By potentially more accurate I should have instead said potentially more precise, since with the 0 to 10 scale it will be easier for an observer to repeatedly read the refractometer index when coffee with the same TDS are tested. Not quite as good as having some sort of vernier scale but with a similar benefit to make readings more precise. By resolution I was referring to the number of digits displayed on the digital meter.

I understand exactly what the author reported in the article I cited. I know the limits of handheld optical refractometers versus more accurate and precise lab instruments, since I began using them in college and afterward in my work.

I really don't have anything more to add to my previous comments and I'm certainly not looking to argue with those (and perhaps you are one) who have invested a lot of money in things like the VST refractometer and believe the money has been well spent. More power to those folk. If someone gave me one for free or they were under $100, I would probably get one. It would certainly be easier to use than the optical device. I really do know how to use my optical one to get the most accurate and repeatable measurements of which the instrument is capable, and that works just fine for me.

jpender wrote: I guess it depends on what you are happy with. If you have a device that has an uncertainty of ±0.1% TDS, for brewed coffee that will translate into an extraction yield uncertainty of about 1%. So suppose you make coffee, measure it and get an EY of 20%. Then you make another cup and get an EY of 21%. You won't know if the second cup was more extracted or if it's just measurement noise.

FWIW: what I'm happy with is using an optical refractometer more as a trending device. When dialing in a new recipe I ultimately go by taste, but since my taste is subjective I can (and do) fool myself into thinking I've gone too far one way or the other at times. Say I'm trying a new pour-over method, or coffee/water ratio for a bean and I think I'm tasting under/over extracted notes. So I pull out my my optical refractometer to see where my trend is heading from sample to sample; say 20, 23, 26% EY. I know that any one of those numbers cannot be relied upon to be accurate. But the trend, coupled with my taste experience, provides some level of correlation to help me offset any sensory bias as I dial-in a recipe.

Is this a reasonable use of an optical refractometer, or am I all wrong here?

Brewzologist wrote:...So I pull out my my optical refractometer to see where my trend is heading from sample to sample; say 20, 23, 26% EY. I know that any one of those numbers cannot be relied upon to be accurate. But the trend, coupled with my taste experience, provides some level of correlation to help me offset any sensory bias as I dial-in a recipe.

Is this a reasonable use of an optical refractometer, or am I all wrong here?

If that resolution still helps you dial in then it's totally reasonable.

I also have an inexpensive refractometer, one with a 0-18% Brix range. My thought was to use it for espresso where less precision is needed. I did an initial test where I pulled three shots. For each shot I prepared filtered samples for both the refractometer and my kitchen oven. In the past I've mailed samples to people with VST refractometers who were willing to test them. My oven dehydration results matched the VST results to within 0.02% TDS. So it's a reasonably accurate, albeit time consuming, method.

When I compared the optical refractometer readings to my oven results the refractometer readings differed by +0.05%, -0.01%, -0.16% TDS. That was actually kind of encouraging. But when I tested the same three samples the next day with just the refractometer they had all shifted by 0.1 Brix. So something I was doing, maybe related to the temperature, was resulting in inconsistency.

I've been meaning to do some more tests but with the pandemic my wife is working at home and she hates coffee. The smell of coffee samples drying in the oven is definitely not her favorite thing!

Does anyone know of a chart to convert refractive index (nD) readings, rather than Brix%, directly to brewed coffee TDS? Most handheld optical and digital Brix refractometers, as many here have correctly pointed out, have a resolution of 0.1 Brix with a precision or accuracy of +/- 0.2. Clearly VST and Atago, which have one decimal place higher resolution readings, use some sort of program that converts nD to TDS/Brix. I've been looking for a good chart that shows how brewed coffee nD varies with TDS because I do have a digital refractometer that gives very precise and accurate nD readings, and is programmable to create different user defined scales to convert these nD readings to any other liquid, such as brewed coffee.

Pressino wrote:Does anyone know of a chart to convert refractive index (nD) readings, rather than Brix%, directly to brewed coffee TDS?

From the VST patent:

%TDS = A + B*T + C*RI + D*T^2 + E*RI^2 + F*T*RI

A = -7.277050E+02
B = -8.726780E-01
C = 5.460650E+02
D = 9.201850E-04
E = -4.681970E-01
F = 6.648126E-01
T = temperature in °C
RI = index of refraction


http://www.freepatentsonline.com/20110246091.pdf

Wow! Thanks for that formula! Much more handy than a simple conversion table and should make it easy for me to program a TDS scale in my refractometer.

Much obliged!

Instead of programming a new user scale in my refractometer, I decided to put the TDS program into an HP-41, which was easier to do. I simply place a few drops of coffee in the well, get a direct readout in nD (accurate as well as precise to 0.0001), and run that number through the calculator and get a TDS value. It would be nice to compare it to direct readings on a VST or Atago, but already I can see it's more accurate and precise than reading Brix % directly off my handheld optical refractometer.

You need one more decimal point of precision to match a VST. An error of 0.0001 means you have a built-in uncertainty in the TDS of ±0.06%. But it's probably better than your inexpensive optical device and maybe easier to use too.

So cool that someone out there is still using an HP-41!

The one I'm using is a Reichert AR200...it is dead on accurate to the 4th decimal place, because calculates to the 5th decimal place and rounds the 5th digit up or down before outputting a reading (i.e. displays 1.3302 if it reads >/=1.33015 and </=1.33024 and displays 1.3301 if it reads >/= 1.33005 and </= 1.33014. It's actually a pretty accurate refractometer (its 3600 cell optical array helps makes it so, as does the internal programming),and this is a higher count array than in the Misco units. I don't know the internal design of the VST, but I'm pretty sure it's going to be fairly close to the VST, as regards accuracy and certainly precision, but I could be wrong. It's fairly expensive, but I got mine for much less than most of those digital Brix refractometers on Amazon and Ebay.

I've been using my HP-41 since college (I got one, a CV, when it first came out). I even used it to do statistical analysis on reasearch I published while in college... Upgraded later to the CX, and have most of the accessories for it. A bit old-school, but I really like it.

I'm not doubting what you're saying about your refractometer. I'm just plugging the numbers into that formula. If the difference in nD is 10^-4 the corresponding difference in %TDS is 0.06-0.07%. The VST has a precision of 0.01% TDS. So it's not likely going to be as close.

Example:

nD -> %TDS

1.33570 - 1.885%
1.33580 - 1.954%
----------------
0.00010 - 0.069%



https://cdn.shopify.com/s/files/1/0092/ ... ometer.pdf