anli said:
IMHO, more strictly speaking, we must say about charge deviation rather total charge itself. Probably, 2 * Pi must be added also
Yes Exactly, because of Linear application is involved not switching one....but still the Igate is much less than Ibase....provided both devices are operated at same frequency say 50Khz....
Here are the results obtained from Sim
Device under test 2SC5200 bipolar and IRFP260N hexfet
Sinewave Frequency = 50Khz , load 4Ohms
With 10A peak Icollector and Idrian, Ibase=1A whereas Igate=25mA....
Its clearly evident that for same output current the Ibase needed by bipolar is much more than Igate needed by mosfet to drive the load at same operating frequency.....
Device under test 2SC5200 bipolar and IRFP260N hexfet
Sinewave Frequency = 50Khz , load 4Ohms
With 10A peak Icollector and Idrian, Ibase=1A whereas Igate=25mA....
Its clearly evident that for same output current the Ibase needed by bipolar is much more than Igate needed by mosfet to drive the load at same operating frequency.....
G.Kleinschmidt said:
Hi Glen,
Please wait! I made a stupid error.
My last THD figueres refer to 8 Ohm, NOT 4 Ohm. It should read as:
C5/6 -> e-Q30 RL = 8R
Vo = 13.8V THD20 = 5.3ppm
Vo = 19.6V THD20 = 12.2ppm
and
C5/6 -> e-Q30 RL = 4R
Vo = 13.8V THD20 = 18ppm
Vo = 19.6V THD20 = 42ppm
I'm really sorry I've made such a terrible blunder.
By means of some tweaking (see schematic), this might further reduced to:
Vo = 13.8V THD20 = 16ppm
Vo = 19.6V THD20 = 36ppm
And by means of TMC even further to (don't tell it your "husband"):
Vo = 13.8V THD20 = 3.4ppm
Vo = 19.6V THD20 = 8.0ppm
In stead of TMC, you might consider NDFL, but, combined with a complementary VAS, it's probably difficult to implement.
Meanwhile, I tried the MJL21193/4 models from Motorala/Modpex.
Not surprisingly, the result was terrible. If you or someone else do have reliable models, I'll have an other try.
Regards, Edmond.
estuart said:
Hi Glen,
Please wait! I made a stupid error.
My last THD figueres refer to 8 Ohm, NOT 4 Ohm. It should read as:
C5/6 -> e-Q30 RL = 8R
Vo = 13.8V THD20 = 5.3ppm
Vo = 19.6V THD20 = 12.2ppm
and
C5/6 -> e-Q30 RL = 4R
Vo = 13.8V THD20 = 18ppm
Vo = 19.6V THD20 = 42ppm
I'm really sorry I've made such a terrible blunder.
By means of some tweaking (see schematic), this might further reduced to:
Vo = 13.8V THD20 = 16ppm
Vo = 19.6V THD20 = 36ppm
And by means of TMC even further to (don't tell it your "husband"):
Vo = 13.8V THD20 = 3.4ppm
Vo = 19.6V THD20 = 8.0ppm
In stead of TMC, you might consider NDFL, but, combined with a complementary VAS, it's probably difficult to implement.
Meanwhile, I tried the MJL21193/4 models from Motorala/Modpex.
Not surprisingly, the result was terrible. If you or someone else do have reliable models, I'll have an other try.
Regards, Edmond.
Oh, no problem at all Edmond
. I really appreciate the time you have put into this. A little boo-boo is hardly a bother! These figures are still looking pretty damn good for what is only supposed to be a simple amp, so I'm not even remotely dissapointed. I did consider using a larger LTP load resistor for biasing the VAS stages for greater LTP gain, but this makes the VAS biasing just a little bit too dependant on the current balance in the LTP's for my liking, so I stuck with 1k as a compromise. I think that your results confirm that this compromise isn't too bad, as the THD is only worsened by a few ppm. This was worthwhile simulating though. However, I really like the improvement gained from the simple TMC modification. Since I'm adding base stopper resistors for the output devices now, I'll do your TMC modification too. At this point I feel that all of your great help with the simulations has safely finalised the design. I think that ~10ppm THD20 is pretty much pushing the performance limit for this basic generic component / symmetrical LTP / miller-comp VAS / EF output amplifier topology, which is all I really wanted to get out of it.
It's a pity about the MJL21193/4 spice models though. I've read someone here complain about these being useless before, but I can't remember exactly who (andy_c ?). Also, unfortunately, I have no idea were accurate models for these parts can be sourced from.
Cheers,
Glen
estuart said:
Edmund,
What I was interested in was the “Sonic Nonsense” comment and another comment somewhere in the middle of this exposition that the only logical follow through for (it seems to me) would be to dump the speakers and the music and replace them with flat screen monitors displaying distortion and spectral noise distribution graphs.
The sonic nonsense is innaresting coming from one who prides himself on his understanding of the finer points of amplifier design. What’s the point of all the splitting parts per million in distortion simulations if it’s all a wash anyway?
In another comment, in another thread discussing cloning a solid silver Japanese amp (which, by the way, I do find to be polishing doorknobs instead opening the door to the next level) the highly technical thread leader mentions listening to his latest silver wound transformer and hearing the sound escape from the speakers for the first time. This got me wondering what the baseline really is.
The discussion panning the concept of resolution as an incomprehensible concept peaks my curiosity in this respect.
For the record, my leanings are far away from the heavily distorted ancient technology, but just as far away from the “we’ll fix it with EC” crowd. I don’t listen to reference recordings (they're boring) and, when the doctors let me and my medication is dialed in, I regularly enjoy the effect of a high resolution system, 30 years in the making of my own design.
Regards, Mike.
IrishboyM4 said:
That's incorrect. In the BJT case when the input signal varies with time, some charge will be stored in the capacitances of the collector-base and base-emitter junctions. Assuming an NPN device, turning off the device requires removing some of that stored charge via the base terminal. This implies current coming out of the base for a short time in the NPN case.
To illustrate this, I've done a sim comparing a complementary BJT output stage with MJL3281A and MJL1302A with a complementary FET output stage consisting of IRFP244 and FQA12P20. Supply voltages are +/- 50V and the input signal is adjusted in each case for an output signal of 40V peak at 20 kHz. Attached is a schematic.
Here is a plot of the base current of the NPN BJT and the gate current of the N-channel FET. Note that the base current of the BJT goes to about -3 mA at turnoff. By contrast, the FET only goes to about -1 mA at turnoff. So the reverse turnoff current of the FET is actually less than that of the bipolar.
The FET circuit has a bias current of 150 mA, while the BJT circuit has a bias current of 107 mA.
The FET circuit has a bias current of 150 mA, while the BJT circuit has a bias current of 107 mA.
Do not forget the fact, that the "input capacitance" depends on circuit connection of the MOSFET.
Yes, you can estimate the effective capacitance in case that you perform a set of measurements with AC voltages of different frequencies and amplitudes. This will fit for the circuit measured.
Yes, you can estimate the effective capacitance in case that you perform a set of measurements with AC voltages of different frequencies and amplitudes. This will fit for the circuit measured.
andy_c said:
Andy,
most of MOSFET simulation models are completely misleading when it comes to the input capacitance behaviour. Simply put, the input capacitance of a MOSFET increases rapidly, almost jumps up from 2 to 5 times of a given in the datasheet value when the Vgs goes through the threshold region. Your simulations are not reflecting this simple fact and in reality the gate currents would be very different from what you've just simulated. Have a look at the total gate charge curves for IRFP244 attached here. The flat part corresponds to the jump in the gate capacitance from about 1nF to about 4 nF and the model you use would not show this behaviour at all.
Cheers
Alex
Re: We all know that Sims arenot 100% accurate
Kanwar, did I say anything about the magnitude?! Try to argue about what is said and not about what you've imagined .
Alex
Workhorse said:Nikitin, but again the magnitude of these gate charging currents is still much lower when compared to BJTs'......
Kanwar, did I say anything about the magnitude?! Try to argue about what is said and not about what you've imagined
.Alex
Your Statement is self contradictory....
This statement clearly states that there might exist or is a DIFFERENCE in the gate currents figures obtained from the Simulation because Devices Models are not correct......when compared to Real worlds ..........
And Nikitin, the Difference could only be seen from the MAGNITUDE of the gate currents involved.....not Imaginations......
This statement clearly states that there might exist or is a DIFFERENCE in the gate currents figures obtained from the Simulation because Devices Models are not correct......when compared to Real worlds ..........
And Nikitin, the Difference could only be seen from the MAGNITUDE of the gate currents involved.....not Imaginations......
Re: Your Statement is self contradictory....
Kanwar, I've just said that the simulation is incorrect because the models are wrong in respect of the input capacitance behaviour. Not more, not less. You did reply that "the magnitude of these gate charging currents is still much lower when compared to BJTs". It had nothing to do with my post - I'm not arguing that point at all. I did reply not to you, but to Andy about his simulations. Why are you trying to argue with me?
Alex
Workhorse said:This statement clearly states that there might exist or is a DIFFERENCE in the gate currents figures obtained from the Simulation because Devices Models are not correct......when compared to Real worlds ..........
And Nikitin, the Difference could only be seen from the MAGNITUDE of the gate currents involved.....not Imaginations......
Kanwar, I've just said that the simulation is incorrect because the models are wrong in respect of the input capacitance behaviour. Not more, not less. You did reply that "the magnitude of these gate charging currents is still much lower when compared to BJTs". It had nothing to do with my post - I'm not arguing that point at all. I did reply not to you, but to Andy about his simulations. Why are you trying to argue with me?
Alex