Why is this zener in series at the base of transistor, and has something failed?

Question

I am debugging an ancient little impact printer and in my probing have found an unexpected voltage level. I'd like to understand this part of the circuit, why it's constructed as it is, and whether what I'm looking at is part of what's wrong, or as expected:

(Detail from the schematic here)

The printer has a number of little solenoids and one of these driver sections for each one, that switches on the 40VDC rail to the coil as shown. The input from the left is a TTL level signal indicating whether the solenoid should be switched on or off. The buffer shown is a 7407.

The transistor is a TIP127, and the diode is a 15V zener. (Resistor is 2.7K fwiw).

When the printer has power, the 40VDC rail is at about 41V (ok). But when I measure the voltage at the bottom of that resistor, it's about 16V. (If I pull the 7407 out it's still 16V, which I assume is because the '07 was in high-z state.)

My questions are:

1. What is the purpose of the zener diode in this configuration? My first thought is that it's for some sort of back flow protection of the TTL circuit, but then I don't expect it to do anything (except drop .7V) in normal operation.
2. Why am I seeing 16V at the bottom of that resistor? I'm not assuming something is "wrong" here but I don't follow how the PNP transistor plus the zener is resulting in this voltage sitting there when the coil is doing nothing-- my sense would be that a TTL level voltage (only) should be turning the transistor on and off as a switch for the coil circuit. Since there are seven of these little compositions, and they all seem to show similar values, I'd be surprised if something's wrong with all of them, but I guess not impossible.

I'm trying to understand here more than I'm trying to debug as such, so thank you for explanations of what this is trying to achieve!

Answer 1

The 7407 is only rated for 30V, so the zener diode allows the transistor to turn off when the output is high. It can be expected to drop at least 10V when the transistor is 'on', likely much more for safety margin.

The 16V you are reading is an 'off' voltage. Something that that is not unexpected. The exact voltage you read when off is not an accurate indication of the zener voltage because it's dependent on leakages and meter loading.

If that is one of those Selectric conversion boxes, they were kind of a pos, tbh.

Answer 2

Check the 7407 datasheet: it is an open-collector buffer with high-voltage output, rated up to 30V.

The supply shown is 40V, however. This would violate the rating.

By connecting a >10V zener in series, the logic-high rating can be respected, while still driving the node solidly low. The logic output can saturate to <1V when low, here putting about 5.5mA through the resistor; in fact, V_OL for 16mA is max. 0.4V so we can be sure it's at least this well saturated.

Note a discrepancy: a zener diode only has significant voltage drop when there's some current flowing; while this can be quite small currents (~µA), it's decidedly nonzero. Does this cause the transistor to turn on when it shouldn't? Fortunately, the TIP127 includes base-emitter resistors, ensuring that a small output-high leakage doesn't turn it on.

That said, the I_OH maximum of 0.25mA well exceeds the TIP127's input sink capacity of, say, 0.6V / 8k = 75µA; I would rather see an external pull-up resistor here. Presumably, the real I_OH doesn't actually get that high, or not at modest temperatures anyway, but that's the limit given in the datasheet and they are allowed to sell you a part with performance that bad; it would be annoying to say the least, to discover a shipment of such outliers in production.

They could've also implemented the level-shifting capability by adding a cascode transistor (right):

^{simulate this circuit – Schematic created using CircuitLab}

or as an inverting stage (left), both of which receive a logic-level input, no buffering required.

These options may've been overlooked, or ignored for various reasons -- perhaps transistors were more expensive (component plus assembly costs) at the time. They're pretty close in cost these days, and with SMTs being pervasive, either solution would be as compact as it would be effective.

Nowadays, we might also consider an IC for the task: high-side load switches are common for automotive use, with robust capabilities, and often handy features like current sensing. Not that these features would've been necessary for a printer, of course, but the electrical requirements of automotive fuel injectors for example are very similar to those of impact pin/hammer type printers back in the day.