cts_casemod,
This is actually a primarily high frequency (RF) analog circuit.
A 50kHz sin wave starts out as a 50kHz square wave in the oscillator on the lower right corner of the board. Signal conditioning in the quad op-amp to the left of the oscillator converts this to two 180 degree out of phase sin waves that are amplified to about 7V p-p in the final amplifiers on the lower left of the board. All this provides the 14V p-p 50kHz sine wave carrier signal for the resolver.
The signals return from the resolver in the form of 50kHz sin and cosine waves in the plug on the upper left corner of the board. (along with the reference signal and the motor temperature signal.) These signals are filtered and amplified with the quad op amp to the plug's right. To the op amp's right is an analog switch that saves just the peaks of the carrier signals. It's triggered by the original 50kHz signal that is delayed and shaped by the quad op amp on the upper-right corner of the board. This op-amp also amplifies and offsets the final output signals, which are sin and cos waves with a frequency 2X the motor rpm. These signal are finally sent out in the 5pin SIL connector on the upper left of the board.
Anyway, as you can see, it's a primarily RF analog circuit.
A lot of my noise paranoia is coming from the app-note that accompanied the output amplifier (the one on the lower left of the board) It HIGHLY recommended AGAINST having any ground plane directly underneith the input and output pins. Also, when considering noise, this board (and possibly the ground plane) will be directly above the microcontroller and it's clock, which are sending out noise at about 30 MIPs in addition to Paul having noise problems at about 125kHz.
My circuit is fairly robust, and shouldn't have weird resonant issues due to this noise, however, it will be sending a critical signal, then filtering/amplifying it. If I can avoid pulling in extra noise or perhaps have a good way of filtering it out, it would make things much better for the AtoD converter in the end.
Regarding power conditioning:
Each IC has at least a 0.1uF ceramic capacitor. IC's that are physically really close share a 0.22uF ceramic capacitor. The final output amp uses both a 1uF ceramic capacitor and a 0.1uF ceramic capacitor.
Both the +5V and +24V inputs use a 10uF ceramic capacitor and a TVS in parallel. Here is where a ferrite or a small 4.7 Ohm resistor in series with the input would be good?
Thanks again for the suggestions; I'm working on the copper pour.
- E*clipse
Quote:
Originally Posted by cts_casemod
Typically the grounds are isolated in the following conditions:
Current measurement;
Galvanic isolation;
Analog and digital logic;
I assume the first is N/A here, so ask yourself if you need electrical isolation. If not, a single ground would be better at shielding noise, that two separate.
The last one is analog and digital logic. I assume most of it is digital, so wont imagine many conflicts there, but review as I am not completely familiar with the design. Generally tough, if this is the case, the positive is isolated instead trough a low value 4.7R resistor or an inductor to cut HF noise, along with some nice, low ESR, ceramic decoupling capacitors.
Isn't the board close to the motor? If so grounding the motor will automatically create a "Faraday cage effect" and better yet if you share both grounds (motor + PCB + Vehicle) unless the board needs electrical isolation, but this is another issue you should avoid, because again, if the GND is shared with the vehicle it will work for you at cleaning noise rather than being an antenna, not to mention the risk for an isolated system if a cable gets caught and makes contact with some path in the vehicle, effectively breaking the isolation barrier.
Ground should be as close as possible to your signals. The larger the gap, the larger the airgap and generally the more coupling (noise) from nearby conductors
|