I would caution against going much above .7 to .75A (350-375mV) if you mounted the diodes to the same heat sinks as the MOSFETs. The higher the temperature the lower voltage the diodes will start to conduct. You have to leave room for increased ambient temperature or running it hard or your cat laying on top of it, etc.. You don't want them conducting at idle, only after some amps are going to the load. Otherwise you could be creating a runaway condition and let out the magic smoke.
If you really want to bump the bias higher then reduce the source resistors a little. (don't forget to reset the bias pots back to 0) Then you can increase the bias more without worry. You could piggyback a third resistor temporarily if you just want to test a higher bias. I tried it and found Mr. Pass' recommended range (.7A / 350mV) to be just right. YMMV
TJ
If you really want to bump the bias higher then reduce the source resistors a little. (don't forget to reset the bias pots back to 0) Then you can increase the bias more without worry. You could piggyback a third resistor temporarily if you just want to test a higher bias. I tried it and found Mr. Pass' recommended range (.7A / 350mV) to be just right. YMMV
TJ
Last edited:
I hope I can explain it well enough; 
Vertical MOSFETs suffer from the same positive compensation issue as BJTs in this regard, as BigE points out, hence the NTC in the bias chain (the negative compensation). The big issue with F5T v2 & v3 is the diodes across the source resistors. If the voltage across the source resistors is high enough at idle to turn on the diodes, the diodes will heat up more which lowers their turn-on voltage further, which turns them on harder, decreasing the source resistance, thus increasing the current to the MOSFETs, which will heat up further and draw even more current.
The cycle continues (very quickly) until either it levels out at some point, or more likely, snowballs out of control and badda boom!
It takes time for the heat inside the devices to make it's way out to the heat sink and to the NTC so it can run away much quicker than the time it takes for the NTC to sense the extra heat and try to compensate.
If you increase the bias past the point where the diodes turn on you will notice it "takes off" and jumps up very quickly. This point is around 400mv, depending on the temperature. The lower the temp, the higher the turn-on point, higher temps = lower turn-on point. If you look at the data sheet for the diode you will see it jumps up very quickly once the voltage gets past the knee.
This is why I advocated for some "breathing room" of the bias. Ambient temperatures vary and if it rises to the point where the diodes are on when they shouldn't be you could end up with a run away.
Besides the higher number of OP devices and increased voltage, this is the what makes the F5T into a v2 or v3. When the amp is under load this is ok. Remember how class A works.
TJ
Vertical MOSFETs suffer from the same positive compensation issue as BJTs in this regard, as BigE points out, hence the NTC in the bias chain (the negative compensation). The big issue with F5T v2 & v3 is the diodes across the source resistors. If the voltage across the source resistors is high enough at idle to turn on the diodes, the diodes will heat up more which lowers their turn-on voltage further, which turns them on harder, decreasing the source resistance, thus increasing the current to the MOSFETs, which will heat up further and draw even more current.
The cycle continues (very quickly) until either it levels out at some point, or more likely, snowballs out of control and badda boom!
It takes time for the heat inside the devices to make it's way out to the heat sink and to the NTC so it can run away much quicker than the time it takes for the NTC to sense the extra heat and try to compensate.
If you increase the bias past the point where the diodes turn on you will notice it "takes off" and jumps up very quickly. This point is around 400mv, depending on the temperature. The lower the temp, the higher the turn-on point, higher temps = lower turn-on point. If you look at the data sheet for the diode you will see it jumps up very quickly once the voltage gets past the knee.
This is why I advocated for some "breathing room" of the bias. Ambient temperatures vary and if it rises to the point where the diodes are on when they shouldn't be you could end up with a run away.
Besides the higher number of OP devices and increased voltage, this is the what makes the F5T into a v2 or v3. When the amp is under load this is ok. Remember how class A works.
TJ
The diodes are in parallel with the source resistors. When they conduct, it it a similar situation to when there are no source resistors at all. In the F5T guide, Nelson Pass writes:
What if you remove the Source resistance altogether?
First off, you get the 44 amps. In addition, with Fets you get an output stage
that will deliver more Class A power at a given bias figure due to the square
law character of the Fets. Unfortunately you also tend to get a thermally
unstable circuit that is prone to bias hogging.
I added the bolding.....bias hogging means that particular MOSfet wants to pass most of the current. BOOM. It is the first to go.
What if you remove the Source resistance altogether?
First off, you get the 44 amps. In addition, with Fets you get an output stage
that will deliver more Class A power at a given bias figure due to the square
law character of the Fets. Unfortunately you also tend to get a thermally
unstable circuit that is prone to bias hogging.
I added the bolding.....bias hogging means that particular MOSfet wants to pass most of the current. BOOM. It is the first to go.
May I ask a stupid question here... having builtv2 myself... why would you want the diodes at all then... meaning:
If you want more power wouldnt it be better to increasd no of parallel output mosfets? Besides the issues above, you would as well have no switching noise of diodes anywhere I would suspect?
If you want more power wouldnt it be better to increasd no of parallel output mosfets? Besides the issues above, you would as well have no switching noise of diodes anywhere I would suspect?
I left the diodes out of my build for a V3. With 4 sets of outputs you do not gain a lot of output power if you are driving typical speaker loads which have a minimum Z somewhere in the 3 ohm range. When I calculated the difference in output power it was not enough to sway me to include the diodes. If you are only gaining 1 dB of power is it worth the risk? People are fixated on the peak output current figures noted by Nelson Pass in the F5 articles. These are very impressive but only possible or necessary if you are welding or driving some very severe loads. My question was how much current do I need to drive a 3 ohm load with a +-45volt supply. The answer is +-15 amps peak. Divide this by 4 sets of outputs and it’s less than 4 amps/device. If memory serves the source R is .5 ohms, so peak Rs voltage would be less than 2 volts. Compared to maybe .4volts for the diodes. So you would loose 1.6x2 =3.2 volts peak to peak of output or about 1.2volts rms. So 30v rms out or 28.8 volts out is 300w vs 272w. A 28 watt difference sounds like a lot but is not likely to be audible at less than 1db difference.
This is just my take on the diodes. Perhaps I have overlooked something that others may wish to comment on. I guess my point is that if the diodes make you nervous, leave them out and the amp will still function just fine. When it comes to an amp going nuclear and putting a rail voltage on the output, I don’t think about the amp damage as much as the expensive speaker that may be connected to it.
This is just my take on the diodes. Perhaps I have overlooked something that others may wish to comment on. I guess my point is that if the diodes make you nervous, leave them out and the amp will still function just fine. When it comes to an amp going nuclear and putting a rail voltage on the output, I don’t think about the amp damage as much as the expensive speaker that may be connected to it.
That is why my build includes the diyaudio store speaker protection circuit.
It has worked twice and saved the speakers.
Both times, I was doing something remarkably stupid.
I have build the f5tv2 with 32v and 0.366v across the 0.5 resistors... and with the diodes.
I measured the psu current and it was around 2.8A if I remember right.
I started now to play aeoundwith my new toy which is a roland quad capture card, pmillets sc interface and dr. jordan measurement software.
i can raise the voltage to approx. 20V (1khzsinus at the output) and than suddenly the THD jumps from 0.5% to 15% or more.
I have measured temperature when I run the amp at 0.2v output vs. 20v output - no difference, around 42 degree celsius.
If i get this right the means I got 20V across my 7.5 ohm measurement resistor = 2.66A * 20V=53W class A, correct?
I wanted to see when actually the diode tbing start to happen and see a steep raise in temperature, but i could notprovokate this... i guess 7.5 ohm is simply to high for this?
I measured the psu current and it was around 2.8A if I remember right.
I started now to play aeoundwith my new toy which is a roland quad capture card, pmillets sc interface and dr. jordan measurement software.
i can raise the voltage to approx. 20V (1khzsinus at the output) and than suddenly the THD jumps from 0.5% to 15% or more.
I have measured temperature when I run the amp at 0.2v output vs. 20v output - no difference, around 42 degree celsius.
If i get this right the means I got 20V across my 7.5 ohm measurement resistor = 2.66A * 20V=53W class A, correct?
I wanted to see when actually the diode tbing start to happen and see a steep raise in temperature, but i could notprovokate this... i guess 7.5 ohm is simply to high for this?
Thanks!
After I posted, I realized that the 4 ohm safety would depend solely on the heat sink ability to manage twice the current as 8 ohms. It has nothing to do with biasing -- all about heat sinking and the VA of the transformer.
====
I have a problem that is not clear as to how to solve it.
I turned the unit on today with P1 P2 turned down to zero ( no smoke YAY!). P3 is set to where is was when it broke. However, all output transistors were removed checked and reused/replaced, but probably not in the same spots.
It looks like an output board does not want to pass the right amount of current. I think these are new Mosfets too. All insulators are Keratherm.
Power supply voltages measure well at the board about 38VDC.
The boards for each channel are wired in parallel.
The gate wiring is twisted together at the Front End board. Gate wiring shows continuity between both boards, ohmed out on unused PCB connections.
Since I have just started biasing, it's not at full power.
The measurements are:
DC Offset: less than 3 mV.
P Channel
Board 1 bias voltage ( with thermistor ): 35 mV Across either pair of source resistors.
Board 2 bias voltage 230 mV ( no thermistor ). Across test points
N Channel
Board 1 bias voltage about 125 mV (with thermistor) Across test points
Board 2 bias voltage 131 mV (no thermistor ) Across test points.
Clearly Board 2 P channel is doing most of the work to balance the N channel current flow.
Any help would be very appreciated. Maybe P Channel Mosfets static damage on Board 1? Guessing, as I don't know the symptoms....
Last edited:
I have taken the diodes out of mine and measured the same power before and after, just needed to dial back p1 and p2 by one turn to have the same bias across 0.5ohm. Speaker resistance is 7 ohm.
I took my thd measuring software and dialed p3 so that h2 is a little higher than h3... lets see how this will sound now... unfortuantely you cant manipulate h3 and h5 with p3...if stays as it is and is normally a little higher than h2 in this amp....but at a very reasonable level of 0.3% thdn until 50 watt... but for sure a different spectrum than a tube amp...
I will listen now and than start to modify with different psu concepts, ww-resistors and feedback... fortunately there are so few parts in it that you can easily play with those options. did you try ferrites instead of grid stopper resistors? Siund much better imho...
I took my thd measuring software and dialed p3 so that h2 is a little higher than h3... lets see how this will sound now... unfortuantely you cant manipulate h3 and h5 with p3...if stays as it is and is normally a little higher than h2 in this amp....but at a very reasonable level of 0.3% thdn until 50 watt... but for sure a different spectrum than a tube amp...
I will listen now and than start to modify with different psu concepts, ww-resistors and feedback... fortunately there are so few parts in it that you can easily play with those options. did you try ferrites instead of grid stopper resistors? Siund much better imho...