I'm just going through the ThermalTrak tread and seems interesting (I'm on page 17right now).
wondering now if there would be a benefit of using a pair of NJL4281/4302 as drivers which of course would remove the whole bias circuit.
what do you guys think ???
this is just and IDEA for now...
would be nice to have some thoughts on this "concept".
wondering now if there would be a benefit of using a pair of NJL4281/4302 as drivers which of course would remove the whole bias circuit.
what do you guys think ???
this is just and IDEA for now...
would be nice to have some thoughts on this "concept".
i think i will build my output stage using the conventional output biasing and use NJL4281/4302 as drivers and output devices.
this way everything is ready to go.
besides theres only a $0.30 difference price wise at my local supplier and its well worth the investment when this whole thing gest sorted out. Besides lookin at the specs the MJL4283 and NJL4281 are pretty much the same besides the diode in the NJL package.
Will post pics of the driver and output stage when i have it completed.
this way everything is ready to go.
besides theres only a $0.30 difference price wise at my local supplier and its well worth the investment when this whole thing gest sorted out. Besides lookin at the specs the MJL4283 and NJL4281 are pretty much the same besides the diode in the NJL package.
Will post pics of the driver and output stage when i have it completed.
fotios said:To be accurate, the only that i consider is missing from this circuit, it is a current mirror (not for the usual B.S. writen from members but for eliminating the DC offset in output, unless you must match the LTP transistors. A trim pot can balance the collector currents of LTP as well. Please MJL give instructions to Ardculda, this is your duty.
The second that is missing is a VI limitter for protecting output transistors from reactive loading. Also this, is very common and easy to implement.
These all are tasks regarding MJL and not me.
Don't worry Ardculda, MJL is a brave man ready to offer his service in everyone that needs it. It is a knight ready to kill the dragon with his amplifier spear.
Hi fotios,
A current mirror, as well as other distortion reducing tactics are wasted here. This is a high power amp and is not meant for extreme fidelity. Stability and reliability are what's important for an amp this size.
DC offset is not a problem (remember, I built this amp on this forum, tested it and it is being used each day). The input pair were not matched and I have less than 10mV offset. No pot required.
I presented the raw amp, without protection circuitry as I don't use anything other than fuses for overload, a crowbar on the speaker itself and a DC detection circuit (R. Slone's) on a separate board.
Adrculda said:
Also what voltage did you design this circuit in mind with ??
i might just want to add one more output transistor on each side as i like round numbers.
And whats the guestimate output on this toy
Hi Adrculda,
The amp is being used at +/-70V rails and load is 4 ohms. I measured ~45Vrms during testing into a 4 ohm dummy and this translates to ~500 watts. It was tested with a 1.6 ohm dummy load, but my transformer lacks the current (only 850VA) and the voltage dropped to ~35Vrms - roughly equal to 760 watts. It did run stable and maintained it's idle current.
There are 8 outputs, the 9th device (the MJL4281A/4302A) are drivers. The MJE15030/31 are predrivers.
If I wanted more power (I will
), I'd bridge 2 of these. The design goal was 1000 watts into 2 ohms, and it has been shown it will do this (with a big enough power supply). The result would be 2000 watts into 4 ohms.
Adrculda said:
The thermal tracks don't offer a significant advantage, especially for the use you have in mind. They have a fairly large disadvantage IMO, the pins are too close together. Not the best situation for a high current amp.
Using them for drivers would be ok, but you still need more support circuitry. 2 diode drops will not give you enough bias - it would be running class B.
MJL21193 said:
Hi Adrculda,
The amp is being used at +/-70V rails and load is 4 ohms. I measured ~45Vrms during testing into a 4 ohm dummy and this translates to ~500 watts. It was tested with a 1.6 ohm dummy load, but my transformer lacks the current (only 850VA) and the voltage dropped to ~35Vrms - roughly equal to 760 watts. It did run stable and maintained it's idle current.
There are 8 outputs, the 9th device (the MJL4281A/4302A) are drivers. The MJE15030/31 are predrivers.
If I wanted more power (I will
Ahh i c...
Thanks for your reply John...
I was thinking of "slightly" modifying it by using 10 output transistors and 4 drivers ( 1 driver for each 5 transistors )
Also this would be done on 2 separate board, so 4 boards in total would be used as a driver/output boards by doing this i can maximise space in the case for the monster toroids ( 2x 1KVA )
also been thinking about what advantage ( if any ) a regulated power supply would have in this application.
As of right now im trying to get the power supply board done ( 10 x 10000uF 80VDC caps ) on EACH power supply as each channel would have its one, but will be running common rail at 75VDC
Keep up the good work!!! POST PICS when you are done!!!!
Don't let the arrogance of older know-it-alls here hold you down. Build as big and crazy as you want. In fact, when you think it's good enough, go even bigger just to prove to em you can!
I built a 350W planned-RMS amplifier, and it ended up being over 400W RMS and a true speaker coil melter!!! Blown two 12inch subwoofers and also blown other low power car speakers just for fun, and ZERO issues from my output stage ever. An intentional dead short to the output stage only pops the fuse. THATS why you overbuild and use OVERSIZED PARTS!
I used 5 pairs of MJL4281/4302 with forced fan cooling for ~350W and I NEVER hear the quiet fans, while other professional amps have used only 4 pairs of lesser devices for 1000W and run hotter! No SOA problems here!
On that scale, you could use 20 paralled output devices for your 1500W+
I got good help from folks here, but also others saying I don't need to use big TO-220 devices for VAS, or such high currents or biasing, and I don't need such high rails, and other nonsense. Others here try to shoot down your brute-force approach, but I say GO FOR IT!!! I can say that the oversized parts NEVER have thermal issues, and they take up tremendous circuit board space.
It's good to think outside the box, instead of just conforming to what others say is right...........you always win in the end every time!
Don't let the arrogance of older know-it-alls here hold you down. Build as big and crazy as you want. In fact, when you think it's good enough, go even bigger just to prove to em you can!
I built a 350W planned-RMS amplifier, and it ended up being over 400W RMS and a true speaker coil melter!!!
Blown two 12inch subwoofers and also blown other low power car speakers just for fun, and ZERO issues from my output stage ever. An intentional dead short to the output stage only pops the fuse. THATS why you overbuild and use OVERSIZED PARTS! I used 5 pairs of MJL4281/4302 with forced fan cooling for ~350W and I NEVER hear the quiet fans, while other professional amps have used only 4 pairs of lesser devices for 1000W and run hotter! No SOA problems here!
On that scale, you could use 20 paralled output devices for your 1500W+
I got good help from folks here, but also others saying I don't need to use big TO-220 devices for VAS, or such high currents or biasing, and I don't need such high rails, and other nonsense. Others here try to shoot down your brute-force approach, but I say GO FOR IT!!! I can say that the oversized parts NEVER have thermal issues, and they take up tremendous circuit board space.
It's good to think outside the box, instead of just conforming to what others say is right...........you always win in the end every time!
EWorkshop1708 said:
The reason why i though to use the NJL4281/4302 is because your drivers always end up sharing the same heat sink as you output transistors and if i could stick the drivers into the middle of the board would be even better as the heat would be transfered from both sides and have a better thermal tracking as compared to having them mounted on the end of the board.
I just got around to be poking around ORCAD an in a sense im a bit dissapointed as the libraries are a bit outdated and missing quite a few components.
As for my earlier post about the power supply i think i'm just going to go unregulated and grossly overbuilt in case i get "bored" and want a higher voltage i don't want to have to pull apart everything because i skimped on the original build.
Currently trying to source 10kuF or 15KuF 250VDC caps
Looking for a total capacitance of 50-60KuF for each channel coupled a 35+A fast recovery rectifiers rated to over 250VDC.
I know I will never end up using it, but why not ??
Some of the price differences are very marginal at times compared to what you are already paying to start off with !!
Hell i seen ppl in car audio even use 400VDC caps in the power supply with just downright crazy ratings !
And lets be honest if yur transistors can take it whats the next part in line that will take the beating ?? you got it !! the power supply .
Also everything you do affect how the amp will function and has its pros and cons. Having a high reserve means its a more linear strain on the toroid, but not as demanding as having too little. Also on the AC side you wont see the current surge as bad as it is a linear draw due to a high capacitance in the supply line to the amplifying circuit.
I have about 6 or 7 ideas on paper right now, but i WILL used the pre-driver section from John's ( Thank you again, your a gentleman and a scholar, where most could do is discourage as where you just pointed me in the right direction ) post which will be on separate board.
I personally find its easier to work with multiple modules rather than with one large entity. this is why i'm keeping the pre-drivers and drivers separated also with being easier to build and upgrade as the parts stay the same only difference is now you can cascade output modules as much as you like as long as your power supply can deliver the current and that means lower impedance can be driven with less heat as the draw is now not divided between 8 output transistors but 16 ( 2 output boards running in parallel )!
As long as your power supply can deliver this setup could be u huge step forward for those that want big power and are on budget as you can step the power up more and more by cascading output boards the only draw back will be space, but i will try and "squish" things together as much as i can with out over heating components.
EWorkshop1708 said:
Unfortunately, 1500W/4R requires about a +/-115V rail - under load. That's just ohms law. Unless you're starting with a 6kVA toroid which is what it would take not to flywheel under that kind of load, it will need to be on the 130-140 range unloaded. High rails are required if you want the power. 1500W/1R is a whole lot easier and only takes a conventional topology with a lot of output transistors. And is less likely to burst into oscillation and/or flames.
The only reason we would be shooting down a brute force approach is if someone was in over his head and was more likely to set his house on fire than set the roof on fire.
Whoa, whoa, whoa ... back up - DO NOT USE fast recovery rectifiers in the mains supply!!!!!! Those are for audiophools (audiophiles) who think the speed of the power supply is important for audio quality It's not. Use standard bridge rectifiers. 35 or 50 amp ones are fine. They take the turn-on surge way better than fast diodes do, especially at the capacitances you're considering, and are plenty fast enough for the application.
About using parallelable modules - be careful here. All your outputs in any given parallel bank need to be in the same thermal environment. And if you split the NPN/PNP bank to separate sinks, they EACH need individual thermal tracking. Which means a diode or two on each. It may take hours, but the temps will drift apart.
About using parallelable modules - be careful here. All your outputs in any given parallel bank need to be in the same thermal environment. And if you split the NPN/PNP bank to separate sinks, they EACH need individual thermal tracking. Which means a diode or two on each. It may take hours, but the temps will drift apart.
Also for a while i been thinkering with the Idea of a variable output power supply using High Power Solid State Relay and few buffering caps on the input of the transformer.
This way the output devices are working at a constant rate where only rail voltage is changed to increase output when needed instead of reducing input signal and therefore raising the noise floor
And thanks to the large capacitor banks i would be using.
What do you guys thins about this idea for limiting power ???
This way the output devices are working at a constant rate where only rail voltage is changed to increase output when needed instead of reducing input signal and therefore raising the noise floor
And thanks to the large capacitor banks i would be using.
What do you guys thins about this idea for limiting power ???
Adrculda said:Also for a while i been thinkering with the Idea of a variable output power supply using High Power Solid State Relay and few buffering caps on the input of the transformer.
This way the output devices are working at a constant rate where only rail voltage is changed to increase output when needed instead of reducing input signal and therefore raising the noise floor
And thanks to the large capacitor banks i would be using.
What do you guys thins about this idea for limiting power ???
If you're looking for a way to limit the rail voltage to only what's needed at any given instant, that's the idea behind the G and H classes. The higher power Crest amps are class G, where the voltage to the output bank varies between half and full voltage like a linear regulator that stays so many volts above the output. starting at the half way point. Some of the Yamaha and Carver amps take it a step further and vary the rail voltage continuously. The bigger QSC amps (and early Carver) are class H, where the voltage is switched in discrete steps. The QSC uses a Transnova output section, which is completely different than the circuits you have been considering, but the rail switches can be borrowed and work just fine with emitter followers like MJL's, Fotois', or any other. These circuits are worthy of study, and you might want to try it out on a smaller (100 to 200 watt) amp *first*.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.