how to design PIN diode driver - high voltage and high speed

[Tiger]

Now i need a high-power T-R switch that can switch state in two
microseconds. Drive outputs: +12 volts(current ,0.5 ampers) , +700
volts , -12volts(-0.5 ampers), -12 volts. Tx path may stand 600
watts
(peak). I want to use PIN driver IC,but don't find IC that may output
+700 volt. So i have to design the switch by myself.
I am going to employ discrete MOSFET as driver . Input to the
MOSFET is TTL logic.
some specification of my switch is :
working frequency band : 3 - 30 MHz
T - R isolation : > -80 dB (when transmitting)

Anyone ever do this? Can i get so fast speed(< 2 microseconds)?
Would you give me some suggestions? Any help will be appreciated!

 

[Tom Bruhns]

Now i need a high-power T-R switch that can switch state in two
microseconds. Drive outputs: +12 volts(current ,0.5 ampers) , +700
volts , -12volts(-0.5 ampers), -12 volts. Tx path may stand 600
watts
(peak). I want to use PIN driver IC,but don't find IC that may output
+700 volt. So i have to design the switch by myself.
I am going to employ discrete MOSFET as driver . Input to the
MOSFET is TTL logic.
some specification of my switch is :
working frequency band : 3 - 30 MHz
T - R isolation : > -80 dB (when transmitting)

Anyone ever do this? Can i get so fast speed(< 2 microseconds)?
Would you give me some suggestions? Any help will be appreciated!

What's the carrier lifetime of the PIN switches you'll be using? That
probably has more to do with how fast you'll be able to switch between
T and R than the FETs you use. What carrier lifetime do you need to
do a good job with your 600W at 3MHz? But I'd also like to know why
you think you need to switch so quickly for that frequency band, since
the transmission modes allowed are not particularly broadband. Also--
why do you need so much isolation? 50dB gets you below 10mW, and I'd
consider it a pretty poor receiver front end that couldn't handle that
without damage. 80dB attenuation won't be as easy over a decade of
frequencies as it would if you were dealing with a single frequency or
a narrow band, though you can add however many stages of low power PIN
or simple diode switches or other RF switches to get to whatever
attenuation you need, after your high power PIN switch. Note that 10
amps in an FET will slew 1000pF 10 volts per nanosecond...

Have you looked for ap notes on the web sites of manufacturers of high
power PIN diodes?

 

[Tom Bruhns]

But I'd also like to know why
you think you need to switch so quickly for that frequency band, since
the transmission modes allowed are not particularly broadband.

(Perhaps your use is something like ultrasound, in which case you
might want something faster than 2usec, but the 3-30MHz is
suspiciously like a spec for HF over-the-air transmissions...)

 

[Tiger]

What's the carrier lifetime of the PIN switches you'll be using? That
probably has more to do with how fast you'll be able to switch between
T and R than the FETs you use. What carrier lifetime do you need to
do a good job with your 600W at 3MHz? But I'd also like to know why
you think you need to switch so quickly for that frequency band, since
the transmission modes allowed are not particularly broadband. Also--
why do you need so much isolation? 50dB gets you below 10mW, and I'd
consider it a pretty poor receiver front end that couldn't handle that
without damage. 80dB attenuation won't be as easy over a decade of
frequencies as it would if you were dealing with a single frequency or
a narrow band, though you can add however many stages of low power PIN
or simple diode switches or other RF switches to get to whatever
attenuation you need, after your high power PIN switch. Note that 10
amps in an FET will slew 1000pF 10 volts per nanosecond...

Have you looked for ap notes on the web sites of manufacturers of high
power PIN diodes?

The carrier lifetime of PIN diode i use is 10microseconds(Typ.) .
Part number is UM4010,manufactured by Microsemi Corp. I have read some
ap notes of PIN Diodes, and got to know that switching speed of my
antenna switch is not only related to carrier lifetime of PIN
diodes ,but also related to driver . A good driver may get faster
switching speed:may up to one fifth or one tenth of carrier
lifetime(please refer to great book "Semiconductor control',writen by
White J F. ,1977) . In order to switch the RF signal(3 - 30 MHz) with
less distortion, accroding to calculation , the carrier lifetime must
longer than 8 microseconds, So i choose UM4010.
The reason of switch speed < 2 microseconds is that transmitting
signal will reach receving antenna in 2 microseconds.
I agree with you , it is not easy to gain 80 dB isolation over a
decade of frequencies. That high isolation may improve the recept
signal to noise ratio £¬avoid Saturation of recept preamplifer. I am
going to take CPW technology to improve isolation besides care of
designing bias circuits.
Yes , My switch is used as a key assembly of ionospheric sounding
system (HF band) .

PS£ºUM4010, Carrier lifetime,10 microseconds; Rs(IF=100 mA@100MHz),
0.5 ohms; Vr(breakdown voltage),1000 volts;
PD(dissipation power),20 watts.
http://www.microsemi.com/catalog/parmlist.asp?P0_CAT=RF&P1_TYPE=PIN&LVP=2&LVP1=0

 

[Tom Bruhns]

The carrier lifetime of PIN diode i use is 10microseconds(Typ.) .
Part number is UM4010,manufactured by Microsemi Corp. I have read some
ap notes of PIN Diodes, and got to know that switching speed of my
antenna switch is not only related to carrier lifetime of PIN
diodes ,but also related to driver . A good driver may get faster
switching speed:may up to one fifth or one tenth of carrier
lifetime(please refer to great book "Semiconductor control',writen by
White J F. ,1977) . In order to switch the RF signal(3 - 30 MHz) with
less distortion, accroding to calculation , the carrier lifetime must
longer than 8 microseconds, So i choose UM4010.
The reason of switch speed < 2 microseconds is that transmitting
signal will reach receving antenna in 2 microseconds.
I agree with you , it is not easy to gain 80 dB isolation over a
decade of frequencies. That high isolation may improve the recept
signal to noise ratio £¬avoid Saturation of recept preamplifer. I am
going to take CPW technology to improve isolation besides care of
designing bias circuits.
Yes , My switch is used as a key assembly of ionospheric sounding
system (HF band) .

PS£ºUM4010, Carrier lifetime,10 microseconds; Rs(IF=100 mA@100MHz),
0.5 ohms; Vr(breakdown voltage),1000 volts;
PD(dissipation power),20 watts.
http://www.microsemi.com/catalog/parmlist.asp?P0_CAT=RF&P1_TYPE=PIN&L...

So the receiving antenna and transmitting antenna are separate, and
about 2000 feet apart?? If that is the case, why do you need a TR
switch? I trust that the transmitter is not so high power that the
receiving antenna is trying to deliver 600 watts to the receiver...and
even if it were, you wouldn't need to reverse-bias the PIN by several
hundred volts. Maybe I'm reading too much into what you wrote.

In any event, it seems like it should be reasonably easy to switch
~700V in well under a microsecond. It's not so different from what
switching power supplies do using half-bridge configurations, and in
your case, you mainly need to supply the current to charge/discharge
capacitance, and the DC current to bias the PIN "on" is modest. For
example, see the data sheet for the Fairchild FQH8N100C 1kV 8A power
mosfet; switching times listed (rise, fall, delays) are in the range
of a few tens of nanoseconds up to 250nsec max.

 

[Tiger]

I am sorry ,because i express unclear. in my system ,Transmitter
and receiver share one antenna.
Accounting for RF choke in DC bias path, A low path filter must
be used. thus in order to get faster switch wave rising edge ,we have
to decrease time constant of circuit . The filter's cutoff frequency
is closer to 3MHz, better rising edge or falling edge of switch wave
we achieve. Do you agree with me?
As to FQH8N100C you mentioned , i am reading it. Thank you for your
advice!

 

[Tom Bruhns]

I am sorry ,because i express unclear. in my system ,Transmitter
and receiver share one antenna.
Accounting for RF choke in DC bias path, A low path filter must
be used. thus in order to get faster switch wave rising edge ,we have
to decrease time constant of circuit . The filter's cutoff frequency
is closer to 3MHz, better rising edge or falling edge of switch wave
we achieve. Do you agree with me?
As to FQH8N100C you mentioned , i am reading it. Thank you for your
advice!

Yes, I would certainly agree that proper filtering to separate the PIN
bias from the RF is a good idea. The faster you need to switch, the
more critical that becomes. It should be an advantage to consider not
only a low-pass filter for the bias but a high pass filter for the RF--
or other methods to reduce the capacitance that the PIN switch must
drive. Of course, be careful about the return path for the current
that charges the capacitance especially on the receiver side, so that
you do not expose the receiver input to excessive current. A power
mosfet that switches several hundred volts in a few tens of
nanoseconds will be generating quite a bit of available energy to a 50
ohm 3-30MHz input! If there exists between the fast PIN driver
circuit and the PIN diodes a low pass filter, and also between the PIN
switches and the receiver input (or transmitter output) a high pass
filter, and the filter passbands do not overlap, then there should be
decent isolation to protect the receiver input. I can tell you that I
also use PIN diode clamps on the input to a receiver I recently put
into production, and they are able to absorb transient energy quite
nicely. They are just diodes normally reverse-biased by several
volts. In a switched system, I would consider driving them actively,
using differential drive so that there is no (or very low) transient
on the RF input that they shunt to ground when they are "on." If the
diodes are matched (as they are in a packaged pair, such as Avago
HSMP-3822) and you drive them to the "off" state symmetrically, they
should inject very low pulse energy into the line they are switching.
The data sheet for the HSMP-3822 has several ideas for getting very
good isolation on the receiver side, once you've gotten isolation from
the high power with a high voltage PIN diode, and there are other PIN
diode ap notes on the Avago site.

Cheers,
Tom

 

[JosephKK]

(Perhaps your use is something like ultrasound, in which case you
might want something faster than 2usec, but the 3-30MHz is
suspiciously like a spec for HF over-the-air transmissions...)

Maybe, maybe not. Frequencies up to 150 MHz are already common for
instruments. 25 years ago that was a bit fancy lab ware, at the same
time 320 MHz was experimental ultrasound. I think experimental in now
around 3 GHz for ultrasound, AFAIK it is primarily used for microcrack
detection in and near metal welds.

 

[AnimalMagic]

Maybe, maybe not. Frequencies up to 150 MHz are already common for
instruments. 25 years ago that was a bit fancy lab ware, at the same
time 320 MHz was experimental ultrasound. I think experimental in now
around 3 GHz for ultrasound, AFAIK it is primarily used for microcrack
detection in and near metal welds.

We work with 3GHz and up to 30GHz all the time, and we buy and use
pre-made pin diode setups. Great for shunting power in rf networks in
test and simulation situations.

Mini-Circuits... guys like that have selections.

 

[goolge]

Yes, I would certainly agree that proper filtering to separate thePIN
bias from the RF is a good idea.  The faster you need to switch, the
more critical that becomes.  It should be an advantage to consider not
only a low-pass filter for the bias but a high pass filter for the RF--
or other methods to reduce the capacitance that thePINswitch must
drive.  Of course, be careful about the return path for the current
that charges the capacitance especially on the receiver side, so that
you do not expose the receiver input to excessive current.  A power
mosfet that switches several hundred volts in a few tens of
nanoseconds will be generating quite a bit of available energy to a 50
ohm 3-30MHz input!  If there exists between the fastPINdriver
circuit and thePINdiodes a low pass filter, and also between thePIN
switches and the receiver input (or transmitter output) a high pass
filter, and the filter passbands do not overlap, then there should be
decent isolation to protect the receiver input.  I can tell you that I
also usePINdiodeclamps on the input to a receiver I recently put
into production, and they are able to absorb transient energy quite
nicely.  They are just diodes normally reverse-biased by several
volts.  In a switched system, I would consider driving them actively,
using differential drive so that there is no (or very low) transient
on the RF input that they shunt to ground when they are "on."  If the
diodes are matched (as they are in a packaged pair, such as Avago
HSMP-3822) and you drive them to the "off" state symmetrically, they
should inject very low pulse energy into the line they are switching.
The data sheet for the HSMP-3822 has several ideas for getting very
good isolation on the receiver side, once you've gotten isolation from
the high power with a high voltagePINdiode, and there are otherPINdiodeapnotes on the Avago site.

Cheers,
Tom

Hi,Tom
Thank you for your reply in my news group"how to design PIN diode
driver - high voltage and high speed ?"! I got so much in your answer!
In my Antenna switch, in order to avoid input overload of my
receiver, a power limiter will be utilized. In your reply , you
mentioned that you also use PIN diode clamps on the input to your
receiver . My question is , is it able to absorb transient voltage >1
mV (the maximum value of receiver's pre-amp is 1mV )? According your
suggestion , i am going to use HSMP-3822 diode pair with 5 volts
forward bias(when transmitting ),-5 volts reverse bias(when
receiving), on the input of my receiver. Do you think the power
limiter plan can absorb the transient energy above 1 mV for the
receiver during both the receiving state "ON" and "OFF"?
Any suggestions will be appreciated!