Semi-blast chiller: read this first
One page for the owner. The full report (41,000 words, every price and source dated 2026-08-30) is the next item in the menu; two red-team reviews and the research memos follow it. This page states what to do and why, with the numbers that matter.
What the two videos got right and wrong
Tech Ingredients zip-tied a 12 V radiator fan inside a home freezer and cut the time to chill 0.8 kg of 50 C water in half (120 to 60 min) and the time to freeze it by two thirds. That result holds up. The fan does not add cooling; it moves heat faster from the food into whatever cold mass is already in the box (frozen inventory, or the packs).
The second video's "goop" (ammonium chloride brine gelled with xanthan, melting at -15.4 C) is a real eutectic and the -15 C plateau is real, but the recipe is 29.5 percent salt where the eutectic is 19.7 percent, so a third of the salt does nothing. You do not need to make it. You bought better packs.
Your situation
- Your working freezers hold -15 to -18 C empty. That is the warm edge of most commercial ratings (0 F is -18 C). Read the model manual and log the box for 24 h before treating it as faulty, and clean the condenser either way; that is the cheapest thing on this page for freezer life.
- You have three Arctic Ice Tundra XX-L packs (10 lb each, 13.6 kg, -15 C). They cannot be charged in a -15 to -18 C box (0 to 3 K of margin; owners report packs slushy after a week at -18 C) and they slowly lose charge stored there. So they live in a separate recharge chest freezer and go into the working box only when a batch goes in. That is the swap bank you proposed, and it is the right design.
- The -23 C ThermoSafe U-tek pack you asked about first works in the same swap bank and takes heat about twice as fast per pack (bigger temperature difference, thinner pouch), but only with a recharge box that proves -28 C or colder, which no ordinary commercial chest freezer is rated for. It is the upgrade to consider after the Tundra packs are tested, not the first purchase.
What the packs will do
- Capacity: Arctic Ice does not publish latent heat. At 220 to 300 J/g the three packs hold 3.0 to 4.1 MJ. A 10 kg batch from 60 C to 5 C is 2.2 MJ (2.9 MJ with margin), so three packs cover one 10 kg batch per charge, and about 10 to 14 kg at most. Packs 4 and 5 add surface, which matters more than mass; two Engel XL 5 lb packs ($29.99 each) expose almost twice the skin of one XX-L.
- Rate: standing on edge, 2 in apart, in the fan stream, the three packs absorb roughly 90 to 225 W at the start of a chill and take 20 to 45 percent of a 10 kg batch during the 90 minutes. The compressor and the frozen inventory take the rest. The packs shorten the inventory's excursion; they do not cap the box temperature and they do not shorten the chill much. The fan shortens the chill: expect 1.3x to 1.6x in a reach-in that already has an evaporator fan, about 2x in a static chest freezer.
- Recharge: 48 h in a still-air chest freezer at -21 to -23 C, 36 h at -25 C or once the melt test proves it. One set of three means one 10 kg batch every other day; a second set of three makes it daily.
- Equipment life: not proven for the compressor. The real levers are a clean condenser, 2 in pans loosely covered (frost and defrost load), and never chilling through a scheduled defrost.
What to buy
| Item | Price seen 2026-08-30 | Note |
|---|---|---|
| Recharge chest freezer: Galaxy CF7 | $299 | must pass a 24 h log at -21 C or colder, return it if not |
| or Atosa MWF9007 | about $506 to $543 | rated to -10 F; price not confirmed on a live page |
| or Galaxy Ultra Cold UCCF8 | $1,369 | rated to -50 F; also covers the -28 C U-tek option |
| Rated fan kit (section 6.3a of the report) | about $240 one fan, $345 two | Sanyo 9WV1212P1J001 IP68 fan, Mean Well GST120A12 listed supply, door switch, relay, diode, timer, 14/2 cable and gland, fuse, guard |
| Video-copy fan kit | about $195 to $205 | GESEXI fan, brick and PWM; only on a dedicated cabinet you accept as altered, never in an R290 or R600a cabinet |
| Probe thermometer | $31 | ThermoWorks ThermoPop 2 |
| Second set of packs, later | about $130 to $140 | when XX-L stock returns, or Engel XL |
No add-on fan goes inside an R290 or R600a cabinet without the maker's written approval, and a reach-in has no owner-installable cable route (a technician fits a bulkhead, or the reach-in gets packs and paddles without a fan). Check the nameplate refrigerant first.
For comparison, a dedicated blast chiller that plugs into 115 V: Avantco SF-3 $2,089 (15 lb per 90 min), Avantco SF-5 $3,789 (31 lb), Atosa EBF-05 $5,707 list (55 lb). Used 5-pan Traulsen and Beverage-Air units run $1,899 to $4,500 on eBay. At 30 kg per batch the SF-5 wins on cost and it is the only option that takes load off the storage freezer instead of moving it to overnight.
This week
- Clean the condenser, read the model manual, log the working box for 24 h.
- Buy the recharge chest freezer, log it 24 h at its coldest setting with the basket inside; keep it only at -21 C or colder.
- Calorimeter test (report section 6.3, Tier D): one charged pack into a cooler with 20 kg of 30 C water, lid on, stir, log until the water stops dropping. A drop to about 13 C means 220 J/g, to about 9 C means 300 J/g.
- Melt test: a charged pack in room air with a probe on its face holds -15 C for 3 to 5 h. Run it after 36 h and after 48 h of conditioning; if the two hold times match within 15 percent, 36 h is enough.
- Fan kit install, then the airflow test (three runs fan off, three fan on, alternated).
- The go/no-go is a real batch at the SOP's mass and pan depth: 70 F by 2 h, 41 F by 6 h with margin, inventory sentinel never above -12 C, no compressor short-cycling, packs recharged on schedule, repeated on a second day. Fail that and the answer is a smaller batch, a paddle or ice bath first, a dedicated cabinet, or the SF-5.
Where the numbers come from
Every figure above is sourced in the full report. Two independent reviews challenged it: an internal three-lens pass (34 findings) and a Codex red-team (30 findings, all applied), both logged at the end of the report with the change made for each.