mje 340/350 = will work in the simulation. BUT , if you either use the real ones
or a better device (3503/1381) , an oscillator or the true performance you will have.
Most models will get you there (basically) , but to simulate the finer points ....
better , more verbose models.
For Andy T's comment , a standard model will work with an EF3 (with predriver stoppers).
When you use the Cordell 2sa/c1381/3503 - njw0281/0302 combo and add the "pre -stopper"
, oscillation will occur at clip.
Most of the time , oscillation WILL NOT occur if a plain ol' sine is simulated. That
is why mine work so well - I test for these "worst case" scenario's.
Below is the best group of models supplied by member Keentoken - our resident
simulator "Guru" !! 😀
PS - the below models will predict the real world outcome to the uV/uA + those
that have tested for THD20K have come within PPM of the simulator results.
OS
or a better device (3503/1381) , an oscillator or the true performance you will have.
Most models will get you there (basically) , but to simulate the finer points ....
better , more verbose models.
For Andy T's comment , a standard model will work with an EF3 (with predriver stoppers).
When you use the Cordell 2sa/c1381/3503 - njw0281/0302 combo and add the "pre -stopper"
, oscillation will occur at clip.
Most of the time , oscillation WILL NOT occur if a plain ol' sine is simulated. That
is why mine work so well - I test for these "worst case" scenario's.
Below is the best group of models supplied by member Keentoken - our resident
simulator "Guru" !! 😀
PS - the below models will predict the real world outcome to the uV/uA + those
that have tested for THD20K have come within PPM of the simulator results.
OS
So I guess I can move slowly to building this amp, but first maybe you guys have some good suggestions on output devices that I could use or input transistors, I know i have to do my homework but you know one person can't know all the devices maybe there is something new or good that I don't know about.
Also , maybe I should post a new thread baout this amp or should I do that after I build one channel , I'm not sure.
Also , maybe I should post a new thread baout this amp or should I do that after I build one channel , I'm not sure.
A few suggestions 😉
- In both models Q4, Q5 are plugged-in upside down. That's why current mirror don't work properly, resulting in high DC offset at the output.
- R14 - too high value (impossible to set quiescent current precisely), also this is a dangerous way of using it - if the central (moving) pin will loose contact, Q7 will most likely close, leading to very high current through the output pairs.
- Option with caps decoupling is a bad idea. Bases are "hanging in the air" DC-wise, ensuring pure class B operation for those transistors and still keeping them in danger of uncontrolled high currents in certain conditions.
- 2N555 are way not the best transistors for the input (high noise, now hfe). I would recommend bc5XX series, adding common-base cascode above them for keeping reasonable Vce (Vcb).
Otherwise, it's a working blameless amp with rather aggressive compensation, which is ok - you can make it a bit lighter later on, allowing better performance close to 20KHz 😉
Cheers,
Valery
Ok , I flipped over the input transistors , somehow when connecting them didnt see that their upside down.
about the decoupling output transistors, maybe i could add two resistors , one from the + or - supply rail and one towards ground and that would create a steady dc bias for the bases while maintaining decoupling from the previous stages? In case something there fails and makes the output fully open as can happen when directly coupled.
Also when driving it from +-70 volts multisim shows about 68volts on the input transistors I guess its near SOA for most bc transistors but arent there some with 80 volt CE and higher values?
about the decoupling output transistors, maybe i could add two resistors , one from the + or - supply rail and one towards ground and that would create a steady dc bias for the bases while maintaining decoupling from the previous stages? In case something there fails and makes the output fully open as can happen when directly coupled.
Also when driving it from +-70 volts multisim shows about 68volts on the input transistors I guess its near SOA for most bc transistors but arent there some with 80 volt CE and higher values?
You can play with this LTSpice file.
Simple blameless with triple emitter follower.
Simulation result:
Phase Margin = 84
Gain Margin = 14
Slew Rate = 73V/uS
THD at 210W/8Ohm, 1kHz -> 0.002251%
THD at 210W/8Ohm, 20kHz -> 0.011880%
PSRR 1kHz -> 97 dB
All transistor model use Bob Cordell's model.
Simple blameless with triple emitter follower.
Simulation result:
Phase Margin = 84
Gain Margin = 14
Slew Rate = 73V/uS
THD at 210W/8Ohm, 1kHz -> 0.002251%
THD at 210W/8Ohm, 20kHz -> 0.011880%
PSRR 1kHz -> 97 dB
All transistor model use Bob Cordell's model.
It might help to explain that LT Spice is a free download program from Linear Technology's website.
Linear Technology - Design Simulation and Device Models
LTwiki-Wiki for LTspice
As someone who only uses it when necessary, I think it is for advanced users with the time to explore and learn alongside electronic design. It's not intuitive or easily configured like Multisim even though the simulation capabilities are so much greater.
Linear Technology - Design Simulation and Device Models
LTwiki-Wiki for LTspice
As someone who only uses it when necessary, I think it is for advanced users with the time to explore and learn alongside electronic design. It's not intuitive or easily configured like Multisim even though the simulation capabilities are so much greater.
yup I downloaded it yesterday , have heard of it before actually , haven't checked out all the simulation possibillities though.
If you understand the theory of circuit design, it would be easy to master LTspice. The harder part would be the software-related part, like how to create a sub circuit etc etc. But for me, I can make sub circuits after struggling in 1 or 2 days, but to understand the theory it took me more than a year because of limited resources on the net at the time and I'm not an EE. Nowadays it is much easier as I have seen so many info on the net, tho still not complete or quite basic.
To run the asc file posted by Bimo you still need a model file. Search for a text file with a name like Cordell_Model.txt which consists of model definition for parts. Put it anywhere (or in LTspice folder) and in your schematic area you have to add something like ".include D:\MyModels.txt".
The problem would be, you have to match the name of the transistors in model file and in schematic. For example in Bimo's circuit the name of a transistor is "BD139_Cordell". This should match with the one in the model file (some may write it "BD139c_c"). That's why Bimo should have also distributed his model file when he shared the asc file.
Good luck, you will be spending more time with simulator instead of with solder 😀
- Status
- Not open for further replies.