Zener Diode Behavior

[Ground_Loop]

I'm troubleshooting a voltage reference in an instrument and my efforts led me to checking a 9V zener diode (part#: Tektronix 152-0174-00).  I lifted one end and attached a curve tracer.  I get the correct reverse voltage drop of 9.0 volts.  However, when I switch to forward bias the diode will not begin conduction until about 95V.  I expected the usual 0.7ish.  Anyone ever seen this behavior in a zener before?

I pulled out a known good device from my stash to check against, and forward conduction began as expected around 0.7V.

For reference this is VR210 from a Tektronix PG506.

[mawyatt]

Likely has a PN junction in series with a ~ 8.2V Zener diode for controlling the TC for compensation of the circuit it's utilized with.

Best,

[Ground_Loop]

Interesting. I've learned my new thing for the day.  Thanks for that.

[TimFox]

Further information on temperature-compensated zener reference diodes:
Motorola made these devices with three choices of the number of forward-biased diodes in series with one reverse-biased zener junction:
1 diode:  usual total voltage of roughly 6.2 to 6.5 V (depending on zener construction and operating current).
2 diodes:  usual total voltage 8.4 to 9.1 V
3 diodes:   total voltage roughly 11.7 V
Zener (actually, avalanche) diodes above 5 V have a positive temperature voltage co-efficient, and the forward-biased diodes have a negative temperature co-efficient, so with an appropriate current the absolute co-efficients (V/K) cancel and the total voltage is constant.
To obtain this cancellation, the device current must give these co-efficients.  The sorting of devices (e.g., 1N821 series) into tempco bins actually represents how close the appropriate current is to the specification current (7.5 mA for 1N821 to 1N829 series).  Back in grad school, where graduate student labor was cheaper than precision parts, the lab next door would buy the cheapest 1N821s and find the appropriate current for each unit to achieve zero tempco, using a Fluke differential voltmeter and a lab oven.

[Ground_Loop]

This one claims to be operating at "exactly 7.5mA."

[mawyatt]

To take the concept of a PN junction in series with a Zener (avalanche as TimFox correctly points out) to extreme, check out the LM399 and LTZ1000 references. These are buried Zeners with transistor base-emitter PN compensation, the "Zeners" have ~ +2mv/C while the Vbe has ~ -2mv/C. The LM399 includes on-chip temperature regulation, the LTZ1000 requires an off-chip Op-Amp but achieves better performance. By burying the junctions below the chip surface they reduced the effects of surface contamination, which helps reduce low frequency noise and drift. Both of these are amazingly good, stable references, the LM399 used in the Keysight 3446X series 6 1/2 digit DVMs (also older 34401) and the LTZ1000 in the Keysight 3458A 8 1/2 digit DVM.

Fun devices to play with if you like precision circuits

Best,

[TimFox]

Motorola also had a series of “reference amplifiers”, with a non-monolithic PN diode and NPN transistor for use in discrete voltage regulators.  Fluke used premium versions in various metrology and calibration gear.