Hi Wendell,
Welcome aboard!
Sounds like you've really gotten involved in a lot of stuff!
As Edmond mentioned, Doug Self's power amplifier handbook is really good to have. I have also enjoyed the books by Ben Duncan and Randy Slone on amplifiers.
Cheers,
Bob
It sure would be great if you could get to RMAF in Denver this October. I cannot say enough kind words about RMAF. It is certainly my favorite.
I'll be at the Montreal show in March, so maybe you can get to that one if you can't make RMAF.
Cheers,
Bob
Hi Bob,
Well, that's a second option for sure. Let's see what the future holds.
Hi Jan,
I dearly wish to have a beer with you again. I'm sure there is plenty to talk about. Seeing you, Stuart and Bob together would very probably be an extraordinary event.
Your web site looks really well polished. It's an excellent site, and builds on it's record of excellent technical content. I thank you for that. I have been able to read some information that wasn't easily found elsewhere.
-Chris
Well, that's a second option for sure. Let's see what the future holds.
Hi Jan,
I dearly wish to have a beer with you again. I'm sure there is plenty to talk about. Seeing you, Stuart and Bob together would very probably be an extraordinary event.
Your web site looks really well polished. It's an excellent site, and builds on it's record of excellent technical content. I thank you for that. I have been able to read some information that wasn't easily found elsewhere.
-Chris
Hi Chris,
Yes, Jan's web site is famously good-looking. Now that the book is just about to bed (last proof check last night, goes to the printer in days) I have to really put in some effort on my site. I don't have a clue about HTML, but I'm hoping to get my son to help me
Cheers,
Bob
I think at least some kind of a get-together at the bar would be a great idea. Of course, Jan and I will have a table in the Atrium where we will be hawking our wares, so some people will probably be hanging out there as well.
The book is going to the printer within days, and I believe copies will be available for distribution mid-to-late September. I'm not sure how long it takes to get to Amazon and then shipped to you, however.
I'll look forward to seeing you and will be honored to sign your copy.
Cheers,
Bob
It's alive!
You're next Bob
jd
This inaugural issue of Linear Audio is going to be a classic. The lineup of authors you have assembled is truly impressive.
BTW, good picture!
Cheers,
Bob
Its really a shame these parts are about to become unavailable. Darn, I mentioned them several times in my book as examples for MOSFET output stages.
I'm curious; does anyone know what companies and amplifiers used these MOSFETs?
Cheers,
Bob
Some Sony ES-products used them a few years ago, don´t know about current models though. Also, I have seen them in at least one model from Swedish brand Bladelius. A real pity if they will be unavailable that soon. Makes me wonder if there are any even better ones on the way to replace them? Otherwise, we could always stick to the IRFP devices, however I really like the bigger Toshiba´s. Seem to be extremely rugged.
Agreed, I really like those big TO-264 packages. One thing that is really important, and sometimes overlooked, is the thermal resistance of the transistor insulator. It is a significant fraction of theta jc. The TO-264 has nearly double the metal contact acrea compared to the TO-247.
The Toshiba devices are better matched and seem better behaved, but I think that with care and careful choice of asymmetrical source resistors, the IRFP devices can be made to perform nearly as well. One may give up a tad bit of power supply headroom due to the higher Vgs in some designs, however.
Cheers,
Bob
Hi Bob,
You still need to cover the variable gate charge in the P channel devices. A low impedance driver circuit would do that (class A-B bipolar drivers anyone?).
Have you ever tried to bias up an output stage with no feedback and drive it, comparing that with a like BJT output stage? I have. The mosfets didn't show well against the BJT parts in that experiment. Just an observation Bob, I'd like to see if you agree or not. Who knows?
-Chris
You still need to cover the variable gate charge in the P channel devices. A low impedance driver circuit would do that (class A-B bipolar drivers anyone?).
Have you ever tried to bias up an output stage with no feedback and drive it, comparing that with a like BJT output stage? I have. The mosfets didn't show well against the BJT parts in that experiment. Just an observation Bob, I'd like to see if you agree or not. Who knows?
-Chris
Hi Chris,
Nonlinear gate charge and gate capacitance is definitely a challenge for both N-channel and P-channel MOSFETs, and the answer mainly lies in providing a good driver circuit capable of both sourcing and sinking current. This is not unlike the situation with BJT output transistors, although in many ways the numbers are not as bad for MOSFETs most of the time. This is one of the biggest reasons I always advocate avoiding MOSFET amplifier designs where the MOSFETs are run right off of the VAS. Things are not as bad with laterals as they are with verticals, at least in the Cgd area. I cover these issues in pretty good depth in the book.
For MOSFETs, it usually means employing class A drivers that are run hotter than one might think is needed for MOSFETs. It also means that care should be taken to avoid running the output too close to the rail, as that is where Cgd rises very rapidly. This may translate to leaving more power supply headroom on the table with MOSFET amplifiers. It may be a fair statement that most of the time a MOSFET power amplifier will be a bit less power-efficient than a BJT power amplifier. There always seems to be a heat vs. sound quality tradeoff in power amplifiers. Of course, class A is the ultimate example of this.
I regularly test amplifiers without feedback around the output stage in order to expose the open-loop distortion of the output stage for both BJT and MOSFET designs. I do this both in the lab and in simulation. I do not run the whole amplifier open-loop, but rather I always put a center tap on the driver bias resistor that runs emitter-to-emitter of the driver, and take the global feedback from there when I want to expose output stage distortion.
You are right. Given the same idle bias and number of output pairs, the MOSFETs always lose out to the BJTs simply as a result of transconductance droop that I described in my original MOSFET power amplifier paper. Transconductance droop is in many ways the opposite of gm doubling. I'm not sure which term was coined first.
That MOSFETs lose out to BJTs is at least true for 1 kHz static crossover distortion when the BJT output stage is properly biased and tested on the lab bench under static conditions and the spectrum of the harmonics is ignored. Take away those caveats and the picture is not so clear, especially at 20 kHz where BJTs suffer from dynamic crossover distortion.
The use of asymmetrical source resistors and multiple pairs of MOSFETs biased a bit hotter really drives down the distortion with a MOSFET output stage, and it stays down at high frequencies and under dynamic operating conditions. Even with only one pair of IRFP240/9240 MOSFETs biased at 150 mA I regularly achieve amplifier THD-20 of less than about 0.03% with a gain crossover frequency of 1 MHz without resort to error correction or fancy feedback compensation.
Cheers,
Bob