Problems with AD5144 SPICE model from analog.com

Edit: I would say half, not three-quarters (0111 1111=127->127/256=0.5)

Hi Romito ,

I'll see what I can find out. Moving this to the LTspice forum.

mike

Hi Romito ,

Yeah, poked around the model and, using 1V source, was able to get 5 megavolts(!) from pin 8 alone ((128/256) * 10Meg) and 2.5MV from from pin 7 alone ((64/256) *10Meg). 1000 0000 and 0100 0000, respectively. 127 gives me 4.96MV, like your result. So, everything seems to work perfectly—if you add a divide-by-a-million behavioral source to your circuit downstream. 

Model is broken.

mike

Great, thank you for your answer ! If I understand  well, there seems to be a wrong factor (around 10 million) coded in the model.

Romito

Hi Romito ,

Correction: a mere 10 million. :-)

If you are so inclined, you might be able to fix it by editing the (probably originally Pspice) model. I might give it a shot, too. My guess is that is coming in here:

g4 24 20 poly(2) (24,20) (22,0) 0 0 0 0 .0004
g5 10 24 poly(2) (10,24) (25,23) 0 0 0 0 1000

v25 25 0 dc 1v

g6 10 26 poly(2) (10,26) (23,0) 0 0 0 0 1000
g7 26 20 poly(2) (26,20) (19,0) 0 0 0 0 .0001

I have to look up the syntax for those arcane SPICE G and E poly sources, which are supported by LTspice, but they are not legit LTspice. I've looked here before:

www.seas.upenn.edu/.../PSpice_ReferenceguideOrCAD.pdf

mike

Hi Mike, ty for your help !

I tried to put a 10MV (10 000kV) DC source to pin 13, only d7(pin 8) connected to 1V and pin 10 to 1V and I got this result:

Maybe pin 13 has its own behavior (like a gain setting ???) on the model but it's not specified anywhere...

I'll follow your advice and try to modify the model to see if I get results.

Romito,

I think this is the answer,

Looks like it needed to modify these parameters (g5 and g6) from 1000 to 0.0001 on the .cir model.

Many thanks MStokowski  to help me to solve this problem.

Romito

Hi Romito ,

Ah, nice. Odds are someone got it backwards. I'll enter your fix in as a model request. That error has been there a while.

mike