roender said:
These are two different things:
In the linear small signal approximation, it can be shown that the output impedance of the emitter follower is equivalent to an inductance and resistance network for gm high enough.
I am talking about non linear distortion at crossover and there you have to analyze using the large signal ( exponential) model.
It can be shown that crossover distortion is minimized for a bias current giving a voltage drop of Vt on the equivalent resistance seen from the load.
This resistance is the sum of the resistance in the emitter including all the parasitic resistance plus the resistance in the base ( including parasitic) divided by beta. This concluse comes from a large signal analysis.
Because the presence of these ill defined resistance, I believe that the best is to adjust the bias to minimize crossover distortion.
JPV
janneman said:
Jan,
As you say the base stopper should define the source.
This is again a good reason to use a topology lile the triple T that gives a very low source resistance.
On another topic, are you living close to the Belgian border?
I wonder if it would be ok for me to come to your place with my tektronic distortion analyzer that I bought on ebay. I could get it calibrated with you audio precision as I understand that you have one.
JPV
JPV
roender said:
This is true, but it is not an issue for static crossover distortion. On the other hand, it might conceivably be an issue for dynamic crossover distortion in the same way that inductance in the external emitter ballast resistor can be degrading.
Let's plug in some simple numbers. Assume all the resistance is in the base in the form of a 5-ohm base stopper resistor. With a transistor beta of 50, this will look like 0.1 ohm at the emitter. If the power transistor is a RET with an ft of 50 MHz, then beta will be 50 up to 1 MHz, where it will then start to decline. This will be the point at which the effective inductance seen looking into the emitter begins to rise. It is the point where the inductive reactance equals the low-frequency emitter resistance seen (0.1 ohm). If I've done the math right, this equivalanet inductance is on the order of only about 17 nH - not much.
There is probably more than this in the wiring and in the emitter resistor, even if it is a non-inductive type.
Cheers,
Bob
janneman said:
Hi Jan,
The simple answer is that we can take it all the way back to the Vbe multiplier. We should assume that the center of the Vbe multiplier impedance is at ground for purposes of this analysis (this is a different issue than the output impedance of the VAS, by the way).
In a more practical sense, recognize that if the driver is biased at 50 mA for each output pair that it serves, then the resistance seen looking back into the driver emitter is about 0.5 ohm for each output pair served. This is likely smaller than Rbb of the output transistor, even if a base stopper is not used.
Notice that a pre-driver biased at 10 mA will have a local output emitter impedance of about 2.6 ohms, which when divided by the beta of the driver will be perhaps 0.05 ohms, which is comfortably less than the emitter output impedance of the driver.
Going back to the base of the pre-driver, we see that the base circuit impedance that it sees from the Vbe multiplier is sufficiently low that its aforementioned emitter impedance of 2.5 ohms will dominate.
Once again, from the viewpoint of impedances to minimize crossover distortion, it is not the output impedance of the VAS itself that counts.
In any case, it can be seen that this sort of thing converges pretty quickly when a Triple is used.
Cheers,
Bob
I'm looking to build a simple and clean sounding amp using 4 pairs of NJL3281/1302 complementary pairs and as driver/biasing stage a LME49830 chip to drive each channel. What do you guys think ??
Hi
My quick answer would be that the LME49830 is a MOSFET driver, and it is not meant to drive BJT ops.
How did you plan to use the TT diodes?
In my opinion there are a lot of discrete amps on this forum using TT that you should have a look at instead of using that chip.
Cheers
yes...
But MOSFET's aren't as linear as complimentary pairs and with the thermal tracking allowing a more stable operation and knowing that they will always be biased correctly regardless of temperature.
I will running them in series in the biasing circuit always allowing the proper bias to be set across to minimize distortion and keep the drivers at their peak performance without overshooting their SOA.
Hi,
You asked a very general question "what do you guys think". Just to get a better understanding of what it is we can help you with, are you comfortable with configuring the bias with the TT diodes? How many diodes are needed, additional resistive elements to get at the target bias voltage? Are you comfortable with calculating the current through the TT diodes and how to relate the TT diode tempco to the output device tempco and how to match those to get good tracking?
jd
That may be the case in some instances, such as low bias
designs, but I don't think it's true as a generalization.
That sounds like a smart idea as lines of transistors on heatsinks have inner and outer devices where inner ones might be warmer than outer ones.
Because the cooler transistors would have diodes that didnt heat up as much and hotter transistors would have diodes that were warmer and so adjusting the bias very accurately.
Ahhh I think I get you. You have a string of TT transistors on a heatsink, and each one adjusts the bias according to HIS temp. Is that what you mean?
jd
That would be simple if you only run a single matched pair, not multiple sets
In my case there will be a the driver IC -> driver MOSFET -> thermal trak output devices.
needless to say it will be a mirrored configuration with drivers between 2 thermal tracks while the IC and its power supply will be housed on a separate PCB with its own heatsink with regulator.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.