Jeff wrote:It was pointed out to me and later confirmed with my Atago PAL-COFFEE that 0.85 is an approximation, one that doesn't hold well over the drip through espresso range. If you're comparing against your own numbers, it's a reasonable approximation. If you're comparing with someone else's TDS numbers then you need to be filtering/centrifuging and using the more complex/standard Brix-to-TDS conversion.

For "me-to-me" comparisons, even Brix is OK.

Yes, that is exactly the spirit of this post. A guidance for one's own brewing reference. In the end, you are the one who taste the coffee. Who else can enforce an exact TDS to be a gold cup? So it is a guidance calibrated to one's own taste.

I have never claim the measurement to be an absolute R.I measurement, so the unit conversion between R.I. to TDS to Brix has no significance in this application.

I just did a few rounds of dilution test. And I am happy to report that a cheap $30 refractometer plus some machine vision can achieve about 0.01% Brix resolution. That is about an order of magnitude better than my eyes can tell. I have done some repetition and data showed that with some careful surface cleaning, temperature stabilization, and loading one can even achieve about less than 0.02% fluctuation among tests of the same sample. The more careful I was, the better the reproduction of the same measured data I could get.

I also did some dilution test by adding 5% of water into my coffee, I can diluted the Brix measurement down for about 4%, which corresponds to about 0.06% Brix, which is well above my detection resolution (0.01% Brix). Hence this simple test verified the proclaimed resolution. The things that need to be improved is on the optics side and sample loading consistency, so it will be more user friendly and resistance to mechanical disturbance.

It is not surprising to achieve the same measuring resolution to those >$300 unit. I can see they are using quite similar approach, based on a simple optics. And they are well worth >$300 price tag considering how much engineering in ME and EE to put them in the nice hand-held package.

The optics used here is pretty rudimentary, hence the limited resolution. But it is appropriate for a consumer-grade device. For chemical/biological sensing, usually less than 1E-6 or 0.0001% is what we need.

Can you elaborate on your dilution proof? I don't understand what you're doing there.

jpender wrote:...this is only in the context of taking numerous readings (he took 35)...

that's a good point, 35 readings is pretty extreme. I'm sure you could figure out a smaller number of samples to get similar accuracy but anything over 3 or 4 is probably a deal-breaker for most. if you need to use TDS for some sort of experiments but can't afford the more expensive refractometers then maybe this is a solution

lessthanjoey wrote:Can you elaborate on your dilution proof? I don't understand what you're doing there.

The concentration and Brix is proportional at small amount of variation. So 5% change on concentration will roughly correspond to 5% change on Brix. Now since I added additional solvent (water) by an amount in mass of 5%, so the concentration was roughly decreased by 5% (or 4.76% if you want more precision). Therefore, the Brix measurement will roughly be reduced by 5% (or 4.23% if you want more precision), which was roughly what I measured. I measured about 4% decreased on my Brix sensor. Since my Brix reading was 1.55%, so 4% decrease would be about 0.06% Brix reduction, and that is well above my detection resolution (0.01% Brix).

I hope that it is not too dense and make sense.

DamianWarS wrote:that's a good point, 35 readings is pretty extreme. I'm sure you could figure out a smaller number of samples to get similar accuracy but anything over 3 or 4 is probably a deal-breaker for most. if you need to use TDS for some sort of experiments but can't afford the more expensive refractometers then maybe this is a solution

You'd have to do a fair amount of experimentation. That blogger didn't sample one coffee 35 times; he sampled 35 coffees of differing strengths once each and compared each one to a measurement from a more accurate device. It's impossible to know from his data whether the mean value was converging in a repeatable way. If he took another 35 samples would he get the same result?

It's also not clear if his protocol was sound. Did he filter the samples? Did he cool them adequately and protect them from evaporation? It's perhaps telling that he mistakenly concluded that a cooling sample results in a changing reading due to evaporation alone.

So who knows? I can imagine you in your kitchen for hours at a time over many days doing tests. And you'd need a reliable reference to validate the accuracy of your results.

Even if you don't care about the accuracy of the value it needs to be repeatable for it to be useful. Getting that answer would take time. What's your time worth?

^^ Are you using refractometer in you brewing process. How about sharing your method instead worrying about other's approach, which you do not have full disclosure to start with? And what is your goal trying to achieve? I think that is more beneficial to everybody.

dukeja wrote:^^ Are you using refractometer in you brewing process. How about sharing your method instead worrying about other's approach, which you do not have full disclosure to start with? And what is your goal trying to achieve?

I have already answered those questions in this thread.

1. No.
2. Oven dehydration. I have described this in detail in another thread on this website.
3. To hopefully find out if what you're attempting has merit.

You are welcome to stay and watch the progress of this little pet project of mine. Apparently, it is only tangent to your purpose, so please be patient and let us make our own meaningful and interested progress here.

What I am interested is
1. Will R.I. measurement (Brix or TDS) useful to my brewing practice and grinder evaluation to ensure cup consistency? => Yes
2. Will such cheap refractometer works for my purpose> => Yes
3. Will I be able to upgrade it with some engineering training I had to a higher level? => Yes
4. May I make a kickstart product out of it to benefit my colleagues? => maybe

I don't know what you mean by "tangent to my purpose". I was sincerely hoping that you would share details. I still hope for that. I'd love to see a photo from your Raspberry Pi camera.

I ordered one of these inexpensive refractometers a few days ago. I think it may work well enough out of the box to be useful for espresso concentration, where a precision of 0.1° Brix is sufficient. If it is possible to get more out of it that would be icing on the cake.

[edited to be less snarky]