Gunns


Chris Bartram G4DGU
 

Hello Dave,

Being negative resistance devices, Gunn oscillator cavity design isn't necessarily intuitive, and putting two into a cavity will probably lead to unexpected results! The guy from Virginia Diodes was probably correct ...

Chris

G4DGU


Paul Evans W4/VP9KF
 

Well, I was the 2T [two terminal] Device Manager at Plessey Microwave, so I should be able to help you :-) My coverage was from wafer selection and test, epitaxy for all the devices we made, cleanroom design and monitoring, process control and modification, chip and wire bonding, packaging, test and installation into product. I also covered varactors, detectors, IMPATTs and TRAPATTs.

First, Gunns are not diodes, but they became known as that pretty universally by mistake in the 60s. Secondly, they only display negative resistance above a certain voltage where RF also starts to be generated under the right conditions. Despite extremely tight measurement and QA I used to say 'every single device we make is different'. I never saw any attempt at combining the output of two devices and I would give it zero chance of working (even if only for a very short time under just the right conditions - mostly thermal). I know of several systems where we would dearly have liked to combine devices outputs for more power!

Before having all the above dropped into my hands, I was an oscillator designer and I can tell you that Gunns have to be in just the right place to achieve oscillation and matching to their (cavity) environment and other [detection or varactor] devices. Redesigning a 22GHz design to get a reasonable number of man-hours to make one work was a case of going back to fundamental design yet making it look the same on the outside.

Taking really cheap diecast oscillator cavities and turning them into transitions seems like making life overly difficult. Try using an established design.
--
73, Paul Evans G4BKI [VP9KF]

P.S. I was fortunate to have studied under Prof. Paul Butler at University of Warwick who 'wrote the book' on hot electron theory which matched John Gunn's achieved best-case Gunn performance, amongst other things.


Andy G4JNT
 

At Malvern, in 1977, they had ten or twenty or so Gunns successfully running together in a development aimed at a personnel detector radar (Doppler, like traffic radars) 
Don't know the exact details, but he said it was quite a simple arrangement, with the Gunns mounted periodically along the waveguide.
I know this coz a fellow MoD student engineer was on the project for his summer placement and knew my interest in Am. microwaves

I'm guessing the Gunns would be mounted a guide wavelength apart.



On Wed, 9 Jun 2021 at 13:16, Paul Evans W4/VP9KF <paul@...> wrote:
Well, I was the 2T [two terminal] Device Manager at Plessey Microwave, so I should be able to help you :-) My coverage was from wafer selection and test, epitaxy for all the devices we made, cleanroom design and monitoring, process control and modification, chip and wire bonding, packaging, test and installation into product. I also covered varactors, detectors, IMPATTs and TRAPATTs.

First,  Gunns are not diodes, but they became known as that pretty universally by mistake in the 60s. Secondly, they only display negative resistance above a certain voltage where RF also starts to be generated under the right conditions. Despite extremely tight measurement and QA I used to say 'every single device we make is different'. I never saw any attempt at combining the output of two devices and I would give it zero chance of working (even if only for a very short time under just the right conditions - mostly thermal). I know of several systems where we would dearly have liked to combine devices outputs for more power!

Before having all the above dropped into my hands, I was an oscillator designer and I can tell you that Gunns have to be in just the right place to achieve oscillation and matching to their (cavity) environment and other [detection or varactor] devices. Redesigning a 22GHz design to get a reasonable number of man-hours to make one work was a case of going back to fundamental design yet making it look the same on the outside.

Taking really cheap diecast oscillator cavities and turning them into transitions seems like making life overly difficult. Try using an established design.
--
73, Paul Evans G4BKI [VP9KF]

P.S. I was fortunate to have studied under Prof. Paul Butler at University of Warwick who 'wrote the book' on hot electron theory which matched John Gunn's achieved best-case Gunn performance, amongst other things.






Andy G4JNT
 

Paul, can you remember if you gave a talk on GaAs to Southampton University electronics students in 1979 or 1980?  Think it was given by a Plessey chap
I remember the talk was 'optional', but of course I went to it.   A few months later, in the RF and Microwaves Finals exam paper was a question "Write an essay entitled 'Gallium Arsenide, the Microwave Material' "
A few buzz-words like 'high-mobility', 'bulk effect' and 'FETs' (very new in in 1980) and there was full marks awarded.  
ISTR getting something like 95% in that paper



On Wed, 9 Jun 2021 at 13:16, Paul Evans W4/VP9KF <paul@...> wrote:
Well, I was the 2T [two terminal] Device Manager at Plessey Microwave, so I should be able to help you :-) My coverage was from wafer selection and test, epitaxy for all the devices we made, cleanroom design and monitoring, process control and modification, chip and wire bonding, packaging, test and installation into product. I also covered varactors, detectors, IMPATTs and TRAPATTs.

First,  Gunns are not diodes, but they became known as that pretty universally by mistake in the 60s. Secondly, they only display negative resistance above a certain voltage where RF also starts to be generated under the right conditions. Despite extremely tight measurement and QA I used to say 'every single device we make is different'. I never saw any attempt at combining the output of two devices and I would give it zero chance of working (even if only for a very short time under just the right conditions - mostly thermal). I know of several systems where we would dearly have liked to combine devices outputs for more power!

Before having all the above dropped into my hands, I was an oscillator designer and I can tell you that Gunns have to be in just the right place to achieve oscillation and matching to their (cavity) environment and other [detection or varactor] devices. Redesigning a 22GHz design to get a reasonable number of man-hours to make one work was a case of going back to fundamental design yet making it look the same on the outside.

Taking really cheap diecast oscillator cavities and turning them into transitions seems like making life overly difficult. Try using an established design.
--
73, Paul Evans G4BKI [VP9KF]

P.S. I was fortunate to have studied under Prof. Paul Butler at University of Warwick who 'wrote the book' on hot electron theory which matched John Gunn's achieved best-case Gunn performance, amongst other things.






Paul G8KFW
 

Hi Paul

I have a number of gun diodes but do not know their polarity can you help me on how to test them and determine their polarity without destroying them

Regards Paul

-----Original Message-----
From: UKMicrowaves@groups.io [mailto:UKMicrowaves@groups.io] On Behalf Of Paul Evans W4/VP9KF
Sent: 09 June 2021 13:16
To: UKMicrowaves@groups.io
Subject: Re: [UKMicrowaves] Gunns

Well, I was the 2T [two terminal] Device Manager at Plessey Microwave, so I should be able to help you :-) My coverage was from wafer selection and test, epitaxy for all the devices we made, cleanroom design and monitoring, process control and modification, chip and wire bonding, packaging, test and installation into product. I also covered varactors, detectors, IMPATTs and TRAPATTs.

First, Gunns are not diodes, but they became known as that pretty universally by mistake in the 60s. Secondly, they only display negative resistance above a certain voltage where RF also starts to be generated under the right conditions. Despite extremely tight measurement and QA I used to say 'every single device we make is different'. I never saw any attempt at combining the output of two devices and I would give it zero chance of working (even if only for a very short time under just the right conditions - mostly thermal). I know of several systems where we would dearly have liked to combine devices outputs for more power!

Before having all the above dropped into my hands, I was an oscillator designer and I can tell you that Gunns have to be in just the right place to achieve oscillation and matching to their (cavity) environment and other [detection or varactor] devices. Redesigning a 22GHz design to get a reasonable number of man-hours to make one work was a case of going back to fundamental design yet making it look the same on the outside.

Taking really cheap diecast oscillator cavities and turning them into transitions seems like making life overly difficult. Try using an established design.
--
73, Paul Evans G4BKI [VP9KF]

P.S. I was fortunate to have studied under Prof. Paul Butler at University of Warwick who 'wrote the book' on hot electron theory which matched John Gunn's achieved best-case Gunn performance, amongst other things.








--
Paul Bicknell G8KFW   South Coast UK


Paul Evans W4/VP9KF
 

Sorry, before my time - graduated in 1979. Strangely Southampton was on my 'list of 5' when applying, but never went for an interview, never mind a talk.

The person who gave you the talk would have been from Caswell Research and could have been Fred Myers (tall, dark haired guy) or Jim Turner (short, orange-ish hair and beard). They were in on making the worlds first MMICs, before USA, etc. It could have been someone else, they were very heavy into GaAs and epitaxy [vapour phase and later MOCVD].

I did use a GAT3 and GAT5 at one time - they were as rare as hens teeth! If their yield was anything like the normal Caswell yield then that's why they were rare :-(

Most Malvern things didn't end well. So the multi Gunn thing was probably abandoned and done an easier way. All I do know is that they didn't exactly have repeatability on their side. The most notable being the long held rumour that one day they made a 'magic' wafer which produced some Gunns that were twice+ as efficient as any recorded before or since [and exceeded theoretical power yield too]. We never saw anything like that in production, or we would have chased it HARD.

By the way, I did noise test measurements on our devices vs. the competition (Alpha, M/A, etc.) and confirmed what our customers were saying when ordering. We were waaaaay ahead on phase noise and stability. It was down to our epitaxy and not our fab (as I later proved). In the 80s all of the temporary traffic light heads were coming from us, radar speed meters (Escort, etc.) as well as buckets full of supermarket door openers! Of course, the military side was a little harder to yield. What few would know is that every wafer was run down the line to full mil. spec. until being dropped in superiority at test to 'commercial use'.
--
73, Paul Evans G4BKI [VP9KF]


 

On Wed, 9 Jun 2021 at 10:41, Chris Bartram G4DGU <chris@...> wrote:
Hello Dave,

Being negative resistance devices, Gunn oscillator cavity design isn't
necessarily intuitive, and putting two into a cavity will probably lead
to unexpected results! The guy from Virginia Diodes was probably correct ...

Chris

G4DGU
 
Yes, I think so too. The guy from Virginia Diodes did not even have to think about it - his answer was immediate. He was not in the least bit surprised.

In contrast, the academic from Scotland who said it was rubbish, was not considering the fact that the DC drive to the diodes was far more than the RF output power.


Graham G3VKV
 

Hi all,
Back in the 80's I used two Solfan iris coupled Gunn oscillators across a magic T to produce more than twice
the output of one unit for wideband FM - a 9mW and 11mW produced 30mW when locked up 
I took the assembly to Crawley Court Round table meeting where it was tested to confirm the results,
The late Tim G3KEU also made one and reported it in the Oct/Nov 1988 Microwave newsletter.
I used mine portable for a number of years. (still got it somewhere)
see
Bob Atkins - KA1GT
                                          Graham G3VKV


 

On Wed, 9 Jun 2021 at 13:16, Paul Evans W4/VP9KF <paul@...> wrote:
Well, I was the 2T [two terminal] Device Manager at Plessey Microwave, so I should be able to help you :-) My coverage was from wafer selection and test, epitaxy for all the devices we made, cleanroom design and monitoring, process control and modification, chip and wire bonding, packaging, test and installation into product. I also covered varactors, detectors, IMPATTs and TRAPATTs.

First,  Gunns are not diodes, but they became known as that pretty universally by mistake in the 60s. Secondly, they only display negative resistance above a certain voltage where RF also starts to be generated under the right conditions. Despite extremely tight measurement and QA I used to say 'every single device we make is different'. I never saw any attempt at combining the output of two devices and I would give it zero chance of working (even if only for a very short time under just the right conditions - mostly thermal). I know of several systems where we would dearly have liked to combine devices outputs for more power!
--
73, Paul Evans G4BKI [VP9KF]

Combining two Gunn diodes to get increased power was definitely published in an amateur journal, but I don't know which. I did not dream it, so it has been done. I would expect there to be injection locking, so they are both on the same frequency.

The guy from Virginia diodes was not at all surprised by this. Virginia Diodes make Gunn diodes at over 1 THz.

I recall giving a talk at our radio club on the diodes once. I had a power supply with an ammeter, and people could see the current rise as the voltage was increased. Then when the current started falling, everyone was perplexed. Negative resistance, or more precisely negative dynamic resistance, is not something most people are aware of.

Dave


Paul Evans W4/VP9KF
 

VERY interesting... especially as Solfan used their own mechanical assembly, but we (Plessey) supplied all their Gunn devices! I never did know why they were the few 'device only' buyers, but it had to come down to price. The target for most of those types of units was barely >$10. The device package for the Gunn cost about GBP2.00 and making the device itself was about the same. That doesn't leave much change to make the assembly and test. I seem to remember the target spec was 10 mW. I think they went out in volumes of up to 5,000 per month at times. Our test was a very quick drop in a test rig, check power out and on to the next. I can only assume that the devices were acting as parametric pumps against each other, in the same way a parametric pump amplifier can be made. Of course, this is all a bit mute now because the use of Gunns has been superceded by FET oscillators. By the time we closed, the first volume DBS LOs were being made in-house using DROs [in fact I was in the last skeleton team of 8 people fulfilling an order for such FETs over a period of 5 months].

--
73, Paul Evans G4BKI [VP9KF]


 

On Wed, 9 Jun 2021 at 19:26, Paul Evans W4/VP9KF <paul@...> wrote:
VERY interesting... especially as Solfan used their own mechanical assembly, but we (Plessey) supplied all their Gunn devices! I never did know why they were the few 'device only' buyers, but it had to come down to price. The target for most of those types of units was barely >$10. The device package for the Gunn cost about GBP2.00 and making the device itself was about the same. That doesn't leave much change to make the assembly and test. I seem to remember the target spec was 10 mW. I think they went out in volumes of up to 5,000 per month at times. Our test was a very quick drop in a test rig, check power out and on to the next. I can only assume that the devices were acting as parametric pumps against each other, in the same way a parametric pump amplifier can be made. Of course, this is all a bit mute now because the use of Gunns has been superceded by FET oscillators.
 
Gunn diodes are still made for THz work. They are not all museum pieces now.

By the time we closed, the first volume DBS LOs were being made in-house using DROs [in fact I was in the last skeleton team of 8 people fulfilling an order for such FETs over a period of 5 months].

--
73, Paul Evans G4BKI [VP9KF]
Dave, G9WRB