HRC Detonation Counter

HRC is an acronym for Honda Racing Corporation.  HRC developed the detonation counter to provide a fast and reliable way to tune grand prix road racers by “removing all human intuition.”  Honda felt this was especially important for the privateer racer. 

Much of what I know about the HRC unit comes from SAE paper 1999-01-3324, Detonation Counter for Production Racing Motorcycle by Soda and Sato.  Although the paper describes a detonation counter specific to Honda's NSR500V, the principles apply to instruments sold for their 125cc and 250cc road racers as well.  I also learned a great deal by dissecting a dead unit and will present my findings below.

Back in the day (and from memory) the HRC detonation counter for a single-cylinder machine cost about $850 USD.  The version for a twin was around $1400 USD.

The SAE paper acknowledges Shindengen Electric Mfg. Co. and NGK Spark Plug Company.  I assume Shindengen manufactured the electronics for HRC.  I know NGK made the sensors because I attempted (unsuccessfully) to buy them.

What is Detonation?

If you have never heard the sound of an engine detonating, consider yourself lucky.  I liken it to the sound marbles make when rattled around in a can.  At least that's the case with a largish bore diameter and significant number of events in a row (see endnote 1).

Detonation is a type of abnormal combustion.  It is characterized by an extremely rapid premature release of energy in what would normally be the last part of the charge to burn (the end-gas).  This causes very high local pressures and the propagation of strong pressure waves across the combustion chamber.  

Detonation is also called “knock” because of the sound transmitted through the engine structure when it occurs.  It's almost like a hammer blow, and can be extremely destructive to the piston and combustion chamber. 

Historically, optimal tuning (ignition timing and jetting) was judged by a skilled mechanic looking at the condition of the piston crown, combustion chamber, and spark plug.  This left a great deal of room for interpretation and required a lot of hard-won experience in the form of destroyed engines.  The detonation counter distilled expert experience and judgment into a few quantitative values.  In the case of the NSR500V, there were four “counts” — the number of detonation events for each cylinder above 50% throttle, and also below 50% throttle.  The ideal situation was to have a few counts per racing kilometer.  Anything less meant suboptimal performance, and anything more could result in engine damage.

According to Heywood's Internal Combustion Engine Fundamentals, detonation produces a shock wave which “…reflect off the walls of the chamber, eventually producing standing waves.  The amplitude of the pressure oscillations builds up as the standing waves are established, and then decays as the gas motion is damped out.”

Below are three graphs from Heywood.  They represent idealized waveforms in a four-stroke engine and are somewhat exaggerated.  Each of the three graphs shows combustion pressure plotted against crankshaft angle.  An important thing to recognize is that any given combustion cycle may (or may not) exhibit detonation.  Identifying and counting the number of detonation events provides an indication of how “close to the ragged edge” the engine is performing. 

Endnote 1. What we actually hear is many, many detonations in a row.  It's probably like sensing a misfire.  Humans only perceive a misfire when it occurs something like 10% of the time.  A single missed combustion cycle every 100 goes completely unnoticed.

These next two scans are also plots of combustion pressure versus crankshaft angle.  They come from the aforementioned SAE paper.  Both represent real-world data from a 2-stroke engine as monitored by a Kistler (misspelled in the figure below) piezoelectric pressure transducer in communication with the combustion chamber.  Each graph shows both the raw combustion pressure (top trace) and the same signal through an electronic bandpass filter (bottom trace).

When the amplitude of the bandpass filter's output is large enough, a detonation event is counted. 

Dissecting an HRC Detonation Counter 

The analog design is surprisingly simple, yet effective.  It comprises two single-sided PCBs, each measuring about 72mm x 45mm.  The upper board contains two Trumeter 7000 LCD counters which provide the numerical readout.  There is also a 3-pin 78L05 5-volt linear regulator on this upper board.  The throttle breakpoint potentiometer (bottom center) and a 2.4V Ni-Cd battery (top center) are also visible.

The lower PCB contain three 3 DIP ICs with their markings obliterated.  Shindengen did not want to make it easy for someone like me to understand how it works.  There are also a handful of passive components.  The active components are:

• LM2901, 14-pin, single-supply quad comparator

• CD4069, 14-pin, CMOS hex inverter

• LMF100, 20-pin, dual switched-capacitor filter

I find it somewhat amusing that the circuit boards are the same inexpensive quality as those found in Japanese consumer electronics of that period.  I had expected them to be FR4 fiberglass.

Circuit Configuration

The detection sensor is a piezoelectric element that fits under the spark plug. 

There is no amplification of the piezo sensor input.  It runs through an R-C antialiasing filter (-3 dB down at 11.3 kHz) into the input of an LMF100 dual switched-capacitor filter.  The filter sections are cascaded.

The LMF100 is configured for divide-by-50 operation and clocked at 1 MHz by a ceramic-resonator oscillator.  This yields a center frequency of 20 kHz (which is further reduced by circuit resistor values).

The output of the cascaded filter is also processed through an R-C filter, this one being -3 dB down at 34 kHz.   

Three gates in a CD4069 hex inverter are used to build the aforementioned oscillator.  Two additional gates provide signal inversion for other parts of the circuit.  One gate is unused.

+12VDC input power (protected against reverse polarity and transients) is regulated down to 5V to run the circuitry.    

A resistor divider creates a 2.5V virtual ground for the LMF100.

One LM2901 comparator provides the amplitude detection threshold for output from the LM100.

One LM2901 sets the TPS threshold for the “above” and “below” counts via the breakpoint potentiometer.  

Two LM2901 comparators are used as buffers to directly drive the Trumeter 7000 LCD counters. 

Both counters are powered via a 2.4 volt Ni-Cd battery (3 diode drops below 5 volts).  The battery allows the LCD to retain their counts for ~30 minutes after the bike's power shuts off.  A single momentary push button zeros both counters. 

A potentiometer allows a variable breakpoint for throttle position to apportion counts.

Each LCD counts detonation in one of two ranges based on throttle position:

• idle - breakpoint

• breakpoint  - WOT

Where “breakpoint” can be 25% - 75% (nominally 50%) of throttle opening and is adjusted via the potentiometer.

Curiously, pin 6 of the LMF100 was left floating.  This should either be tied to circuit V+ or V- (which then partly determines the filter's operating mode).  Left floating, it behaves as if tied to V- (which I determined via testing).  Tying pin 6 to V- connects each filter's second summing point to AGND.

I don't think leaving pin 6 floating was an oversight, as it is directly adjacent to a V- PCB trace.  Again, this seems like an attempt at obfuscating the circuit's operation.

Circuit Testing

I attempted to measure the frequency of HRC's bandpass filter, but the detonation counter was inoperable.  Although I purchased the device as non-functional, it's also possible I damaged something during depotting.  So I replicated the LMF100's circuitry to measure its performance.  Reverse-engineering is not a completely straightforward task.  Each section of the LMF100 can be configured to operate in any of 12 modes.  The LMF100's datasheet is fairly sparse.  A good reference appears to be National Semiconductor's 1985 Switched-Capacitor Filter Handbook.  Someday I may update this section with more information.

My performance measurements for the entire system showed a center frequency of about 13.3 kHz.  It was -3 dB down at 8.6 kHz and 25.7 kHz.  This yields a geometric center frequency of 14.8 kHz.  The overall circuit gain was 0.39 (equivalent to an attenuation of 2.5).

Honda's SAE paper says the NSR500V's detonation counter has a bandpass center frequency of 13 kHz.  I was a bit surprised to measure a center frequency of only 13.3 kHz for the smaller-bore RS125's counter.  The filter may even have been designed for 13 kHz, and my use of 5% resistors to replicate the circuit could explain the small difference. 

In any case, because the NSR500V's bore is bigger than the RS125 (68 mm versus 54 mm) I had expected the RS125 detonation counter's bandpass filter would be tuned to a higher frequency.  See the section titled Theoretical Determination of Detonation Frequency at the end of this page for more information.

Spectrum Analyzer Data

The chart below comes from HRC's SAE paper.  It shows the signal from the piezoelectric sensor in the frequency domain.  Quoting the paper, “It is understood that the spectrum of detonation is between 10 kHz and 20 kHz.”  

The X-axis is frequency plotted on a logarithmic scale.  The peaks circled are thus in the region of 10–20 kHz. 

It is also interesting to note the far left peak at 0.1 kHz (100 Hz).  Although the resolution is very poor at that end of the plot, it would seem to show the engine is running no faster than 6000 rpm (100 pulses per second equals 6000 pulses per minute).

TPIC8101 Knock Sensor Interface

Texas Instruments introduced their TPIC8101 knock sensor interface IC around 2003.  As of 2026 it is still being sold, and quite inexpensively I might add (the 1-piece price is under $8).  Although this is not part of the HRC detonation counter, it was something I investigated to accomplish the same task.  However, when I was unable to purchase NGK's piezo sensor, I lost all interest in the project.  It's certainly possible other automotive knock sensors could be employed instead, albeit with more engineering work.  I'll leave some information here for those interested.

The TPIC8101 has two sensor inputs and communicates with a microcontroller via the SPI bus.  It does not function as a standalone device and must be used with a microcontroller.  The bandpass filter is programmable giving a choice of 64 frequencies ranging from 1.22 kHz to 19.98 kHz.  This makes the SPIC8101 applicable to a wide variety of bore sizes.  Note that the sensors must be of the “flat” (also known as non-resonant) type.  TI even sold an evaluation module (TPIC8101EVM) for about $150.

Theoretical Determination of Detonation Frequency

I had originally intended this section to be featured more prominently in the write-up.  But I did this work over 20 years ago and have been unable to determine where I originally found the equation I'll be presenting.  I had attributed it to Charles Draper, a remarkable genius who entered college at the age of 15. 

But I can say with certainty it does not appear in Draper's 1928 MIT master's thesis, A Method for Detecting Detonation Waves in the Internal Combustion Engine.

It's also not found directly in Draper's 1934 NACA Report No. 493, The Physical Effects of Detonation in a Closed Cylindrical Chamber.  (Although it may be possible to derive such an equation from that report.)

The mystery equation relating frequency of detonation to bore diameter is:  f = k1,0,0 * Cs / pi * B

Where:

f is knock frequency in kilohertz

k1,0,0 is 1.841  (“wave number for the first radial mode of vibration”)

Cs is 900 m/s (approximate speed of sound considering a combustion chamber temperature of 2000 K)

pi is, of course, ~3.1416

B is the bore diameter in millimeters

Below are some cylinder bore diameters, and the detonation frequency as predicted by that equation:

• Honda RS road racers: 54 mm = 9.8 kHz

• Yamaha short-stroke TZs: 56 mm = 9.4 kHz 

• Honda NSR500V: 68 mm = 8.7 kHz 

• Tul-aris 800: 85 mm = 6.2 kHz 

• Hypothetical large piston: 108 mm = 4.8 kHz

The above frequencies agree with information presented by Heywood on abnormal combustion (and more specifically, measuring knock severity.)  Heywood writes, “The filter is set for the first transverse mode of gas vibration in the cylinder (in the 3-10 kHz range, depending on bore and chamber geometry).”

Unfortunately, the 8.7 kHz prediction for the NSR500V is not at all close to the 13 kHz bandpass filter HRC described in their SAE paper. 

Furthermore, the bandpass filter I found inside an RS125's detonation counter was centered around 13.3 kHz.  Again, this is far from the 9.8 kHz predicted by the equation. 

I must assume that HRC knew a good deal about this and built something that performed as intended.  My current working theory is that the prediction equation was based on 4-stroke combustion chamber geometry (and old designs at that) and a 2-stroke's vastly simpler combustion chamber produces higher-frequency detonation waves.