opamp matching +/- input impedance necessary for buffer?

Status
Not open for further replies.
Jump to Latest
#21
Hi Bill, I think I have your work email address, is that OK? I assume you are still there? last email was in sep 2006!!! you had changed jobs by the looks of it. I'll send you a PM to check 🙂 The email address you have for me is no longer active.

Yes if you are using a generic opamp in LTspice then it is an ideal opamp so totally devoid of any real world characteristics, TI has a lot of models available so the OPA627 should be on their site I suspect. The file I send you should help with working out how to do that I suspect 🙂 I've been through the same, started off with what ever was available, and just did ac analysis, then moved on to getting models for the parts I have and doing transient analysis. It's a powerful tool provided you recognise it has limitations and is a bit too perfect 🙂

Tony.
 
#24
Spice op amp macromodels seldom use good input Q models - often no nonlinear junction C at all so they do not give good indications of the cm input nonlinearity distortion

http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf has measurements - the OPA627 is unusually good with its Difet input Q

you can get around the problem by "cascoding" the whole input op amp by bootstrapping power supply pins at considerable circuit complexity

Supply Bootstrapping Reduces Distortion In Op-Amp Circuits | New operational amplifiers optimized for high-performance audio and ultrasound applications combine extremely low total harmonic distortion plus noise (THD+N), -130 dB, with large output vo
 
#25
Perhaps you could use an ideal high-gain controlled voltage source combined with two big junction diodes as the op-amp model, one big junction diode from the positive input to the negative supply and an equally big junction diode from the negative input to the negative supply. Choose the junction diodes to have a capacitance of a few pF. It won't match the real op-amp very well, but the trends will be the same.
 
#26
jcx said:
Spice op amp macromodels seldom use good input Q models - often no nonlinear junction C at all so they do not give good indications of the cm input nonlinearity distortion
http://electronicdesign.com/article...tstrapping-reduces-distortion-in-op-amp-.aspx
Click to expand...

This model may be better than most...

extract from the E version of the model:

Code: 
* DIFF INPUT CAPACITANCE
CDIF  1  2  8.0E-12
* COMMON MODE INPUT CAPACITANCE
C1CM  1  99 7.0E-12
C2CM  2  99 7.0E-12
* INPUT VOLTAGE NOISE
VN1 61 0  0.6
VN2 0  62 0.6
DN1 61 63 DY
DN2 63 62 DY
EN 64 1  63 0 1
* INPUT CURRENT NOISE
RN1 0 65 60.3865
RN2 65 66 60.3865
RN3 66 0 120.773
RN4 0 67 60.3865
RN5 67 68 60.3865
RN6 68 0 120.773
******************
Click to expand...
Tony.
 
#29
I have just done another active line level crossover network again. The result to me is very obvious. The DC offset of the opa627 increases along with the difference increase of the +/- input impedance. However, when it is used in a buffer, i.e. a piece of wire goes from the -input to the output, the DC offset remains low, even though the +/-input impedance is not matched.

Can somebody give an explanation?
 
#30
If the amp has a gain of 10, the DC offset will be multiplied by that. So if a opamp with a gain of 10 has a 6mV offset, if it is then configured for unity gain it will have an offset of .6mV.

PS. The OPA637 model is much more detailed... Here is a copy I modified to have the GBP of the OPA627 (but I may have messed something else up in doing so):

Code: 
.SUBCKT OPA627_k  1 2 3 4 5
C1   11 12 2.5E-12
C2    6  7 16.75E-12
DC    5 53 DX
DE   54  5 DX
DLP  90 91 DX
DLN  92 90 DX
DP    4  3 DX
EGND 99  0 POLY(2) (3,0) (4,0) 0 .5 .5
FB    7 99 POLY(5) VB VC VE VLP VLN 0 552.6E6 -60E6 60E6 60E6 -60E6
GA    6  0 11 12 1.810E-3
GCM   0  6 10 99 2.868E-9
ISS   3 10 DC 486.0E-6
HLIM 90  0 VLIM 1K
J1   11  2 10 JX
J2   12  64 10 JX
G11 2 4 POLY(4) (10,2) (11,2) (4,2) (66,0) 0 1E-12 1E-12 1E-12 1.6E-6
G21 1 4 POLY(4) (10,1) (12,1) (4,1) (68,0) 0 1E-12 1E-12 1E-12 1.6E-6
R2    6  9 100.0E3
RD1   4 11 552.6
RD2   4 12 552.6
RO1   8  5 54
RO2   7 99 1
*  RP    3  4 4.286E3
RSS  10 99 411.5E3
VB    9  0 DC 0
VC    3 53 DC 2.700
VE   54  4 DC 2.700
VLIM  7  8 DC 0
VLP  91  0 DC 55
VLN   0 92 DC 55
* OUTPUT SUPPLY MIRROR
FQ3   0 20 POLY(1) VLIM 0  1
DQ1  20 21 DX
DQ2  22 20 DX
VQ1  21  0 0
VQ2  22  0 0
FQ1   3  0 POLY(1) VQ1  5.38E-3  1
FQ2   0  4 POLY(1) VQ2  5.38E-3 -1
* QUIESCIENT CURRENT
RQ    3  4  7.5E4
* DIFF INPUT CAPACITANCE
CDIF  1  2  8.0E-12
* COMMON MODE INPUT CAPACITANCE
C1CM  1  99 7.0E-12
C2CM  2  99 7.0E-12
* INPUT VOLTAGE NOISE
VN1 61 0  0.6
VN2 0  62 0.6
DN1 61 63 DY
DN2 63 62 DY
EN 64 1  63 0 1
* INPUT CURRENT NOISE
RN1 0 65 60.3865
RN2 65 66 60.3865
RN3 66 0 120.773
RN4 0 67 60.3865
RN5 67 68 60.3865
RN6 68 0 120.773
******************
.MODEL DY D(IS=1E-15 AF=1 KF=73.4E-18)
.MODEL DX D(IS=800.0E-18)
.MODEL JX PJF(IS=500.0E-15 BETA=3.37E-3 VTO=-1)
.ENDS
Unfortunately it seems nonlinear input capacitance is still not modeled. However if you look at the datasheet the inputs are cascoded so the input capacitance should be linear...

- keantoken
 
#32
I have now tried matching two of the opamps at DC using the method prescribed in one of the early pages.

I found that for opa627 buffers the DC offset is the smallest when the negative input is connected to the output via wire, i.e. no matching is required. I am happy for this finding.
 
Status
Not open for further replies.
Top Bottom