Making —Yellow watercolour paint



Current procedure — last updated 7. August 2022

  1. Mix the watercolour paste
    1. Wear a mask. Go outdoors - this can be very messy.
      Pour the medium into the stainless steel container first.
    2. Then pour in the powdered pigment.
      Use a hand whisk to spin the pigment into a nasty paste. Yellow is heavier and less prone to creating dust.
    3. Start dispersing, using a dispersion disc in a drill press.
    4. Cover the bowl while dispersing to limit the dust.
  2. Mill the paste
    1. Set both rollers to 6mil. Set the fast roller a little narrower.
    2. Prepare a large bowl to use. Find a stand to raise up the bowl to catch the paint to reduce splashing.
    3. Use a toothbrush to clean the rollers during the runs.
    4. Put through the mill 10 times.
    5. Test for grind at the end of run #4, #6, #8, and #10.
  3. Package into pans
    1. Line the bottom of 2 cookie sheets with newspaper
    2. Arrange the 600ish empty 2ml half-pans into rows
    3. Fill the pans to the top using a pipette set to 2.2ml
    4. Allow to dry, and then top off
    5. Allow to dry, and then package into wax paper bags

Drapak’s manufacturing notes

6-7. August 2022 - 2 batches of 1000ml medium to 160g of pigment
Py74 powdered arylide yellow pigment: 320 grams × $0.080/g = $25.60
Watercolour medium 2000ml × $26.73/4 cups (1000ml) = $53.46
Number of pans filled 700 pans × $0.08/pan Cost of pans = $56.00
Total for materials = $135.06 Material cost/pan = $0.193
Total time 630 minutes
Final grind size 10-20µm
Time spent per pan 0.90 minutes × labour rate $0.445/min= $0.401
Total cost/pan = $0.593

I learned a lot from this run.

For mixing and pre-dispersion, I learned that for yellow, it may not matter a whole lot. I don't think that using the shaker bottle would be helpful for this pigment, but I don't think that using the drill press for pre-dispersion mixing helps a huge amount, either. I would say simply dump the pigment into the beaker and mix away.

Milling was a different story. Using the hopper with a medium nozzle made a big difference. Part of the problem in the past with yellow was that flocculated particles would settle to the bottom of the bowl when I was pouring the paste into the mill. These particles would be dry in the centre, and so when they would shear in the rollers they would suck up water and create a dry paste.

With the hopper, I can give the paste a quick stir and pour it in quickly. As it goes through the hopper, the paste that settles is also partially mixed with the wettest paste, so the flocculated particles have lots of water to draw on. The paste gets a little wetter as it goes, so any particles that remain in the rollers from earlier in the run can be washed out.

The upshot is that yellow mills faster and with higher quality and less frustration.

Dispensing was frustrating at first. I was trying to top off some of the partially filled pans from the last run. This required 1.300ml, but with only a 1.000ml max pipette and a 2.000ml min one, this meant needing two shots per pan to top off. God it tool forever. However, after a while, I realized that I can draw a 10ml of paint into the large pipette and then use that to top off several pans at once. That probably cut my dispensing time by 30-50%.

I need to create a name tag for the foil pans to help distinguish which pans are for school and which are for Drapak. It took a little head scratching for cyan. Everything else is so well labeled.

I also would like to create a dispensing robot to help me dispense the paste into the pans. The time it takes is considerable, and it is the least enjoyable part of the process. I am starting to research things like peristaltic pumps to be able to help me build one.

I increased the number of runs through the mill to 10. The paint tests suggest that the extra runs are worth it. I am also thinking of making a standard template for testing the paint that can be archived.

I also tried to do two batches at once to see if it saved any time. The answer is certainly yes. Much of that extra time was taken up in this run because of trouble with pipetting, but I think I have that figured out now. I think I will now try to do double runs where possible.

23-29. July 2022 - 1000ml medium to 200g of pigment
Py74 powdered arylide yellow pigment: 200 grams × $0.080/g = $16.00
Watercolour medium 4 cups × $26.73/4 cups (1000ml) = $26.73
Number of pans filled 375 pans × $0.08/pan Cost of pans = $30.00
Total for materials = $72.73 Material cost/pan = $0.194
Time to mix 0 minutes
Time for pre-dispersion 90 minutes
Time to mill 180 minutes
Time to dispense 80 minutes
Total time 350 minutes
Final grind size 10-15µm
Time spent per pan 0.93 minutes × labour rate $0.445/min= $0.414
Total cost/pan = $0.608

I tried a smaller amount of pigment for making this batch of yellow. I also tried to mix it longer on the drill press. Of these, I would say that using the smaller amount of pigment was the most useful.

Measuring and mixing the batch was easy and pretty clean. The pigment has very little airborne dust. It is there if you do the work in sunlight, but it is not much.

I actually tried mixing the pigment for much longer. I probably mixed the materials for three hours or more. The paster was a little smoother, but not much.

The problem seems to be that the pigment really wants to flocculate. It is very grainy, and even after being mixed with a dispersion blade for hours, there was still a lot of fine grains. These grains tended to settle to the bottom of the paste. When milled, it tends to build up in the slow rollers, and create a lot of build up on the middle and slowest rollers. However, as the mill operates, the grains break down nicely.

I used a toothbrush a lot while milling to clean the rollers, which helped. Setting the blade bevel down on the apron also helped, because it kept the mill fairly low maintainance, and I could focus on cleaning the rollers. But the gumminess of the grains slowed the milling down. Adding water to the gummy mixture lubricated and sped things up, though. I probably added a 300-400ml in water as the proces continued.

Dispensing was reasonable, but a little time consuming due to the thickness of the paste. However, it was easier than before because of the extra water.

TODO: Change the pigment amount to 170g. Add water to the recipe. Don't bother with dispersion mixing. Consider using a ball mill. Consider additional milling on the three-roll mill.

21-23. July 2022 - 1000ml medium to 170g ( + 70g = 240g) of pigment
Py74 powdered arylide yellow pigment: 270 grams × $0.080/g = $19.20
Watercolour medium 4 cups × $26.73/4 cups (1000ml) = $26.73
Number of pans filled 360 pans × $0.08/pan Cost of pans = $28.80
Total for materials = $74.73 Material cost/pan = $0.208
Time to mix 0 minutes
Time for pre-dispersion 60 minutes
Time to mill 180 minutes
Time to dispense 90 minutes
Total time 330 minutes
Final grind size 7-15µm
Time spent per pan 0.917 minutes × labour rate $0.445/min= $0.408
Total cost/pan = $0.62

The mixing of this was tricky because of the math. Normally I would use 500ml of pigment to 1000ml of medium - 1:2.

However, because of the difficulty that came with the last run, I measured out 333ml to get a 1:3 ratio. This worked out to be 170g.

Whisking this into the medium was clean and easy, and I did it in the kitchen sink with no mess.

I put the paste to mix on the drill press for 60 minutes. The paste was very liquid, so I had to bring the press down to the lowest speed.

I am letting the mixture sit over night to stiffen up.

The next day, the paint had not stiffened up. I started to doubt myself, so I added enough pigment to get the mix up to 500ml. This was an additional 166ml, which ended up being 70g. I mixed it for a short period of time, and it did not seem to thicken. I began to think that last time I had such trouble because I had accidentally doubled to measures.

When I began to mill the liquid, It started OK, but the paste was a little grainy, and it was much thicker at the end of the run. The first run lasted an hour, and there was a great deal of difficult thick paste.

On the second run, I mixed the thick parts roughly into the thinner parts of the paste. I also used an old toothbrush to clean the second roller and the top of the apron while working. This helped things, and the second run was only 30 mins.

The paste got better and better, and the final runs were about 15 mins each. On run #4, I tested the grind, it was 10-15µm. The paste started to get thinner as I kept going.

In the last runs, the paste got faster and cleaner, but the paste was getting thicker and thicker. My guess is that the pigment is generally OK with mixing with water, but strongly flocculates. On the initial grinds, the granules at the bottom of the paste were large, and when they were ground, more water was needed to coat the newly smaller particles. So they got very very dry.

Then as the paste kept getting ground, the particles got smaller and smaller and needing more water. So the mix gets thicker and thicker.

Interesting. Next time, I am going to try using only 200g of pigment.

1. March 2022 — reworking the pigment-medium mix for yellow

I needed to top off the huge batch of yellow pans that are drying out. However, I am worried that the yellow may be somewhat starved for medium, so I added 2 cups of a fresh batch of watercolour medium to the remaining yellow pigment and mixed it on the drill press.

The result was very watery, but I filled the pans anyway. Interestingly, the pigment and the water are separating as the pans dry. The pigment is behaving a little like clay, and the paint is starting to crack. We will see what the mixture is like once I do further tests.

This batch would have a balance similar to 1:3 pigment:medium.

21. February 2022 — Finishing off yesterday’s nightmare

850 pans @ $61.61 = $0.072/pan

Final grind size after 8 passes through the mill: 15-20um

Once I added the additional batch of watercolour medium, things went a lot better. The total volume of the mixture was huge, and I had to split it into two batches. The final tally was this:

  1. 8 cups of watercolour medium
  2. 4 cups of yellow pigment
  3. 5 cups of additional water

I am changing the final recipe to be 2:1:1 medium:pigment:water. Hopefully, this will help make the batches less crazy next time. I want to keep an eye on the amount of medium needed for the next time, since this batch was a little thin and does not perfectly bind to the paper.

The grind itself was excellent. Eight passes through the mill was wonderful. Making sure the thickness of the paste was like honey was very helpful. I tightened the gap on the fast rollers, and it increased the speed of the grind. When I put the knife a little tighter to the roller, this helped as well, and in the final runs, there were three separate streams of pigment going into the bowl. Based on a suggestion from my wife, I brought the bowls up higher until the apron in order to reduce splashing, using a couple of bread pans. It would be very handy to have a wooden platform for this to bring the bowl up even higher and perhaps include a builtin splash guard. Total time was about 4 hours including cleanup.


20. February 2022 — Milling the previous day’s pre-mix

Holy smokes! What a nightmare! The pigment and water sat overnight and became a very very thick paste. I ran it through the mill twice, and it took forever to go through the mill. It was a thick, butter-like paste, and very very dry. During the third run, I whisked 4 cups of water into it, and this helped a lot. During the fourth run, I added one more cup of water, for 5 cups in total. The fourth run ended up creating a runny mix that splashed as it went into the mill. What a mess! It took 2 ½ hours to run the material and clean up. The final grind size was 25 - 35 um. Goodness!

I am pretty sure that what happened was that the yellow pigment swelled as it absorbed the water. And it kept drinking it up. I have been mixing the yellow dispersion from Kama Pigments 1:1 in the past, which would have already swelled to accept the extra water. I think what I have to do is mix in a second batch of medium to make a mega sized batch of yellow. I will still mill it eight times and see how the grind changes.

19. February 2022 — mixing 4 cup portions = 8 cups at 1:1

I made a 1:1 mix of dry pigment to medium. I started with the mixing bowl filled with medium, and then dumped the cups of yellow pigment into an area where they would be sucked into the mixing vortex. I set out eight small pieces of hardware to help me keep count as I mixed the cups into the medium. The process was quite quick, perhaps about 20 mins. There was very little dust involved, and cleanup around the mixer was very fast: maybe 5 mins. I am letting the mixture sit for a day before milling.

11. Oct 2021 — ½ cup total at a 1:1 mix

Mixing: The pigment mixed slowly into the medium. It did not want to wet at first, and then ended up with a paste that was thicker than expected.

Milling: I set the roller gaps originally to 15 mil each. I tried to set the rollers to 10mil each, but the bronze shoes squealed pretty badly. I did not have any trouble with pigment falling through the rollers. A fair amount of the pigment dried on the rollers as well.

Mixing TODO: medium first, and mix in pigment. Allow to sit for a while to see if it wets better over time. Mixing a larger batch would make a lot of sense.

Milling TODO: try starting the mill off at 15 mil gaps on the slow side, and 10 mil on the fast side. Using a spatula to scrape out the mixed paste helped a lot.