Ni circuit design suite 11 bug?

[bwana1]

Hi all,
I'm a retired engineer who is willing to learn using NI CDS11.
As I'm used to start bottom-up I'm trying to simulate some simple circuits at first.
So I began simulating a simple op-amp follower, powered by a single supply from
-VCC to +VCC following the supply midpoint-value obtained with two 10k resistors.
The initial idea was to use it to obtain an analog ground for a battery operated equipment.
I have used TI TL08x op amps all the life long, so I was very disappointed when I saw lots
of errors coming out from simulation. At first it seemed that it wasn't allowed to put analog
ground symbol on a generic node like the op-amp output because virtual instruments gave
faulty readings, even using floating instruments as a tester. My second move was then to
stick it to the most negative potential, the battery (12V) minus. This time the circuit began to
work but behaving erratically, halting on errors that could not be recognized. I then replaced
the TL084 with a CA3130 and all worked well at first.
I made some other test with other largely used op amps and saw that these coming from 741
were more prone to simulation crash. I thought that the SPICE model was broken, and
downloaded some up to date from TI site. A visual cross inspection showed that the SPICE
models loaded were OK.
It seems to me the simulator has some bug dealing with those op-amps, can anybody please
indicate me a possible error or an alternative very common 4 x jfet op-amp building block that
works instead?
Best regards

 

[keith1200rs]

I don't know what NI CDS11 is, but it sounds like you are battling with buggy software instead of learning. Unless you have a particular reason for wanting to stick to that simulator, there are others around that are a lot more stable. Don't bother with LTspice either - it has huge problems converging. The demo version of SIMetrix is fully functional, just limited by circuit size. https://www.simetrix.co.uk/

Mind you, if you have bought the other software - hassle the company who you bought it off and get some support.

Keith.

 

[bwana1]

I don't know what NI CDS11 is, but it sounds like you are battling with buggy software instead of learning. Unless you have a particular reason for wanting to stick to that simulator, there are others around that are a lot more stable. Don't bother with LTspice either - it has huge problems converging. The demo version of SIMetrix is fully functional, just limited by circuit size. https://www.simetrix.co.uk/

Mind you, if you have bought the other software - hassle the company who you bought it off and get some support.

Keith.

I'm very surprised, I thought that it was clear from the title I referred to National Instruments Circuit Design Suite 11 that it doesn't just look the last in its class...
specially because Analog Devices itself relies on a custom version to evaluate its own products!

 

[keith1200rs]

Ah - most people refer to it as Multisim, I think. Anyway, I don't know its pecking order in the world of simulators but it sounds like it has problems. For many years Pspice was the standard and that was what I used but support was lousy and bugs would take 6 months to get fixed. Not much use if it crashes everytime you try to run a particular model. I was recently evaluating Tanner EDA for SOI design and that crashed everytime I tried to run with SOI models.

My suggestion is get support from NI and if you get nowhere, try another simulator.

Keith

 

[bwana1]

My suggestion is get support from NI and if you get nowhere, try another simulator.

Keith

Yes, but first I have to be sure it's not my fault.
In the meantime I'm trying some other AD quad op-amps (OP482, OP490, AD648, AD704, AD713, ADA4062-4) and definitely there is quite a different behaviour in respect to TI's TL08xx that regularly crash. Now I'll do some little step further, simulating a difference amplifier using AD704, and if I succeed I'll then try simulating a log amp.
Thank you for your support.
Al

 

[keith1200rs]

If you let me know the circuit & model I can try it on my simulator. It may not prove anything but you never know. Analog Devices models were notorious for convergence problems in the past, but are pretty good now, for example. There are some duffers around - some with simply incorrect node connections.

Keith

 

[keith1200rs]

I have done some simulations with the TL084 models and found no problems with my simulator, although I am not necessarily using the same models or circuit as you.

Keith.

 

[bwana1]

Thank you for your support, Keith.

I apologize having not replied before. I have a "database invalid size" error that I wasn't able to fix even by a wipe-all disinstallation (registry included) and successive reinstallation, though it appears it makes no harm.
In the meantime I arrived to obtain meaningful behaviour by connecting ALL package units to power lines, though it would be enough to have connected a single unit.
I attach now two snapshots of my actual simulation that seem to work, but even so it happens that the simulation crashes without an evident specific reason. I think this simulator is buggy, and the fact you succeeded with TL084 is the evidence. The circuit I post would be the frontend for a microwave WSWR meter, that should continue with a log section to make the division in the formula. I know it's a dated solution, but it intrigues me, even because dealing with offsets and drifts in a log amplifier is not that simple, and requires a careful design. For sure, offsets and drifts of my beloved TL 084 far exceed what can be accepted, in any way I think I'd use an amplifier like this in my sketch, any suggestion is welcome.
Best regards
Al

 

[keith1200rs]

Al,

Just one thought - in general, simulators don't like capacitors directly across voltage sources - it usually gives infinite current and a divide by zero error. So, C1/C2 across V1 could cause a problem. Put a small resistor in series with V1 (unless J1 has a finite resistance).

I wouldn't expect that to be the cause of your problem as it exists whatever opamp you use, but you never know.

Keith.

Added after 21 minutes:

Al,

I'm not sure which analysis you are doing, but a DC sweep might be informative.

Keith.

 

[bwana1]

Hi Keith,
The DC sweep you posted agrees with my expectations as I supposed an input signal no more than .4Vp. OK about capacitors, I have no experience with simulators and have put them as usually they are inserted. I don't understand the meaning of the green transient, the blue one would be the output at startup when V1 is already on, like voltages even in my own simulation.
I didn't insert any oscilloscope in my test because they appeared to increase the test instability.
The analysis I'm doing, as I have no experience with sims, mirrors what I could have done with real instruments on the bench, that's my main experience, just think, turn a knob, look, think.
And from this viewpoint the circuit I sent you looks good, but I'm aware that with simulators one should perform quicker, extensive and more definive analyses, that's the reason I want to learn how to.
And all your suggestions are welcome!
If your experience suggests another simulator as easier to learn, let me know. In the past I saw another promising lab software on Linux KDE, its name was kdelabs, but that too had crash issues.
Best regards
Al

 

[keith1200rs]

Sorry, I should have said - the green trace was with a different input voltage.

I am not a fan of simulators which try to make it "user friendly" by making things look like oscilloscopes & signal generators - the schematic soon gets cluttered.

I have used Microcap, ICAP and Pspice in the past and now use SIMetrix. I think once you know how to use one you will not find it too difficult to use another as they are all based on Spice and so have a lot of similarities. I chose SIMetrix due to the support. It is pretty stable now, but if it ever crashes and I need it fixed, I usually get a fix within a day. That compares with up to 6 months even for critical bug fixes with most other companies. It seems pretty good at convergence - something that LTspice seems particularly bad at.

I wouldn't suggest changing now unless your crashes stop you from working.

Keith.

 

[bwana1]

Hi Keith,
I can come back only now cause the bad wireless connection I have in my village specially when it's raining.
Thank you for your suggestions, I'll look to the sims you say.
In my present experience with NI multisim I thought a lot of times it's a crap, though I can't figure how such a high-end (I'd say price) can be a crap. So I say myself that I am the stupid and not the others, but don't know how long I'll say that.
By the way, avoiding single op-amp modules that seem more prone to crashes, in those days I succeeded in simulating all the VSWR meter signal conditioning, I attach here the schematics. The main interesting circuit, the log multiplier-divider comes from that Holy Bible that's the old Burr-Brown op-amp application handbook published by Mac Graw Hill in the early seventies, when I suppose you as me were young men.
Great cookbook! hit at the very first shot!
I'd like some comment of you
Best regards
Al