High fT at low current can mean it will switch faster (this would be good for crossover).Well, one can argue for and against which is more preferable, having high ft down to the lowest currents (which may be important for lower crossover distortion?), or at the other side of the spectrum
Lower fT at high current means more phase shift, you have to compensate the amplifier more to avoid oscillation at high current, so lower ULGF, use less feedback, which increases distortion in all conditions, so not good.
Low current fT doesn't seem to matter much for ULGF and stability. In fact you get the highest bandwidth when the output voltage and current are zero.
There is a practical issue too: paypal have ceased to operate in Russia, hence bank transfer fee, plus 30€ shipping for 4 transistors, making them as expensive as a tube of Sankens 😉
More important question is how to use such a faster transistors.
And it’s quite complicated and much more important - how exactly to make a broadband output stage for picking all the benefits from fastest transistors.
And it’s quite complicated and much more important - how exactly to make a broadband output stage for picking all the benefits from fastest transistors.
Or you might look into Silicon Carbide Mosfets, they have very low gate capacitance and low On resistance so should be very fast in switching. And they can handle a lot of heat and current.
I haven't looked at many datasheets but do they give any data for linear operation? That seems to be the issue with newer mosfets. They will switch large currents but can't handle much current in the linear region.
Hmmm. No. Definitely no.Nope. The amplifier must be stable at the worst case operating point of output devices. That's low Vce and high current.
Worst case is a reactive load with ohase at something like 60 degrees. So max current at a 2/3 of power rails. This is a hardest circumstances.
This is a very different thing.This worst case point determines what the open loop unity gain frequency and compensation will be. If you compensate an amplifier only according to the output devices operating point at 0V output and no current, then ULGF will be too high, and it may oscillate near clipping.
Talking about EF3 intrinsic stability of such a stage becomes main frequency bandwidth limiting factor.
Some useful info:
https://www.diyaudio.com/community/...ing-transistors-ok.352876/page-3#post-6181849
There is a formula for calculating current slew rate at http://www.cordellaudio.com/papers/another_view_of_tim_II.pdf
In this it is mentioned the capacitance needing to be discharged by the driver transistors to turn a power transistor off can be measured in microfarads. One could look at switching power devices and the spec t off as well as driver stage currents.
In this it is mentioned the capacitance needing to be discharged by the driver transistors to turn a power transistor off can be measured in microfarads. One could look at switching power devices and the spec t off as well as driver stage currents.
As usual Mr Cordell delivers the goodsThere is a formula for calculating current slew rate at http://www.cordellaudio.com/papers/another_view_of_tim_II.pdf
In this it is mentioned the capacitance needing to be discharged by the driver transistors to turn a power transistor off can be measured in microfarads. One could look at switching power devices and the spec t off as well as driver stage currents.
The nice thing is: Cbe due to charge storage is proportional to current, but gm is also proportional to current. So, with increasing collector current, the same AC base current will result in a smaller Vbe swing, which is compensated exactly by an increase in gm. So AC current gain stays quite the same over a wide range of collector currents. I mean it has a pole but it doesn't move too much. But Cbe also contains a fixed capacitance, and that reduces AC current gain at low current. And at high currents it runs out of steam, which also reduces gain.
Once switching spikes are eliminated, the remaining crossover distortion is due to the drop in gm at low output current, which means the drivers have to swing Vbe faster to follow the signal in the crossover zone. This occurs with all transistors (fast or slow) unless there is enough bias to plug the gm hole at zero current. And at low current the fixed part of Cbe is still there, so when Vbe has to swing harder, the drivers also have to deliver more current to Cbe to swing through the crossover quickly.
Who's stopping you from doing politics ?

Do Class AB bipolar transistors work as switches or as analog units?
I thought switching transistors were different from bipolar audio transistors.
Please clear my confusion.
I thought switching transistors were different from bipolar audio transistors.
Please clear my confusion.
ST said in their TDA series (2030/2050) data sheets that the unit should not be driven at a slew rate higher than 15V/microsecond, or instability will result in audio use.
IIRC one of the chips had a flat response to 250 kHz, and was used in servo mechanisms and ultrasonic equipment as well as audio amps.
So a fast transistor pair, like the higher resolution LCD, has a limitation in real life, you cannot hear or see the difference respectively beyond a certain limit.
IIRC one of the chips had a flat response to 250 kHz, and was used in servo mechanisms and ultrasonic equipment as well as audio amps.
So a fast transistor pair, like the higher resolution LCD, has a limitation in real life, you cannot hear or see the difference respectively beyond a certain limit.
Any switch will exhibit a non infinite slew rate...so the square wave slope still leaves some room for analogue. You need positive feedback to make a square wave out of a sine wave...thus any switching process has a slew rate and a defined speed.
Transistors intended for switching duty often have characteristics that make them SUCK for audio use.Do Class AB bipolar transistors work as switches or as analog units?
I thought switching transistors were different from bipolar audio transistors.
Please clear my confusion.
Good, but lack an answer why?Transistors intended for switching duty often have characteristics that make them SUCK for audio use.
Switching devices usually optimized for large currents and switching freq's in range 10-50 kHz. Keeping in mind 3-5 Amp of current capability of modern MOSFET drivers we more or less understand.
Audio specific usually suffer lack of bandwidth and driving current, really doesn't need such a low Rdson mighty crystals.
SOA is terrible for switching transistors. Even if they sounded beautiful you’d be constantly blowing them up. Unless you oversize the output stage enough to erase any gains obtained by the speed.
And many switching types have single digit hFE. And they are almost ALL lower than Sanken, Toshiba, OnSemi audio types even if they manage to get out of the single digits.
What about Cascoding them with more traditional (slower) high power transistors to stay at 10 VDS or so? I'm thinking switching MOSFETs rather than switching BJTs.SOA is terrible for switching transistors. Even if they sounded beautiful you’d be constantly blowing them up. Unless you oversize the output stage enough to erase any gains obtained by the speed.
What about bd249C/BD250C?SOA is terrible for switching transistors. Even if they sounded beautiful you’d be constantly blowing them up. Unless you oversize the output stage enough to erase any gains obtained by the speed.
Composite cascodes don’t lend themselves to building traditional complementary EFs. Application space is different.What about Cascoding them with more traditional (slower) high power transistors to stay at 10 VDS or so? I'm thinking switching MOSFETs rather than switching BJTs.
General purpose POS’s - ie, TIP types. Mediocre as a switch, mediocre as an amp. If your circuit is “good enough” with those, you don’t need to be concerned with output transistor fT.What about bd249C/BD250C?
- Home
- Amplifiers
- Solid State
- Which is the fastest NPN/PNP pair power transistor for output stage?