Why we usually set ac=1v?

[emado2]

This voltage does not affect the analysis , it is just a scaling factor

 

[vivek_raj_verma]

hi there,

AC = 1 volts eases the computation.

Gain = vout/vin and u can see when u write vdb(vout) it gives u gain (Vout/vin) at that node. It does not affect any other thing. u can write whatever u want here but in that case u have to see how can u extract gain from the simulator u r using???

 

[wszywps]

Then '1' in 'AC 1' doesnot mean 1v, it just a unit 1.

i agree

 

[hung_wai_ming@hotmail.com]

I think the conclusion is that always we should set ac=1 for ease of gain computation. Anything we can set definitely, but to be honest, it is no use as everyone has the habit to check the Bode plot directly instead of re-calculating it if we don't set ac=1. It does not need to relate to any input voltage range to decide if we need to set ac=1 or ac=10, always we should check transient with real amplitude input instead of ac analysis

 

[airboss]

ummm,

i've never thought about this question before. so i ran simulation in Spectre for an op which works well, i mean, an op that is stable and meets the gain spec. then i replace "ac magnitude" with "2", the op amp isn't stable anymore.

so, based on the discussion above and the simulation, should we come to the conclusion that, small signal gain is the Vop/Vip. we can use any number for vip but to make life easier, just use vip=1. however, when vip>1, stability might not maintain anymore?

 

[safwatonline]

ummm,

i've never thought about this question before. so i ran simulation in Spectre for an op which works well, i mean, an op that is stable and meets the gain spec. then i replace "ac magnitude" with "2", the op amp isn't stable anymore.

so, based on the discussion above and the simulation, should we come to the conclusion that, small signal gain is the Vop/Vip. we can use any number for vip but to make life easier, just use vip=1. however, when vip>1, stability might not maintain anymore?

if u put AC=2 , then u don't compare to unity gain freq. (agin crossover) , u have first to find the loop gain (i.e. divide the output by the input) then find the PM of the loop gain not the output. ofcourse the OPA will be stable if it was stable in the first case (u just changed the input !!!)

 

[wonbef]

setting ac=1v is to set a reference point only. of course, you can set other value.

 

[shaikhsarfraz]

Hi,
We set ac=1V only to make the calculations simple for ourselves.

i.e ac analysis is done primarily to find the loop characterisitcs.

Gain = 20 log (Output/Input )

If input equal to 1, then gain is 20log (output).

However one question still remains:

ac analysis is a small signal analysis.
Here the tool lineraizes the MOSFET's around some operating points.
Now if the stimulas is as large as 1V, is it not that the transistors will go into triode.

Is there something in built in the tools (SPECTRE), say some algorithms, which makes the devices insensitive to such large voltage swings.
Because theoritically for small signal analysis the input stimulas should be around .25 V, else it may cause deviation from non linearlities.

Regards
Sarfraz

 

[faizalism]

Ok... good information about the small signal analysis......What about this DC voltage

vip vip gnd dc 'vdd*0.5' ac 1.0

Do we need also biased with DC voltage? Because this voltage will affects the gain for the amplifier. I'am still can't simulate my differential amplifier because this problem. Even can't find best reference for AC simulation for differential amplifier (2 inputs, 2 outputs).

thanks,

 

[shaq]

Ok... good information about the small signal analysis......What about this DC voltage

vip vip gnd dc 'vdd*0.5' ac 1.0

Do we need also biased with DC voltage? Because this voltage will affects the gain for the amplifier. I'am still can't simulate my differential amplifier because this problem. Even can't find best reference for AC simulation for differential amplifier (2 inputs, 2 outputs).

thanks,

Because we need all the transistors bias at sat. region.

 

[shaikhsarfraz]

Hi,
For biasing all the devices, we need to give a dc shifted small ac signal.

you can say something like


1.2 V (dc) + .2 V (ac)

Regards
Sarfraz

 

[haley_8]

it actually doesn't matter what value you use for AC, but 1V is just an easy number to work with.

 

[smilelangjun]

No matter what the values are, u can see the results are the same.

 

[shaikhsarfraz]

Thats true that the tools give us the results even if we put ac as 1V.

However theoritically speaking these high value of small signal ac can cause problem, and there can be error in the final results.

Sarfraz

 

[cinvo]

just for gain=Vout

 

[HenNavarre]

one times any value is easy to figure out so that why we use 1, AIC is hard enough to add unnecessary complications

 

[analog_prodigy]

i think the simulations work even with ac = 0.1-1v. we give ac =1V just for convenience. that's it
u can give any value

 

[faizalism]

actually it is for simplicity.......if u used other values, u need to subtract the gain(dB) of the output with the input gain (dB).
Av(dB) = Vout(dB) - Vin(dB)

so it is up to you, dude.

 

[suria3]

ummm,

i've never thought about this question before. so i ran simulation in Spectre for an op which works well, i mean, an op that is stable and meets the gain spec. then i replace "ac magnitude" with "2", the op amp isn't stable anymore.

so, based on the discussion above and the simulation, should we come to the conclusion that, small signal gain is the Vop/Vip. we can use any number for vip but to make life easier, just use vip=1. however, when vip>1, stability might not maintain anymore?

if u put AC=2 , then u don't compare to unity gain freq. (agin crossover) , u have first to find the loop gain (i.e. divide the output by the input) then find the PM of the loop gain not the output. ofcourse the OPA will be stable if it was stable in the first case (u just changed the input !!!)

Hi,

I dont understand when you say that "divide the output by the input) then find the PM of the loop gain not the output". Normally, we break the loop, then find the gain , and find the PM at the unity gain frequency. Please explain here.

Thanks,
Suria.

 

[windknows]

that is the point

Hi,
We set ac=1V only to make the calculations simple for ourselves.

i.e ac analysis is done primarily to find the loop characterisitcs.

Gain = 20 log (Output/Input )

If input equal to 1, then gain is 20log (output).

However one question still remains:

ac analysis is a small signal analysis.
Here the tool lineraizes the MOSFET's around some operating points.
Now if the stimulas is as large as 1V, is it not that the transistors will go into triode.

Is there something in built in the tools (SPECTRE), say some algorithms, which makes the devices insensitive to such large voltage swings.
Because theoritically for small signal analysis the input stimulas should be around .25 V, else it may cause deviation from non linearlities.

Regards
Sarfraz