Push-Pull ClassA has a constant power draw on the two supplies.
The Power supplied equals the power dissipated in the amplifier plus the power dissipated in the load.
If the load is reactive then it dissipates less power and that results in more power dissipation in the amplifier. See ESP's explanation of dissipation and reactive loading.
Thanks for reading my posts. They are not for my benefit.
NOTE !!!!!
The Power supplied equals the power dissipated in the amplifier plus the power dissipated in the load.
If the load is reactive then it dissipates less power and that results in more power dissipation in the amplifier. See ESP's explanation of dissipation and reactive loading.
Thanks for reading my posts. They are not for my benefit.
NOTE !!!!!
is completely different from the oft repeated claim for ClassA where we see an alleged constant current draw. This is usually wrong.
Last edited:
Andrew, I don't think we are at the same place in the discussion. You are not reading my posts either - especially in context of your own, so I don't think there's much point in prolonging this. I have a serious question that needs an answer, it is clear that this is not a conversation that will lead to one.
Good day to you.
Good day to you.
Apropos of cascoding the F5 Turbo front end:
Since the original, playing with this sort of circuit and with different devices and
such, I decided that there is a stability advantage to having some resistance
off the Drains of the input devices when cascoding. This means 100 to 200
ohms (or so) between the Drains and the Emitters of the cascode transistors.
I have seen several examples of cascoded F5's tend toward oscillation, often
fixed by a lag resistor (Drain to Gate) on the outputs, but traceable to the
low impedance seen by the input device Drains (the Cascode).
Particularly with Lateral output Mosfets.
So if you run into this issue, you can try this. Or try it anyway...
Since the original, playing with this sort of circuit and with different devices and
such, I decided that there is a stability advantage to having some resistance
off the Drains of the input devices when cascoding. This means 100 to 200
ohms (or so) between the Drains and the Emitters of the cascode transistors.
I have seen several examples of cascoded F5's tend toward oscillation, often
fixed by a lag resistor (Drain to Gate) on the outputs, but traceable to the
low impedance seen by the input device Drains (the Cascode).
Particularly with Lateral output Mosfets.
So if you run into this issue, you can try this. Or try it anyway...
All this oscillation talk has me a bit nervous. From what I understand doing a search on the topic I wont know that there is oscillation until I see burned Mosfets and gate stopper resistors.
Checking the published circuits I see the official F5 circuit showing 100r gate stopper resistors and all other versions of the F5 show 47.5r. Not sure why they were reduced. Given the oscillation issue is it prudent to increase them to 100r? I am in the process of building an F5TV3 with Toshiba Mosfets 2SK1530/2SJ201so I can easily switch them in at this stage if that is advisable.
Also wondering how to add the 100-200r resistor suggested by Mr. Pass using the store boards. Any ideas?
Thanks.
Nash
Checking the published circuits I see the official F5 circuit showing 100r gate stopper resistors and all other versions of the F5 show 47.5r. Not sure why they were reduced. Given the oscillation issue is it prudent to increase them to 100r? I am in the process of building an F5TV3 with Toshiba Mosfets 2SK1530/2SJ201so I can easily switch them in at this stage if that is advisable.
Also wondering how to add the 100-200r resistor suggested by Mr. Pass using the store boards. Any ideas?
Thanks.
Nash
Noob in Town. Starting F5T V2
Until now I have only lurked on the list but this is the first of many comments/questions from me.
This will be my first DIY amp since building a pair of dual-mono Borbely 60W MOSFET amps in the early 80s. I added some high current voltage regulators and even adapted the servo circuit from Borbely's Servo 100 to my amp. It served me well until it stayed behind in one of my moves.
Recently I purchased the DIYAudio boards for the F5T and have begun to collect parts for my build. Input FETs from Spencer, and 8 (probably fake) 22000MF 63V electrolytics from some fleaBay seller in some city I can't begin to pronounce....
One of my concerns it the availability of suitable output devices. Has anyone considered the use of SEMELAB BUZ901/BUZ906 devices in suitable packages? They are a bit expensive but come in a dual die package which seems to make them very robust. Max operating temp of 150C and a power disssipation of 125 W. I don't know enough (yet) to know if they will work, let alone figure out what adaptations will be necessary to make them happy but if anyone here has any experience with these devices, or can offer any insight I would be pleased to hear from you.
Until now I have only lurked on the list but this is the first of many comments/questions from me.
This will be my first DIY amp since building a pair of dual-mono Borbely 60W MOSFET amps in the early 80s. I added some high current voltage regulators and even adapted the servo circuit from Borbely's Servo 100 to my amp. It served me well until it stayed behind in one of my moves.
Recently I purchased the DIYAudio boards for the F5T and have begun to collect parts for my build. Input FETs from Spencer, and 8 (probably fake) 22000MF 63V electrolytics from some fleaBay seller in some city I can't begin to pronounce....
One of my concerns it the availability of suitable output devices. Has anyone considered the use of SEMELAB BUZ901/BUZ906 devices in suitable packages? They are a bit expensive but come in a dual die package which seems to make them very robust. Max operating temp of 150C and a power disssipation of 125 W. I don't know enough (yet) to know if they will work, let alone figure out what adaptations will be necessary to make them happy but if anyone here has any experience with these devices, or can offer any insight I would be pleased to hear from you.
BUZ 901/906 are available but are different pinout.
Irfp240/9240 are available and not too expensive. You can buy enough to get some matched parts.
http://www.mouser.com/ProductDetail...=sGAEpiMZZMshyDBzk1/Wi1F3z9PgzPBn8rpRYOMim0Q=
http://www.mouser.com/ProductDetail...=sGAEpiMZZMshyDBzk1/Wi1F3z9PgzPBnuHmAlkPuz34=
Irfp240/9240 are available and not too expensive. You can buy enough to get some matched parts.
http://www.mouser.com/ProductDetail...=sGAEpiMZZMshyDBzk1/Wi1F3z9PgzPBn8rpRYOMim0Q=
http://www.mouser.com/ProductDetail...=sGAEpiMZZMshyDBzk1/Wi1F3z9PgzPBnuHmAlkPuz34=
Last edited:
I'm setting the bias on my F5T and I'm getting different voltages across the top and bottom
source resistors. It's on the order of a 50 mV difference. The offset jumps around a bit
but is within +/-5 mV of 0. I've measured across the 220 ohm feedback resistors and there is less than a mA going there. So how can I have different currents?
I've checked the 1 ohm resistors and they are the same on top and bottom.
I'm using the same meter to measure both voltages.
Has anyone else seen this?
source resistors. It's on the order of a 50 mV difference. The offset jumps around a bit
but is within +/-5 mV of 0. I've measured across the 220 ohm feedback resistors and there is less than a mA going there. So how can I have different currents?
I've checked the 1 ohm resistors and they are the same on top and bottom.
I'm using the same meter to measure both voltages.
Has anyone else seen this?
Different voltages across the source R are normal. Component tolerances and the different characteristics of the N- and P- channel parts is why you get slightly different readings.
The offset and overall bias is what counts - so I now measure bias current off the supply rail resistor in the CRC, just find it easier to deal with. The front end draws a very small amount of current, insignificant compared to the overall bias figure so this reading works reasonably well for me.
The offset and overall bias is what counts - so I now measure bias current off the supply rail resistor in the CRC, just find it easier to deal with. The front end draws a very small amount of current, insignificant compared to the overall bias figure so this reading works reasonably well for me.
Probably the biggest impediment to building this amplifier is overthinking things. The outputs used (or the alternative irf 240/9240s) are cheap, tried, tested and true in this design (especially with the temp compensation and all that) so just give it a whirl
Last edited:
I figured out why I was getting different drops across the N and P mosfet source
resistors. The problem was my P devices weren't perfectly matched. I was just
measuring across the TP3 and TP2 test points. The 2nd device was drawing
less. So, I was measuring the low current device of the two N and the high
current device on the two P. The offset was zero because the total
current of the two N equals the two P devices.
resistors. The problem was my P devices weren't perfectly matched. I was just
measuring across the TP3 and TP2 test points. The 2nd device was drawing
less. So, I was measuring the low current device of the two N and the high
current device on the two P. The offset was zero because the total
current of the two N equals the two P devices.
This is wrong !!!!!!
For a single pair output stage the upper device current is the same as the lower device current.
The voltage drop across the upper source resistor is the same as the voltage drop across the lower device source resistor.
Any difference in these two measured voltages is down to inserting DIFFERENT source resistor values. Match your source resistors - accurately !
Then you will find that the source resistor voltage drop DOES NOT get affected by N & P channel device mismatching.
When you come to doing a many pair output stage, you still NEED to macth your source resistors if you intend using the resistor voltage drops to determine the relative currents through the paralleled devices.
Matching of the devices helps a lot in getting the devices running at similar temperatures.
One unmatched device in a set of four (2pairs), could end up with that one running cool or hot in comparison to it's partners. One hot running device, could lead to failure of that hotter device taking out a speaker.
Last edited:
Andrew - I agree the statement you quoted was wrong. You're also right that
the current thru connected N/P pairs must be the same to get an offset
of 0V. My problem was with having a pair of P channel mosfets that had
different currents. So, one N/P pair was drawing less current than the
second N/P pair. I just happened to measure the drop across the P device
in one leg and the N device in the other leg. I mounted the circuit
boards to the heatsink such that TP2 and TP3 of the N and P boards were
for different legs. I'm using 1% source resistors so they are very well matched.
the current thru connected N/P pairs must be the same to get an offset
of 0V. My problem was with having a pair of P channel mosfets that had
different currents. So, one N/P pair was drawing less current than the
second N/P pair. I just happened to measure the drop across the P device
in one leg and the N device in the other leg. I mounted the circuit
boards to the heatsink such that TP2 and TP3 of the N and P boards were
for different legs. I'm using 1% source resistors so they are very well matched.