no.ozonek said:
Sustained beta devices is applied to the high current ability, in particular the droop, or lack of it, in hFE at currents approaching 50% of max IC.
Hi ozonek and megajocke !
I believe that sustained beta means that beta does not drop at HIGH currents.
Thorsten L.
I believe that sustained beta means that beta does not drop at HIGH currents.
Thorsten L.
Which means it does not load up the drivers and increase beta related distortions (even though in itself it does not suffer these distortions).
The real issue with beta droop happens in the preceeding stages.
The real issue with beta droop happens in the preceeding stages.
I have designed the NJL parts into a Class A amp, and find they are excellent as the temperature sensing diode allows very accurate and stable biasing of the output devices.
We are looking at using them for some AB designs in the future due to the sensor feature.
Many AB amp designs monitor the heatsink temperature, and the outcome is that the amp can end up overbiased when playing hard or it has to be set up underbiased to avoid thermal runaway as the heatsink and chip temperatures don't track properly over time.
Sensing the tab temperature eliminates the greater part of this error and delay by at least an order of magnitude.
Sanken made parts with the same idea but tied the diode to the base of the darlington output, so they can't be connected into a VBE multiplier as easily.
Stable gain in the output and driver stages makes a difference to the SQ particularly in dynamics. Having used Sanken parts in an upgrade over older TIP parts (that had a 'drooping' gain) the feedback from several customers was of cleaner output at high levels and better dynamics.
Seems to make sense that you will get better linearity when the voltage amp isn't working into a non linear load...
The SOA of mosfets looks good on paper, however we have had problems getting them to share current, particularly at high Vds.
They are ideal for switching amps but the other issues such as varying transconductance, parasitic oscillation at high Vds, a lack of complimentary devices, and Vgs spread make them difficult to use in mass produced product.
That's probably why there are so many manufacturers sticking with Bipolar output transistors, due to the predictable and repeatable specifications.
We are looking at using them for some AB designs in the future due to the sensor feature.
Many AB amp designs monitor the heatsink temperature, and the outcome is that the amp can end up overbiased when playing hard or it has to be set up underbiased to avoid thermal runaway as the heatsink and chip temperatures don't track properly over time.
Sensing the tab temperature eliminates the greater part of this error and delay by at least an order of magnitude.
Sanken made parts with the same idea but tied the diode to the base of the darlington output, so they can't be connected into a VBE multiplier as easily.
Stable gain in the output and driver stages makes a difference to the SQ particularly in dynamics. Having used Sanken parts in an upgrade over older TIP parts (that had a 'drooping' gain) the feedback from several customers was of cleaner output at high levels and better dynamics.
Seems to make sense that you will get better linearity when the voltage amp isn't working into a non linear load...
The SOA of mosfets looks good on paper, however we have had problems getting them to share current, particularly at high Vds.
They are ideal for switching amps but the other issues such as varying transconductance, parasitic oscillation at high Vds, a lack of complimentary devices, and Vgs spread make them difficult to use in mass produced product.
That's probably why there are so many manufacturers sticking with Bipolar output transistors, due to the predictable and repeatable specifications.
thorstenlarsen said:
Yes, it was maybe a poor choice of words from my side. I was referring to the drop in current gain and speed that happens above around half max Ic. So with low currents I meant less than 7 amps... Maybe not the best terminology but I would probably call a 200W per channel amp "low power"...
You have to watch out: hfe and speed drops a lot at high current. At max Ic speed for many sustained-beta types is *lower* than MJ(L)2119x types and current gain similar. If you put transistors in series in high power amps for lower voltage and thus better second breakdown immunity or do class G you will find you need quite a lot of base current with low impedance loads as you usually have less pairs (about half) in parallell than in a class B amp of the same power!
megajocke said:
Anyone who makes a large class G amp with only two stages of current gain deserves the mess he makes for himself.
When using 3 stages, those NJW0xxx devices look like the ideal choice to drive a bank of old-schoold devices like the 2119x. Then you can use a nice, fast device with low Cob in front of that and still drive a welding rod if you wanted to.
Yes... 😀
I'm going to use MJ2119x as outputs and FJL4x15 or possibly FJA4x13 as drivers in my amp. Predrivers will probably be MJE3x0 or 2SC4793 and its complementary.
I'm going to use MJ2119x as outputs and FJL4x15 or possibly FJA4x13 as drivers in my amp. Predrivers will probably be MJE3x0 or 2SC4793 and its complementary.
wg_ski said:
As old as it is.... the tought of an arc welding rod midrange driven by an array of old school To-3 still sounds sexy...
K-amps said:
K, you forgot your sedative again.
Is it me, or does that 3 resemble something when i turn my head just a teeny weeny bit.
It is the Financial ruin we are bringing to the world.... the guilt is just disorienting Jacco...
I mean it was just like yesterday when I ordered trays of the MJ21195/96... then no one wanted T-O3's... so I sold em... now they say it is sexy again (once I got hooked onto sustained beta devices...)...
I mean it was just like yesterday when I ordered trays of the MJ21195/96... then no one wanted T-O3's... so I sold em... now they say it is sexy again (once I got hooked onto sustained beta devices...)...
In my valve+BJT power amp project I will use 16 pairs of FJL4x15 output devices. The peak current will be about 2A with 4 ohms load.
Sajti
Sajti
K-amps said:
I mean it was just like yesterday when I ordered trays of the MJ21195/96... then no one wanted T-O3's... so I sold em... now they say it is sexy again (once I got hooked onto sustained beta devices...)...
Back when we only used four per amplifier, TO-3's were no hardship. With six, eight, twelve in parallel you have to re-think how much machining you're willing to do by hand. So the T0-264 wins. I only keep TO-3's around to repair or rebuild legacy units.
I'm not necessarily hooked on sustained beta. The 21194 is just fine by me. It's when manufacturers are concerned with saving fifty cents per amp by not buying predrivers that sustained beta outputs are necessary. Drivers (or pre-drivers) almost always benefit, but sustained-beta 1-2A TO-220's have been around for decades. The main reason I'll use a C5200 over 21194 is cost. They are dirt cheap these days. Not every project needs the highest power units available.
other things being constant... doesn't adding more stages (pre-drivers pre-pre drivers etc. increase the incidance of oscillations caused by the high gain feeback loop to make the amp's come closer to unity gain? if I sound gibberish... think compensation..
"Pre" pre drivers? Never needed 4 current gain stages before, but there's a first time for everything!😎
In practice, a Lin front end, even one with "enhancements" like cascoding and current mirrors, is pretty darn stable even with a triple darlington. Triples get their bad reputation from doing weird things like CFP drivers and having voltage and current gain in the OPS. But those oscillations are almost always LOCAL and won't respond to loop compensation techniques.
One trick I use to get a bit better phase margin is to bypass the OPS for the lead compensation cap. Tap the cap from the VAS or the first current gain stage. You don't have the extra HF phase shift of the outputs in the loop anymore. It makes a triple have just as much margin as a double. This even works with ultra-high-gain front ends. My favorite toploogy for pro sound amps is a triple OPS, a degenerated darlington VAS (so I can use an MPSA42's low Cob with a transitor with good SOA and high Cob), a simple phase inverter and an OP AMP for the front end. No need to match diff pairs or play games to get the currents balanced. The op-amp designer did that for me. It has a lot of loop gain, but I can always get them stable.
In practice, a Lin front end, even one with "enhancements" like cascoding and current mirrors, is pretty darn stable even with a triple darlington. Triples get their bad reputation from doing weird things like CFP drivers and having voltage and current gain in the OPS. But those oscillations are almost always LOCAL and won't respond to loop compensation techniques.
One trick I use to get a bit better phase margin is to bypass the OPS for the lead compensation cap. Tap the cap from the VAS or the first current gain stage. You don't have the extra HF phase shift of the outputs in the loop anymore. It makes a triple have just as much margin as a double. This even works with ultra-high-gain front ends. My favorite toploogy for pro sound amps is a triple OPS, a degenerated darlington VAS (so I can use an MPSA42's low Cob with a transitor with good SOA and high Cob), a simple phase inverter and an OP AMP for the front end. No need to match diff pairs or play games to get the currents balanced. The op-amp designer did that for me. It has a lot of loop gain, but I can always get them stable.
When topologies get so logically perfect... they sometimes sound like crap 😉
(not saying yours does.... but I have yet to come across the silver bullet) ....
(not saying yours does.... but I have yet to come across the silver bullet) ....
I've had exactly the same topology give drastically different results. Ranging from easy-to-construct with good but not fantastic results, to haphazard wiring which is marginally stable and sounds like ***, to put together with great care, with both channels on one PCB, true star ground to a single lug, heavy gage power wires at right angles to signal flow, with the end result rivaling $10000 amps. Once you get to a sufficient level, the exact circuit topology is far less important than how it's put together.
Hello Ozonek,
I tried to send you a PM but your email is not on.
Would you mind to set it on receive?
Best, Arjen.
I tried to send you a PM but your email is not on.
Would you mind to set it on receive?
Best, Arjen.