Forums

PIN diode

Started by Unknown October 19, 2021
On Wed, 20 Oct 2021 18:30:45 -0400, legg <legg@nospam.magma.ca> wrote:

>On Wed, 20 Oct 2021 13:09:01 -0700, John Larkin ><jlarkin@highland_atwork_technology.com> wrote: > >>On Wed, 20 Oct 2021 14:50:53 -0400, legg <legg@nospam.magma.ca> wrote: >> >>>On Tue, 19 Oct 2021 19:20:04 -0700, jlarkin@highlandsniptechnology.com >>>wrote: >>> >>>>I want to inject a 100 ps test pulse into a 50 ohm transmission line, >>>>tee-wise, sometimes, from a 25 ohm source. So I need a series switch. >>>> >>>>I'd never paid much attention to PIN diodes... they are RF stuff. But >>>>this one is shocking: >>>> >>>>https://www.mouser.com/datasheet/2/249/MADP_008120_12790T-1921620.pdf >>>> >>>>2 ohms on, 0.14 pF off. >>>> >>>>And that's packaged. Chip and beam-lead parts are even better. >>>> >>>>Of course, in the long-honored RF tradition, there are no DC specs. No >>>>hint of the forward conduction curve. A tiny note suggests that 10 mA, >>>>1 volt might happen. >>> >>>What's so special? >>> >>>A simple re-sort of smd marking files by function, voltage and >>>current, pulls up quite a few similar or superior devices. >>> >>>http://ve3ute.ca/query/smd_PIN_V_A_210810E.zip >>> >>>RL >> >>The amps and volts numbers appear to be abs max, forward current and >>reverse voltage. That tells nothing about the conduction curve. >> >>RF! > >You mean the 'R' values . . . ?
What is the hfe column? Some values seem to be ohms (RF resistance?) and some are volts and some are in RT's and one is RfT! RF!
> >I'd assumed you were praising C and R specs. You can sort for >those. They're a start. For a nominal V/I plot, you'd have to >get the datasheet.
That was my point. PIN diode data sheets don't seem to have V/I data.
> >Nominal plots aren't much use in design, where every part used >has to work.
No data at all is better? That's all over RF data sheets. -- If a man will begin with certainties, he shall end with doubts, but if he will be content to begin with doubts he shall end in certainties. Francis Bacon
Gerhard Hoffmann wrote:
> Am 20.10.21 um 17:59 schrieb Jeroen Belleman: >> On 2021-10-20 17:42, Dimiter_Popoff wrote: >>> On 10/20/2021 5:20, jlarkin@highlandsniptechnology.com wrote: >>>> I want to inject a 100 ps test pulse into a 50 ohm transmission line, >>>> tee-wise, sometimes, from a 25 ohm source. So I need a series switch. >>>> >>>> I'd never paid much attention to PIN diodes... they are RF stuff. But >>>> this one is shocking: >>>> >>>> https://www.mouser.com/datasheet/2/249/MADP_008120_12790T-1921620.pdf >>>> >>>> 2 ohms on, 0.14 pF off. >>>> >>>> And that's packaged. Chip and beam-lead parts are even better. >>>> >>>> Of course, in the long-honored RF tradition, there are no DC specs. No >>>> hint of the forward conduction curve. A tiny note suggests that 10 mA, >>>> 1 volt might happen. > > No. the S-parameter tables give the conditions used for the measurement. > >>> Thanks for posting this John. I had been looking for something like >>> that for a pulse generator I might want to build one day (after >>> nearly 30 years this day may be coming closer...) and so far I >>> had seen only parts with 1-2 V reverse voltage ability; this one >>> seems to handle way more than I need. >> >> You do realize that it's not the PIN diode that makes the pulse, >> right? A pin diode is merely a switch, and not even a very fast >> one at that. > > and, for the LTspice fans:&nbsp; spice cannot simulate it. > > Gerhard
PIN diode switches exploit the same charge-storage phenomenon that's responsible for the reverse recovery behaviour of the diode. It's a transport phenomenon, governed by an integral equation that isn't reducible to a system of ODEs, so you're right, SPICE can't simulate it for real. (The reason is that the diode has internal state that isn't instantaneously visible at the pins.) It can be phonied up over some range of parameters, of course. Some years ago the late lamented JT, some others, and I had a big argument here on this very point. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Clifford Heath wrote:
> On 21/10/21 6:18 am, John Larkin wrote: >> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >> wrote: >> >>> Am 20.10.21 um 19:52 schrieb John Larkin: >>>> On Wed, 20 Oct 2021 18:05:10 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >>>> wrote: >>>> >>>>> >>>>> No. the S-parameter tables give the conditions used for the >>>>> measurement. >>>> >>>> >>>> It's the only suggestion of DC operating point in the entire document. >>>> Slightly better than nothing, I guess, generous by RF standards. >>>> >>>> This PIN data sheet has a Vf/If curve, absurdly wrong in the RF >>>> tradition. >>>> >>>> https://www.digikey.com/en/products/detail/infineon-technologies/BA895E6327/12109715 >>>> >>>> >>>> The general tend seems to be high Vf and high reverse leakage. >>> >>> There is capacitance vs. backward voltage&nbsp; and resistance vs. forward >>> current. On page 1.&nbsp; What else could one want? >> >> Voltage drop vs DC current. Diodes usually specify that, unless they >> are "RF" parts. >> >>> Getting 20 mA forward current through a xyzzy diode is for beginners. >> >> Beginners with big power supplies. > > RF switches typically need to stay turned on for the whole cycle. If you > want to pass 10mA of RF current, you need 20mA of DC, or more, depending > on linearity requirements. > > Not very familiar with class-A, are you John? RF is almost all class-A. > > CH
That would be so with a Schottky or very fast junction diode. The point of using PIN diodes is that even with a DC current much less than the peak RF current, an RF half-cycle transfers only a small portion of the stored charge in the junction, so the diode hardly notices. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Am 21.10.21 um 01:15 schrieb John Larkin:
> On Thu, 21 Oct 2021 00:22:10 +0200, Gerhard Hoffmann <dk4xp@arcor.de> > wrote:
>> Am 20.10.21 um 23:34 schrieb Clifford Heath: >>> On 21/10/21 6:18 am, John Larkin wrote: >>>> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk4xp@arcor.de>
>>>>> Getting 20 mA forward current through a xyzzy diode is for beginners.
>>>> Beginners with big power supplies.
>>> RF switches typically need to stay turned on for the whole cycle. If you >>> want to pass 10mA of RF current, you need 20mA of DC, or more, depending >>> on linearity requirements.
>> There is no way around the 20mA. It is needed that the serial resistance >> goes below 2 Ohms.
> A pin diode doesn't quit conducting when the current reverses. That's > the point of the thick I region with a long charge recombination > lifetime. A little DC current can switch a lot of RF current.
> The Macom part that I cited is under 2 ohms RF resistance at 100 MHz > and 10 mA DC. The RF current is limited by power dissipation. Which, > of sourse, they don't specify.
> Without useful data and without Spice models, I suppose people design > around these parts by fiddling.
you won't get a spice model because spice does not understand carrier lifetime which is crucial for PIN diode operation. You can easily get a Keysight ADS or a MWO model. But getting ADS or MWO will set you back > $100K unless you know the net and have enough criminal energy. ;-) Gerhard
On Wed, 20 Oct 2021 21:02:16 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>Clifford Heath wrote: >> On 21/10/21 6:18 am, John Larkin wrote: >>> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >>> wrote: >>> >>>> Am 20.10.21 um 19:52 schrieb John Larkin: >>>>> On Wed, 20 Oct 2021 18:05:10 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >>>>> wrote: >>>>> >>>>>> >>>>>> No. the S-parameter tables give the conditions used for the >>>>>> measurement. >>>>> >>>>> >>>>> It's the only suggestion of DC operating point in the entire document. >>>>> Slightly better than nothing, I guess, generous by RF standards. >>>>> >>>>> This PIN data sheet has a Vf/If curve, absurdly wrong in the RF >>>>> tradition. >>>>> >>>>> https://www.digikey.com/en/products/detail/infineon-technologies/BA895E6327/12109715 >>>>> >>>>> >>>>> The general tend seems to be high Vf and high reverse leakage. >>>> >>>> There is capacitance vs. backward voltage&#2013266080; and resistance vs. forward >>>> current. On page 1.&#2013266080; What else could one want? >>> >>> Voltage drop vs DC current. Diodes usually specify that, unless they >>> are "RF" parts. >>> >>>> Getting 20 mA forward current through a xyzzy diode is for beginners. >>> >>> Beginners with big power supplies. >> >> RF switches typically need to stay turned on for the whole cycle. If you >> want to pass 10mA of RF current, you need 20mA of DC, or more, depending >> on linearity requirements. >> >> Not very familiar with class-A, are you John? RF is almost all class-A. >> >> CH > >That would be so with a Schottky or very fast junction diode. The point >of using PIN diodes is that even with a DC current much less than the >peak RF current, an RF half-cycle transfers only a small portion of the >stored charge in the junction, so the diode hardly notices. > >Cheers > >Phil Hobbs
What's extra weird to me is that a regular diode would have a dynamic resistance around 5 ohms at 5 mA DC. But that Macom PIN diode is about 2 ohms at 5 mA. The mental model of a regular diode with an unusually wide I region doesn't seem to work. -- Father Brown's figure remained quite dark and still; but in that instant he had lost his head. His head was always most valuable when he had lost it.
Am 21.10.21 um 03:12 schrieb Gerhard Hoffmann:
> Am 21.10.21 um 01:15 schrieb John Larkin: >> On Thu, 21 Oct 2021 00:22:10 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >> wrote: > >>> Am 20.10.21 um 23:34 schrieb Clifford Heath: >>>> On 21/10/21 6:18 am, John Larkin wrote: >>>>> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk4xp@arcor.de> > >>>>>> Getting 20 mA forward current through a xyzzy diode is for beginners. > >>>>> Beginners with big power supplies. > >>>> RF switches typically need to stay turned on for the whole cycle. If >>>> you >>>> want to pass 10mA of RF current, you need 20mA of DC, or more, >>>> depending >>>> on linearity requirements. > >>> There is no way around the 20mA. It is needed that the serial resistance >>> goes below 2 Ohms. > > >> A pin diode doesn't quit conducting when the current reverses. That's >> the point of the thick I region with a long charge recombination >> lifetime. A little DC current can switch a lot of RF current. > >> The Macom part that I cited is under 2 ohms RF resistance at 100 MHz >> and 10 mA DC. The RF current is limited by power dissipation. Which, >> of sourse, they don't specify. > >> Without useful data and without Spice models, I suppose people design >> around these parts by fiddling. > > > you won't get a spice model because spice does not understand > carrier lifetime which is crucial for PIN diode operation. > You can easily get a Keysight ADS or a MWO model. > > But getting ADS or MWO will set you back > $100K unless you know the > net and have enough criminal energy. ;-) > > Gerhard
To make the Ingrid: I would not call it fiddling, applying the S-parameters and assume 20 mA DC through the diode, or -10V across. Gerhard (Ingrid was a women in the German sub-internet who was famous for endless auto-follow-ups where you could follow every step of her thinking...) And I already proposed QUCS studio as a simulator that also handles S-parameters. < https://dd6um.darc.de/QucsStudio/download.html >
On Thursday, October 21, 2021 at 12:28:23 PM UTC+11, jla...@highlandsniptechnology.com wrote:
> On Wed, 20 Oct 2021 21:02:16 -0400, Phil Hobbs <pcdhSpamM...@electrooptical.net> wrote: > >Clifford Heath wrote: > >> On 21/10/21 6:18 am, John Larkin wrote: > >>> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk...@arcor.de> wrote: > >>>> Am 20.10.21 um 19:52 schrieb John Larkin: > >>>>> On Wed, 20 Oct 2021 18:05:10 +0200, Gerhard Hoffmann <dk...@arcor.de> wrote: > >>>>> > >>>>>> No. the S-parameter tables give the conditions used for the measurement. > >>>>> > >>>>> It's the only suggestion of DC operating point in the entire document. > >>>>> Slightly better than nothing, I guess, generous by RF standards. > >>>>> > >>>>> This PIN data sheet has a Vf/If curve, absurdly wrong in the RF tradition. > >>>>> > >>>>> https://www.digikey.com/en/products/detail/infineon-technologies/BA895E6327/12109715
Which curve are you objecting to? The one at the bottom left of page 3 seems perfectly sensible to me.
> >>>>> The general tend seems to be high Vf and high reverse leakage. > >>>> > >>>> There is capacitance vs. backward voltage and resistance vs. forward > >>>> current. On page 1. What else could one want? > >>> > >>> Voltage drop vs DC current. Diodes usually specify that, unless they > >>> are "RF" parts. > >>> > >>>> Getting 20 mA forward current through a xyzzy diode is for beginners. > >>> > >>> Beginners with big power supplies. > >> > >> RF switches typically need to stay turned on for the whole cycle. If you > >> want to pass 10mA of RF current, you need 20mA of DC, or more, depending > >> on linearity requirements. > >> > >> Not very familiar with class-A, are you John? RF is almost all class-A. > > > >That would be so with a Schottky or very fast junction diode. The point > >of using PIN diodes is that even with a DC current much less than the > >peak RF current, an RF half-cycle transfers only a small portion of the > >stored charge in the junction, so the diode hardly notices. > > What's extra weird to me is that a regular diode would have a dynamic > resistance around 5 ohms at 5 mA DC. But that Macom PIN diode is about > 2 ohms at 5 mA. > > The mental model of a regular diode with an unusually wide I region > doesn't seem to work.
Your mental model seems to leave out the capacitative element of the PIN diode impedance - it's clearly not a resistive impedance, and talking about "dynamic resistance" does make it obvious where your problem comes from. If you'd been able to type "dynamic impedance" the weirdness might not have been there. -- Bill Sloman., Sydney
On 21/10/21 12:02 pm, Phil Hobbs wrote:
> Clifford Heath wrote: >> On 21/10/21 6:18 am, John Larkin wrote: >>> On Wed, 20 Oct 2021 20:24:22 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >>> wrote: >>> >>>> Am 20.10.21 um 19:52 schrieb John Larkin: >>>>> On Wed, 20 Oct 2021 18:05:10 +0200, Gerhard Hoffmann <dk4xp@arcor.de> >>>>> wrote: >>>>> >>>>>> >>>>>> No. the S-parameter tables give the conditions used for the >>>>>> measurement. >>>>> >>>>> >>>>> It's the only suggestion of DC operating point in the entire document. >>>>> Slightly better than nothing, I guess, generous by RF standards. >>>>> >>>>> This PIN data sheet has a Vf/If curve, absurdly wrong in the RF >>>>> tradition. >>>>> >>>>> https://www.digikey.com/en/products/detail/infineon-technologies/BA895E6327/12109715 >>>>> >>>>> >>>>> The general tend seems to be high Vf and high reverse leakage. >>>> >>>> There is capacitance vs. backward voltage&nbsp; and resistance vs. forward >>>> current. On page 1.&nbsp; What else could one want? >>> >>> Voltage drop vs DC current. Diodes usually specify that, unless they >>> are "RF" parts. >>> >>>> Getting 20 mA forward current through a xyzzy diode is for beginners. >>> >>> Beginners with big power supplies. >> >> RF switches typically need to stay turned on for the whole cycle. If >> you want to pass 10mA of RF current, you need 20mA of DC, or more, >> depending on linearity requirements. >> >> Not very familiar with class-A, are you John? RF is almost all class-A. >> >> CH > > That would be so with a Schottky or very fast junction diode.&nbsp; The point > of using PIN diodes is that even with a DC current much less than the > peak RF current, an RF half-cycle transfers only a small portion of the > stored charge in the junction, so the diode hardly notices.
Yes, thanks, JL also pointed that out. That must also be relevant to how the Skyworks limiter diodes work too. Once they turn on, they stay on in both directions for a number of cycles at least. CH
On Wednesday, October 20, 2021 at 6:02:35 PM UTC-7, Phil Hobbs wrote:

>
The point
> of using PIN diodes is that even with a DC current much less than the > peak RF current, an RF half-cycle transfers only a small portion of the > stored charge in the junction, so the diode hardly notices.
And it's an important diagnostic tool, to be able to sweep (slowly) the bias while probing the RF capacitance; it tells you the doping profile of the junction. 'Hyperabrupt' doping gives the biggest capacitance variation, good for varactors.
On 10/20/2021 4:41 PM, Clifford Heath wrote:
> On 21/10/21 3:00 am, jlarkin@highlandsniptechnology.com wrote: >> On Wed, 20 Oct 2021 18:42:53 +0300, Dimiter_Popoff <dp@tgi-sci.com> >> wrote: >> >>> On 10/20/2021 5:20, jlarkin@highlandsniptechnology.com wrote: >>>> I want to inject a 100 ps test pulse into a 50 ohm transmission line, >>>> tee-wise, sometimes, from a 25 ohm source. So I need a series switch. >>>> >>>> I'd never paid much attention to PIN diodes... they are RF stuff. But >>>> this one is shocking: >>>> >>>> https://www.mouser.com/datasheet/2/249/MADP_008120_12790T-1921620.pdf >>>> >>>> 2 ohms on, 0.14 pF off. >>>> >>>> And that's packaged. Chip and beam-lead parts are even better. >>>> >>>> Of course, in the long-honored RF tradition, there are no DC specs. No >>>> hint of the forward conduction curve. A tiny note suggests that 10 mA, >>>> 1 volt might happen. >>>> >>>> >>>> >>> >>> Thanks for posting this John. I had been looking for something like >>> that for a pulse generator I might want to build one day (after >>> nearly 30 years this day may be coming closer...) and so far I >>> had seen only parts with 1-2 V reverse voltage ability; this one >>> seems to handle way more than I need. >> >> Yeah, the specs shocked me, after struggling to do this with a >> schottky. The carrier lifetime could be a problem with long pulses. >> >> I love pulse generators. Let me know if I can help. >> >> There are all sorts of cheap fungens and scopes and such, but pulse >> generators are generally still mediocre and expensive. I want to do >> one myself some day, a really fast version of our DDG. > > There is plenty of excitement around the NanoVNA and the TinySA. > It's time to do a TinyTDR. Bet you'd sell many thousands. If you do the > pulse gen and the sampler I'll do the rest. > > Clifford Heath
The NanoVNA does TDR. <https://www.youtube.com/watch?v=9thbTC8-JtA>