Recovery time of TR switch

Hello, I've been working on a small NMR system using a single coil for transmit and receive at frequencies near 10MHz. I'm now trying to optimize the system so I can switch between transmit and receive faster. I have been using a combination of PIN diode SPDT switch circuits, PIN diode SPST switches, and IC analog switches. I want 70dB isolation from the transmit to receive port when in transmit mode, so I've been using two stages. The PIN diode circuits are nice in that they can give the lowest loss (like -0.2dB per stage) on the receive side, and thus the better overall noise figure, but integrated analog switches seem to give better recovery time. Here's an example of the PIN diode switch I'm working with at the moment:


The first stage is a SPDT switch, and the second is a SPST switch, both using lumped element quarter wave impedance transformers. Diodes D1 through D4 are all MA4P7470F-1072T PIN diodes. Explanations for the operation of both switch types can be found here: **broken link removed**

The circuit can switch from receive to transmit very quickly (few us), but when going from transmit to receive, I see ringing on the output at my 10MHz frequency of interest, which lasts around 30-50us before dropping below my noise floor. This ringing persists even if I am not transmitting any thing, and the coil is not attached (I just put 50ohm terminators on the TX and Coil ports). Therefore that transient ringing must originate in the TR switch itself during the switching event, I think.

I'd like everything to be quiet in 10us or less. Is there anything I can do hear to either damp the ringing, or reduce its amplitude, without severely impacting the performance of the circuit (mainly its loss from the coil port to the LNA port)?

Thanks in advance

So going from recieve to transmit you see ringing. But where, at the LNA output?

You problem is a combination of those giant inductors (which tend to ring when you abruptly change the current going thru them), and the fact that your shunt diodes have no obvious way to dischage themselves when going from forward to reverse bias. there is always a 2nd pin diode in the bias path that might turn off first, keeping the first diode still partially on.

I would suggest a spice analysis of the entire circuit.

biff44 said:

So going from recieve to transmit you see ringing. But where, at the LNA output?

Click to expand...

Sorry, I mistyped. I see it when going from transmit to receive (both control signals to the switch drivers go low to high, PIN diodes go from forward to reverse biased). I can roughly see it after the LNA (16dB) and another 45dB of amplification on the scope at around 10MHz. It's much easier to see after downconversion to baseband. Signal level I'm talking about is on the order of microvolts, referred to the coil port, so it's impossible to see without substantial amplification.

You problem is a combination of those giant inductors (which tend to ring when you abruptly change the current going thru them)

Click to expand...

Those inductors do ring a bit, but not at 10MHz, which is what I care about.

and the fact that your shunt diodes have no obvious way to dischage themselves when going from forward to reverse bias. there is always a 2nd pin diode in the bias path that might turn off first, keeping the first diode still partially on.

Click to expand...

Well each pair of diodes in each switch circuit shares a DC current path, so their bias current should always be equal. But I do notice that during reverse bias, the reverse voltage across the diodes is not equal. But they should still be at high impedance, especially at low signal levels, right?

I would suggest a spice analysis of the entire circuit.

Click to expand...

I've tried LTspice (though I can't model PIN diode behavior), but I can't get it to replicate any ringing at 10MHz. It shows some ringing at around 4MHz, but it dies within a few us.

I should also mention that the LNA is tuned to 10MHz, and is noise matched, not impedance matched. At first I thought it was ringing in the LNA match network, but I've tried replacing it with a broadband LNA block and it didn't help.

I also facing the same kind of the problem. I used the diode of MA4P7104F-1072F. In my case the frequency is 205MHz and output of the PA(Tx path) i have a circulator, where circulator main port goes to antenna and isolated port goes to limiter and limiter is designed by help of the above mentioned diode. when i switch from Tx path to receive path i see for 30uS some noise of 205MHz comes . Generally these noise are in microV you can not detect. For seeing the same we need to have subsequent gain of minimum 20dB. So can you please share with me how did you solve this problem.

Thanks
Sanjaya

Its your emitter followers, when they turn on, they pull loads of current through the diode, unfortunately when they turn off, they can not discharge the RF decoupling capacitors. Try putting a nice low emitter resistor in, say 470 ohms, or to the max current the transistors will take, or drive the diodes from a complementary pair to push- pull the current.
Frank