Moteurs à carburateur et ZF 8HP : faire dialoguer un vieux système d'alimentation avec une boîte moderne

Here is the thing nobody tells you before you start: the 8HP is smarter than your engine.

A carburettor engine tells the world almost nothing. It makes vacuum, it spins at some speed, and it delivers fuel because physics said so. An 8HP transmission controller, by contrast, expects a running commentary. It wants to know how much torque the engine is making right now, how fast that number is changing, and what the driver is asking for.

Put those two things together and the mechanical conversion is the easy half. The hard half is building an engine data story convincing enough for the gearbox to trust.

This guide covers how to do it properly, and the specific places carburettor 8HP builds go wrong.

Why the 8HP is different from an old automatic

An old three or four speed hydraulic automatic shifted on pressure and governor weights. It did not know or care how much torque your engine made. Line pressure was set by a cable or a vacuum modulator and the box simply did its thing.

The 8HP does not work that way. It is a torque-modelled transmission. Its control strategy continuously estimates engine torque and uses that estimate to decide:

• How hard to clamp each clutch pack during a shift
• How long the shift should take
• How the torque converter should behave
• When to release and apply the next gear

If the controller's torque estimate is wrong, shift quality degrades even when the mechanical installation is perfect. That is the single most important idea in this article. You can have a flawless adapter, perfect converter spacing and a beautifully mounted gearbox, and still end up with a car that shifts badly because the numbers feeding the controller are nonsense.

What the controller actually wants

Input	What the controller uses it for
Engine speed	Referencing converter slip and shaft speed relationships
Throttle position	Driver demand and rate of change of demand
Manifold pressure	Engine load, and therefore torque estimate
Torque estimate	Clamp pressure and shift timing
Rate of change	Anticipating the shift, not reacting to it

The last row is the one people miss. It is not enough for the controller to know the engine is at 4,000 rpm. It needs to know it is arriving at 4,000 rpm quickly.

Meet the engine where it is

A modern EFI engine hands all of this over for free. The ECU already computes load and torque, and it already knows throttle position. A CAN bus message passes it along.

A carburettor engine has none of that by default. Depending on what you have, there may be:

• No ECU of any kind
• No throttle position sensor
• No manifold pressure sensor
• No CAN bus
• Possibly only a tachometer signal, and a slow one at that

So your job is to manufacture the missing data. In practice that means fitting sensors that a carburettor engine never had, and making sure the signals you do send are fast and clean enough to be believed.

Fit the three sensors that matter

Throttle position sensor

You need a TPS on the carburettor linkage or throttle shaft. On most installations this is straightforward, and aftermarket TPS units designed for carburettor conversions exist.

What actually matters is signal behaviour, not the brand:

• It returns to the same closed-throttle value every single time
• It reaches a repeatable wide-open-throttle value
• It has no dead spots through the travel
• The linkage does not flex or have slop
• It is mounted rigidly so vibration does not change the reading

A TPS that drifts or has dead spots is worse than no TPS at all. The controller will believe the driver is opening and closing the throttle when they are not, and it will make shift decisions on that false information.

Manifold pressure sensor

Fit a MAP sensor to a true manifold vacuum source.

This catches people out, because a carburettor has several vacuum ports and they do not all mean the same thing. A ported vacuum signal, taken from above the throttle blade, rises and falls with blade position rather than representing actual engine load. You want full manifold vacuum, below the throttle blade.

Get this wrong and the torque model is wrong at exactly the point where it is compressed the hardest: low throttle and cruise.

Engine speed

Engine speed is the most demanding of the three. The controller is comparing engine speed against transmission input speed in real time, and it needs that comparison to be valid during the fastest changes the engine will ever see.

Which brings us to the problem that breaks more carby 8HP builds than anything else.

The RPM resolution trap

Some ignition systems and aftermarket ignition boxes output a tachometer signal that updates only a couple of times per crankshaft revolution. At a steady idle, that looks perfectly fine on a laptop.

During a fast shift, it is nowhere near fast enough.

Here is the mechanism. The transmission input shaft speed sensor is typically a proper toothed wheel, updating many times per revolution. The engine speed signal, if it comes from a slow tacho output, updates far less often.

During hard acceleration or a fast rpm rise, the gearbox input speed sensor delivers many fresh readings while the engine speed signal is still reporting an older value. For a brief window the controller sees:

• Gearbox input speed climbing rapidly
• Engine speed apparently lagging behind

The controller concludes the relationship between the two is implausible. It may think the engine is dragging behind the gearbox, or that a speed mismatch exists that does not. Shift control degrades.

Nothing is mechanically wrong. The engine is not slower than the gearbox. The signal is simply not updating fast enough to be trusted.

How to think about it

The useful way to reason about this is as a race between your signal update rate and your shift duration. A shift event that lasts a few hundred milliseconds needs engine speed data that refreshes many times within that window, not twice per revolution.

Source	Typical resolution	Suitable for 8HP?
Tacho output from an ignition box	Very low, often 1 to 2 pulses per revolution	Only if the controller is confirmed to handle it on your engine
Distributor trigger	Low to moderate	Marginal
Crank trigger wheel	High, many teeth per revolution	Yes, this is the target

Best practice for engine speed

• Use a crank trigger wheel with genuinely high tooth count
• Use a shielded signal wire, properly grounded
• Match the sensor type to what the controller expects
• Ensure stable sensor power and ground
• Apply filtering only where the controller documentation calls for it

Do not plan on a slow tacho feed unless you have specifically confirmed it works on your combination. It is the single most common cause of a carby 8HP that shifts badly for no apparent reason.

Building a torque model for an engine with no torque data

The 8HP controller needs an estimate of engine torque. With no ECU, you have to give it one.

The good news is that a standalone transmission controller can calculate torque internally from the signals you feed it. That is the approach to aim for.

Your torque model does not need to be mathematically exact. It needs to be close enough that the gearbox shifts correctly, so that any learning or adaptation is trimming small errors rather than compensating for a model that is fundamentally wrong.

What to base the model on

Factor	Why it changes the model
Engine displacement	Sets the overall torque ceiling
RPM	Torque varies across the rev range
Manifold pressure	Primary load indicator on a naturally aspirated engine
Throttle position	Driver demand and the rate of change
Expected peak torque	Anchors the top of the table
Camshaft profile	Changes low-load behaviour dramatically
Idle vacuum	Tells the controller what "light load" actually looks like
Aspirated or boosted	Fundamentally different load relationships

The camshaft point is where a generic model falls over. A mild engine idles with strong vacuum. A big-cam engine idles with weak, lumpy vacuum and behaves oddly at low load. If the controller assumes stock-vacuum behaviour, the torque estimate goes wrong precisely in the low-throttle range that dominates takeoff, creeping in traffic, garage shifts, and gentle driving.

Many standalone controllers include a TPS-based torque ceiling feature for exactly this situation, letting you cap the torque value at various throttle positions on an engine with poor vacuum signal. If your engine has a big cam, use it.

Keep both MAP and TPS

It is tempting to lean on MAP alone, since it represents load. Do not.

MAP tells the controller how loaded the engine is. TPS tells it what the driver is asking for. Two situations can show nearly identical manifold pressure but completely different driver intent:

• Light throttle, climbing a hill
• Throttle opening rapidly under acceleration

Manifold pressure alone does not distinguish those fast enough to be useful for shift timing and shift pressure. Throttle position does.

The limitations you cannot engineer away

This is the honest part, and it is where a carby conversion differs irreversibly from an EFI one.

No automatic downshift blipping

With a drive-by-wire throttle, an ECU can blip the throttle on a downshift to match engine speed to the next gear. With a carburettor and a mechanical cable, the throttle belongs to the driver's right foot. The controller cannot open it.

If you want downshift blips, you need a mechanical throttle actuator for the controller to command. That is a real project of its own.

Limited torque reduction during shifts

On an EFI build, the ECU can reduce torque during a shift by retarding ignition, cutting fuel, closing the throttle, or dropping boost. Reducing torque during the shift reduces clutch load and makes the shift easier on the hardware.

A carburettor engine has far fewer levers:

Strategy	Available on a carby engine?
Cutting fuel	No, the carburettor is mechanical
Closing the throttle	No, unless you add an actuator
Dropping boost	Only with an electronically controlled boost strategy
Retarding ignition	Possibly, if you run an ignition system that accepts a retard input
No torque reduction at all	Yes, and this is the common answer

The last row is worth dwelling on. A properly configured controller can manage the clutch packs well through the shift without any torque reduction being available, and this is how a large number of carburettor 8HP builds run successfully. It is not automatically a problem.

It is, however, a reason to be deliberate about the rest of the setup. If you cannot trim torque during a shift, the quality of your engine speed signal and your torque model matters even more, because the controller has less margin to work with.

If your ignition system does support a retard input, that is the one torque reduction lever most likely to be available to you. Check whether your controller can command it before assuming it is out of reach.

Controller approaches differ, so choose deliberately

Not every controller handles a carburettor engine the same way, and the difference is architectural rather than a matter of settings.

Standalone controllers with internal torque calculation work from the signals you supply and build the torque estimate themselves. This is the most direct route on an engine with no ECU, because you control exactly what data they receive.

Controllers that emulate CAN bus data to satisfy a factory TCU work differently. Here the factory transmission control logic is retained, and the controller's job is to convince it that it is still installed in a production vehicle. This can work well, but it means the factory logic still expects a believable engine environment, including behaviours your carburettor engine may not be able to produce.

That is not a reason to avoid the approach. It is a reason to think carefully about whether the data you are sending is complete enough to satisfy logic that was written for a different engine entirely.

The practical guidance: whichever route you take, the torque model and the quality of your engine signals are the deciding factors. Solve those first and the controller choice becomes much easier to evaluate.

Commission the signals before you drive it

Most carby 8HP problems are signal problems, and signal problems are cheap to find on a laptop and expensive to find on the road.

Work in this order:

1. Verify the TPS through its full travel. Confirm it returns to the same closed value

every time and has no dropouts.

1. Verify the MAP reads a sensible idle vacuum and responds cleanly to throttle changes.

Confirm it is on true manifold vacuum, not ported.

1. Verify engine speed resolution. Check the update rate, not just the number. Confirm it

tracks a rapid throttle blip without lagging.

1. Check the relationships. With the car stationary and the engine running, the controller

should see engine speed and transmission input speed make sense together.

1. Build and sanity check the torque model against what the engine actually is, including

cam profile and idle vacuum behaviour.

1. Confirm the remaining inputs and outputs before the first drive: shifter, brake pedal,

paddle or manual mode, ignition wake-up, reverse light output, mode inputs.

1. First drive on light throttle. Log the data. Look for implausible speed relationships or

torque estimates before you ask the gearbox to do anything hard.

Diagnostics: symptom to likely cause

Symptom	Check first
Shifts flare or feel vague under hard acceleration only	Engine speed signal resolution and update rate
Poor low-speed behaviour, awkward creeping and garage shifts	MAP source (ported vs manifold) and idle vacuum assumptions in the torque model
Random or inconsistent shift behaviour with no mechanical fault	TPS condition, dead spots, linkage flex, mounting rigidity
Torque estimate looks plausible at wide open throttle but wrong at cruise	TPS not contributing, or torque model built only from MAP
Everything fine at steady rpm, falls apart during rapid changes	The RPM resolution trap, full stop
Gearbox behaves as if the engine is lagging it	Engine speed signal update rate versus input shaft sensor resolution

The mechanical side still has to be right

None of the above replaces the fundamentals. Regardless of fuel system, an 8HP conversion still needs:

• A correct engine-to-transmission adapter plate for the exact engine and 8HP combination
• Correct crank, flexplate and torque converter spacing
• Correct starter motor engagement
• A suitable transmission crossmember
• Tailshaft or driveshaft work
• A properly planned transmission cooler and oil line setup
• Shifter or CAN keypad setup
• Wiring, power supply, fusing and grounds done properly
• Correct controller selection for your variant
• Correct transmission generation and valve body compatibility

Get the mechanical side wrong and no amount of torque modelling saves it. Get the mechanical side right and the control side is the difference between a car that drives beautifully and one that never quite behaves.

Frequently asked questions

Can you run a ZF 8HP behind a carburettor engine? Yes. It is a well-established conversion. The mechanical side is comparable to any other 8HP swap. The work is in supplying the engine data the controller needs to build a torque model.

What sensors does a carburettor engine need for an 8HP swap? At minimum, a fast engine speed signal, a throttle position sensor on the carburettor linkage, and a manifold pressure sensor on a true manifold vacuum source.

Why does the RPM signal resolution matter so much? The controller compares engine speed against transmission input speed in real time. A slow engine speed signal cannot keep up during rapid rpm changes, making the two look inconsistent even when the hardware is fine.

Can I use a tacho output from my ignition box? Only if you have specifically confirmed the controller handles it on your engine. A crank trigger with high tooth resolution is the reliable choice.

Does a carburettor 8HP need torque reduction during shifts? Not necessarily. Many carby 8HP conversions run without any torque reduction and shift consistently. If your ignition system supports a retard input, that is the most likely torque reduction option available to you.

Can the controller blip the throttle on downshifts? Not with a mechanical throttle cable. Automatic blipping requires an actuator the controller can command.

Next step

If you are planning an 8HP behind a carburettor engine, the sensor and controller decisions are worth getting right before you buy anything. Tell us your engine, cam profile, induction setup and intended use, and we will help you work out which controller approach and which sensor setup actually suits it.