john curl said:
Mr. Curl I found some 22R0 resistors and used them for now. I can tell you that it made a significant difference the the sound though. Smoother, more detailed and less fatiguing, top to bottom! 10R0 are on order!!!
Thanks, for the info!
It makes me wonder why more designers don't run more current through the drivers? Maybe it doen't make a significant measured difference?
I am also starting to think that it is a lot harder to make a good class A/B amp than a class A amp.
Thanks, Don
There are any number of reasons that people don't push things harder:
--It produces more heat, which in turn can require larger, more expensive parts.
--A lot of people calculate the slew rate at something like 20kHz and use the bare minimum current required. However, if you believe (as I do, and as [I believe] John does) that wide bandwidth is important, then it makes no sense to starve the circuit at a mere 20kHz. My bench prototype of an all-FET JC-3-ish circuit rolls off at 220kHz at the moment (only one pair of output devices--no that's not practical in the real world, I'll be adding another pair, perhaps two more) with no feedback. Small signal distortion, incidentally, is around .05%...again, with no feedback.
Gotta go, one of the twins is in trouble.
Grey
--It produces more heat, which in turn can require larger, more expensive parts.
--A lot of people calculate the slew rate at something like 20kHz and use the bare minimum current required. However, if you believe (as I do, and as [I believe] John does) that wide bandwidth is important, then it makes no sense to starve the circuit at a mere 20kHz. My bench prototype of an all-FET JC-3-ish circuit rolls off at 220kHz at the moment (only one pair of output devices--no that's not practical in the real world, I'll be adding another pair, perhaps two more) with no feedback. Small signal distortion, incidentally, is around .05%...again, with no feedback.
Gotta go, one of the twins is in trouble.
Grey
I currently have a front end with a single output pair on the bench. So far things look pretty good. I'll be adding another output pair as soon as I can get time. Performance will deteriorate a bit. I'll see if I can find things to fine tune. Two output pairs are practical for a 25W amp, particularly if you don't intend to drive 1 Ohm loads; it's a matter of quiescent bias vs. load vs. heat dissipation. If you want monster bias it might be wiser to go with three pairs in the output.
I have another, higher drive current front end I want to try which substitutes MOSFETs for JFETs at one point in the circuit. Whether more drive current for the outputs outweighs the presence of the MOSFET remains to be seen. It's all a question of tradeoffs.
At the rate things are going, Mike's bipolar output version will be ready long before my MOSFET version. It looks as though the parts he has chosen are widely available. He's also sticking to the original topology--my circuit departs somewhat, particularly in terms of feedback and the choice of FETs all the way though.
Grey
I have another, higher drive current front end I want to try which substitutes MOSFETs for JFETs at one point in the circuit. Whether more drive current for the outputs outweighs the presence of the MOSFET remains to be seen. It's all a question of tradeoffs.
At the rate things are going, Mike's bipolar output version will be ready long before my MOSFET version. It looks as though the parts he has chosen are widely available. He's also sticking to the original topology--my circuit departs somewhat, particularly in terms of feedback and the choice of FETs all the way though.
Grey
There are two good reasons for paralleling output devices in a low voltage amp. The ideal bias point is usually the PEAK BETA of the output device. If it is 1A, for example, then 1A is the best bias point for that device, but that is only 16W into 8 ohms. If you want more idle current, then paralleling is probably better, than just operating the device above its peak beta.
Second, the extra output devices s spread the heat more uniformly across the heatsink.
Second, the extra output devices s spread the heat more uniformly across the heatsink.
I'm looking at 1 to 1.5A bias per MOSFET. I don't recall that the JC-3 schematic specified rail voltages, but I'm looking at something on the order of 27 to 28V rails. I'm using TO-247 outputs and looking to keep Pd below 30W per device for reliability's sake using passive heatsinks.
Given the rail voltages and the fact that I'm using vertical MOSFETs in the output, I'll lose a bit to Vgs, but I'm still calling it a 25W amp in round figures.
I'm looking at genuine, government inspected, accept no substitutes, pure class A, so it comes down to a question of what my load impedance is. Do I design for the Bohlender-Graebner RD-75s that I'm using for midrange drivers or for the Magnepan ribbon tweeters I'm using for tweeters? The RD-75s are a little over 5 Ohms--a cinch. My daughter could build that one. The ribbons are something like 2.3 Ohms. Okay, that's gonna require more bias.
Honestly, given that the RD-75s are in the upper 80s at 1W, I'll probably put the single-ended amp on the tweeters, which means more bias is required. Then the bridged version--circa 100W/8 Ohms--goes on the RD-75s, where it will deliver something more along the lines of 160W...fairly reasonable for my listening habits.
If I decide to hook this into my water cooled system, it's Katie bar the door...I routinely run TO-220 output devices at thirty-odd watts each. TO-247s? Ha! Just watch me!
Grey
Given the rail voltages and the fact that I'm using vertical MOSFETs in the output, I'll lose a bit to Vgs, but I'm still calling it a 25W amp in round figures.
I'm looking at genuine, government inspected, accept no substitutes, pure class A, so it comes down to a question of what my load impedance is. Do I design for the Bohlender-Graebner RD-75s that I'm using for midrange drivers or for the Magnepan ribbon tweeters I'm using for tweeters? The RD-75s are a little over 5 Ohms--a cinch. My daughter could build that one. The ribbons are something like 2.3 Ohms. Okay, that's gonna require more bias.
Honestly, given that the RD-75s are in the upper 80s at 1W, I'll probably put the single-ended amp on the tweeters, which means more bias is required. Then the bridged version--circa 100W/8 Ohms--goes on the RD-75s, where it will deliver something more along the lines of 160W...fairly reasonable for my listening habits.
If I decide to hook this into my water cooled system, it's Katie bar the door...I routinely run TO-220 output devices at thirty-odd watts each. TO-247s? Ha! Just watch me!
Grey
Hello Mr Curl,
You Have said that you like to have all FET amp but you worry when some one shorted the output the output device will blown away.
let say I take the risk what is your sugestion for output FET ?
What is more important in your opinion in front end the topology or the device chosing ?
Thanx
You Have said that you like to have all FET amp but you worry when some one shorted the output the output device will blown away.
let say I take the risk what is your sugestion for output FET ?
What is more important in your opinion in front end the topology or the device chosing ?
Thanx
Someone on this thread was going to put a fet output version together. I did it more than 20 years ago, and it works great! You don't have to worry about breakdown, that was only for power supply voltages over +/- 35V that fets tend to go down in a short circuit, faster than the circuit breaker or fuse.