Help me with shifting a signal!

level shifters

hello,
i need to shift a signal with VCM=1.6V and swing 0.4Vpp to Vcm=0.6 with the same swing? note this signal may have frequnecy 1.25GHz,
i want it by using the least # of stages"one will be gr8".
also i thought of using a cap. to block DC then a potential divider to set the new CM but this will be power consuming.
thnx

Re: level shifters

safwatonline said:

hello,
i need to shift a signal with VCM=1.6V and swing 0.4Vpp to Vcm=0.6 with the same swing? note this signal may have frequnecy 1.25GHz,
i want it by using the least # of stages"one will be gr8".
also i thought of using a cap. to block DC then a potential divider to set the new CM but this will be power consuming.
thnx

Click to expand...

Hi safwatonline,
I think you have described the best technique in your question.
Use a capacitor to block DC and then set the required CM with a resistor, a simple C-R network. The required CM voltage will be applied to one end of the resistor and the other end of the resistor is connected to the node where you want the CM shift.

Disadvantages of the network :
1) It acts as a high pass filter and does not allow low frequency components, so if you are worried about frequency components at the output of the network, set the pole of the network accordingly.
2) Since you are connecting the resistor to another node, make sure its value doesn't affect your other circuit characteristics.

I have used this technique for designs upto 4 GHz frequency without any problem. If the resistor is connected to a MOS transistor input, power consumption is almost zero as gate current is zero. If the resistor is connected to a BJT input, power consumption is base current * R, which again is small depending on the value of beta of the technology.

Can you elaborate what other power consumption is there in this circuit? If you are thinking of generating the CM voltage from a resitive divider from supply voltage , then you can always scale the values of resistors to reduce power consumption.

Bharath

Re: level shifters

hello Bharath,
thnx for ur reply,
actually u also described my problem , as the HPF may affect the circuit performance and to decrease its cuttoff frequency this will introduce huge caps and resistances which i cannot afford their areas . note alos that small resistors will consume some large currents.
BTW can i generate the biasing voltage i need "without using bandgap or voltage divider", i was thinking to pass some current into a diode connected MOS and use its Vds as my biasing voltage , is that methode OK or it will not be efficient with process variation.
the main problem is difference between the two voltages(1.6,0.6) which will need many stages of high speed buffers "high current consumption" (note also the large swing),
so i was wondering if there is some other technique that i dont know
regards,
a.safwat

Re: level shifters

safwatonline said:

hello Bharath,
thnx for ur reply,
actually u also described my problem , as the HPF may affect the circuit performance and to decrease its cuttoff frequency this will introduce huge caps and resistances which i cannot afford their areas . note alos that small resistors will consume some large currents.
BTW can i generate the biasing voltage i need "without using bandgap or voltage divider", i was thinking to pass some current into a diode connected MOS and use its Vds as my biasing voltage , is that methode OK or it will not be efficient with process variation.
the main problem is difference between the two voltages(1.6,0.6) which will need many stages of high speed buffers "high current consumption" (note also the large swing),
so i was wondering if there is some other technique that i dont know
regards,
a.safwat

Click to expand...

The pole of the HPF depends on your application, ie, how low a frequency do you want to pass through and for most applications, the value of caps/resistors will not be huge as you describe them to be. Again, it is totally application dependent.

Using the Vgs of a diode connected resistor as a biasing voltage is usually a bad practice, as it varies a lot over process and temperature variations (mainly caused due to the threshold voltage mismatch, vgs adjusts itself such that the required current flows through). If you can tolerate that variation (typically around 200 mV), then you can use it.

Another technique is to use source follower (Vgs shift) or emitter follower (Vbe shift), which also has extra power consumption, depending on the frequency of operation.

Bharath

level shifters

a.safwat

Hi!
Here is a suggestion to generate the biasing voltage i need "without using bandgap or voltage divider".
A inverter with power 1.2V which connects input with
output can porvide an about 0.6V reference.

BR

eric

level shifters

thnx eric , but how can i get the 1.2 supply,
i am working with umc 0.18u with supply 1.8 V