Hi,
I have some problems to simulate PSU circuit in MC 12. This circuit working OK in LT Spice, working OK in real world (used for amplifier), but i cant made valid sumulation. It has voltage doubler which feed MOSFET regulator for input stage and VAS stage of a amplifier.
Circuit is in attach. If someone have little time, please help.
Used transormer has double 36VAC secondaries. I need for output stage arround 50V, for input + vas arround 25mA...
Thanks in advance,
I have some problems to simulate PSU circuit in MC 12. This circuit working OK in LT Spice, working OK in real world (used for amplifier), but i cant made valid sumulation. It has voltage doubler which feed MOSFET regulator for input stage and VAS stage of a amplifier.
Circuit is in attach. If someone have little time, please help.
Used transormer has double 36VAC secondaries. I need for output stage arround 50V, for input + vas arround 25mA...
Thanks in advance,
My pleasure. Now that the sim works, you can replace the 36V AC sources with a transformer. To be simulated with coupled inductors, each with DC resistances you have to measure. What matters is the voltage over the primary series resistor: issues like peak current. With MC I also use simulated loads (ultra simple power amps) to verify dynamic behavior of power supplies (both linear and switching). As a result, construction always works.
The saying goes, "practice makes the master" and although making the simulation as "real world" as possible initially takes effort, after a while it becomes a habit. The only "negative" is, simulations take more time to run. The positive side is to discover issues that weren't immediately clear at the design phase. Measurement of transformers can reveal that under resistive load, they can deliver up to 3 times the rated (capacitor input) power.
There is a reason that people payed $4700 for Micro-Cap and LTSpice was free. Engineers are not stupid, but if you have to pay for something, free is better. It is ironic that LTSpice is not SPICE based but is their own brew of math made simulation. Micro-Cap is a SPICE based tool with real models which can be used at different levels of complexity. LTSpice parts are behavior based - meaning - worthless to simulate expect to get the nice pictures that you expect and will get. Put 400 V on a pin that would blow up - and the part will run fine.
Now that Micro-Cap is free - it is crazy to stay with LTSpice.
There are traps with LTSpice, first you have to consider - the only reason that they came up with this 'spice like' simulation software is so that they did not have to share their models with others. It is not SPICE based. In fact, if you are using their parts in the simulation - they are behavior based - meaning that they give you the behavior with the correct input.
Just try to make an LTSpice show that one of their parts can go unstable - you just can not do it. It provides nice pictures but that's about it. It is made for speed as well, so if you have an inductor - make the dc resistance equal to zero ohms - this way it will force the tool to treat it as a real part and not an ideal Norton source.
Stop using it and use Micro-Cap. it is a SPICE based tool with a huge library and is not designed by a chip manufacturer - which has a terrible interface in comparison to a polished product like Micro-Cap.
Now that Micro-Cap is free - it is crazy to stay with LTSpice.
There are traps with LTSpice, first you have to consider - the only reason that they came up with this 'spice like' simulation software is so that they did not have to share their models with others. It is not SPICE based. In fact, if you are using their parts in the simulation - they are behavior based - meaning that they give you the behavior with the correct input.
Just try to make an LTSpice show that one of their parts can go unstable - you just can not do it. It provides nice pictures but that's about it. It is made for speed as well, so if you have an inductor - make the dc resistance equal to zero ohms - this way it will force the tool to treat it as a real part and not an ideal Norton source.
Stop using it and use Micro-Cap. it is a SPICE based tool with a huge library and is not designed by a chip manufacturer - which has a terrible interface in comparison to a polished product like Micro-Cap.
Fascinating discussion. I recall some guys using SPICE back in 1984 at RCA. Obviously has a good long track record (and a good long library by now).
Obviously, there were not many that could be convinced to pay $4700 for MicroCap
or it would be still around.
Programs that are no longer supported are dead and they soon start to smell.
Bad timing for a fresh start with a program that has just been declared dead,
not exactly a brilliant idea to invest your time when you're standing at the grave.
Just wait when Microsoft forces the next WIN10 update down your throat.
Where did you get that drivel that LTspice is not Spice based or that "its" parts
are behaviour-based? It uses the models you provide. Just don't hope that
somebody will give you a transistor level model of their newest DC/DC converter.
> Now that Micro-Cap is free - it is crazy to stay with LTSpice.
No. It's crazy to catch a falling knife.
(more drivel, FUD and fakes deleted)
If I had to select a new simulator to replace LTspice it would be
NGspice. But for now I have to re-learn ADS.
Gerhard
Actually the company did great and the man who owned it decided to give it away after deciding to retire. Pretty noble. KTSpice would never compete for 30 years if they had asked for near the money MicroCap brought. If you want to stay with a non SPICE based tool then do, but educate yourself and ask LT if the remit tool is SPICE based. Then you will believe. Yes, it can import SPICE models but it translates them. LTSpice parts ARE behavior based - simple as that! Again ask LT, educate yourself instead of spilling dribble - and showing your level of knowledge. Or lack of.
I don't think that a MicroCap fanboi is qualified to teach me about spice.
We had to learn the innards of spice in the late 70s and had to write the
important algorithms ourselves before we were given the spice sources.
And we had to pass the tests.
I have compiled the Fortran and C versions myself, on the VAX
and later on Unix/Linux. And yes, I have Microcap too on a virtual machine.
But not on the active one.
And if you google LTspice, already the first result contradicts you.
Gerhard
We had to learn the innards of spice in the late 70s and had to write the
important algorithms ourselves before we were given the spice sources.
And we had to pass the tests.
I have compiled the Fortran and C versions myself, on the VAX
and later on Unix/Linux. And yes, I have Microcap too on a virtual machine.
But not on the active one.
And if you google LTspice, already the first result contradicts you.
Gerhard
Thrilled
I just discovered this three days ago. As a long time LTSpice user, I downloaded Micro-Cap, and have been at it since. I have to say I'm rather blown away by the whole experience, positively.
Of note are the ability to change component values and instantly see bias current and voltage changes. This alone is not only efficient but also lots of fun.
The THD analysis saves a lot of time I'd otherwise spend iterating in LTSpice to get the whole picture. That said, compared to LTSpice, the Y-axis on THD plots will take some getting used to (for me).
The way in which DC results are presented with all voltages, currents, and/or dissipation shown right on the schematic at once, is great. Hey, if LT has this feature I've not seen it.
I still love LTSpice though. And LT has some critical things. For example, if MOSFETs are important to you, to my knowledge, the latest and greatest models are the subthreshold MOSFET models which simulate the exponential (as opposed to square law) behavior under lower Vgs conditions. See Kean's models e.g. Thanks to KeanToken and others.
Does Micro-Cap support subthreshold MOSFET models? If Micro-Cap doesn't now, it may never.
One downside to Micro-Cap might be that too much is done under the covers. One example is the loop_gain probe. In one case, I saw that conditions were favorable for oscillation in the MHz region and yet it reported 89 degrees of phase margin because it found the first crossing in the KHz region where gain was below unity at 180 degrees of PM. _Please_.
Why not use both? That's what I plan to do, at least for now.
I just discovered this three days ago. As a long time LTSpice user, I downloaded Micro-Cap, and have been at it since. I have to say I'm rather blown away by the whole experience, positively.
Of note are the ability to change component values and instantly see bias current and voltage changes. This alone is not only efficient but also lots of fun.
The THD analysis saves a lot of time I'd otherwise spend iterating in LTSpice to get the whole picture. That said, compared to LTSpice, the Y-axis on THD plots will take some getting used to (for me).
The way in which DC results are presented with all voltages, currents, and/or dissipation shown right on the schematic at once, is great. Hey, if LT has this feature I've not seen it.
I still love LTSpice though. And LT has some critical things. For example, if MOSFETs are important to you, to my knowledge, the latest and greatest models are the subthreshold MOSFET models which simulate the exponential (as opposed to square law) behavior under lower Vgs conditions. See Kean's models e.g. Thanks to KeanToken and others.
Does Micro-Cap support subthreshold MOSFET models? If Micro-Cap doesn't now, it may never.
One downside to Micro-Cap might be that too much is done under the covers. One example is the loop_gain probe. In one case, I saw that conditions were favorable for oscillation in the MHz region and yet it reported 89 degrees of phase margin because it found the first crossing in the KHz region where gain was below unity at 180 degrees of PM. _Please_.
Why not use both? That's what I plan to do, at least for now.
Yes Micro-Cap has all the capability of using different levels of SPICE models. Double click on the MOSFET and below the basic description window and at the top of the list of parameters you will see the LEVEL selection. There is a pull down selection where you will find Schichman-Hodges, Grove-Frohman, BSIM1-4, and a bunch of others. You can also import your own very easily.
Drivel indeed, LTSpice is based directly on the public domain code that Berkeley released decades ago.
You are so young, as a freshman in university I compiled the FORTRAN source on a PDP-11/34 in almost 100 overlays, since it had so little RAM. Eventually I did the same with SUPREM-4 too.
I accidentally came across a fact that I did not like with MC. I wanted to put two sinusoidal voltage sources with 180 degree phase difference on the circuit. The developers made such a source, but with the phase expressed in radians. The funny thing is that when I made netlist, I found standard sources with phase in degrees in the text. Why did you make radians? And that's not all. I put one source with phase 0, copied it and changed the phase to 3.1415. I counted and got the same signal on both sources. I made a netlist and found that the description of the sources is exactly the same. I went to the editor of the first source and found that the phase instead of 0 became 3.1415. Why did you do that?
Hey, we did our NR, nodal analysis, companion network, linearisation around o/p etc
on a Telefunken TR4, the first _commercial_ microprogrammed machine that was
withdrawn from military service. It had a dual card reader, a Buroughs(?) disk with a
head per track that could run open. The platter was > 1m dia, the motor used 3 phase
current with a leather belt as transmission. All OC604, AC122 transistors.
Anybody here wants Ge transistors, old stock, but guaranteed genuine?
Microprogram store was abt. one cubic meter of AA112 like diodes.
In a sense, it was my first PC. The operators ran it from 9 to 5 and some of us EE
students were allowed to use it outside that time.
< Telefunken tr4 - Google Search >
There was also a TR440 that no one was allowed to touch, and a Cyber 176
later in Berlin. Big number cruncher with weird word length and character set.
Our group in Berlin also ran the first Unix on this side of the pond
on a PDP11/40E. There were only 5 11/40E ever, the E stands for
"microprogrammable" and we had 2 of these.
cheers, Gerhard
Last edited:
