SPICE macromodel for Harris HA9P5002 seems to have a problem

[jmelson]

Hello, all, I'm trying to get a SPICE macromodel for the Harris HA9P5002 unity-gain buffer to work.
Here's the model from the Renesas web page :
*
*  COPYRIGHT (C) 1991 HARRIS CORPORATION
*  ALL RIGHTS RESERVED
*
*  HA-5002 MACRO-MODEL
*  REV:  12-2-93 to fix CP value
*  D.W. RIEMER
*
*  PINOUT: IN VCC VEE OUT
*
.SUBCKT HA5002 3 4 5 2
.MODEL D1  D  IS=5.57E-11 N=1 RS=10
.MODEL D2  D  IS=3E-11 N=1 RS=10
*
*  INPUT STAGE
*  OFFSET VOLTAGE MAY BE VARIED BY
*  ADJUSTING THE VALUE OF SOURCE VIO.
*
CI 3 0 +1E-12
LI 3 12 +2E-09
VIO 12 18 +3E-03
IP 4 7 +1E-05
CP 4 7 +8E-15
IN 13 5 +7E-06
CN 13 5 6.66E-15
DO2 11 13 D2
DI2 18 13 D2
DO1 7 11 D2
DI1 7 18 D1
*
*  SINGULARITES
*  GAIN MAY BE ADJUSTED BY CHANGING
*  THE VALUE OF EP1 (0.995->?)
*
EP1 15 0 11 0 0.995
EP2 17 0 14 0 1
EP3 16 0 6 0 1
GP4 0 1 8 0 1
RP1 15 14 600
RP2 17 6 350
RP3 16 8 100
RP4 1 10 1 
CP1 14 0 +1E-12
CP2 6 0 +1E-12
CP3 8 0 +1E-12
CP4 1 0 +2E-09
LP4 10 0 +2E-09
*  OUTPUT STAGE
IPS 4 5 +8.3E-03
EO 19 0 1 0 1
RO 19 9 3
LO 9 2 +1E-09
CO 2 0 +1E-12
.ENDS HA5002


And here's a link to a horrible scan of the app note :
https://www.renesas.com/br/en/software/mm5002.html

What I am getting is that the input pin seems to be tied to a stiff voltage source centered between the rails, and the output follows that.
I've gone through the whole netlist and can't find anything wrong.  There are current sources to the + and - power supply which feed a diode
current switch (I'm probably calling that the wrong name).  The output of the current switch feeds a voltage-controlled voltage source,
so I'd assume that has an infinite input resistance on the control input.  I can't see anything with any of this that would jam the input at,
in this case, 2.5000 V (I'm working with a single-ended 5 V supply in this case.)

I'm using Protel 99SE for schematic entry and simulation.  While slightly quirky, it generally works just like SPICE.

Can anybody see anything wrong here?

Thanks,

Jon

[iMo]

Hmm, it works in LTspice..

[jmelson]

Hmm, it works in LTspice..

Did you just use an LTspice library component, or did you enter the SPICE .SUBCKT from my message?  Maybe somebody fixed the problem with
the manufacturer-provided model.

Thanks,

Jon

[iMo]

I've copied/pasted your text.

[jmelson]

I've copied/pasted your text.

Yes, there appears to be something wrong with the Protel 99SE simulator.  I still haven't gotten the HA9P5002 model to work, but I did get another similar part to work that was already in the library.  But,when I put TWO parts in the schematic, everything goes wrong.  So, I have an OP275 op-amp that works fine by itself.  And, I have the (ancient) EL2001 buffer that works fine.  But, when I put the buffer after the op-amp, the op-amp does NOT work.
I've been trying to find where things go wrong, but not having much luck with it.  I've done some modest simulations before and they worked fine, but this one is giving me fits!

Thanks for testing the model for me.  I may have to move this project over to LTspice.

Jon

[jmelson]

OK, so I got the models into LTspice (Funny, the OP275 is an Analog Devices part, they own LTspice now, but it isn't in their library.)  And, it works in LTspice.

So, my original attempt was with the simulator in Protel99SE, which I've used for years.  If you can get the schematic part to call up the model subcircuit, then it has always worked for me before.  But, on this one, somehow, the presence of the the HA9P5002 in the circuit fouls up the OP275 somehow.  I have not been able to figure out what in the model is causing it.

But, since LTspice works (although I HATE their schematic entry) I can get someting accomplished.

Thanks,

Jon