Big Trak Motor Drivers

The 75494 (discussed on the previous page under Big Trak Brawn) isn’t powerful enough to drive the Big Trak’s main motors. For that job, discrete (stand alone) transistors are necessary.

Each motor can go forward, reverse, or coast, allowing the Big Trak to go forward, backward, spin left, spin right, and stop. Brake mode is not actually supported, electrically and mechanically speaking. However, due to the friction of the gearbox, the coast mode actually acts like a brake.

The motor drivers are ingenious and little perplexing. Before opening up the hood, I was expecting to see a motor driver chip or four transistors and four diodes for each motor. Boy was I surprised!

Schematic of the Big Trak motor driver circuit with bipolar transistors. Only a half bridge is needed per motor due to intelligent use of a split power supply.

Schematic of the Big Trak motor driver circuit with bipolar transistors. Only a half bridge is needed per motor due to intelligent use of a split power supply.

The first sneaky thing is that the Big Trak engineers managed to use only two output transistors per motor. Thus, the Big Trak has a half bridge, instead of a full bridge per motor, yet can still drive each motor forward, coast, and reverse.

The key to this accomplishment was center-tapping or splitting the battery pack into 6 V and 3 V. One motor wire is always connected to 3 V. The other motor wire is either electrically disconnected (no transistors on) for coast mode, electrically switched to 0 V (9113 NPN transistor on) for one direction, or electrically switched to 6 V (9112 PNP transistor on) for the other direction. That’s a great trick!!

One potential drawback to this approach is that the motors can only receive half the battery pack voltage at one time (6 V to 3V or 0 V to 3 V; which is 3 V either way). But, that’s actually fine if you want to use a motor rated at half the voltage.

Another potential drawback is that, if the designers aren’t careful, half the batteries will drain faster than the other half. For example, if the forward command engages both PNP transistors, then both motors will draw only from the second pair of D cells (6 V to 3 V portion of the battery pack). If the forward command is used more than the backward command (which it usually is), then one pair of cells will be exhausted first.

Like in most robots, since one motor faces the opposite direction than the other, it would be perfectly natural to engage one PNP transistor for one motor and one NPN transistor for the other motor. In that case, each motor would draw from a different pair of cells, balancing battery usage for forward motion. Spinning left or right would consume power from only one pair of cells at a time, but as long as you turn right as often as you turn left, it would balance out.

Unfortunately, instead of flipping transistors to compensate for the motors facing opposite directions, the Big Trak designers flipped motor wire connections. As such, the Big Trak sucks power (300 mA to 500 mA under load) from only one pair of D cells when driving forward. If your Big Trak drives forward more than backward (which it usually does), then one pair of D cells will be exhausted first.

Other Motor Driver Components

There’s a non-polarized ceramic capacitor (C6, C12) across each motor to consume spikes and reduce electrical noise. R6, R12, R13, R15 are current-limiting resistors on the base of each transistor. These parts are ordinary and appropriate.

Now here’s where I start to get confused. R5 and R11 are presumably pull-up resistors to make sure the PNP transistors are off by default. R14 and R16 are pull-down resistors to make sure the NPN transistors are off by default. The higher pull-down value (15 kilohm versus 10 kilohm) is probably to avoid overwhelming the weak output of the TMS1000 microcontroller.

I don’t understand why pull-up and pull-down resistors are even necessary on current-driven devices (bipolar transistors). Do they provide value as spike reduction paths?

C13 and C15 are equally mysterious. I can’t decide if they exist for spike reduction on the lines that connect to the sensitive DMOS microcontroller (as opposed to the more rugged bipolar driver chip) or if they are there to absorb the output of the microcontroller on power up.

Based on oscilloscope tests, The TMS1000 outputs seem random at power up. Therefore, perhaps sometimes the outputs will randomly specify that both transistors turn on for a single motor. Such a condition would result in a brief short circuit (bad). Perhaps the capacitor delays the turn on time for at least one transistor long enough for the microcontroller to switch to the correct output state?

The TMS1000 had an odd way of specifying outputs, so I would think the output matrix could have been set for the Big Trak to prevent it from allowing both transistors to be enabled at the same time. If so, then maybe the capacitor delays the turn on time for the transistor so that the motors won’t jerk at power up?

Another mystery of this motor driver is the lack of diodes across the transistors to protect them from motor spikes. (The lamp also lacks a diode across it.) I mean, every classic schematic with a bipolar transistor and an inductance-based device shows a diode across either the device (single transistor designs) or the transistors (more than one transistor). Are the capacitors across the motor leads enough? Were discrete diodes expensive enough to be worth leaving them out? An oversight?

I was unable to locate datasheets or specifications for the 9112 and 9113 transistors. But, if you need to, it seems like you could replace them with 2907A and 2222A transistors respectively. Don Kerste would probably recommend power bipolar transistors for better performance, like the Zetex 718 and Zetex 618 respectively.

Left: Standard 9112 and 9113 transistors in TO-92 packages. Middle: 2N6715 in a TO-237 package. Right: 9112 and 9113 in TO-237 packages.

Left: 9112 and 9113 transistors in TO-92 packages. Middle: 2N6715 in a TO-237 package. Right: 9112 and 9113 in TO-237 packages.

Half or more of the Big Traks had 9112 and 9113 transistors in standard TO-92 packages. I was surprised to find a 2N6715 transistor in a TO-237 package (metal heat sink fin at top) substituted for the 9113 in a “newer” (higher serial number) Big Trak. Then, I was surprised again by finding both the 9112 and 9113 in a TO-237 package in an even “newer” Big Trak. And finally, I was surprised yet again to find the 9112 and 9113 back to standard TO-92 packages in the “newest” (highest serial number) Big Trak.

My first hypothesis is that the transistors were overheating and so the designers switched to a package with better heat dissipation, the TO-237. However, that doesn’t seem to fit with the return to a standard TO-92 package in the later Big Traks.

My second hypothesis is that they simply ran out of stock of the 9112 and 9113 transistors in TO-92 packages. So they substituted the 2N6715 transistor for some 9113 transistors, and then eventually had to purchase 9112 and 9113 transistors in a different package. I don’t know if this is what happened.

The Big Trak was designed to have motor-driven accessories controlled by the IN and OUT commands on the keypad. Let’s see how those worked...