Looking for a switch circuit

[simon_z]

Hi guys,
I am designing a transceiver kind of circuit. During transmitting, the LC tank which has very high AC power (high voltage >= +/-500 Volt) has to be isolated by a switch from the receiver side. On the other hand, during receiving period, the receiving signal is very weak so that the on-state resistance of the switch must be very small ( < 1 ohm). Meanwhile, the receiving cycle happens right after the transmitting. Therefore, the switch time must be short ( < 10 us).
I was going to use a solid state relay to fulfill the switch function. But I have been having trouble to find one which can bear such high voltage. I also thought about two MOSFETs connected back to back to bear AC voltage. But the driving circuit is an issue.
Do you guys have any suggestion? Any idea would be greatly appreciated.
Thank you for your help in advance.

Simon

 

[emontllo]

Hello Simon,

I would suggest to make your own Solid State Relay using some Photovoltaic output (voltage) optocouplers like the toshiba TLP190B. It is capable of easily driving MOSFETs simply and with isolation.
You can make a similar circuit like this in order to have bidirectional SSR with two back to back mosfets:
File:SolidStateRelay-Diagram.svg - Wikipedia, the free encyclopedia
You can find 600V low RDSon MOSFETS (<0.5 Ohm) like IPB60R380C6
Consider protecting gate with some TVS or Zener to avoid overvoltages. Do not use gate pulldown resistors since the photovoltaic optocouplers can provide small currents (40 uA!).
Good luck,

Ernest

 

[betwixt]

Or possibly consider using PIN diodes but I lack experience in these so I'm not sure if they are available with sufficient voltage rating or Ron. You don't mention what frequecy your transceiver works at, if it is more than a few tens of MHz you might find you get insufficient capacitive isolation if you use normal MOSFET switches.

Brian.

 

[simon_z]

Thanks Ernest and Brian. Your helps are great.
Currently I am looking to make an isolated driver circuit my own, as Ernest suggested.
Meanwhile, I found the other plan: **broken link removed**. This plan should be faster. But I am not sure the high potential isolation is good enough. I am doing some simulation to verify it. I will let you know how it goes.
To Brian: the conducting signal has frequency from 1MHz to 6MHz. You are right about the PIN diodes. That is the typical way of the doing the RF power isolation for my application. Since I designed a new power amplification way for it, I figure it should be equipped with a new isolation method. ^_^. I will try this as well.

Thank you for the great suggestions and helps.

Simon

 

[mtwieg]

Well pin diodes are pretty damn effective as RF switches. But ones that operate at frequencies below 10MHz are uncommon, and you want high voltage as well... might be hard to find an appropriate device.

When you say isolation, do you mean the switch has to be isolated, or are you just talking about the isolation provided by the switch to the load in the off state?

 

[FvM]

1 to 6 MHz is a lower frequency than I initially expected, so there's at least a chance. The high voltage however will cause difficulties to achieve sufficient isolation. I really wonder, if a low resistance RX swicth is a reasonable approach. Are you sure, that impedance matching is required for the receiver?. Similar applications in ultrasonic transceiver RX switching are mostly using a series capacitor and a voltage limiter as the first "switching" stage. The capacitor can be made part of the tank circuit.

With MOSFET switches, you would at least need a T-circuit to achieve sufficient isolation. The first series switch would introduce a rather large parallel capacitance if a low Rdson is required. Convenient driving methods like photovoltaic couplers are definitely too slow.

 

[simon_z]

Well, when I talk about isolation, it means that how well the switch can block the high power from receiver side. However, if the back to back MOSFETs switch is used, the isolation between the sources to the driving circuit is also required because the voltage on the source would follow the high-power side voltage during the negative half period when the switch is in off state.

To FvM: low resistance is because the receiving signal is very weak, almost at the same level of the thermal noise of 50ohm resistor. Large resistance would introduce more noise and comprise SNR. I wonder the "switching" stages structure you mentioned. Do you have any schematics? Another question, the T-circuit you mentioned, does it mean two mosfets back to back or something else?

Also, I checked some photovoltaic couplers' driving ability. Turns out it is too slow for the big power mosfet which Qg is around 50nC.
How do you think of the transformer coupled driving circuit I found? Does it sound reasonable?

Anyway, thank you guys for all the help.

 

[FvM]

One switching stage idea would be based on a series capacitor, but it's not compatible with low noise 50 ohm receiver matching. So the back-to-back MOSFET switch seems the only suitable option. I would assume a T switch topology, with the high voltage switch represented by back-to-back MOSFET. However, did you calculate the expectable switch capacitance and respective reactive currents in TX mode?

Transformer coupling will be in fact the obvious driving method for the HV switch. But it's not easy. Also crosstalk may be a problem for at low RX voltages.

 

[emontllo]

If high speed MOSFET switching is desired I would suggest using fast isolated MOSFET drivers.
I have worked with Silicon Labs MOSFET dirvers and are very fast (ns), check this one for instance Si8233BB. In this case you need an isolated power supply (1/4 W would be enough) to power the driver side of the IC. Some MOSFET drivers have their own power supply like the ones from Analog ADUM series.

Ernest

 

[simon_z]

So glad that I chose to post the question here. You guys have been enormously helpful. Thank you all for the great ideas.

TO Ernest: I checked out Si823x as you suggested. I think it might be the plan for me. About the ADUM series, could you please be more specific on which ones have the isolated power supply?

Thank you.
Simon

If high speed MOSFET switching is desired I would suggest using fast isolated MOSFET drivers.
I have worked with Silicon Labs MOSFET dirvers and are very fast (ns), check this one for instance Si8233BB. In this case you need an isolated power supply (1/4 W would be enough) to power the driver side of the IC. Some MOSFET drivers have their own power supply like the ones from Analog ADUM series.

Ernest

---------- Post added at 06:48 ---------- Previous post was at 06:41 ----------

I found the isolated driver from ADI. It is ADUM6132.

Simon

 

[FvM]

Suggesting Analog Devices iCoupler devices raises the question, if a DC/DC converter directly connecting to the hot RX input node will be compatible with claimed receiver sensitivity?

The isoPower devices are operating the DC/DC converter around 180 MHz switching frequency. AN-0971 is showing the implications in an impressive way. It may be the case, that the unusual high switching frequency allows to cut the huge interfering current injected by the iCoupler with a low-pass. But the application note suggests, that there are subharmonic components, too.

Of course, a similar problem arises with more conventional DC/DC converters as well. A transformer with double inter-windings shield will be required to drain the injected currents. I don't think that you'll find them with a catalog device.

 

[betwixt]

I know this may sound silly but have you considered using conventional power diodes as PIN switches? I have used them as signal switches for this purpose at low frequencies before but not in an application where T/R change-over had to be very fast. They actually have quite good isolation when reverse biased.

Brian.

 

[simon_z]

Bad news guys. Just did the simulation with back to back mosfet switch. MOSFET's output capacitance, Coss, which is around 200pF usually, cannot provide big isolation for signal has frequency larger than 1MHz. Gave up on it. A whole week wasted on this. Any suggestion to get rid of Coss?

To Brian: I will have to go back to the PIN plan. The reason I didn't prefer to use PIN based switch was it requires a quarter wave transmission line. For my application, 1MHz to 6MHz, the line is pretty long. Do you have any circuit which doesn't need this transmission line?

Thank you all for the help. Hope my failing experience could save you some time in future.

Simon

 

[betwixt]

I'm not sure why the transmission line is needed. I would guess in the design you have seen, the transmission line is being used as choke to let DC through but block the RF. This is of course very practical when the wavelength is very short.
The PIN diode or an ordinary diode used in the same way is just a controlled impedance, in reverse bias is behaves like varicap with low capacitance and in forward bias it behaves like a (near) perfect DC voltage drop with low impedance to AC. I think at frequencies up to 6MHz you can probably get away with conventional wound chokes instead of the transmission line.

Brian.