forward. However, the easy way out is not lost upon me... returning both the empirical table result and the UNESCO result with a disclaimer and expanding the columns exactly following Johan's predictable plotted values.

Thank you for the NIST document, I have not seen that yet.

I guess my only real burning question is what the heck Johan exactly did to derive his table.

It may be easier to experimentally test some extreme points on the Jochan table than to try to determine if what he did was correct. If those extremes are correct, the whole table is likely pretty good.

IMO, I would:

1. make some 35 ppt seawater, and test it at multiple temps. You can even buy it if you don't want the stress of making it.
2. Cut that seawater with RO/DI to the lowest point in his table (sg = 1.020) and test it at multiple temps (or buy it).

If that holds up, I'd accept the whole table.

There are lots of good standards available from Osil (though I do not know how much they cost). American marine/Pinpoint sent me a couple years ago that I used to calibrate/check my conductivity meters:

Seawater Standards & Salinometers Archives

Salinometers & Seawater Standards for salinity calibration, nutrient analyses, laboratory determination of Dissolved Oxygen & quality assurance.

osil.com

" Please note that in the US, Guildline Instruments is the exclusive distributor of IAPSO Standard Seawater and all other Seawater products that are sold from OSIL."

https://www.guildline.com/oceanography/iapso-standard-sea-water/iapso-p35-standard-sea-water-and-salinity-packs

 

[BeanAnimal]

It may be easier to experimentally test some extreme points on the Jochan table than to try to determine if what he did was correct. If those extremes are correct, the whole table is likely pretty good.

IMO, I would:

1. make some 35 ppt seawater, and test it at multiple temps. You can even buy it if you don't want the stress of making it.
2. Cut that seawater with RO/DI to the lowest point in his table (sg = 1.020) and test it at multiple temps (or buy it).

That is somewhat the plan I had settled on if I decide to go through with it.

I was going to setup a temperature control. Mix seawater and adjust it at 25C to match exactly 1.0264 on the instrument, assuming that the instrument is properly calibrated. Load that into the test cylinder with a volumetric flask and take readings at specific temperatures. Return to 25C, add 25C RO/DI with a volumetric flask and compare the math with the scale and the table. Readings would be done in Sg and then after, converted to PSU using UNESCO equation of state. All sounds easy enough. Just time consuming.


Maybe I can pawn this off on our two chem lab enthusiasts....
@Dan_P @taricha Interest in saving my sanity?

 

[Randy Holmes-Farley]

That is somewhat the plan I had settled on if I decide to go through with it.

I was going to setup a temperature control. Mix seawater and adjust it at 25C to match exactly 1.0264 on the instrument, assuming that the instrument is properly calibrated. Load that into the test cylinder with a volumetric flask and take readings at specific temperatures. Return to 25C, add 25C RO/DI with a volumetric flask and compare the math with the scale and the table. Readings would be done in Sg and then after, converted to PSU using UNESCO equation of state. All sounds easy enough. Just time consuming.


Maybe I can pawn this off on our two chem lab enthusiasts....
@Dan_P @taricha Interest in saving my sanity?

That sounds like a good plan to me.

That's very roughly how I tested swing arm hydrometers for temp correction.

Used 35 ppt seawater at various temps to check how well they corrected for temp effects (quite well, though in absolute terms the values were off).

 

[BeanAnimal]

I am editing this post... I grabbed another copy of the Johan sheet (this time directly from the PDF Randy published in reefkeeping)

Upon graphing these values, they are much more linear on both directions (temperature and sg). This changes a good bit and makes things a bit less confusing and is a lesson in disguise.

Garbage in, garbage out. It would appear that copies of the PDF running around are not the original PDF.

For those of you who do not know. PDFs do NOT contain "text". A PDF is an image. The "text" that can highlight, copy, paste, etc. Is a "layer" of the PDF that is build using (usually) OCR. This can be embedded or done on the fly in PDF readers. You can NEVER trust what you copy/paste or parse out of a PDF to be wholly accurate. It is OCR either on the way on or on the fly on the way out.

I still want to move forward with comparing this to UNESCO and will report back.

Question: Shall I edit the prior posts based on the poorly parsed data?

 

[taricha]

Maybe I can pawn this off on our two chem lab enthusiasts....
@Dan_P @taricha Interest in saving my sanity?

So I recently dipped a little toe into salinity. I was interested in a method that seemed like it would be fairly simple, not too expensive, repeatable, tied to fundamentals, and accurate enough to check other standards/devices against.

https://www.reef2reef.com/threads/p...r-measuring-density-specific-gravity.1066348/

I found the above to be good enough that I feel confident evaluating other standards or devices with it.
It's not really what you are after, but it might help keep you from falling too far into crazy town.

Here is a quick compare at 25C for the Johan's (Empirical), New (BRS) and UNESCO

Question: Shall I edit the prior posts based on the poorly parsed data?

I would be curious to see the table quoted above with your correct values instead.

 

[Reefahholic]

I’ve been using this one. Not sure who made it or how accurate it really is.

 

[Reefahholic]

This question appears to be one of those many aspects of science that we just accept, but when one digs down to first principles, is way more complicated than it initially seems. Given how complicated it is, I have no confidence that any company that sells into the hobby gets this exactly right both for temp and salinity effects (and completely ignoring the chemistry variations reefers experience at a fixed salinity). Good enough? Maybe or maybe not, depending on what they actually did.

To generate such a table perfectly, one needs to know the density of seawater as a function of temperature at a bunch of different salinities, and the thermal expansion coefficient of the hydrometer at different temperatures, and other effects, such as accounting for the meniscus (not just knowing where to read, but the added weight of it on the hydrometer if curving up or the subtracted weight if it curves down.

Here's a NIST document with some crazy complicated equations. See Appendix B especially. Equation B7 is pertinent. Appendix A is all about the surface tension effects. Have fun. lol

Loading…

www.govinfo.gov

"from which it is clear that the expansion of the liquid as well as the expansion of the hydrometer glass must be taken into account. "

"When aqueous solutions contact a clean, glass hydrometer, the rising meniscus can alter the hydrometer’s equilibrium position by the equivalent of up to 0.2 % reduction of the liquid density."

Here's a literally worked out example:

Example A3:
Given: A 1.0000 to 1.0500 sp gr 60/60 °F hydrometer was calibrated by the comparison method and at a reading of 1.0350 its correction is -0.0008. The surface tension of the liquid used during calibration was 68 mN/m. The hydrometer mass is 50 g and its stem diameter is 0.5 cm. The hydrometer is being used to measure the density of a liquid with a surface tension of 33 mN/m and the reading of the hydrometer in the liquid under test is 1.0350. What is the specific gravity of the liquid under test?

Ouch! That’s a brain tweaker for sure. So many variables.

Is it really worth jumping down that rabbit hole or just knowing it’s more about stability than the most precise number?

The Ocean fluctuates a lot, but my OCD also wants to be as close as possible to 1.0264.

I often wonder how far we’re really off even with the Tropic Marin hydrometer which is the best I can find for accuracy. All of the calibration solutions sold in the hobby are garbage. I bought 3-4 different calibration solutions one time, and they all read differently.


This is about as good as we can do, but you still need an accurate chart.

 

[BeanAnimal]

I will post back here once I get everything gathered.

The long of the short is that the table that Randy published years ago (provided by Johan Thelander) is close enough. It has some anomalous data points but is overall pretty close. It appears to be have empirical origins or input that is likely a result of the device's real world thermal behavior.

The table does not exactly match a straight Density conversion table for Sg based on standard seawater equations, but it is close. The mismatch is likely because the mathematically derived table is "ideal" and does not consider the real world properties of the device.

The table provided by BRS is not useable and should be fully ignored.

So in short, either the table or a straight mathematical correction can be used. I will post some data points for both when I get a chance.

To answer your question, none of it is that important with regard to keeping a healthy reef tank. somewhere near a few PSU either side of 35 is likely fine. Using a hydrometer at 24 instead of 25 and not even bothering to temperature correct is likely close enough too... This while exercise is just to answer my unanswered questions, more than anything else.

 

[Reefahholic]

I will post back here once I get everything gathered.

The long of the short is that the table that Randy published years ago (provided by Johan Thelander) is close enough. It has some anomalous data points but is overall pretty close. It appears to be have empirical origins or input that is likely a result of the device's real world thermal behavior.

The table does not exactly match a straight Density conversion table for Sg based on standard seawater equations, but it is close. The mismatch is likely because the mathematically derived table is "ideal" and does not consider the real world properties of the device.

The table provided by BRS is not useable and should be fully ignored.

So in short, either the table or a straight mathematical correction can be used. I will post some data points for both when I get a chance.

To answer your question, none of it is that important with regard to keeping a healthy reef tank. somewhere near a few PSU either side of 35 is likely fine. Using a hydrometer at 24 instead of 25 and not even bothering to temperature correct is likely close enough too... This while exercise is just to answer my unanswered questions, more than anything else.

Agreed. I’ve chased many rabbits down some long holes only to realize later it wasn’t worth the effort.

 

[Randy Holmes-Farley]

Is it really worth jumping down that rabbit hole or just knowing it’s more about stability than the most precise number?

This is more of a science detail thread than one designed to alter what typical reefers do. BeanAnimal states that up front.

 

[Randy Holmes-Farley]

The table provided by BRS is not useable and should be fully ignored.

That part is certainly useful! I'll start a thread at BRS to see what can be done.

 

[Randy Holmes-Farley]

Do you have a link to the BRS table? I cannot seem to find it there now.

 

[Reefahholic]

This is more of a science detail thread than one designed to alter what typical reefers do. BeanAnimal states that up front.

Yeah, definitely a complicated process if you want to account for ALL the variables. Best of luck!