Cool stuff - I did a video on Tunnel Diodes several years ago. Their are remarkably difficult to characterize in the negative resistance reading. They had traditionally been used in very high frequency oscillators, as well as in the trigger circuits of some of the older Tektronix analog scopes. I suspect what you're seeing in no-mans-land is due to oscillations (the SMU likely doesn't like seeing a negative resistance).
I have some Ge devices somewhere in my junk pile. I built a wide frequency range tunnel diode dip meter for testing a 400 MHz receiver back in the mid 60's. Worked great. I accidently built a great FM transmitter at about the same time. I was trying to build an oscillator to tune over the FM radio band. Turns out my breadboard was so microphonic it worked great as an FM transmitter just by talking to it. I also have a UHF tv tuner that I bought to update a standard tv back in the day before they were built in. It uses a single tunnel diode as oscillator and down conversion mixer into a low pass filter feeding a standard ribbon cable to the tv. I could play with the LC and "nudge" the frequency range all the way up to 2 GHz according to my lecher line detector. Tunnel diodes were somewhat challenging but fun to play with.
The GE transistor Manual, 7th edition (1964) 450.37 8/64 850M has a whole chapter on them, including many practical circuits. This book really shows my age and pre-dates ISBN numbers of any kind whatsoever.
I still have my copy as well. Also has the transistor test set build instructions in that edition. Probably one of the best transistor manuals of that era.
My memory from engineering school is that the curve is a third order function. The Source meter result looks like 4th or 5th order,... I have no explanation other that it is oscillating. You could put a scope probe on the diode while testing with the Source meter. They are interesting. Another interesting device is an Avalanche transistor,...
Around 50 years ago, I read about tunnel diodes used as microwave oscillators. Don't remember any graphical representation, so this brought some memory, and has me fascinated. Thank you.
You ask why ... many years ago, I measured these things and got the same graph. Why ? because I measured the parallel connected resistance of the power supply and the negative resistance of the diode. only when I used a nicd battery and a 10 ohm potentiometer as an adjustable voltage source, I obtained a perfect theoretical graph. Regards .
Here is the Math: For parallel resistors we have: Req= (R1*R2)/(R1+R2) Say now R1 ABS(R2) (i.e. ABS means Absolute Value) In that case both numerator and denominator become negative and therefore Req becomes positive (division of two negative numbers is positive) So you can measure Req and since your R2 is known positive value which you shunt R1 ( your diode ) with it. you can calculate R1(your diode) equivalent resistor. Here some more homework for you: 1. Use the circuit of Esaki oscillator to create crude oscillator. 2. Put this diode (with correct biasing) in parallel with parallel LC tank to increase the Q of the LC tank ( as long as your negative R is less then parasitic R of LC tank) the negative R of the diode cancels part of positive parasitic R of the LC thank. 3. You can choose biasing of the diode to control Q of the parallel LC tank. 4. The circuit above become oscillator if ABS(R1) > ABS(R2)
Interesting critters. Only time I've had anything to do with them was nearly 50 years back when I was in USAF electronics. Some gear had them mounted in a tuned cavity oscillator that fed into a waveguide mixer. We tore a bad one apart and it was sort of empty inside, except you could tell - very high precision empty - the surfaces were nearly polished mirrors.
They do very fast switching and they were used for that also in the trigger circuit of old oscilloscopes. The drawing of the curve is correct, the element is just connected with inductive wires, so it oscillates. With your rf equipment it will be easy to measure the signal frequency with a non contact probe. The signal of that type which might have a rather high capacity of its own should be in the low GHz range if not below. ..., (I just checked the data sheet 50pF.)
The issue is the negative resistance. Here's an idea. Swamp it with positive resistance. Trace a 10 ohm resistor, then trace the diode in parallel with the resistor and subtract the current measured from the first trace
We currently make germanium planar tunnel diodes. They make very high freq. oscillators when operated in the transition area just where the negative resistance begins. Usually they are used in the region before that where the tunneling current predominates. They are used in high speed detectors.
@@larrybud imagine you needed to track a high velocity object and make corrections to the trajectory of a device designed to intercept said object. A high frequency super sensitive detector might come in handy.
Tektronix used them as Trigger detectors in their scopes at about the transition time from tubes to transistors. The special feature was exactly the negative resistance and a definite switch with appropriate hysteresis. By the way, they are called Esaki diodes by the name of their inventor.
@@InssiAjatonin the Tek trigger circuits the tunnel diodes were used just shy of the negative resistance region, biased so that the when signal reached the trigger level the diode snaps over the negative resistance region to the higher voltage level. The diode does the very quickly which makes a very sensitive and stable trigger circuit.
The negative resistance region is unstable. The only way you can have the diode functioning there is if you add some series resistor (which effectively makes it a positive resistance, and then it is not so interesting). You can actually see on your curve tracer (the very dim traces) that the voltage jumps to higher value with a current above the threshold. And it will again fall suddenly to a lower voltage value when the current falls down below a value. In between those two limiting current values, you have two possible voltage values. So it will function as a bistable circuit with just a diode and a series resistor. Or you can get it to oscillate with an inductor, biased to the unstable (negative resistance) point.
Uses include microwave amplifiers, relaxation oscillators, fast flip flops, rising/falling edge sharpening, fast triggers. The negative resistance region is unstable. Forcing its voltage to be in this region causes the current to switch back and forth between the two stable branches of the graph. The meter then performs a time average of these currents to report.
The missing negative portion of the curve has an interesting similarity to the temperature-vs-heat-transfer curve between a solid surface and a boiling liquid. (See the Leidenfrost effect.) IIRC is took a long time to figure out how to get a ridged enough temperature control to plot the full curve.
A tunnel diode is use in the Tek 1502 TDR to produce the high speed transient. I don't think tunnel diodes have been made in many years. There were a lot of papers written on them back in the 1960s.
@@IMSAIGuy haha your good mood and curiousity inspires, It's not all that serious in the end. I'm not professionally active in electronics, I just like figuring stuff out and playing. Fun thing is that logic is useful in my profession which is a bit high level. Anyhow, I hope you keep enjoying your channel and congrats on the succes with the 100/k mark, which is remarkable in these youtube times :) Greetings from the Netherlands (yeah yeah, Philips and ASML)
Flat region in negative dynamic resistance region is averaged value of oscillations ( check your programmable meter datasheet for details how it filters and averages input ). Curve tracer shown do not average that much - take a look at diode during test using HF oscilloscope.
I remember back in the 80s, seeing a circuit in an ancient electronics book for a wireless transmitter "bug" that used a tunnel diode. My freind and I were left wondering where to get one, and what the heck it even was.
I have never played with tunnel diodes, but do have a nice book about them. "Tunnel Diode Manual" circuits, applications, specifications, published by GE, 1961. The tunnel diode curves in this book show a smooth, mostly straight line down from the first peak. That part is, of course, the negative resistance regime. Near the 0 current line the curve is smooth and its slope gradually becomes positive again, transitioning into the exponential regime. I spent some time using Kiethley and NI instruments like yours to test quantum dot image sensors, which also operate by tunnelling. I don't think there is anything wrong with your method. The East Block diodes you have exhibit a strange hysteresis property. They might do something interesting in a tradional tunnel diode oscillator circuit, but I suspect they are more like a snap recovery diode. You should get some genuine US or Japan manufactured tunnel diodes. GE made Germanium tunnel diodes back in the day.
I think the unexpected kink in the curve can be elminated by heavy capacitive filtering. Without it, it will oscillate always. If this curve is the true one for this diode, it is very curious and is caused by the manufacturing specifics. That would also indicate why the oscillator is not working in a stable way and creates strong harmonics and nosie-injected behavior.
Really interesting! The curve tracer had very thin/dim nearly-horizontal lines that seemed to suggest that the I/V curve differs when the voltage is increasing vs. decreasing... kinda like a capacitive effect, or maybe magnetic hysteresis. I tried to guess which parts came from which sweep-direction, and was surprised to see from the software test that my expectation was literally cut in half... half-correct, and half wrong. I'm curious if the software-test would show different results if the voltage-sweep was reversed. (sweeping from higher voltage to lower, as opposed to sweeping reverse-voltage)
Might try a series RC in parallel with the diode, maybe 50ohm and 10pf, something like that to snub the oscillation on the curve tracer or Source meter.
The higher flat region in the AVERAGED voltage region of the RF/microwave oscillations. The drop to the lower flat is just where the oscillation ends after the stray capacitance charge has been depleted. That all is just my reasoning, or “educated guess”.
As far as I know about devices with negative differential resistance, you can either have voltage controlled devices that have a well defined current for each voltage point, and there are current controlled devices that have a well defined voltage for each current point. It would be very interesting to see how the results differ if you use the sourcemeter either in current- or voltage mode and compare the resulting curves.
In the forward direction there are three voltages that have the same current. One in the forward tunneling region, one in the negative resistance region and one in the normal forward region.
@@Bdix1256 Exactly, so the measurement done by IMSAIguy should be the right way because for each voltage there is only one defined current and I think that´s what he measured. Anyhow, because of the confusing results it would also be interesting to set the current and measure the voltage, maybe it lights up a little bit the "forbidden area".
I don’t know about the static sensitivity of this Russian GaAs tunnel but in general tunnel diodes are not really static sensitive (at least not in the way a mosfet is). They are just sensitive to excessive current. They are also not sensitive to radiation. I say this even though our data sheet does claim they are static sensitive. I think it’s just so the customers treat them nicely 😉
@@Bdix1256 You may be right , I think about reverse voltage. Remember , two diodes stop working couse touching and manipulating in hands . The junction capacity is low , this devices easly go up to 8 Ghz .... This diodes work in russian osciloscopes producing short pulses for synchronisations.
Wikipedia chapter and verse on tunnel diodes. A similar device the gunn diode is used in microwave oscillators before more conventional.semiconductiors became àvailable9 at microwave frequencies.
How does the curve tracer perform the sweep? Does it send out a voltage sweep and measure the current or does it send a current sweep and measure the voltage? When looking at the trace there seems to be a faint, very thin line crossing over from the breaking point horizontally to the normal diode curve, which suggests that the curve tracer sweeps a current. That way I don’t expect to see the drop in the middle as in that case the current would have to drop. The other test sweeps a voltage and in that case the current can drop as the voltage increases. How is the reverse behavior of this diode? Looks like it would make a perfect very low forward voltage rectifier for small signals like in a crystal set.
The switching occurs far faster than the scope can respond. It is usually specified in picoseconds. Note also that at low currents, the diode looks like a near short circuit.
"Negative resistance" seems like a bit of a misnomer here, creating a feeling of mystery where there should be none. The absolute resistance value always remains positive, what goes negative is its local derivative with respect to V, i.e. dR/dV. It's very weird behaviour (as with many nonlinear problems in general), but it's not breaking physics. As to why it doesn't appear to stop conducting nearly as quickly as it starts to... couldn't guess. Maybe it's interacting with the equipment, or maybe it's been damaged ¯\_(ツ)_/¯
@@edgeeffect No doubt they were made by the Russians. Working with gallium arsenide is a very expensive process. Not likely to be made by second sources.
What would that do? You need to run the ferrite on the VNA, not the curve tracer. It's a frequency dependent part, not a voltage dependent part. But yes, in general doing some analysis on ferrites would be interesting.
@@gorak9000That sudden jump in current at the discontinuity in the curve creates high frequency harmonics, no? Maybe I don't quite understand how the curve tracer works. Kind of new to this stuff, particularly at this level of detail, but find it fascinating.
@@AdmiralQuality Oh, you mean put a ferrite on the leg of the tunnel diode and run it on the curve tracer. I thought you meant use the curve tracer to show the effectiveness of a ferrite bead as a separate concept. Yeah, a ferrite might get you something on the curve tracer on the tunnel diode, but I suspect the Keithly that he used later has way better control of the source, and he was getting a decent graph out of it.
@@gorak9000 Yeah! I have no idea what the Keithly is exactly or how it works. I only discovered this channel recently and have a LOT of material to review. :)
@@AdmiralQuality a SMU (source measure unit) is like a very accurate (and usually four quadrant, so both source and sink both current and voltage) power supply that can also measure very accurately as well. So you can do voltage or current sweeps, and measure the other parameter. You can also set "compliance" values for either voltage or current too. My first exposure to one was in a Keithly 4200 SCS (semiconductor characterization system). With the preamps on some channels of the SMUs in that box, you can measure down to pA or fA - if your test setup is capable of preventing stray leakage currents enough to measure such small currents. Even moving the test leads from the SMU to the device can cause stray currents because of the triboelectric effect of the center conductor slightly rubbing on the dielectric in the cable caused by movement! A Keithly sales rep asked me what I was trying to measure, took one look at my setup, and laughed and laughed at me because my setup was so poor, but my supervisor wouldn't buy what was required to do it the right way, so what was I to do?? That's grad school for you!
For the non-curve-tracer testing that you performed... The test which used the Keithley 2400 to source voltage and measure current... Please do an additional video on how a hobbyist with limited lab equipment might set up a work-around to do the same test. I have a Siglent SDM3065X DMM (Kelvin,) SDS1204X-E scope, SPD3303X PSU, Rigol DG 1032Z 30MHz Function Generator (trying to upgrade it to 60MHz) - lots of additional/duplicate/analog equipment, including counters - an old laptop to tie it all together and to run the software. Please include details on the needed software. Also, is this really a more general question of how to fake a curve tracer? One that is better than the Asian kits, or the DIY "Octopus" circuits? Thank you, in advance!
The reverse bias region is actually where this device is generally used. It is sometimes referred to as a “back diode “. The response in this region is linear.
I searched digikey, there's many, sold by "marketplace" sellers not digikey themselves. Prices range from 250 and go up north of 1600 PER EACH depending on the frequency. The most expensive ones operate at 18GHz. So I'm going to go out on a limb here, and say "yes, they're rare". What did you try to buy one when you were 14 for? Seems like they were used as oscillators for radar back when 2GHz was "super ultra high frequency"
YT censors and deletes comments like crazy these days - you will conform to their unknown non-written and random rules, or you will disappear, capeesh? There are many channels that I don't even bother commenting on anymore - most of my comments disappear into the ether. Sometimes I can see them when I'm logged in, but in a not-logged-in incognito window, they're nowhere to be found
For a couple weeks, I was even getting pop up messages telling me my (completely innocuous) comments had been removed because they violated "community standards", and that if I did it again, my yt account would be suspended for 24 hours! I think I even was suspended for 24 hours and couldn't comment on anything once. Then suddenly that all stopped, and it went back to just silently deleting comments. Their AI has transcended human intelligence and now is way smarter than the meat bags that actually watch the videos and write the comments!
Your power supply testing to creat a curve is really wrong! Your PS isn't very fast in the current limit mode, because there is a relatively large capacitor at the output to smooth the voltage. B) your ps can't handle the negative resistance correctly. It expects the current increases with the voltage. A curve tracer is much faster to react because it doesn't have such a capacitor. I have never seen a tunnel diode myself. But my mentor told me in early 80s that by applying different voltage and settiing up currecly the nomad zone, i could have a VCO to fine tune the oscillation frequency. But not much. Tunnel diodes were essential elements in microwave and radars in 50s to 80s.
