Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

MMIC Bias Tee Snubber Protection

Status
Not open for further replies.

gianbo85

Newbie level 6
Joined
Dec 25, 2009
Messages
13
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Location
Milan, ITALY
Activity points
1,437
Hi Everybody,

I am Gianluca, an EE Master's student from Italy.

This is going to be my first post in this forum but I have read you for quite a while now.

I am designing an ultrafast pulser to trigger a photodetector. The last stage of this pulser is an MMIC PA from Hittite.

I developed a Bias Tee to power it up with a 400ma @ 12V bias.
The bias tee is quite a common one with a DC blocking capacitor and two inductors in series.

My problem is that if a failure of the 12V line occurs, the inductors in the bias tee will just boost their voltage, damaging my precious MMIC.
How can I protect the IC RF out? I didn't want to use a zener diode because it will introduce a capacitance on the RF pin that needs to have a rise time of less than 200ps!

Any ideas?

Happy Christmas to everyone and thanks in advance for any help

Gianluca
 

gianbo85 said:
Hi
My problem is that if a failure of the 12V line occurs, the inductors in the bias tee will just boost their voltage,

Could you explain your schematic a little better. I do not understand why the inductors would "boost" their voltage in a passive bias tee.

Merry Christmas to you, and happy holidays to all!

Rich
 
Last edited by a moderator:

This is a simple schematic of my design. I am at my parents' home now and I don't have the values of L2 and L3 but they are around some tens of uH.

It's my first time designing a bias tee but if the inductors behave like in switching power supply,if the voltage on the positive rail is disconnected (user fault or failure) then applying v = L * di/dt th voltage will raise to thousands of volts! Am I right?

How to design a snubber circuit for the last inductor without having to connect a diode on the RF pin of the power amplifier? I'd like to keep the capacitance of that pin as low as possible because it has to go from 0V to 7-8V in 200 ps!

I really need some help here... maybe I am missing something but I am pretty new to microwave design.

Gianluca
 

I agree, a 1 mH inductor with 0.4 A thru it will not be happy if you suddenly opened the power supply connection. Those are pretty big inductors! Why are they so big in value? Usually circuits with 200 pS rise times use small inductors, like 2 nH size. Are you driving a laser or LED?

In any event you are going to need to add a fast schottky diode between the 1 mH inductor terminal and ground on the amplifier side, oriented so that it does not allow a negative bias to temporarily go into the amplifier. I might also put a zener diode in the same location to keep the DC voltage clamped at whatever maximum voltage the PA could handle. Then maybe a small capacitor, like 100 pF, in the same location, just to keep the rise time slow enough to let the diodes do their job.
 

Hi,
And I think the coupling C of 10 uF may be over dimensioned (depend on your needed F-low value), or you need some RF C(high-Q Caps) parallel with them too...
K.
 

The values of L and C are so high because the low freq of the bias tee is around 100khz. Basically the waveform coming out of this MMIC amplifier is a squarewave that must have:

- 200ps rise time

- no oscillations (or 1% limited ones)

- 40 MHz period

- duty between 10% and 90%

- the on period voltage level must be as flat as possible

All of this is to enable an ultra fast photo diode.

Going back to the bias tee. The 1mH inductor is not a problem, I can put a fast diode across it like in switching power supplies. The problem is L3.

As I already said I don't like putting a zener diode between the RF pin of the power amplifier and ground. It adds a lot of capacitance and I don't know if I can acheive 200ps rise time with that much capacitance on the RF pin.

so.... How to snubber the voltage on L3?? It's about 3-33uH (yet to be determined).

The 10uF capacitor is actually 1uF (the 10uF is used elsewhere in the design) and is a fingered one by Murata. Very high Q and specific for microwave (up to 2-3 Ghz).

Thanks for the replies. Now, what about the problem with L3? How to clamp the voltage there without a zener diode?

Gianluca
 

Explain the part about "enable an ultrafast photodiode". Photodiodes do not take 400 ma of current.
 

Hi,
I like repeat biff44`s question:
Are you driving a laser or LED (because you like speek over photodiode too)?
Ciao!
K.

Added after 5 minutes:

Just about "The 10uF capacitor is actually 1uF, Very high Q and specific for microwave (up to 2-3 Ghz)".

If you have really 200ps(10 to 90%) rise time: you must have a system bandwith of min./ca 10GHz!
Sorry, but I don`t belief, that some uF capacitors are good elements at these frequency....
K.
 

As I have writted in my previous posts, this bias tee needs to bias a power amplifier:

HMC637LP5 (https://www.hittite.com/content/documents/data_sheet/hmc637lp5.pdf)

The power amplifier ouputs on a 50ohm line that goes to a photodetector.

The photodetector is a SPAD (Single Photon Avalanche Photodiode) that can be enabled (gated) by exceeding a certain voltage given by the sum of a constant polarization on one terminal (in my case around -30V) and a square wave on the other terminal. The square wave is generated by the RF Power Amplifier.

As per datasheet, the HMC637 power amplifier needs a 12V - 400mA bias to output 1W needed by the photodiode.

The HMC637LP5 is a very delicate component and we already broke three of them just by wrong soldering temperature and ESD. Since it is very expensive, my professor asked me to add some overvoltage protection on its pins. How to find a way not to use a zener diode to clamp the voltage? It will add unecessary capacitance on that pin that needs to rise 7-8V in 200ps.

Any other question is more than welcomed. Thanks for your help

Gianluca
 

Seems a little unusual way of driving--using an RF amplifier. If it were me I might be tempted to use something like a digital gate made of discrete components, or a laser driver IC.

You can accomplish your goal by not allowing the +12 volt power supply from suddenly being shut off. How about just putting a big capacitor (10,000uf) where the 1 mh coil joins the 12 volt rail. That way, the 12 volt supply can not be instantly removed--it will take some time for the capacitor to bleed down in voltage. With the slow ramp-down of voltage, there will not be a big spike generated in the coil. You will need to keep the gate bias on all the time the capacitor is bleeding off!!!

You might need a 0.5 ohm series resistor between the power supply and the big capacitor to limit in-rush current.

As far as blowing up the amplifier, you might want to build a power supply "sequencer" circuit. Such a circuit uses a series fet or transistor in the 12volt line that will not allow 12 volts to be applied to the ampliier until the proper bias voltage on the gate is applied first, and will shut off the 12 volts if the gate supply is interrupted. You would want some R-C time constant slow down in the circuit to, once again, keep the 12volts from turning off suddenly, or maybe a big capacitor on the gate bias input also, and the sequencer turning the drain bias off well before the gate bias capacitor voltage is discharged.

Added after 8 minutes:

As far as your 200 ps rise time, there are pulse shaping circuits that the UWB people use to make transistions quickly from a slow input waveform. You might want to add such a circuit between the power amplifier and the photodetector.
 

Hi gianbo85,
Is your SPAPD from Hamamatshu or EG&G or Princenton?
Is these a project at Desy please?
I can not understand your 1 mH need_why is it with for ex. 1-10 uH or so not OK?
You can not apply some "active quenching"?
Then is the veryhighspeed switching of power out of way...
Do you thees script:
https://www.physics.ucsd.edu/~tmurphy/apollo/doc/Dautet.pdf
K.

Added after 4 hours 39 minutes:

Hi gianbo85,
What do you belief please, can be helpful this script four your solution finding?:
"Monolithic active quenching + picosecond timing circuit suitable for large-area single-photon avalanche diodes,
A. Gallivanoni, I. Rech, D. Resnati, M. Ghioni, and S. Cova , Milano":
www.opticsinfobase.org/abstract.cfm?URI=oe-14-12-5021
K.

Added after 57 minutes:

As biff44 spoke over that too:
The LMH6533 laser-diode driver (not the highest speed) features switching rates of less than 0.8 nsec and contains two high-current outputs for reading and writing 650-nm DVD or 780-nm CD lasers. It achieves read, write, and erase ...

And what about :
AVX, MultiGuard multilayer ceramic transient voltage suppression arrays-or similar products?
For ex. the AVX "Transfeed" transient voltage suppressors have 200-250psec respond times:)
Please refer to side 41:
www.mouser.com/catalog/supplier/library/pdf/AVXTVS.pdf
I can remember over some similar high speed Comm-line semiconductor transient suppressor diode-arrays from Vishay too...
K.
 

Thanks biff44. I like the idea of the big capacitor. I will try some spice simulations to see what size I need to limit the overvoltage to 20V (the maximum that the PA can take).

I will also go through some AVX and Vishay datasheets... there might be some very low capacitance diodes.

I have already looked at some laser diode drivers but I can't find some with 8V of voltage range and such fast rise times.

I already thought about some kind of sequencer to make sure that the +12V rail is not applied without the gate signal. If I'll have any doubt in the future I will ask you :)

This is a thesis project I am doing at Politecnico di Milano, a fellow student is trying to use a digital gate using BJT and at the moment he acheived 500ps - 1ns rise time so it's not enough for my specs. I already have a sample board from Hittite and I know that the amplifier works well: no oscillations, fast rise time etc. I'll use this one for the moment.

I am planning to use a pulse shaper before the amplifier. It will be a Ge-Si digital gate from On Semiconductor. They used it in my lab and it behaved well.

My SPAD photodetector is a custom one. We produce it in our lab so no Hamamatsu or other companies... by the way, the article you posted by Ghioni, Cova etc... well, they are my professors :)

I will try with the capacitor idea at the moment. If I will need more help I will write back soon. Thanks again to everybody for the help!

It's still early but.... happy New Year!
 

Hi gianbo85,
I think that reason of your rise time problems is missing a proper bypass C over L3/L2...
The AVX-TVS Catalog at sides ~50-55 are interessant too:)
I readed in detail the datasheet of your Hittite: I belief that you must build a Pwr-sequencer...
You can find some commercial ICs too.
Fine, that you knows the people:)

By the way, can you send me please in PM your SPAD specs_my group has some interest on those products -for research projects...

Thanks & Happy New Year!
K.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top