Designer’s Guide Community

Ricky Chen

Community Member

Offline

Posts: 49

LO to IF Feed-Thriugh issue
Mar 26^th, 2012, 3:21am

Dear All,
In my circuit, I have a single balanced passive current driven Mixer, followed by a TIA (20MHz RC BW). The mixer switch is a NMOS with 20/0.13u size.
The LO is a (ideal) square wave 2GHz with 40 pS rise and fall time and rail - rail swing is 0.6V to 1.2 V.
The input current signal is a sine wave of 2.001GHz with peak amplitude of 1nA. ( I have attached the circuit)

I should get a 10 MHz sinusoid at the o/p of the TIA. But due to the LO feed-through the signal is getting completely swamped at the TIA o/p.(Attached here)
When I used IDEAL SWITCH I got the desired o/p (attached here).
But with 10fF capacitor between LO and TIA input making my signal completely swamped.

I am wondering why the LO which is at 2 GHz frequency is not getting filtered at the TIA o/p. I have used a 5pF capacitor at the TIA input but still it is NOT getting filtered.
Can anybody tell why this is happening and what can be done (without switching to double balanced mixer toplolgy) to address the issue.

Back to top

MIXER-TIA.JPG

IP Logged

Ricky Chen

Community Member

Offline

Posts: 49

Re: LO to IF Feed-Thriugh issue
Reply #1 - Mar 26^th, 2012, 3:21am

Ricky Chen wrote on Mar 26^th, 2012, 3:21am:

Dear All,
In my circuit, I have a single balanced passive current driven Mixer, followed by a TIA (20MHz RC BW). The mixer switch is a NMOS with 20/0.13u size.
The LO is a (ideal) square wave 2GHz with 40 pS rise and fall time and rail - rail swing is 0.6V to 1.2 V.
The input current signal is a sine wave of 2.001GHz with peak amplitude of 1nA. ( I have attached the circuit)

I should get a 10 MHz sinusoid at the o/p of the TIA. But due to the LO feed-through the signal is getting completely swamped at the TIA o/p.(Attached here)
When I used IDEAL SWITCH I got the desired o/p (attached here).
But with 10fF capacitor between LO and TIA input making my signal completely swamped.

I am wondering why the LO which is at 2 GHz frequency is not getting filtered at the TIA o/p. I have used a 5pF capacitor at the TIA input but still it is NOT getting filtered.
Can anybody tell why this is happening and what can be done (without switching to double balanced mixer toplolgy) to address the issue.

Back to top

TIA_OUTPUT_IDEAL_SW.JPG

IP Logged

Ricky Chen

Community Member

Offline

Posts: 49

Re: LO to IF Feed-Thriugh issue
Reply #2 - Mar 26^th, 2012, 3:22am

Ricky Chen wrote on Mar 26^th, 2012, 3:21am:

Dear All,
In my circuit, I have a single balanced passive current driven Mixer, followed by a TIA (20MHz RC BW). The mixer switch is a NMOS with 20/0.13u size.
The LO is a (ideal) square wave 2GHz with 40 pS rise and fall time and rail - rail swing is 0.6V to 1.2 V.
The input current signal is a sine wave of 2.001GHz with peak amplitude of 1nA. ( I have attached the circuit)

I should get a 10 MHz sinusoid at the o/p of the TIA. But due to the LO feed-through the signal is getting completely swamped at the TIA o/p.(Attached here)
When I used IDEAL SWITCH I got the desired o/p (attached here).
But with 10fF capacitor between LO and TIA input making my signal completely swamped.

I am wondering why the LO which is at 2 GHz frequency is not getting filtered at the TIA o/p. I have used a 5pF capacitor at the TIA input but still it is NOT getting filtered.
Can anybody tell why this is happening and what can be done (without switching to double balanced mixer toplolgy) to address the issue.

Back to top

TIA_OUTPUT_MOS_SWITCH.JPG

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: LO to IF Feed-Thriugh issue
Reply #3 - Mar 26^th, 2012, 5:41pm

Hi Ricky,

the TIA is only going to act like a proper filter up to the loop-gain bandwidth of the op-amp. The 5 pF looks like it is helping though. Just taking a cap divider, 0.3Vpk LO gets divided by the 10fF and the 5pF giving around .6mV. So its somewhere around your answer and still much larger than the 2uV output you are looking for...

It seems that maybe your input is just too small.

regards,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

Ricky Chen Community Member Offline Posts: 49	Re: LO to IF Feed-Thriugh issue Reply #4 - Mar 26^th, 2012, 7:16pm Thanks a lot Aaron. Can you tell what generally one expects the input current into the TIA value ( current assuming a Gm stage sitting before the TIA) for receivers like GSM....
Back to top	IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: LO to IF Feed-Thriugh issue
Reply #5 - Mar 26^th, 2012, 7:17pm

If you look at the DFT or FFT of the steady state portion of the output you will probably be able to see the 10MHz signal even in the presence of the LO leakage.

Your LO drive is an ideal source that can provide a large amount of current to the gate cap of the switches. In a practical implementation there will be a buffer driving the gates with finite slew rate (even at 40ps rise/fall time). A realistic buffer may provide slightly less LO feedthrough?

Back to top

IP Logged

Ricky Chen Community Member Offline Posts: 49	Re: LO to IF Feed-Thriugh issue Reply #6 - Mar 26^th, 2012, 7:22pm Thanks RFICFUDE, But as you see, as Aaron mentioned the with 0.3 vpk LO (mine is though 1.2v Pk swing EVEN FAR MORE ) one is getting 0.6mV which will swamp the output due to the desired input. If u see the DFT then the 10MHz component will be very less as compared to the LO leakage..Which is NOT desired to be..
Back to top	IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: LO to IF Feed-Thriugh issue
Reply #7 - Mar 26^th, 2012, 10:28pm

Hi Ricky,

RFICDUDE was right. Its true that the 10MHz component is much smaller, but you can always amplify it while filtering the LO signal (or simply not amplifying the LO). That's probably why he mentioned the DFT.

You should look up some product datasheets to see what the typical minimum detectable signals are like. For example, for -110 dBm, the input voltage for a 50-ohm system is 1uVpk. So as a complete guess, if a typical LNA provides 14 dB voltage gain into a 200-ohm load, then 200*Gm=5, so Gm=0.025. Therefore the current would be 0.025*1u=25nA.

regards,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

Forum Administrator

YaBB Administrator

Offline

Posts: 145

Re: LO to IF Feed-Thriugh issue
Reply #8 - Mar 26^th, 2012, 11:34pm

You have a 1.2V LO and a 5uV input signal, and you have basically one stage of filtering. It is not at all surprising that your LO feed through is larger than the desired IF output. The feed trough is only 3mV not 0.3V, which suggests that the suppression of the LO is 55dB, which seems pretty good. Your 5pF capacitors are connected to virtual grounds, so they will not be much help in suppressing the LO if the TIA still has much gain at the LO frequency.

-Ken

Back to top

IP Logged

Ricky Chen Community Member Offline Posts: 49	Re: LO to IF Feed-Thriugh issue Reply #9 - Mar 26^th, 2012, 11:36pm HI Aaron, Thanks for reply. I was suspecting that the LO-leakage can be suppressed by the subsequent filtering and Gain Stages. But now a day people are directly connecting the Mixer to the RF input instead of using LNA. I wonder how that fares for signal like GSM of -99dBm.... Regards, Ricky
Back to top	IP Logged

Ricky Chen

Community Member

Offline

Posts: 49

Re: LO to IF Feed-Thriugh issue
Reply #10 - Mar 26^th, 2012, 11:38pm

Forum Administrator wrote on Mar 26^th, 2012, 11:34pm:

You have a 1.2V LO and a 5uV input signal, and you have basically one stage of filtering. It is not at all surprising that your LO feed through is larger than the desired IF output. The feed trough is only 3mV not 0.3V, which suggests that the suppression of the LO is 55dB, which seems pretty good. Your 5pF capacitors are connected to virtual grounds, so they will not be much help in suppressing the LO if the TIA still has much gain at the LO frequency.

-Ken

Thanks Ken

Back to top

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: LO to IF Feed-Thriugh issue
Reply #11 - Mar 27^th, 2012, 3:00am

Quote:

Your 5pF capacitors are connected to virtual grounds, so they will not be much help in suppressing the LO if the TIA still has much gain at the LO frequency.

Actually I think if the TIA inputs are virtual grounds at the LO frequency, then the TIA will filter the LO at the output. If they are not, then the 5pF capacitors will help.

Quote:

But now a day people are directly connecting the Mixer to the RF input instead of using LNA. I wonder how that fares for signal like GSM of -99dBm....

I'm not sure if it is done in commercial products. Maybe ones where performance requirements are very relaxed. Leaving aside the noise performance, there is not much isolation from the LO to the antenna, so the unwanted LO emmission would be very high.

regards,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: LO to IF Feed-Thriugh issue
Reply #12 - Mar 27^th, 2012, 4:18am

Here is something else you can consider.

Currently you have one switch per LO phase connected to each input of the amplifier. This guarantees the LO leakage will be differential which is not desirable. You may be able to convert some or most of the differential LO leakage to common mode if you balance the circuit (in respect to the LO leakage) by adding two more switches (one per LO phase) where the RF side is connected to a dummy load (closely resembling the load on the RF input port) and the other side is connected to opposite input of the amplifier. If the circuit is symmetric then the LO leakage will be common mode. It will still be there and will respond to the common mode response, but it should be much smaller in the differential response.

This is essentially implementing a double balanced mixer where only one RF input is used and the other input is connected to a dummy load to balance the RF input load.

Back to top

IP Logged

Ricky Chen Community Member Offline Posts: 49	Re: LO to IF Feed-Thriugh issue Reply #13 - Mar 27^th, 2012, 4:41am Thanks RFICDUDE. This is one way to make the LO leakage common mode for Single balnced Mixer and reject at the TIA output. However the CLOCK drive is now doubled.
Back to top	IP Logged

RFICDUDE Community Fellow Offline Posts: 323	Re: LO to IF Feed-Thriugh issue Reply #14 - Mar 27^th, 2012, 5:15pm As they say, "there is no free lunch." I can appreciate that the added cost in current consumption may be excessive for a single balanced application.
Back to top	IP Logged