Design a Non-Contact RPM meter

Question

I am currently reverse-engineering a non-contact RPM sensor module that uses a Bosch knock sensor ( 
 Knock Sensors (0261231046) - Knock Sensors - Bosch Auto Parts ) 
 and a Condenser MIC output as the input. The knock sensor and MIC signal is fed into 2 separate op-amp 
 LMC66x CMOS Dual Operational Amplifiers datasheet (Rev. D) 
 , and the op-amp output goes to a CODEC 
 AD1845 , DigChip http://www.digchip.com 
 for processing. 
 My goal is to understand the op-amp circuitry and determine the appropriate configuration needed to condition the knock sensor and MIC signal before feeding it to the CODEC. I am a bit confused about where to start and what concepts I should study regarding op-amp. and there is a charge pump as well, 
 LT1054 Switched-Capacitor Voltage Converters With Regulators datasheet (Rev. G) 
 Could anyone guide me on: 
 How to analyze the existing op-amp circuit? What op-amp configurations are commonly used for piezoelectric sensors such as knock sensors? What topics should I study to better understand this signal-conditioning stage? How can I determine the expected signal levels and bandwidth requirements before the CODEC input? 
 The ultimate task is to redesign this board while maintaining identical functionality and output behavior as the existing device, so understanding the analog front-end is critical. 
 Any suggestions on the right approach, reference materials, or debugging techniques would be greatly appreciated. 
 AI is confusing with too much data. Not getting a clear path.

geralds · Accepted Answer

Sorry to tell you: you cannot rework this schematic as well we cannot. 
 AD1845 Datasheet and Product Info | Analog Devices 
 The CODEC is obsolete. 1. you need the values of all components 2. the ADC-Codec has a firmware you cannot read out because security key. 3. the PCB is a multilayer board and also the parts are very close placed to each other; you cannot follow the tracks. 
 Have you a component checker with you can readout its values? - I'm sure you haven't.

michaelkellett · Answer

From your picture of the pcb it's an old design. 
 So my next question is - why do you want to copy it ? 
 Pretty much all the parts are old (starting with the now obsolete AD1845). I'm sure that whatever processor they used will be a 25 year old design. 
 So if I were your I would start by writing a specification of what the new instrument must do. 
 If there is something really clever that the existing one does, and you can't work out how, then maybe it would be worth seeing how they did it. 
 If you have a working board it will be easier to measure what it does than to try and reverse engineer it. 
 I know the Bosch sensor of old and designed a competitor to it (of which over 2E6 were sold). You can easily simulate it with a signal generator with a 1nF cap in series. (I'm going on 25 year old memory here to check its capacitance with an LCR meter). 
 It's a piezo electric accelerometer with a fairly rough HF performance, very dependent on how it's attached to the machine. It can work at quite low frequencies (as low as 1Hz but the amplifier deign gets tricky). It was intended to work with a band pass filter centred at around 5kHz. 
 Inject a signal in place of the knock sensor and measure the signal at the input of the Codec. (You can get data sheets for the AD1845). The nominal bandwidth of the knock sensor is up to 10 - 20kHz. The low end is defined by the sensor source capacity interacting with the amplifier load resistance. The signal conditioning will likely be a band pass filter but the high and low frequency cut offs could be anywhere in the range from 10Hz to 10kHz - depending on the strategy they use to determine RPM from vibration. 
 Measuring the analogue system bandwidth will give you some good ideas about how the system works. 
 If you ever get to designing an input ciruit then beware that the knock sensor might have a working output level of mV but can put out 100V signal levels if you hit its base with a spanner. If you disconnect one and heat it up to 100C or so and then short it out you can get a decent spark ! It can take out pretty much any unprotected op amp. 
 Please let me know how you get on. 
 MK

acdc90 · Answer

I can remember fixing a timing light for diesel motors which had a clamp on strain gauge which clamped on to pipe from injector pump. 
 I expect you want to use on comon rail engine, so clamp onto the comon rail and devide signal by the number of cylinders to feed a counter 
 a microphone will not do the job

Design a Non-Contact RPM meter

Top Replies