A change coming for my audio sensor: I’m thinking that I don’t need to know the amplitude of sounds from both sensors individually; rather, the most important thing is the timing of the sounds. So, I’m planning to add/change some circuitry to do the following:
Sum the two ‘ears’ into a single analog signal for relative amplitude measurement (so that when I yell at him, he knows that I’m mad!).
Add a comparator (probably LM339) to each of the audio channels, and use these two signals to drive digital inputs on the PIC (instead of the two A/D inputs I’m using now). When a signal is seen on either ear, then go do an A/D reading to find out the amplitude.
These changes will free up an A/D input (that’s good, because I was wanting another one). And, the digital comparator signals can trigger interrupts on the PIC, making it much easier to measure the time difference between the ears without having to poll the A/Ds. The LM339 will have 2 comparators left over to use for something else. Not having to poll the ears is really good, because I plan for this PIC to also handle IRPD scanning, which involves driving a servo and such.
That sounds fantastic Pete, I like the timing idea because with my sound sensor, it reacts differently when the mic elements are pointed to different objects. for example, if one mic is pointed to a close flat object it tends to hear the sound better than the mic pointed to to the actual source. Sound bouncine off of other objects can be an issue, therfore, timming can help reduce this because the mic that receives the sound first will be the mic that is closest to the sound source.
I totally agree with Mike, I am drooling over your test equipment. I wish I had vintage test equipment, not because it’s vintage but because making do with a ten dollar multimeter really hampers progress.
Since you guys like the test gear so much, here’s what’s in the picture (okay, so I’m bragging a bit 8) ).
Clockwise from lower left:
An HP VHF signal generator (I forget the number). I do a lot of radio-related projects, so an RF generator is essential.
Behind the smaller sig. gen. is a big HP618D VHF generator.
Tek 475A o-scope with an HP 100 Mhz probe. A top-of-the-line analog scope about 20 years ago.
Under the scope is an HP frequency counter. Counts up to 500 Mhz (more with optional plug-ins), and 8 (or is it 10?) digit resolution. It was a top-of-the-line counter about 25 years ago. I paid $30 for it last year (I had to repair the power supply).
Simpson 260 VOM. A ‘classic’ meter.
Under the Simpson is an HP3446 DVM, from the 80’s.
A Weller soldering station (it’s a military surplus model).
On the far right: A variable DC power supply, with an HP 5V supply sitting on top of it.
Not seen in the picture:
A GenRad RLC bridge.
A Fluke DVM.
A Heathkit transistor tester.
A military tube tester.
A bunch of vintage test gear that I don’t actually use much.
I just uploaded the latest schematic for my sensor board: geocities.com/saipan59/robots/sensor.jpg
Note the “servo connector” - the sensor board controls an HS-55 micro servo, for panning the IRPD sensor back and forth. I don’t have the IRPD hooked up yet.
The other new part is the LM339 comparator in the audio circuit.
I made some progress on the batteries and the “movable COG” concept.
For batteries, for now I’m using six AA NiMH cells (2500 mA-hr). The cells are mounted on the guts of the charger board. With a couple of minor mods, the charger will now handle 6 cells instead of 4.
The battery+charger combo is ‘suspended’ inside the chassis, and can slide back and forth on two 1/8" brass tubes. I just need to hook up the wiring, and a mechanical linkage to the servo, then I’ll be able to move the battery pack forwards/backwards as part of the walking gait.
I had previously thought I would use N-gauge model railroad tracks and wheels, but this new approach is a lot better I think.
Sounds like it will work out nicely. I have also considered this; however I am going to move the battery bay and electronics bay as 1 unit. It will have an extra 2 servos for an X and Y type of movements. Good luck and get that picture up before I start cutting into my chassis
The first pic is looking into the side of the chassis. In the upper-left is the servo. In the left-middle is the lever-arm that moves the batteries and charger. You can see parts of the two brass tubes that are the ‘slider’ for the assembly to move on.
The second pic is a bottom view, showing the lever-arm. The brass tubes are sticking out on the left.
I’ve been creeping along slowly on this, but here’s an update:
The sensor board is coming along. The PIC code now can ‘scan’ for obstacles using an HS55 servo with an IRPD sensor mounted on it. It takes 40 IRPD readings across the range (about 150 degrees). Next step is to get the IRPD data sent to the main PIC via I2C.
Meanwhile, I’m thinking of changing my bot’s basic locomotion. Although I got the 4-legged walking to work OK, it’s agonizingly slow. It would spend a good portion of it’s battery power just going from one room to another, and after a few weeks I’m thinking it would start wearing out servos and such.
SO, I’m thinking about converting it into “a wheeled bot with legs”. Put a classic 3-wheeled setup on the bottom (2 driven wheels, plus an omnidirectional caster), and make it’s 4 legs become more like “arms”. Perhaps add a gripper to one of the ‘arms’. Perhaps put a couple of types of sensors on each ‘arm’.
Mike: It’s hand-soldered on perfboard. I doubt that I’ll make a PCB for it, because the one prototype is all I expect to need.
Nick: The sensor is a 16F88. The main PIC is an 18F8720 (now obsolete - the current equivalent is 18F8722), so it’s pretty much top-of-the-line in the 18F series, but it doesn’t have specialized peripherals like USB and such.
I’m trying to decide how to do the motors.
I could rob two of the 475 servos from the ‘hind legs’, and modify them to continuous-rotation. But I worry about the longer-term durability of servos being loaded like that. And I definitely don’t want the servo’s output shaft to be supporting weight, so I would need to provide the wheels with their own shafts and bearing.
I’ve got a pair of decent gear-head motors that would work well I think. The only hassle there is I need to add an H-bridge driver, and SW to do PWM.
