active LPF using lmh6629 op amp

Hi,

The LMH6629 is not unity gain stable and would be unstable if operated this way. The minimum stable gain of the LMH6629 is 4V/V (default LLP-8 package condition, COMP pin can be pulled high for 10V/V minimum) or 10V/V for the SOT23-5 package. You could:

a) Change your operating conditions to match the minimum gain requirement

b) Use external compensation to possibly operate at a closed loop gain of 1V/V

c) Use a different device which is unity gain stable

Hope this answers your question. If you want to go with option "c" above, let me know some of your requirements (supply voltage, noise requirement, swing, load, etc.) and I might be able to recommend a more suitable part to you.

Thanks,

Hooman

Hi Hooman

regarding option C you suggested :

since that i already have a ready PCB i am looking for an op amp with sot 23-5 package . it should have GBW product of at least 450 - 500 MHz , stable with low gain (< 4V/V), doul  voltage supply 3.3v or 5v . it need to drive 50 ohm load  noise less then 5nV/sqrt(Hz) or max noise figure of 5dB when terminating 50ohm.   . 

regarding option B you suggested , let say i want to have gain of 4 what are the values of te resistor and cap between vin_p and vin_n?

thanks

idan shmuel  

Hi Idan,

Regarding option c:

OPA694 fits the LMH6629 SOT23-5 package and is low noise, high bandwidth and unity gain stable but needs 7V total supply voltage minimum. LMH6703 (8V minimum supply) and OPA695 (5V minimum supply) in SOT23-6 are your other choices (the shutdown pin can be left floating for normal operation with the other pins matching the LMH6629 pin out) but LMH6703 requires +/-5V supplies. All these devices are current feedback type and thus require the use of a specific feedback resistor (please see the respective datasheets to determine this value).

Regarding option b (assuming you like to set the closed loop gain to +4V/V):

The exampled worked out on page 18 of the LMH6629 datasheet is for the case of 3.7V/V (resulting in Rc= 56ohm, Cc= 33pF) with R_s = R_T= 50ohm:

For your particular circuit, if you set the closed loop gain to 4V/V (R_F= 249, R_G= 82.5, both stages), simulation shows that the following components should give  you adequate phase margin (> 50 degrees):

R_c= 41.2, C_c= 47pF

Hope this answers your questions?

Thanks,

Hooman

Hi Idan,

I must say that I find the range of components you've chosen odd (e.g. 33ohm resistor in front of the 2nd stage). In fact, when I simulate your original filter, I get time domain oscillations no matter what closed loop gain I set for the LMH6629 or how I choose to compensate it externally!

I've re-evaluated your filter with what I feel are more realistic resistor values for a 1.8MHz filter as shown below (Butterworth, Sallen-key):

If necessary, you could compensate the circuit above with the Rc, Cc values I sent you yesterday, which would make the design more robust. Time domain simulation of the filter above is stable (vs. your original circuit which is not).

I've also attached the FilterPro design if you choose to open it and edit it to your liking.

Hope all this helps.

Thanks,

Hooman

5305.LMH6629 Filter 1.8MHz Butterworth 5_31_12.zip

in lmh6629 application notes there is a reference to app note QA-21 that deals with stabilizing this amp.   i took my original filter pro design which was in origin for unity gain and simulated with same components just now i let the first stage with gain of 10 and the second stage is with unity gain, the spice ac sweep simulation was stable.

so what i did wrong becuase in reality it dint work ?

Hi Idan,

Your frequency response plot ends at ~50MHz. When I simulate with your latest design, I see the instability / frequency response peaking at ~1GHz. So, if you extend the upper reach of your frequency response you'll probably see the peaking that I'm seeing. Transient simulation shows an oscillation frequency of 600MHz for me.

Regards,

Hooman

hmm i guess you are right small noise comming at high freq cause this amp be unstable,