I did with an older LM49810 prototype which had bigger heatsinks and it was noticeable with some quiet passages (I'm thinking in Dvorak symphony nº 9 Adagio, i'm sure you will recognize the passage if you listen to it), but the character of the amplifier didn't change (to my ears) at louder volumes since mosfets overlap better than bjt's so they don't have gm-doubling distortion.
do IRF mosfets need thermal compensation
I know that IRFP mosfets have a positive tempco, but i've been looking at it's gate voltage vs drain current plots and noticed that the difference between current at 150 and 25 ºC shows an increase of some 400 mA worst case. This means that, when powered from 55V rails this would give an increase in dissipation of some 22W, but the heatsink will increase its dissipation from 0 to 30W (3W/C Heatsink to air), I'm not saying that i'm planning to run the transistors at 150ºC, but rather that this compensating effect would prevent them to reach that temperature.
Furthermore, since the transistors are showing a transconductance of more than 4 S everywhere, the increase in current will give a decrease in current of about 0,3 A with 0,25 ohms degenerating resistors, reducing the effect to an increase in dissipation of only 0,1A*55V = 5,5W.
If i'm right bipolars show a huge tendency to thermal runaway due to theyr very high transconductance. Is this correct or i'm going to burn my house?
I know that IRFP mosfets have a positive tempco, but i've been looking at it's gate voltage vs drain current plots and noticed that the difference between current at 150 and 25 ºC shows an increase of some 400 mA worst case. This means that, when powered from 55V rails this would give an increase in dissipation of some 22W, but the heatsink will increase its dissipation from 0 to 30W (3W/C Heatsink to air), I'm not saying that i'm planning to run the transistors at 150ºC, but rather that this compensating effect would prevent them to reach that temperature.
Furthermore, since the transistors are showing a transconductance of more than 4 S everywhere, the increase in current will give a decrease in current of about 0,3 A with 0,25 ohms degenerating resistors, reducing the effect to an increase in dissipation of only 0,1A*55V = 5,5W.
If i'm right bipolars show a huge tendency to thermal runaway due to theyr very high transconductance. Is this correct or i'm going to burn my house?
Re: do IRF mosfets need thermal compensation
Can you post something with reference marks to show where you are taking your numbers from?
I haven't seen your graphs, but I doubt you're reading them correctly.ionomolo said:
Can you post something with reference marks to show where you are taking your numbers from?
You are right about me not reading the datasheet properly. In fact I did it properly but the n-channel one has a much higher tempco and i did only look at the p-channel plots.
But the question is still there. Thermal runaway is caused by the transistor increasing it's power generation faster than the heatsink can dissipate and the degenerating resistor can block, not simply because it's increasing it's current.
The red line shows the effect of heating the device 125 º. Assuming a 2 ºC/W thermal resistance with natural convection at the heatsink, which should be reasonable for a 100W amp then increase in dissipation is about 63W, but increase in heat generation is increased by 55W. This does not mean that 150º is the point where i plan to operate it, but that increase in dissipation + degenerating resistors would counteract thermal runaway.
But the question is still there. Thermal runaway is caused by the transistor increasing it's power generation faster than the heatsink can dissipate and the degenerating resistor can block, not simply because it's increasing it's current.
The red line shows the effect of heating the device 125 º. Assuming a 2 ºC/W thermal resistance with natural convection at the heatsink, which should be reasonable for a 100W amp then increase in dissipation is about 63W, but increase in heat generation is increased by 55W. This does not mean that 150º is the point where i plan to operate it, but that increase in dissipation + degenerating resistors would counteract thermal runaway.
Attachments
Hi,
a ten fold increase in bias current and now you're left with no reserve in output stage dissipation capacity. i.e. the amp has no capacity left to output any signal.
It might survive the high bias, but it can no longer operate as an amplifier.
BTW,
the heatsink will probably be running at 20 to 30 Cdegrees below the Tc and this could be another 10 Cdegrees below Tj.
So, for Tc=150degC, the heatsink could be at 110degC to 120degC.
What is ambient now? and is deltaTsink-amb?
a ten fold increase in bias current and now you're left with no reserve in output stage dissipation capacity. i.e. the amp has no capacity left to output any signal.
It might survive the high bias, but it can no longer operate as an amplifier.
BTW,
the heatsink will probably be running at 20 to 30 Cdegrees below the Tc and this could be another 10 Cdegrees below Tj.
So, for Tc=150degC, the heatsink could be at 110degC to 120degC.
What is ambient now? and is deltaTsink-amb?
You are right about that. I will add that Vbe multiplier in the circuit, but i've found that resistor biased amps sound better than vbe-multiplier-biased ones. This might as well be due to the lenghts of wire requierd to place the Vbe multiplier in the same heatsink than the output stage.
Use different output devices if you don't want to use a Vbe multiplier. Renensas 1058/162 or Magnatec devices do not need thermal compensation. Check the AN-1645 app note from National for info on these devices and biasing.
-SL
-SL
The problem of the hitachi/renesas ones is that there are a lot of fakes arround. That is not a problem with the IRF ones.
Now i've taken the compensation scheme from the LME49830 demo board. 😀.
The last question for my LME49810 "final" amp is wheter to use a DC servo or an input coupling cap. The dc servo is recommended in the LM4702 AN-1490, saying that "the input cap had an undeniable effect" The problem i see from the DC servo is that output impedance from an opamp may be even less linear than the impedance of a capacitor, so it may sound even worse, of if the impedance raises too much at high frequencies it may set the noise gain over the stability margin of the LME49810.
Anybody has done both and can compare stability and sound?
Now i've taken the compensation scheme from the LME49830 demo board. 😀.
The last question for my LME49810 "final" amp is wheter to use a DC servo or an input coupling cap. The dc servo is recommended in the LM4702 AN-1490, saying that "the input cap had an undeniable effect" The problem i see from the DC servo is that output impedance from an opamp may be even less linear than the impedance of a capacitor, so it may sound even worse, of if the impedance raises too much at high frequencies it may set the noise gain over the stability margin of the LME49810.
Anybody has done both and can compare stability and sound?
baker clamps
hi there....I was wondering if you chip-guys (and maybe girls) have found the baker clamp audibly useful. A good test would be to make two identical amps one with a 49810 (or03) and the other with a 49811 and compare the two with the same program material at the same level scrunched up against the rails...Is ther a difference? I am curious. till the next post....so long
hi there....I was wondering if you chip-guys (and maybe girls) have found the baker clamp audibly useful. A good test would be to make two identical amps one with a 49810 (or03) and the other with a 49811 and compare the two with the same program material at the same level scrunched up against the rails...Is ther a difference? I am curious. till the next post....so long
The LM49811 specs much worse than the LME49810 and will obviously sound worse. Clipping looks anything but soft on the scope so i belive that it's more to give a fast recovery from clipping than to give tubelike clipping (ie prevent rail sticking). It's not the worst clipping amp i've ever heard.
Wrong!
http://hhscott.com/_!images/TUBE/em87_Dual.jpg
EDIT: The local parts supplier gave me an entire bag of these so i'm going to build a 10 frequency steps vu-meter to give a bit of vintage look to this solid-state beast.
http://hhscott.com/_!images/TUBE/em87_Dual.jpg
EDIT: The local parts supplier gave me an entire bag of these so i'm going to build a 10 frequency steps vu-meter to give a bit of vintage look to this solid-state beast.
I belive that these are not hard to find, it's a pity they add some HV hazard to the amp.
Back to the original topic, i built my LME49810 amp to revive the feeling i got the first times i went local beerish concerts (50K people town, expect terrible sound quality at outrageous spl). The LME did the thing. While playing quiet it sounded like conventional hi-fi, but when i pushed it there was a point when music started to have the "vibrant" magic sound of those live (amplified with terrible quality) concerts. It took some time to me to realize that the "magic" begins with the clipflag glowing! It's the first time i've seen a true-to-life reproduction and i even plan a regulated psu to provoke that effect at lower levels. Nothing to do with overdriven small plastic loudspeakers.
Conclusion: The lme49810 clips nicely.
Back to the original topic, i built my LME49810 amp to revive the feeling i got the first times i went local beerish concerts (50K people town, expect terrible sound quality at outrageous spl). The LME did the thing. While playing quiet it sounded like conventional hi-fi, but when i pushed it there was a point when music started to have the "vibrant" magic sound of those live (amplified with terrible quality) concerts. It took some time to me to realize that the "magic" begins with the clipflag glowing! It's the first time i've seen a true-to-life reproduction and i even plan a regulated psu to provoke that effect at lower levels. Nothing to do with overdriven small plastic loudspeakers.
Conclusion: The lme49810 clips nicely.
ionomolo said:
I've noticed a similar thing on a chip-amp I've built,and been using based on the TA8261AH chip ( Datasheet )
If you push it well into clipping (the clipping LED on solid!) it still dosn't sound "bad"...just sorta compressed,None of the 'usual' hard clipping artifacts..(unless you REALLY crank it up,past 11)
what a load of codswallop.ionomolo said:
You belong in a loony bin for promoting that philosophy.
A bit harsh... 95% of the people in audio are trying to revive their first hi-fi experience, i belive there is nothing wrong about trying to revive the first lo-fi one, and this excludes those who build tube amps... These are doing the same thing, only they are a bit older 😀.
I belive that designing amplifiers is a superior art to building musical instruments as it's not about making things sound right but rather about taking to the present things that happened in the past. That's not that bad, and it's less of a problem if there is a switch to turn it off.
I havn't said i don't like accurate reproducton, but sometimes i like some "magic" that amplifying the signal with sub-ppm thd does not have.
