The models we use in LT are (as I understand it) basic "1st order" models and don't represent fully the way a real device will behave. For example a real device with an AC signal across it will experience changes to junction capacitance, changes to gain and so on. Those parameters (again, as I understand it) aren't reflected in the models.
If you look at the data sheet for an LM4562 opamp and look at the method of testing distortion you will see that it says it is below current measurable limits. To get the real figure needs a simple trick... look at the data sheet to see how its done. (The same trick for the simple one transistor circuits won't work.) What I'm getting at is that in LT, its easy to get a circuit to perform "better" than the best currently available... so something 'aint right somewhere
What LT can do when used at this level is to guide you so that you can see what direction changes are heading, what the effect of certain changes are and so on but the reality won't exactly match the simulation. If it did then we would all be making world beating amplifier designs.
If you look at the data sheet for an LM4562 opamp and look at the method of testing distortion you will see that it says it is below current measurable limits. To get the real figure needs a simple trick... look at the data sheet to see how its done. (The same trick for the simple one transistor circuits won't work.) What I'm getting at is that in LT, its easy to get a circuit to perform "better" than the best currently available... so something 'aint right somewhere
What LT can do when used at this level is to guide you so that you can see what direction changes are heading, what the effect of certain changes are and so on but the reality won't exactly match the simulation. If it did then we would all be making world beating amplifier designs.
Hi and thanks a lot again for the very valuable reply
I understand that ... i dont understand
I mean ... as usual i am trivializing the issue
But ...
1) can we say that some parts have better performance of others for some application ? like to give a gain voltage of 2 to the incoming signal ?
2) can we say that taken a component its performance varies a lot in function of the working conditions ?
I have one big problem ... to understand the optimum working point for a bjt from the data sheet of the part , given that this point do exist of course
It is very clear from the sim that the performance change a lot with Vce and so on ...
I think that this point should be easy to spot in a data sheet
What i have to look for ? in which graph this point can be spot ?
Like working conditions that give max linearity to the device (i am guessing here)
I think i have to read to book first for sure ... i am sounding much like a neanderthal man, electronics wise.
Kind regards, gino
I understand that ... i dont understand
I mean ... as usual i am trivializing the issue
But ...
1) can we say that some parts have better performance of others for some application ? like to give a gain voltage of 2 to the incoming signal ?
2) can we say that taken a component its performance varies a lot in function of the working conditions ?
I have one big problem ... to understand the optimum working point for a bjt from the data sheet of the part , given that this point do exist of course
It is very clear from the sim that the performance change a lot with Vce and so on ...
I think that this point should be easy to spot in a data sheet
What i have to look for ? in which graph this point can be spot ?
Like working conditions that give max linearity to the device (i am guessing here)
I think i have to read to book first for sure ... i am sounding much like a neanderthal man, electronics wise.
Kind regards, gino
I haven't got all the answers, that's for sure.Yes, some parts do perform better in certain applications, some will reliably show that improvement in LT and others will not.
When you increase the supply, the change in voltage across the transistor relative to the DC conditions becomes small and that makes a big difference to performance.
Two important characteristics are "early effect" and the "Cob" (collector output impedance).
Yes, some parts do perform better in certain applications, some will reliably show that improvement in LT and others will not.
When you increase the supply, the change in voltage across the transistor relative to the DC conditions becomes small and that makes a big difference to performance
Hi and thanks and yes this is what i am seeing
Problem is that i find much more easier to read an op-amp data sheet than a bjt data sheet ... in the op-amp most of the graph are very clear
This is very difficulf for me now. I have to leave it for the future after reading the book.
Thanks a lot for the very interesting paper attached.
Unfortunately the only thing clear to me is the conclusion
There is more than enough to get worried ...
but i will read about this Kind regards, gino
Remember that good circuit design "designs out" the effect that temperature and so on will have on a circuit. That is why a simple circuit like the Radford can work as well with a transistor with a gain of 50 or 500. It doesn't matter, it doesn't alter the AC gain of the circuit or the alter DC operating conditions.
Thanks for the advice. Let's say that a very basic strategy to overcome the temperature effect may be is to use, as i intend to do, more powerful transistors well withing their SOA ... much well within actually. And maybe even heavily heath-sunk ?
Like the 2SC2238B for instance.
http://www.jmnic.com/pdf/2sc/2SC2238A.pdf
Then if one will be not enough i will pass at two ... but no more.
Or two or nothing .... when i look inside a line preamp and i see even more than 20 active devices per channel ... i wonder if this is really the right approach.
My simple mind would suggest me to use the very few devices in their optimum working conditions,
conditions that at the moment i have no clue which they are but i have the feeling that can be seen in a dataheet somewhere. Back to play with sims.
Kind regards, gino
Optimum conditions depend on your own specifications and intended use.
If you want a single transistor preamp (and the downside to that is that it inverts the signal) then you need to consider whether it can drive the intended load... and to make sure it can means you have to decide on the operating point to achieve that. Its no good saying you get optimal noise performance for example at xyz bias when that level of bias won't even drive a damp piece of string. Or conversely, if the optimum point is with it drawing a few hundred milliamps, then that too would be a compromise too far.
There are no rights and wrongs... the whole design has to be considered as a whole.
If you want a single transistor preamp (and the downside to that is that it inverts the signal) then you need to consider whether it can drive the intended load... and to make sure it can means you have to decide on the operating point to achieve that. Its no good saying you get optimal noise performance for example at xyz bias when that level of bias won't even drive a damp piece of string. Or conversely, if the optimum point is with it drawing a few hundred milliamps, then that too would be a compromise too far.
There are no rights and wrongs... the whole design has to be considered as a whole.
Hi and thanks again
Just to explain i would like to make a line preamp like the single mosfet Bride of Zen but with a power bjt
I do not know if it is possible
the load should be a tipical solid state power amp ... like a 20k load (maybe with some capacitance) with 2V of sensibility
Sources will be only digital or a phono stage separated ...
If this is not possible i will stick to the BOZ that i quite like ... it has a nice dimension in sound, it is not flat at all
I have already some boards for that
But BJTs are more brutal ... i have in mind a 20W to220 from Toshiba or similar
That i can use at 60V let's say
But i have no model of those
Here is where i am lost ... if the fft is good does this not imply that the bjt will work ?
The bjts i have in mind are commonly used as drivers for output stages in high level power amps
So they should drive an input stage of a power amp without any problem as well
a much easier task i guess
Actually i have an amp insanely low at 5K .... but i would be willing to put it in the garbage bin.
For me the preamp makes the sound ... just a honest and robust power amp will do the rest.
As i said already when there is the review of a cheap couple, let's say Rotel, it is almost always the preamp that fails to convince
Usually the sound is flat and hard
The evidence is that good solid state preamp are rare and usually hugely expensive.
But they make the sound ... they give the 3d effect for one.
Thanks and regards, gino
Last edited:
Just looking at the fft isn't always going to guarantee the best (sonic) performance because there will always be something "different" that you perceive as "better".
I think you are going to have to trust your ears and try building a few different designs tbh. All the single ended designs (as opposed to long tailed or differential input stages) tend to have that "musical" and non fatiguing sound to them in my experience. And it seems like that is what you like too.
Driving 5k to a few volts is no big deal, its only a milliamp or so at the kind of levels a power amp input would need.
I think you are going to have to trust your ears and try building a few different designs tbh. All the single ended designs (as opposed to long tailed or differential input stages) tend to have that "musical" and non fatiguing sound to them in my experience. And it seems like that is what you like too.
Driving 5k to a few volts is no big deal, its only a milliamp or so at the kind of levels a power amp input would need.
Thanks again and i think you are perfectly right
I like for instance the BOZ that is single ended and i think has indeed some 2nd order distortion.
I have seen that there are some models of drivers in the Cordell's file
They are to126 type .. i will start with one of them i think
Thanks a lot again. Kindest regards, gino
Hello ! i got distracted for a moment
You know what ... now that i have discovered that for my dream bjts there is no model around i am a little
And how can be that for the TOTL devices there is no model
Maybe i will discover that there are no models for the sublime 200W Sanken multi-emitters as well ??? Unbelievable ...
I still think that a spice model should be made available by the manufacturer together with the datasheet ... it is just a .txt file if i understand correctly
When you read Bobs book you will see that you can create your own models
When you read Bobs book you will see that you can create your own models
Hi again Mr Mooly
Well this i could not imagine really ... now i understand because they do not provide them
I was thinking that the very demanding task is to fill in a datasheet with all the curves coming from actual measurements
and that the parameters inside a model come quite easily from the datasheets
I just stop here. I need to see this book now for sure that should arrive soon
I am sure i will be able to keep also the 2nd very low ... like -100dB at least !
with just traces of 3rd ...
Thanks again and kind regards, gino
Last edited:
Hi and i have a basic question.
If a circuit simulates does this guarantee that the actual circuit will work ?
This is very important for me because sometimes instead of the sine Vout i get a window with voltages and currents and i suppose the circuit is wrong
Like the one attached for instance ... will it work ?
Thanks a lot. Regards, gino
If a circuit simulates does this guarantee that the actual circuit will work ?
This is very important for me because sometimes instead of the sine Vout i get a window with voltages and currents and i suppose the circuit is wrong
Like the one attached for instance ... will it work ?
Thanks a lot. Regards, gino
No, definiately not. Component tolerances or a model that is not describing the component accurately enough can simulate well but might be disasters in the real world.
So the answer is actually "it depends.."
But something that wont work in a simulation as intended I wont bother transfer to the real world.
So the answer is actually "it depends.."
But something that wont work in a simulation as intended I wont bother transfer to the real world.
You have a fundamental error in your circuit. Can you see what it is ?
(you also need to look a your simulation settings for distortion. 10kHz signal, fourier analysis set for 20)
Hi and thanks a lot for the valuable reply.
It is a condition necessary but not sufficient ... i see
Thanks again, gino
Hi, of course not
yes .. i have run the simulation and the 10kHz FFT ... it looks fine
In general is there a way to run a sure check on a circuit in general to understand if it works ?
Thanks a lot, gino
Hi and sorry and please let me elaborate a little
If i understand well the only critical components are the active ones
A passive components must only be sized correctly but it will work anyway
A sim program should be able to give a warning, like i got many times, when the working conditions calculated for an active components are out of the acceptable range
Like a check function of the circuit
Is there a specific function for this in LTSpice ?
or a command able to perform this check ?
That would be very nice indeed
Thanks again, gino
If i understand well the only critical components are the active ones
A passive components must only be sized correctly but it will work anyway
A sim program should be able to give a warning, like i got many times, when the working conditions calculated for an active components are out of the acceptable range
Like a check function of the circuit
Is there a specific function for this in LTSpice ?
or a command able to perform this check ?
That would be very nice indeed
Thanks again, gino
Last edited:
Hi, of course not
Well it might seem to work in simulation but it shouldn't
Your PNP input transistor is back to front. Turn it around and tweak the bias to get symmetrical clipping.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.