Well, sometimes it blows 4A fuses when turned on... I think it has to be with the voltage multiplier, its transistor probalby can't turn on as fast as it should, so as it's opened current through power transistors (IRF640) is too high, could be that?
Any ideas will help, of course, and I'm thinking on removing voltage multiplier and leave a single preset.
Any ideas will help, of course, and I'm thinking on removing voltage multiplier and leave a single preset.
Change to BC546, even with bias at minimum, still blows both fuses (used to be only the -40V rail). Measured output mosfets with diode tester and are not shorted, all of them measured 502 mV. also measured other transistors and none of them are shorted. Don't know really what the problem is...
Mine is running perfectly with MJE340 instead of BC546 and no other changes.
Hugh here is Quasi Nmos200 Quasi's web page : https://sites.google.com/site/quasisdiyaudiosite/nmos-series/nmos200-1
Marc
Hugh here is Quasi Nmos200 Quasi's web page : https://sites.google.com/site/quasisdiyaudiosite/nmos-series/nmos200-1
Marc
That's shorted
.They should measure open with red to drain, black to source.
The opposite way you will find a parasite diode (built-in) which will give around 650mV.
To be certain, pull them from the PCB and measure them "outside".
The Vbe multiplier is never "too slow".
Before pulling them, check that drains are not shorted to the heatsink.
PS: you *did* use nipples on those exposed IRF640 metal tabs, didn't you?
SLD82,
Here is the schematic (I understand this is OK to publish since it's on the public domain).
If the Vbe multiplier BC546 is substituted BD139, the bias voltage at the output gates will be slightly increased because the BD139 has MUCH less beta than the BC device. This would be usually 80 compared with 300, increasing the voltage dropped across the base and collector. This increases the bias voltage considerably which drives the output very hard.
The reason is that the bias generator has around 5.3 volts as designed, and the source resistors are very small at 0.22R. If a small increase from 5.3V to say 5.5V the additional bias will be impressed - anything up to 10A - upon the source resistors, hugely increasing the quiescent, in your case, blowing the fuses.
Vbe multipliers are extremely tetchy, and should has be built as designed to deliver the intended quiescent. You should be able to correct this using a BD139 by reducing the base/collector resistor from 470R to 390R. This is a guestimate, you will need to experiment with the pot at half set, ie 100R.
Cheers,
Hugh
Last edited:
Yes, in general cases this is correct, but in this particular amp the Vbe biasing string is incredibly low resistance.
Idle current across the Vas transistor T7 (and its current source T6) is almost 14 mA.
The biasing string current is conventionally chosen as 1/10th of that, meaning around 1.4 mA would be fine.
Why "1/10"?
So we can use a real world biasing transistor, and not worry too much about its actual Beta/Hfe .
Beta 100 would mean we'd have a base current 1/100 of the collector current , or 1/10 of the chosen biasing string current, for an error of 10%.
Which is no problem at all, that's why part of the biasing string is an adjustable trimmer.
And if we have Beta 500 , as in a BC546C , no big deal, instead of 10% error we would have 2% .
BUT in this particular amp, they did not choose a bias string current of around 1.4 mA, they chose ridiculously high 7mA.
Why ridiculous? Aren't they erring on the safe side?
Well, respect to Beta, maybe, ......, but now *half* of the Vas current is wasted running the biasing string (R13/VR2/R14) , instead of fully passing through Thermal sensor T8.
I bet that throws accuracy away.
Not dissing AKSA's calculation, which is fine in a "normal" amp, but in *this* case I would not lower even more any of those resistors.
FWIW, in my quite similar architecture 300W/4 ohms MosFet Bass amp I sense with a BC556 (yes, a PNP) , and the biasing string is around 7K total , 10X higher than the meager 710 ohms shown here.
Works like a charm, under heavy stage conditions.
*Literally* Heavy, most of my customers are Metal Bass players.
Idle current across the Vas transistor T7 (and its current source T6) is almost 14 mA.
The biasing string current is conventionally chosen as 1/10th of that, meaning around 1.4 mA would be fine.
Why "1/10"?
So we can use a real world biasing transistor, and not worry too much about its actual Beta/Hfe .
Beta 100 would mean we'd have a base current 1/100 of the collector current , or 1/10 of the chosen biasing string current, for an error of 10%.
Which is no problem at all, that's why part of the biasing string is an adjustable trimmer.
And if we have Beta 500 , as in a BC546C , no big deal, instead of 10% error we would have 2% .
BUT in this particular amp, they did not choose a bias string current of around 1.4 mA, they chose ridiculously high 7mA.
Why ridiculous? Aren't they erring on the safe side?
Well, respect to Beta, maybe, ......, but now *half* of the Vas current is wasted running the biasing string (R13/VR2/R14) , instead of fully passing through Thermal sensor T8.
I bet that throws accuracy away.
Not dissing AKSA's calculation, which is fine in a "normal" amp, but in *this* case I would not lower even more any of those resistors.
FWIW, in my quite similar architecture 300W/4 ohms MosFet Bass amp I sense with a BC556 (yes, a PNP) , and the biasing string is around 7K total , 10X higher than the meager 710 ohms shown here.
Works like a charm, under heavy stage conditions.
*Literally* Heavy, most of my customers are Metal Bass players.
So how does one further optimise the Vbe stage in the Nmos200 and its many variants, including the Nmos350 Mk II in which the B-C resistor of the Vbe is increased to 680E and the 200E Trimpot is moved to the B-E string with the variable pin shorted to one end pin of the trimpot, which is in series with the 47E resistor and the Vbe transistor has been changed to a BC550C type?
Can't talk about NMos 350 (although I guess it will be quite similar to my own 300W )
In the case of the NMos 200 , if the original values work, continue with them.
My *personal* preference would be to double or quadruple those values, as I mentioned before, but choosing one or the other is your option.
I'm not trying to preach anything here
)In the case of the NMos 200 , if the original values work, continue with them.
My *personal* preference would be to double or quadruple those values, as I mentioned before, but choosing one or the other is your option.
I'm not trying to preach anything here
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.