about k value (stability)!

[zyunknown]

Hi,
I have a question on the stability analysis on the designs of two LNAs.

The bandwidth requirement of one LNA is 50MHz-1GHz. Another one is from 10GHz-16Ghz.The attached two diagram is my K value stability analysis.

Would u pls. give me some advice on if my staboty analysis is OK?

Thanks in advance!

 

[BigBoss]

Your circuit may oscillate around 6GHz and it can saturate your circuit even tough this frequency is out of band..
An LNA should be uncondictionally stable over very large band otherwise after LNA and Mixer, some spurs will influence your design.

 

[zyunknown]

Your circuit may oscillate around 6GHz and it can saturate your circuit even tough this frequency is out of band..
An LNA should be uncondictionally stable over very large band otherwise after LNA and Mixer, some spurs will influence your design.

Thanks, but i still wonder about the frequency range in which the K value should be simulated and focused on. For the LNA with the band of 50M~1G, what kind of frequency range should be noticed? 1M~10G is ok or not?

 

[BigBoss]

Thanks, but i still wonder about the frequency range in which the K value should be simulated and focused on. For the LNA with the band of 50M~1G, what kind of frequency range should be noticed? 1M~10G is ok or not?

I generally check the stability factor up to 100GHz for an LNA.100Ghz may seem too much but if there is an oscillation somewhere else, this carrier can force the LNA into saturation and that's why linearity will drop very much.

 

[zyunknown]

Thanks, but i still wonder about the frequency range in which the K value should be simulated and focused on. For the LNA with the band of 50M~1G, what kind of frequency range should be noticed? 1M~10G is ok or not?

I generally check the stability factor up to 100GHz for an LNA.100Ghz may seem too much but if there is an oscillation somewhere else, this carrier can force the LNA into saturation and that's why linearity will drop very much.

Thank you! Still two more questions:
1.How about the stability in low frequency? Should i consider the stability from DC to the high band?
2. What kind of K value will be OK? I think it is not enough that K is just greater than 1, how about being greater than 1.05 or 1.1?

 

[DoYouLinux]

Hi,

About the stability factor, what kind did you use ?

If it is Rollet K factor, consideration is like digital, K > 1 = stable, K < 1 = unstable. There is no differences between K = 1.01, 1.1 10 or 100. They just tell you that your amplifier is stable.

If it is 'mu' factor, the greater the mu factor, the more stability is your circuit.

Hope this helps,

DYL

 

[BigBoss]

Thank you! Still two more questions:
1.How about the stability in low frequency? Should i consider the stability from DC to the high band?

Yes, the stability should be considered from DC to very high frequencies.

2. What kind of K value will be OK? I think it is not enough that K is just greater than 1, how about being greater than 1.05 or 1.1?

if the stability factor is equal to 1, in this case the system "may" oscillate with some load and source impedances (conditional stability). Under this circumtance, some fine tuning may help you...

But better case to prevent the stability factor approaching to unity and keep far it away as much as possible.

But another remark: Less than unity doesn't mean that the circuit will absolutely oscillate.But this is valid around k=1 zone..
But the stability less than much from unity, the circuit will potenatially be unstable...

 

[tomhive]

Thank you! Still two more questions:
1.How about the stability in low frequency? Should i consider the stability from DC to the high band?

Yes, the stability should be considered from DC to very high frequencies.

2. What kind of K value will be OK? I think it is not enough that K is just greater than 1, how about being greater than 1.05 or 1.1?

if the stability factor is equal to 1, in this case the system "may" oscillate with some load and source impedances (conditional stability). Under this circumtance, some fine tuning may help you...

But better case to prevent the stability factor approaching to unity and keep far it away as much as possible.

But another remark: Less than unity doesn't mean that the circuit will absolutely oscillate.But this is valid around k=1 zone..
But the stability less than much from unity, the circuit will potenatially be unstable...

Hi,
I wants to add one point..
Make sure K>1 for DC to your 10*band atleast....Bcos generaly higher frequenscy oscillation doesnt falls in your band of interest.but lower will fall in your band...lets say higher order harmonics
Thanks

 

[zyunknown]

Thank you all! I've got it!