# Wrong output in HSpice

#### ashuembed

##### Newbie level 6
Hi
I am new to HSpice and simulating a simple CMOS inverter, the netlist is as follows,
Code:
Inverter Circuit

M1 OUT IN VDD VDD CMOSP L=0.18u W=1.8u
M2 OUT IN 0 0 CMOSN L=0.18U W=0.9U
VDD VDD 0 1.8
VIN IN 0 0 PULSE 0 1.8 2n .5n .5n 7n 20n

.OPTIONS LIST NODE POST
.TRAN 200p 20n
.PRINT TRAN V(IN) V(OUT)
.LIB "tsmc_018um_model" CMOS_MODELS
.END
when I simulate it using HSPICE J-2014.09-SP2 all is well, output is coming as expected in CosmosScope

also in avanwaves output is coming as expected,

but in my another lab where HSPICE L-2016.06-SP1 is installed, simulating the same netlist with HSPICE L-2016.06-SP1 output on cosmosScope is comming like this,

even on changing the x-axis to (0 to 20n) the output is,

on avanwaves when i try to open the .tr0 file it gives error,

I am unable to figure out what is happening,

- - - Updated - - -

thanks

#### dick_freebird

Maybe the problem is "another lab" and not the software
version. Inspect the log file for warnings about missing
.LIB target. Inspect the target model files to be sure they
are where and what you think they are.

#### ashuembed

##### Newbie level 6
Yes sir I have ensured that the LIB file is OK, actually yesterday night I simulated with HSPICE L-2016 in my first lab, actually both labs has HSPICE L-2016 but still no luck the same wrong output. Then I simulated on three different computers with same netlist and model files but HSPICE J-2014 gives correct output and HSPICE L-2016 wrong output. And no body in the lab including the instructors is able to fix the problem.

thanks

##### Super Moderator
Staff member
Looking at your 3rd scope image... Its scale markings do not look the same as the other scope scales. Its time frame appears to be seconds, not nanoSec.

As something to try...
Rather than import the old netlist drawn in the old version...
Draw the same circuit new in the version of software you say you must work in. Its netlist possibly has some subtle differences.

Simplify it as much as you can. It may help if you start with a more basic circuit, and get the timeframe correct, and the tilmestep correct, and the scope scale correct. Etc.

Every simulator has its quirks. Software developers seem to avoid making their program easy to use. (Notice that your instructors cannot find the problem?) Of course software writers need to compete and add features that users demand. One has to be a longtime user to get to know the quirks.

#### ashuembed

##### Super Moderator
Staff member
Is the load a capacitor by itself? That can give a simulator trouble.

Try changing the load to a resistor. Compare the simulator results. See if the problem goes away.

Then change the load to series RC. Compare all simulator results.

#### ashuembed

##### Newbie level 6
Great thanks for your reply, now I am trying the simplest, voltage divider
Code:
*** TOP LEVEL CELL: R_divider{sch}
R0 vout vin 10k
R1 0 vout 10k

* Spice Code nodes in cell cell 'R_divider{sch}'
v1 vin 0 sin(0 5 1k)
.OPTIONS LIST NODE POST
.tran 1u 5m
.print tran v(vout) v(vin)
.END
output of hspice l.2016, on cosmosscope

after scaling the x-axis,

awaves display,

these above are ALL WRONG OUTPUTS.

output of hspice j2014, on cosmosscope

awaves display,

these above are ALL CORRECT OUTPUTS.

##### Super Moderator
Staff member
For further assistance it may require someone familiar with your setup and your simulator. Example, 'Unknown file error' may have to do with the way HSpice is installed on your computer, rather than your circuit or component models. Or the way one version of HSpice handles files, compared to another version.

Nowadays simulators have become many and various. Each has its quirks no matter which simulator you use. You need to try this and that in order to cope with the unexpected.

Those sloping plots imply that the component models are more complex than they need to be for building a 'simple cmos inverter'. Did you try using a single transistor to invert a signal (common emitter mode)? That works too. Did you achieve successful plots in all your labs?

#### ashuembed

##### Newbie level 6
Till now I have tried 10 examples, the last is the above voltage divider, it has no transistor or any other device model involved. All examples are showing the same behaviour, they are running well with hspice j2014 but not with hspice l.2016.
It is very strange and embarrassing situation.

##### Super Moderator
Staff member
Although I lack hands-on experience with HSpice, I get the idea that software developers like to change things around from one version to the next. Perhaps it's their way of gaining continued revenue from their customer base.

Did you locate a message board for HSpice?
Did anyone present a similar problem as yours?

#### ashuembed

##### Newbie level 6
I could not locate a message board for HSpice, is there any way or forum, email etc to ask this problem directly to the synopsys hspice team? I could not register to the synopsys community because Non-customer community registrations have been temporarily paused.

All the students, teachers and users of HSPICE L2016 I met till now in my friend circle report the same problem, all are saying that HSPICE L2016 has some bug and produces corrupt simulation output files, and the left it and lost the hope completely.

##### Super Moderator
Staff member
It is very strange and embarrassing situation.
Embarrassing for someone but remember you are not the one responsible.

About computer programs there's a saying: 'If it's difficult to use it's not the user's fault, it's the programmer's fault.'

they are running well with hspice j2014 but not with hspice l.2016.
I have to work in HSPICE L-2016
no body in the lab including the instructors is able to fix the problem.
It's hard to believe these statements must co-exist. Someone who's just learning should not have to run up against this. What simulator did others in your circle turn to? After trying a few simulators (free, online, interactive, non-interactive), you should gain a perspective to make comparisons and contrasts.

And, did you test various circuits with real hardware, and real oscilloscope?

#### ashuembed

##### Newbie level 6
Actually In my EDA work group there are two labs one is 'design lab' another is 'implementation lab'. my design flow is covered using 4 different EDA tools linked together, and only hspice l.2016 is linked to these softs so i am only to use hspice l.2016 only, also I don't have the login access of the design lab which has both hspice2014, 2016, I work in 'implementation lab' becoz i have to use some tools other than the hspice also which are not in the design lab.

so i have to use hspice2016 only.

##### Super Moderator
Staff member
Sorry, I know nothing about this. HSpice ought to have a help number, if it's that prominent.

Surface thinking might lead you to tell the boss 'his choice' of software gives you problems, but he'll probably say 'find a way around it.' You want to stay in his good graces, and to avoid looking helpless. Don't confront him yet. Instead, can you run the simulation in the 2014 version so you get good results... Then attach the good printout to a schematic in the 2016 version?

It's a workaround, and it shows resourcefulness on your part.

Or, what happened to the 2017 or 2018 or 2019 version? Who says you can't use something more modern?