Which charcoal is correct for bakhoor

This page answers one sourcing question: which charcoal specification belongs on a purchase order for a bakhoor line, and why the answer most supplier catalogues give is the wrong one. We manufacture bakhoor-grade coconut shell charcoal for OEM buyers at container scale, so we are not a neutral party here. The mechanism is therefore set out in full so that you can check it, and the test in §5 runs on your current supplier’s stock without involving us at all.

§1 · First, decide which burn you are buying for

Every specification decision downstream depends on a split that supplier catalogues rarely make explicit: is the coal heating a resin, or is it baking a molasses?

In bakhoor and oud use, the coal sits beneath a resin-bearing chip in an open mabkhara and does one thing — supply heat. The aromatic fraction leaves the wood as vapour; the wood itself is meant to survive the process rather than burn. The coal is a heat source, not a fuel. What you are actually buying is a temperature band held steady, plus the absence of anything the coal contributes to the smoke on its own account.

In shisha use the coal is a fuel. It drives heat through foil or a heat-management device into a molasses bed and has to keep doing it for the length of a session. Sustained thermal output is the product. Every rung up the fixed-carbon ladder buys more of it, which is why that ladder exists, why it is priced as a quality grade, and why it is printed on almost every coconut cube specification in circulation.

The two jobs optimise in opposite directions. If your quotation, your specification sheet and your supplier’s sample notes are all written in shisha language, you are being sold up a ladder that was not built for your product. That is the argument of this page, and §3 is where it lands.

If you are sourcing for shisha, barbecue or restaurant fuel, we are not your supplier and this page will not help you. We make bakhoor-grade cube only, under our customers’ brands, and only by the container.

§2 · What bakhoor actually asks of a coal

Four properties decide whether a coal is correct for a bakhoor line, ranked here by how much of the outcome each one determines.

1 · Peak surface temperature

What you sell is an oil and resin fraction that leaves agarwood and blended bakhoor at a modest temperature. Drive past that band and the volatiles flash off at once — an intense first minute, then very little — while the substrate itself begins to combust and lays an acrid woodsmoke note over the profile the blender designed. A useful reference point comes from the one place in this craft where temperature is set rather than inferred: controlled electric agarwood burners are run near 150 °C. That is roughly the band the resin wants. It is not a claim about the surface temperature of any coal — a lit coal runs far hotter and is moderated on its way to the chip by distance, by the burner and by its own ash. Peak surface temperature is what decides how much moderating the rest of the system has to do.

2 · Thermal stability across the session

A bakhoor burn is a plateau, not an event. Heat has to work inward before anything aromatic happens at all — published electric-burner practice allows 15–20 minutes for heat to reach the resin inside a chip — and then the temperature has to stay put while the fraction leaves. Our criterion for a bakhoor-grade cube is therefore a usable plateau of 60–120 minutes, which is a different specification from maximum total burn time and frequently in tension with it. A cube that runs hot early and then collapses gives your customer a scorched first chip and a cold second one, and they will report those as two unrelated complaints.

3 · Ash behaviour

The white-ash wait is not ritual. The standard instruction — light the coal, wait about 10 minutes until a white coat has formed, and only then place the chip — exists because that ash layer is an insulating buffer between the coal’s peak and the resin. Reading it as a thermal control rather than as a custom is our inference, but it is a short one: it explains why going too early scorches the chip and wastes material, which is the outcome every guide warns about. That standard figure is a habit rather than a measurement, and it is not the figure our own cube needs. §5 has ours, per format, and the gap between the two is the most useful thing on this page.

Which makes ash behaviour, rather than ash percentage alone, the property that decides a bakhoor burn. Ash that forms evenly and holds together is a working temperature control. Ash that packs dense, or sheds and re-exposes bare coal underneath, removes that control and puts the chip back onto the peak. How long the layer takes to form is a separate question and not a quality one — §5 is where that is measured, and where we correct what this page first said about it. Ash behaviour is also the property that makes the test in §5 possible, because unlike fixed carbon it is visible to the naked eye.

4 · Olfactory neutrality at ignition

The coal must add no note of its own in the first minutes, while the chip is still cold and anything in the smoke reaches the room unmasked. Two things break this. One is an ignition accelerant, which off-gasses by design — see §3. The other is a binder that was never fully gelatinised in production, which can ferment in transit and release a sour note at light-up. The second is both more common and much harder to trace, because it develops after the coal leaves the factory and the symptom points at the blend rather than at the coal. Our position is that this is a paste temperature problem in extrusion rather than a binder ratio problem, and that a sour smell present before ignition points to transit rather than to production. The mechanism, and a check you can run on a carton you already hold, are on why some cubes smell sour before they are lit.

We do not describe our cube as odourless. The word is not measurable, and a supplier using it is telling you nothing you can verify. The publishable form is no detectable off-odour at ignition under a named test, with the test named — and ours is on that page.

§3 · Our position: fixed carbon is a variable to control, not to maximise

For bakhoor, a higher fixed-carbon figure is not an upgrade. It is a variable with a correct band, and specifying the top of the ladder works against the product you are selling.

That contradicts essentially every coconut charcoal specification sheet in the category, so here is the reasoning in full, with its weakest link marked rather than buried.

The mechanism

Fixed carbon is the fraction of the briquette that burns as solid carbon once the volatile matter has gone. Raising it, and correspondingly lowering volatile matter, raises the sustained surface temperature of the burning cube and lengthens the phase in which the cube runs at its most aggressive. That is combustion fundamentals and we state it as fact.

The requirement it collides with

Bakhoor needs the opposite: a moderate band, held. Excessive heat degrades the fragrance rather than releasing it, because the oils leave too fast and the substrate starts contributing its own combustion products on top of them. This is consistent across consumer and trade guidance on burning oud, it matches what blenders report, and we also state it as fact.

What we conclude from those two, and how strongly

Put them together and a cube optimised for maximum fixed carbon is optimised against the bakhoor requirement. That conclusion is ours. It is reasoning from two well-supported statements, not a measurement: we have not measured chip-side temperature against fixed carbon across grades, we have not found anyone who has published such a measurement, and we do not claim a threshold above which a cube becomes unusable. We are specifically not saying that high fixed carbon ruins bakhoor. The defensible form of the claim, and the one we will defend, is the heading of this section. If the measurement gets run and comes out the other way, this section is wrong and we will change it.

What to write on the purchase order instead

Stop specifying a fixed-carbon floor as though it were a quality grade. Specify instead: a fixed-carbon band with an upper bound as well as a lower one; the test method and issuing laboratory for that figure; ash behaviour described in terms of formation time and coherence, not percentage alone; moisture; and olfactory neutrality at ignition under a named test. A supplier who can answer only the first of those is quoting you a shisha cube, whatever the heading on the document says. Our own figure, with its basis and the report it came from, is on the specification page, which carries a method and a source on every row — one lot measured rather than a band promised, which is the distinction the second item in that list is asking your supplier for.

One option to rule out before you compare: quick-light discs are the same argument in a cruder form. A quick-light carries an ignition accelerant distributed through the briquette, and that accelerant off-gasses at light-up — precisely the minutes when the chip is coldest and contributing nothing of its own to mask it. Whatever comes off reaches your customer first and then sits underneath the blend. For a premium oud product this is not a trade-off with an upside: the convenience is bought entirely at the expense of the thing being sold. Every serious guide in the category already says so and we claim no credit for it. It appears here only so that it is ruled out before the comparison below. Ruled out for a premium line, that is, and not in general — instant-light won this market for reasons that are real, and which customer each product is actually for is a page of its own.

§4 · Four options, compared on what actually decides the purchase

Specifications alone do not settle this, because the same number means different things depending on what the coal was built to do. Each column below answers four questions instead: who the option is for, who it is not for, what trade-off it accepts, and how you can verify the answer yourself.

The row to read twice is who it is not for. A supplier carrying all four of these products will never print it, because every cell in that row costs them a sale. Ours costs us one too: if you burn on an electric plate, the first column is not for you either, and §6 explains why.

The substitution most of our buyers arrive already making has a page of its own: using shisha cubes for a bakhoor line.

§5 · The time-to-ash test — our method, and what it measured

This is our own proposed test. It is not an industry standard, it is not a laboratory method, and it has not been validated against an instrument. We have now run it, which changes what we can publish and not what the method is: a measurement does not make a protocol standard, and this one is still ours. We publish it because it is falsifiable, because it costs nothing, and because it settles the argument above on your bench rather than in our copy.

What we measured

Minutes from ignition to full white-ash coverage, three repeats per format, one bench, one session. The figures sit beside the burn duration for the same format, because the two together are what a format commits your customer to — how long they wait before the chip goes on, and how long they get afterwards.

The middle column is ours and this page owns it. The right-hand column was measured under a different protocol and belongs to another page — the protocol behind the burn durations, and what those figures do and do not mean, which is where they are explained rather than here. They are set side by side because a buyer choosing a format is choosing both at once.

Two things in that table changed on 2026-09-23 and both are corrections rather than additions. The disc row used to be one row: it carried 12 minutes and was labelled simply “disc”, and we have since established that the figure was measured on the 33 mm. The 40 mm is five minutes slower. Until the diameters were pinned down the two readings looked like a disagreement about one number rather than two numbers about two sizes, and this site publishes which of those it turned out to be. The 40 mm figure also carries one repeat rather than three, because it came out of the single-disc comparison on the quick-light page rather than out of the three-repeat round, and a cell that did not say so would be borrowing the other figures’ conditions.

The second is the burn-duration column, which still has one disc figure because nobody recorded the diameter of the disc it was measured on. It is a real measurement of a real disc and it keeps its place; what is not known is which of the two sizes it describes, and rather than split it into a pair of invented figures to match the shape of the column beside it, the cell says what is missing. That also retires a sentence this page used to carry — that the two columns rank the same way. Across the three formats they did. With the disc split they cannot be compared at all on that row until the burn duration is measured per diameter.

Against those, a shisha-grade cube from another Indonesian producer reached full coverage in 8 minutes. That one is a cube, so it reads against our cube rather than against the table as a whole: less than half the time, under the same protocol, on the same bench, on the same evening.

What the result means

Ash appears when surface carbon has been consumed. A cube high in fixed carbon and low in volatile matter establishes glowing surface combustion quickly and ashes over fast. Ours does not, and that is deliberate. At 78.2% fixed carbon and 13.7% volatile matter — both with a basis, a method and a report number on the page that owns them — and carbonised at 400–650 °C, our cube spends its first minutes driving off volatiles before the surface settles into a steady glow. The ash layer builds slowly because of that rather than in spite of it.

So time-to-ash is not a quality metric. It is a usage parameter: a property that tells your customer when to place the chip, and tells you nothing about whether the cube is any good.

When we first published this test we read a long ash wait as a sign of a cube running hot; we had that backwards. The measurement runs the other way, and the reading above is the corrected one.

Why the formats differ — our reading, not a measurement

A disc presents more surface per unit of mass than a cube, so its surface carbon is consumed sooner and the ash layer completes earlier. That is our explanation for the order in the table and we have not measured it: we hold four figures and a reason we find plausible, which is not the same thing as having shown the reason to be right. If you run the test and the order comes out differently, the figures stand and the explanation is what falls.

The two diameters are the closest thing to a test of it that we have, and they went the way the reading predicts: a wider disc of the same thickness carries less surface for its mass than a narrower one, and the 40 mm took longer than the 33 mm. That is one comparison of one pair on one bench and it is not a demonstration — it is the explanation surviving the first occasion it could have failed, which is worth exactly that much and no more. It is also why the 40 mm sits close to our cube while the 33 mm sits far below it, so “a disc ashes over sooner than a cube” is a statement about the 33 mm before it is a statement about discs.

What this gives you

The correct ash wait is specific to the cube, and to the format. A customer who applies a shisha habit — eight or ten minutes — to a cube that needs eighteen is putting bakhoor onto a surface still driving off volatiles.

That is the acrid first impression, and on this reading it is caused by the instruction rather than by the coal. The standard practice a customer inherits allows about 10 minutes, which is close to right for the comparison cube and roughly half of what ours needs. It is also the most commercially useful thing in this measurement, because it is preventable with one line of print on a box — and that line is not the same on a cube box as on a disc box, nor the same on a 33 mm box as on a 40 mm one.

Why four figures rather than one

A single span covering every format would have averaged away the only part of this a buyer controls. Format is chosen on the purchase order and a wait time is printed on a carton, so one range across three formats would have been accurate and useless. It would also have hidden the thing worth seeing, which is that the order matches the burn durations rather than cutting across them.

What the figures are not

Three repeats per format, one bench, one session. The protocol does not control for draught, cube placement, ignition consistency or ambient conditions, and those move this figure more than anything else on the list — so these are what one bench produced and not a tolerance we hold to. The comparison is one cube, one grade, one producer: it is not a survey of the category, and we do not name the producer.

What you need to run it yourself

A standard mabkhara, a stopwatch or a phone, the cube you currently buy, and a candidate cube. Nothing else, and no laboratory.

Procedure

1. Same room, same ventilation, no fan and no draught. Airflow moves this figure more than any other uncontrolled variable, so both cubes must see identical conditions or the result means nothing.
2. Ignite one cube by your normal method. Start the clock at the moment the cube is self-sustaining — glowing without an applied flame — not when heat was first applied. This is the step people get wrong, and it is worth agreeing on before you compare figures with anyone else.
3. Stop the clock when white ash covers the whole visible surface, with no black patch remaining on the upper face.
4. Record the figure. Photograph the cube once 10 minutes have elapsed, whether or not it has finished, so that you hold a fixed point of comparison as well as a number.
5. Repeat three times per cube and use the median, not your fastest run.
6. Run the candidate cube identically, in the same session, in the same room. If you are comparing formats rather than suppliers, run each format separately — that is what the table above is.

How to read your own result

The figure is a setting, not a score. What you do with it is print it: the minutes your customer should wait before the chip goes on, for the format in that box. A cube that ashes over slowly is not a worse cube — it is a cube with a longer correct wait, and the complaint everyone blames on the coal is what happens when that wait is printed nowhere.

Comparing two cubes tells you something narrower than it used to. If your current supply and a candidate reach full coverage at noticeably different times under identical conditions, the two are not interchangeable behind the same printed instruction, whatever else is equal between them. That is a packaging consequence before it is a quality judgement, and we are no longer willing to read it as one.

What it does not tell you

Nothing about ash content by mass, nothing about moisture, and nothing about fixed carbon as a number — all three need a laboratory and a stated method, and no stopwatch substitutes for one. It also does not transfer between burners: a figure from your mabkhara is not comparable with a figure from ours, which is why we ask for your burner type alongside any figure you send us.

What would falsify this

Two things, separately. If a bench running this protocol finds our three formats ashing over in a different order, the surface-area reading above is wrong and we will say so here. If cubes of materially different fixed carbon reach full coverage in the same time under identical conditions, then ash timing does not track the proximate figures at all, and more of this section goes than that. We would rather be shown either than neither: send us your figures with your burner type and your format and we will publish the aggregate, including the runs that disagree with us.

§6 · Where this does not apply

Three limits, stated because a specification argument that claims to apply everywhere is not a specification argument.

Electric burners

None of this page applies. There is no coal, no ignition event and no ash buffer; the temperature is set on the device and held there, which is exactly why electric practice is useful to quote as a reference point and useless as a market for us. If your line is built around electric burners, your sourcing problem is the chips and the blend, and we do not supply into that use case at all.

Self-igniting and pre-blended tablets

Where the fragrance is compounded into the tablet itself, the coal is not a separate variable and there is nothing here to specify. The argument above holds only when the heat source and the aromatic material are two different objects that you purchase separately.

Shisha, barbecue and restaurant fuel

For those applications the fixed-carbon ladder is right and the reasoning on this page runs backwards. They appear here as comparison only. We do not manufacture for them, and a supplier serving both markets from a single specification is serving one of them badly.

§7 · How we verify, and what we have not published yet

What we do not publish yet

The laboratory figures that headed this list are now published: ash, volatile matter, fixed carbon, moisture and calorific value, each with its basis, its method and its report number. They live on that page and on no other page of this site — everything here references it rather than restating a value, so that there is never a second version to contradict the first. What is still outstanding is shorter than it was.

Shorter again since that sentence was written. §5 now publishes our own time-to-ash figures, one for each format, measured under the protocol on this page rather than estimated from what we expected them to be. That leaves one item, and it is the one we would hold another supplier to first.

• Certificate identifiers. We name a document only together with its issuer and its date, which is also the standard we would suggest you hold every other supplier to.

Batch dependency

Specification values for charcoal are batch-dependent. Where figures are published, the governing figures for your shipment are the ones on the certificate of analysis issued for your lot, not the indicative values on anyone’s website. How we handle sources and evidence generally is set out in our source and verification policy, and the scope of what this site does and does not represent is in the disclaimer.

Reviewed by Mohamed Noor, Director — 23 September 2026

§8 · What supports each claim on this page

Most of the claims below are not measurements. Dressing them as measurements would make the page easier to read and harder to trust, so here is the basis for each load-bearing claim and the form in which we are willing to state it. One row is withdrawn and stays visible as withdrawn, because that is the only way a ledger is worth keeping.

§9 · Next step: request the sample, and run the test on both cubes

The useful next step is not a call. It is 1 kg of our bakhoor-grade cube on your bench, beside the cube you currently buy — what ships, and on what terms, is set out on the sample request page. The sample itself is free of charge; courier terms are on the request page.

What to do with it

1. Run §5 on both cubes, three repetitions each, same room and same session.
2. Light one of each cold, with no chip on it, and smell the first minute of each. That is the olfactory neutrality check, and it needs no equipment.
3. Then run your own blend over both and judge the first chip specifically, because the first chip is where scorching shows before anything else does.

What to send us

Your burner type, the market you ship into, and the time-to-ash figure you measured on your current supply. A sample request that arrives with a measured figure gets a technical answer from production rather than a quotation from sales, because we can see what you are comparing against.

Our minimum order is 1 × 20 ft FCL, set out with the rest of how an order works end to end. We manufacture under your brand only. We do not sell retail, we do not sell single cartons, and if your programme is smaller than a container we are the wrong factory — we would rather tell you that now than after you have sent us a brand brief.

Request the 1 kg sample, and see what arrives with it