Yes -but you don't necessarily need to make all the passes at the same time on the same day.
You can focus on the "hearts" fraction in a single pass, knowing that you're not getting the greatest yield. Collect the tails fraction (the heads pass is through a classifier) over a series of roast/crack/winnow steps, and then run several batches of tails through at once.
I think the mistake people make is that they think they have to do all the passes from a single roasting batch before they process the next roasting batch. You have to re-set the winnower for each pass.
Far more efficient if you run multiple batches through at the same setting.
Also - the feeding mechanism is what takes time if you're hand feeding the winnower. Fine some way to automate the feeding and you can get the process started and walk away from it. That's where you'll pick up efficiency after pre-classifying.
after months of trying to find the right size mesh in NZ I came across this http://polynet.co.nz/pages/quadra and I think its just profit for classifying before winnowing.
I am playing with designs for a winnower which pre classifies the mix and then passes into several gravity separation winnowers (classic down pipe with air draught type). I saw in one of Brooklyn Cacao's videos that they are using a gyratory (vibratory) sieve to classify, I imagine this is before putting the mix through the vortex winnower. Just wondering if anyone has any experience with these sieves? They look like they would be ideal for a winnower setup.
Classifying after cracking and before winnowing is a great step to increase the efficiency of the winnowing process. I have done it manually as making (or buying) the screens is significantly less expensive than buying a vibratory screener from Kason, Russel-Finex, or elsewhere.
Brooklyn Cacao uses (or at least used; I haven't seen the winnower working close-up for over a year) the vibratory screener because the integrated cracker was over-cracking and cracking unevenly. Putting the entire mass into the winnower resulted in very low yields and/over very low throughput. Things may have changed in the last year in those last two regards, I don't know. I see they are now offering a multi-stage vortex winnower but at something like $70,000 (or so I have been told) I think there are far less expensive ways to achieve similar or greater throughput.
One key is to run as much classified material through the winnower at a time as possible. Constantly changing the settings (the amount of vacuum) leads to inefficiencies.
You want to scalp off the large pieces (only partially cracked, flats), then pass through all the fines (and the germ). Work with the "hearts" fraction.
You can also pre-classify before roasting and before cracking to achieve more consistent results.
For those who have used a dust deputy for winnowing, what horse power shop vac did you use? I am trying to set up now but am not getting enough upward airflow in my vortex to suck the husks upward into the second bucket. Thanks for your help.
I'm not even using a shop vac -- just a regular old vacuum cleaner. There were three mods I had to make to the Dust Deputy to get it to work as an effective winnower. First, you have to make some kind of provision to get airflow through the bottom of the DD (I do this by placing a couple of pieces of polycarbonate (1/2" x 1.5" x 18") between the top of the bucket and the lid -- it lets air flow into the bottom of the DD funnel). Second, you have to extend the exit funnel (I just built a quick extension funnel with poster-board and tape), and third, you want to restrict the airflow on the inlet side when you are feeding. Both of these adjustments serve to increase the vertical airspeed which helps to get the husks going up while not being so fast that the nibs don't go down...it's a matter of experimenting a bit and making fine adjustments. Good luck!!!