CMOS to Microstrip Impedance Matching

[nlantz]

I need to somehow take the output of a CMOS IC and transmit that signal over a 50 Ohm Microstrip line.
The output impedance of the CMOS IC is about 15 megaohms. I see a lot of examples for going from 75 ohms to 50 ohms or similar but nothing on something of this magnitude. Any ideas?

CMOS (High-Z) ---> ? ---> Microstrip (50 Ohm)

 

[Svein]

I need to somehow take the output of a CMOS IC and transmit that signal over a 50 Ohm Microstrip line.
The output impedance of the CMOS IC is about 15 megaohms. I see a lot of examples for going from 75 ohms to 50 ohms or similar but nothing on something of this magnitude. Any ideas?

CMOS (High-Z) ---> ? ---> Microstrip (50 Ohm)

You are wrong. The input impedance of a CMOS IC is on the order of 15MΩ. The output impedance varies with the IC chosen, but I do not think you can achieve 50Ω. The closest you can get is probably something like this: http://www.onsemi.com/PowerSolutions/product.do?id=MC74AC245. Otherwise, consider using a bipolar pair between the CMOS and the microstrip.

 

[tech99]

I need to somehow take the output of a CMOS IC and transmit that signal over a 50 Ohm Microstrip line.
The output impedance of the CMOS IC is about 15 megaohms. I see a lot of examples for going from 75 ohms to 50 ohms or similar but nothing on something of this magnitude. Any ideas?

CMOS (High-Z) ---> ? ---> Microstrip (50 Ohm)

CMOS is a logic system, so it is a switch, not a linear source. The problem is similar to that of matching a Class C amplifier to a line. Is your frequency high enough that microstrip matching is necessary?

 

[CWatters]

I suspect there is more to the original question but.. Have you tried a series resistor..

http://web.cecs.pdx.edu/~greenwd/xmsnLine_notes.pdf

Depending on what's at the other end of the line you may need to look at parallel termination at the destination.

 

[nlantz]

The input impedance of a CMOS IC is on the order of 15MΩ. The output impedance varies with the IC chosen,

I thought that was wrong also. Unless the gate of the MOSFET is the output a high impedance output doesn't make sense. And the gate should never be the output. That's the output impedance that the company told me though. I am going to measure it myself when the evaluation kit comes in.
From my understanding of CMOS, the output should only be high-z for a very small time during a switch between states. And output impedance depends on the MOSFETs at the output and their configuration. Which is proprietary, hence why I asked the company.

Is your frequency high enough that microstrip matching is necessary?

The system is operating at 2 Gb/s so we are on the threshold a bit. It's something that I am concerned about though. I am treating this as an RF system. If only so I don't look like a dummy at design review.

I suspect there is more to the original question but.. Have you tried a series resistor..

http://web.cecs.pdx.edu/~greenwd/xmsnLine_notes.pdf

Depending on what's at the other end of the line you may need to look at parallel termination at the destination.

I haven't tried anything yet because the part is still in the mail. This looks like a good thought. Thanks for the link! I will look more into this.

 

[tech99]

I thought that was wrong also. Unless the gate of the MOSFET is the output a high impedance output doesn't make sense. And the gate should never be the output. That's the output impedance that the company told me though. I am going to measure it myself when the evaluation kit comes in.
From my understanding of CMOS, the output should only be high-z for a very small time during a switch between states. And output impedance depends on the MOSFETs at the output and their configuration. Which is proprietary, hence why I asked the company.The system is operating at 2 Gb/s so we are on the threshold a bit. It's something that I am concerned about though. I am treating this as an RF system. If only so I don't look like a dummy at design review. I haven't tried anything yet because the part is still in the mail. This looks like a good thought. Thanks for the link! I will look more into this.

The CMOS device may have an "on" resistance of perhaps 100 ohms, but it might be better to swamp this out with a series resistor as it is not well defined. As the line cannot be much more than about 100 ohms, you will than need a shunt resistor as well to define the line driving impedance, maybe 50 ohms. So you end up with a potential divider and some loss.

 

[nlantz]

The CMOS device may have an "on" resistance of perhaps 100 ohms, but it might be better to swamp this out with a series resistor as it is not well defined. As the line cannot be much more than about 100 ohms, you will than need a shunt resistor as well to define the line driving impedance, maybe 50 ohms. So you end up with a potential divider and some loss.

Some loss is OK. The next stage is a stage is actually a step attenuator. I am estimating it will need to be at about 6dB nominally. I could just dial that back a bit and we would be fine.

 

[berkeman]

The system is operating at 2 Gb/s

What CMOS logic gate are you using to drive a 50 Ohm microstrip TL at this frequency?

 

[nlantz]

What CMOS logic gate are you using to drive a 50 Ohm microstrip TL at this frequency?

I don't have any info about the gates inside the IC.

 

[tech99]

I need to somehow take the output of a CMOS IC and transmit that signal over a 50 Ohm Microstrip line.
The output impedance of the CMOS IC is about 15 megaohms. I see a lot of examples for going from 75 ohms to 50 ohms or similar but nothing on something of this magnitude. Any ideas?

CMOS (High-Z) ---> ? ---> Microstrip (50 Ohm)

Having given it a bit more thought, I think it is only necessary, in order to prevent reflections, to match the receiving end of the line. This can be done with an attenuator, as you mention. At the sending end, the device is driving a low impedance, so it is possible it will exceed its ratings for dissipation and supply current. In such a case, it will be necessary to use a series resistor to bring the device within safe conditions. Check the overall loss due to the resistor (potential divider action) and the attenuator.

 

[CWatters]

I think Berkeman meant which CMOS logic family? or is it some custom IC?