PIN diode used instead of SRD

[BlackBishop]

Hi - this is my first question posted on this forum - and I have more to go later

I saw some docs and pages on the web about using cheap PIN diodes instead hard to get and expensive SRDs but they were quite general.
You may already guessed it is to be used as pulse generator with very short rise / fall times.

I don't know how fast PIN diodes could be in such application. Is it possible to go 50ps or lower or is that SRD only region?

Assuming that I have spec for SRD - there are junction capacitance, carrier life time, rise time and breakdown voltage.
For example: 0.15pF, 4ns, 30ps, 14V.
Quick analisis of SRDs show those with lower capacitance and lower carrier life time have shorter rise times.
Is it enough to pick those PINs with the lowest junction capacitance and carrier life time and expect decent rise times?

 

[jiripolivka]

I think you are in an error . PIN diodes have the I intrinsic layer to slow their RF response. Therefore they behave in RF circuits as very poor detectors, rather they are used as variable resistors. This property makes them good variable attenuators or switches.

SRD or avalanche diodes behave quite differently. They accumulate charge in one current direction and discharge it fast in the reverse direction. This property offers to generate harmonic signals, and therefore SRD diodes are used as frequency multipliers.

Maybe you should mention what is the purpose you intend to use one of those devices. PIN and SRD diodes are simply quite different animals.

 

[vfone]

Some kind of epitaxial PIN diodes have characteristics which are very similar to those of expensive Step Recovery Diodes, and particularly when are used in frequency multipliers they can replace SRD.

**broken link removed**

There are a lot of Agilent application notes showing this.

 

[jiripolivka]

I have read the avago data sheet but there is no mention recommending the PIN diodes as frequency multipliers.
SRD diodes are expensive for a reason; their snap-discharge does generate harmonics much better than any other kind of diode.
If you want to make a frequency multiplier for a low cost, then test any other diodes in that function. With some selection work and a good luck, many switching, mixer and detector diodes do function as good multipliers.

I do not know your specification but at Spacek Labs. we use our Schottky diodes as frequency multipliers up to 110 GHz. We have diode doublers, triplers and quadruplers, see "www.spaceklabs.com". The same diodes are excellent in low-loss balanced mixers.

We have used SRD diodes in varactor frequency multipliers but as those devices are "parametric" and function only at one frequency set and are critically dependent upon input power level, they are no more used.

You can try a PIN diode as a frequency multiplier; I would not lose time in that as an ordinary diode is usually better.

 

[BlackBishop]

Ok - it doesn't have to be PIN diode, could be anything obtainable.
What I try to do is sampler so it needs a generator able to produce short pulses with sharp edges.
Most circuits I saw used SRDs - but very few companies still produce them, almost impossible to find where to buy online.
Tunnel diodes seem ok too but again - this is legacy technology - even though they are quite cheap on ebay it is limited resource.
Than I read some Agilent application notes about using PIN as frequency multipliers - I have it somewhere on my disk but can't find it right now.
I even have some agilent HSMP PIN diodes laying around (but those "wrong" ones with 200ns life time :evil
As far as I remember they didn't go above 10GHz.

So here is what I would like to find:
- something capable of generating step pulses - the shorter the better
- quite low repetition rate - 1MHz is fine - anything between 100kHz - 100Mhz is fine
- pulse strong enough to drive quad / octal shottky diodes bridge
- parts available on market and acceptable prices - it is not going to be government sponsored hi-tech project but something for personal and educational use - so if something goes wrong I don't want to cry for a whole day that I broke expensive part that is no more produced

This may help you to help me.

 

[jiripolivka]

From your response it sounds like you need a sample-and-hold circuit. I do not work in this area, sorry for now.
Somewhere deep in my records I saw an interesting paper by an Italian author about how he built an experimental UWB "radar" from a scratch. He used standard fast TTL devices, some RF transistors and switching diodes, to generate fast sampling pulses.
In his UWB transmitter he generated 'half-sine' pulses, in the receiver, a similar keyed diode bridge was used to detect and process the signal.

Please try to google UWB signal generation and reception. Maybe you can find something useful.

The fastest-response PIN MMIC I know is one by Agilent, operating at Ka-band, with < 4 ns response. Maybe Hittite has something even better.

 

[BlackBishop]

what I interested in is generation of monopulse which seems harder than multiplying frequency by for example diode bridge
yeah - there are NLTLs out there but lets keep close to ground

BTW - here is an example use of PIN diodes as freq multipliers

 

[jiripolivka]

Thank you for the reference! It is about a quite low frequency output; I used in 1980 a 74F04 with an OCXO input at 26 MHz; my spectrum was a bit better I remember. I think any Schottky detector or mixer diode can generate a better spectrum.

If I have time, somewhere I have a copy of that "UWB intro" paper. Maybe there are more papers on how to generate an UWB monopulse pattern. Please see "www.picosecond.com", they have interesting white papers and products in this UWB field.

 

[FvM]

AN 1054 is in fact the only paper I'm aware of, that's talking about PIN diodes behaving quite similar to SRD. But it's not talking about pulse generation and respective parameters. As long as you don't have other references, I don't see any other option than trying some of the said pin diodes.

Do I guess right, that you most likely face difficulties to do this, because you don't have access to the required measurement equipment?

 

[jiripolivka]

You are right, UWB pulses need expensive equipment to generate and observe. In my work I prefer to use frequency spectrum before time response. Mathematically both are equivalent, so if my transmission system has an overall bandwidth of 1 or 10 GHz, I know it can transmit pulse pattern of 1.0 or 0.1 nanosecond. Our customers then sent us eye patterns showing that our system can transmit 2 or 5 Gb/s data while we have no test equipment here. But it works.

 

[biff44]

The "I" region in a pin diode can act a little like the drift region of an SRD. Obviously the SRD is optimized for step currents, while the PIN diode is optimized for something entirely different.

The thing that makes the big voltage spike is the series inductor leading up to the diode, not the diode itself. What you need is a device that will conduct for a small portion of the half-cycle of the input sine wave, and then have the current abruptly go to zero (like when the last electron in the "I" region is either collected or recombines on its own). Then the inductor generates a big voltage spike.

To some extent, the width of the voltage spike is controlled by that inductor, not the diode.