Hi
You may be able to simulate an AC analysis with transistors operating at higher currents by forcing an unusual bias condition e.g. d.c. offset that would put one of the output stages into its higher current mode to perform the D.C. setup.
The difficulty I see is that you will need to run simulations at several currents and for positive and negative outputs to turn on the different output halves. Transient simulation covers all these, but as a start if you run simulations under quiescent conditions, half and full power out, + and -, that may reveal the behaviour of the amplifier.
John.
You may be able to simulate an AC analysis with transistors operating at higher currents by forcing an unusual bias condition e.g. d.c. offset that would put one of the output stages into its higher current mode to perform the D.C. setup.
The difficulty I see is that you will need to run simulations at several currents and for positive and negative outputs to turn on the different output halves. Transient simulation covers all these, but as a start if you run simulations under quiescent conditions, half and full power out, + and -, that may reveal the behaviour of the amplifier.
John.
I haven't dove into the TMC (yet) and that's why I simulated OLG in conventional way. I tried both positions of resistor - before and after Tian's probe (I use the probe from Frank Wiedmann's page). I modified slightly the Vostro's schematic in order to easily conduct the tests (picture 1). Neither ways showed me some meaningful results. The try with resistor after the probe (picture 2) yields more promising results (unity gain at 844 kHz and phase shift -113 deg.). The other (picture 3) (with resistor before probe) - unity gain at 197 kHz/phase shift -90 deg. Results form transient analysis with pulse signal at the input, however, are unpromising (picture 4). I think the result from picture 4 is closer to the OLG yielded from the try with resistor before probe. The other thing I observed is that putting the load after the probe gave wrong results too.
May be Dave Zan will explain more about that strange results.
May be Dave Zan will explain more about that strange results.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I followed Bob Cordell's recommendation to go TMC and to take the collector of the Darlington VAS directly to ground. In my Reasonable amp project, it dramatically reduced the distortion.
I am a few steps behind you following along with Tian the probe. Today's project is to catch up some.
Just curious, your stopper resistors all seem large to my inexperienced eye. Can you give me a hint to what you are addressing so I can look into it further? I was putting real estate on the PWB for them, but found in my simulation they to not be necessary for stability and seemed to increase distortion. Many moons ago, buy luck or stupidity, I did fix an unstable output with 1 Ohm stoppers right on the pin. I learn something from just about every design posted on this forum. Or as Tom Lehrer said: " When in doubt plagiarize. That's why God gave you eyes. Let no ones work evade your eyes" 😀
An Evening Wasted with Tom Lehrer - Wikipedia, the free encyclopedia
(For those too young or of different cultures to get the reference)
I am a few steps behind you following along with Tian the probe. Today's project is to catch up some.
Just curious, your stopper resistors all seem large to my inexperienced eye. Can you give me a hint to what you are addressing so I can look into it further? I was putting real estate on the PWB for them, but found in my simulation they to not be necessary for stability and seemed to increase distortion. Many moons ago, buy luck or stupidity, I did fix an unstable output with 1 Ohm stoppers right on the pin. I learn something from just about every design posted on this forum. Or as Tom Lehrer said: " When in doubt plagiarize. That's why God gave you eyes. Let no ones work evade your eyes" 😀
An Evening Wasted with Tom Lehrer - Wikipedia, the free encyclopedia
(For those too young or of different cultures to get the reference)
I am sure I don't have sufficient operands specified. I tried to follow the procedure on the Spectrum whitepaper, but there are too many differences between their SPICE and LT. Trying to follow this thread, and I know I am missing something.
When I say lost, OK, I can easily understand why my looking across the input pair will only tell me if it is stable at quiescent, which is not a lot of help. Although it should be closer than I think I am seeing so I don't understand when I make it less than 20 degrees past when gain drops to zero, it is totally out of whack. It is hard to think all the Cob's, etc change that much. These are called transistors, not variactors.
My current best guess its to put a 2 Ohm load on it and send in a 20K at the input sufficient to cause clipping. This seemed to be more informative than removing parts to put in a square wave (which changed the phase response anyway)
So, we have this concept I don't even get the basis for. We are inserting a voltage and current source into the VAS. By doing some iterations, this is going to give us something that tells us what is going on. Great, but I don't understand how one point tells us the delta between two other points. I am not that knowledgeable yet.
So, I plug in a voltage and current source.
I do not understand the SPICE syntax where the big long summation is picked up. What makes SPICE look at it to presumably generate new graphs?
The parameters for the voltage and current sources as shown above have variables not defined anywhere I can find.
Do I ground the amp input?
My most recent SPICE model is posted over in my first amp protection thread, sans what I was trying to add for this test.
Heck, I did not know until I looked here how to shift the temp for devices. I was looking for a dialog box to put parameters in, not knowing I actually just edit the component name text adding a variable.
When I say lost, OK, I can easily understand why my looking across the input pair will only tell me if it is stable at quiescent, which is not a lot of help. Although it should be closer than I think I am seeing so I don't understand when I make it less than 20 degrees past when gain drops to zero, it is totally out of whack. It is hard to think all the Cob's, etc change that much. These are called transistors, not variactors.
My current best guess its to put a 2 Ohm load on it and send in a 20K at the input sufficient to cause clipping. This seemed to be more informative than removing parts to put in a square wave (which changed the phase response anyway)
So, we have this concept I don't even get the basis for. We are inserting a voltage and current source into the VAS. By doing some iterations, this is going to give us something that tells us what is going on. Great, but I don't understand how one point tells us the delta between two other points. I am not that knowledgeable yet.
So, I plug in a voltage and current source.
I do not understand the SPICE syntax where the big long summation is picked up. What makes SPICE look at it to presumably generate new graphs?
The parameters for the voltage and current sources as shown above have variables not defined anywhere I can find.
Do I ground the amp input?
My most recent SPICE model is posted over in my first amp protection thread, sans what I was trying to add for this test.
Heck, I did not know until I looked here how to shift the temp for devices. I was looking for a dialog box to put parameters in, not knowing I actually just edit the component name text adding a variable.
plus the big string does not even have matched () if that matters. Looking at the same string in Spectrum. Don't understand it. At least it tells me what the source variables should match to to.
Last edited:
I stripped it down to very basic. Nothing optimized, but it should suffice for getting the syntax correct. It is half done. I'm basically lost.
HELP! I think I am getting close, but I am not there yet. I looked at the example in LTSpice and what I can find here along with some hints from Spectrum. At least it runs. Makes no sense yet.
Error log says I have bad symbol in my func statement, yet if I cut and paste it as a plot definition, I get a plot that does not complain.
Please ignore the actual amp, I just stripped a bunch of stuff out to make is simpler. What is important is it is TMC.
Error log says I have bad symbol in my func statement, yet if I cut and paste it as a plot definition, I get a plot that does not complain.
Please ignore the actual amp, I just stripped a bunch of stuff out to make is simpler. What is important is it is TMC.
I've been reading this thread with great interest. As I am also trying to wrap my brain around that whole tian probe concept, and so far I am still totally lost (I'm not the only one apparently, I have good company)
I think I can understand this somewhat, but then I am looking at doing this on an amp design that is 100% symmetrical, so there are 2 TMC loops on the symmetrical vas. Then in this case, how can we insert just one single probe into 2 loops????
This will be very educational.
I think I can understand this somewhat, but then I am looking at doing this on an amp design that is 100% symmetrical, so there are 2 TMC loops on the symmetrical vas. Then in this case, how can we insert just one single probe into 2 loops????
This will be very educational.
- Status
- Not open for further replies.