Diode-connected transistor has negative resistance?

[merlinb]

I needed a CA3086 transistor array but they've got crazy expensive, so I tried subbing a HFA3046. Long story short, it didn't work, but that's not really my question.

I discovered that with the base and collector connected together, the CA3086 behaves like a normal diode like you'd expect.
But the HFA3046 has negative resistance! I have attached my measured DC I/V curves for diode-connected, and for just the BE junction alone.
How is this possible? Has anyone else seen diode-connected transistors exhibit negative resistance? Is there anything on the datasheet I should look for if I wanted to exploit/avoid this feature in future?
Datasheet: https://docs.rs-online.com/8edb/0900766b8002512d.pdf

[RoGeorge]

I've heard of negative resistance when the C and E terminals are swapped (transistor "reversed"):
https://www.eevblog.com/forum/testgear/fooln-around-with-dso-awg/msg4422982/#msg4422982

How does the plot looks like if you reverse C with E?

[mawyatt]

Somethings wrong here, your graph shows a BE only junction conducting ~0.25ma at 400mV Vbe. Unless this was conducted at a very high temperature like 100C, this is not possible with a proper silicon bipolar small junction transistor.

Best,

[2N3055]

Static damage?
These are 8GHz transistors, small geometries, easy to damage.

[merlinb]

Static damage?
These are 8GHz transistors, small geometries, easy to damage.

Not impossible I suppose, though I did get the same behaviour from all four transistors in the package, and I tried two separate packages

[AnalogTodd]

Static damage?
These are 8GHz transistors, small geometries, easy to damage.

Not impossible I suppose, though I did get the same behaviour from all four transistors in the package, and I tried two separate packages

Which curve did you take first? If you are getting as much current as you show for the BE junction, the diode junction should show at least as much current.

[tggzzz]

I needed a CA3086 transistor array but they've got crazy expensive, so I tried subbing a HFA3046. Long story short, it didn't work, but that's not really my question.

I discovered that with the base and collector connected together, the CA3086 behaves like a normal diode like you'd expect.
But the HFA3046 has negative resistance! I have attached my measured DC I/V curves for diode-connected, and for just the BE junction alone.
How is this possible? Has anyone else seen diode-connected transistors exhibit negative resistance? Is there anything on the datasheet I should look for if I wanted to exploit/avoid this feature in future?
Datasheet: https://docs.rs-online.com/8edb/0900766b8002512d.pdf

Show us a picture of your experimental setup, complete with schematic. That can be very informative and help rule out (or in!) some possible causes.

In the absence of a photo, even a written description and hand-drawn schematic are better than nothing.

[PCB.Wiz]

How is this possible? Has anyone else seen diode-connected transistors exhibit negative resistance?

Something is awry in your measurements.
Take the plots in two directions up and down, as the two way sweep can expose hysteresis from effects like self heating.

The using-collector curves should always be lower in voltage / higher in current than base only connection - see their data. Zero negative resistance there .
NPN VBE junctions are very well behaved, but GHz RF parts will be very small and very good at picking up stray RF.
 

[tggzzz]

NPN VBE junctions are very well behaved, but GHz RF parts will be very small and very good at picking up stray RF.

Depending on the details of the circuit, not only picking up RF. Hence my request for photo plus schematic.

[T3sl4co1l]

Try again with a ferrite bead on the base terminal.

Tim

[merlinb]

OK here's a simple setup -it doesn't get much simpler. Please excuse the crudity of the diagram I didn't have time to paint it.
PSU through a 3k resistor to the 'diode'.
Monitoring the voltage across the diode.
I slowly wind up the supply voltage.
In this instance with a 3k load, at around 700mV across the diode it breaks into some rather nice MHz oscillation. Average voltage across the diode (see scope reasout) falls until we exit the negative resistance region.

[tggzzz]

OK here's a simple setup -it doesn't get much simpler. Please excuse the crudity of the diagram I didn't have time to paint it.
PSU through a 3k resistor to the 'diode'.
Monitoring the voltage across the diode.
I slowly wind up the supply voltage.
In this instance with a 3k load, at around 700mV across the diode it breaks into some rather nice MHz oscillation. Average voltage across the diode (see scope reasout) falls until we exit the negative resistance region.

That, as I and T3sl4co1l suspected, is the Big Hint.

Your schematic doesn't show most of the components in your circuit. If you insert vaguely suitable values and simulate it, you will see what's likely to be happening.  To see it in the circuit, you will need a higher bandwidth scope

Alternatively, to see the kind of thing that is happening, have a look at https://entertaininghacks.wordpress.com/2024/03/16/practical-traps-with-a-one-transistor-audio-amplifier-solderless-breadboards-and-oscilloscopes/

[merlinb]

No, it's a classic relaxation oscillator. Adding 22nF across the diode makes a nice 50Hz triangle, for example -you don't need high bandwidth to measure that.
The question is why does this happen with this particular transistor type, when most transistors behave like a textbook diode?

[tggzzz]

No, it's a classic relaxation oscillator. Adding 22nF across the diode makes a nice 50Hz triangle, for example -you don't need high bandwidth to measure that.
The question is why does this happen with this particular transistor type, when most transistors behave like a textbook diode?

50Hz is, of course, suspiciously close to mains frequency in the UK.

The time constant of 22nF and 3kohms is 15kHz. That is a long way from 50Hz; to get to 50Hz, you would need to increase the resistor.

If you increase the capacitor to, say, 47nF, does the relaxation frequency reduce to 25Hz?

Hence I'm unconvinced it is a classic relaxation oscillator.

What tests have you done to verify it isn't also oscillating in the UHF/microwave regions?

What happens when you try T3sl4co1l's suggestion of a ferrite bead on the base? If you haven't got a ferrite bead, insert a 100ohm resistor close to the transistor's base.

[Terry Bites]

Classic "Negistor" behaviour, see attached.

More bipolar mood swing:
https://hackaday.com/2021/06/10/an-explanation-of-a-classic-semiconductor-riddle/
After trying it out, I mentioned this wonderistor a fellow engineer and he bet me £20 that it wasn't true and that I'd been april fooled.
Quids in, 1996 quids too!

[Xena E]

I've been trying to search for some reasonable citations for the following, but have drawn a blank!

Because the B/C connected transistor is capable of pulling down its collector further than the base junction with regards to the emitter, any signal that is impressed between the two, (on that "short" circuit), will appear amplified at the collector. At HF that short circuit is an inductor for instance.

One prospect who applied for a position with my company, was a student who was asked to produce something they had made: She produced just such a circuit that was as I remember a silicon diode, silicon bjt, couple of capacitors and a coil of wire, the basic circuit was that of an ampified diode but given a load of a 300 ohm telephone earpiece resolved a local radio station at good volume... it utilised the above effect.

I guess what I'm saying here is the form of this phenomenon reported by the OP, one must not take it that a DC circuit connection form is going to behave in a textbook manner as you would expect, I recreated what the student had built, however with a uhf capable transistor and stripline inductors. It basically jammed terrestrial TV in the local area.

As for reversed supplies, early transistors collectors/emitters were designated as "preferred", and would often work either way up in a circuit!

Best regards...

X

[tggzzz]

Because the B/C connected transistor is capable of pulling down its collector further than the base junction with regards to the emitter, any signal that is impressed between the two, (on that "short" circuit), will appear amplified at the collector. At HF that short circuit is an inductor for instance.

Diode-connected transistors are used in many ways, e.g. current mirrors.

Naive explanations of how simple transsitor circuits operate can - with more detailed analysis and understanding - be seen to be too simple, with "surprising" consequences.

I guess what I'm saying here is the form of this phenomenon reported by the OP, one must not take it that a DC circuit connection form is going to behave in a textbook manner as you would expect, I recreated what the student had built, however with a uhf capable transistor and stripline inductors. It basically jammed terrestrial TV in the local area.

Inspired by Horowitz and Hill, here's (a VHF radio example) "one I made earlier"[1]: https://entertaininghacks.wordpress.com/2024/03/16/practical-traps-with-a-one-transistor-audio-amplifier-solderless-breadboards-and-oscilloscopes/

[1] a catch phrase in the UK.

[Xena E]

Because the B/C connected transistor is capable of pulling down its collector further than the base junction with regards to the emitter, any signal that is impressed between the two, (on that "short" circuit), will appear amplified at the collector. At HF that short circuit is an inductor for instance.

Diode-connected transistors are used in many ways, e.g. current mirrors.

Naive explanations of how simple transsitor circuits operate can - with more detailed analysis and understanding - be seen to be too simple, with "surprising" consequences.

Thanks. And thanks for the link, interesting.

Yes, indeed that's so.

The young lady to whom I referenced, (Amie), had investigated the circuit, described it correctly, and published it on an Internet forum called TheRadioBoard.rb.

Quite refreshing to meet a (young) person with such an in depth understanding. I think she was 11 when she applied the principle to a reflex/regenerative radio.

I went on an Internet search...

A guy called Mike from Hawaii presented a refinement of the design and used it as single transistor DX radio, calling it the Amidyne.

https://groups.io/g/TheNewRadioBoard/message/863

Unfortunately in searching for the original posts that Amie made I found that they are now gone because of the original RB forum going permanently off line about three years ago.

There also used to be a paper floating about on the Internet written by one Charles Wenzel that had examples of the "short circuit amplifying" technique. He though seems to have unfortunately spoiled his credibility in the world of technology because of certain quirky aspects of his persona.

X

[tggzzz]

Classic "Negistor" behaviour, see attached.

More bipolar mood swing:
https://hackaday.com/2021/06/10/an-explanation-of-a-classic-semiconductor-riddle/
After trying it out, I mentioned this wonderistor a fellow engineer and he bet me £20 that it wasn't true and that I'd been april fooled.
Quids in, 1996 quids too!

That PDF has the base open and voltages >7V, and "avalanche multiplication effects".

The OP's measurements have collector open (or connected to base) and voltages <0.8V.  Hence not the same.