0.39Vf for the source resistor bypass diodes indicates that they are just starting to turn on while the output current is zero.
When the output current increases slightly above zero, the diodes will very effectively bypass the source resistors.
Now you are relying totally on the Lateral mosFETs' negative temperature coefficient to try to maintain close matching of drain current. The diode bypass still has a positive temperature coefficient.
Some will run hotter than others. Some will eventually overheat, even though the heatsink is at what seems like a sensible temperature.
I recommend a lower quiescent Vf until you are sure the amplifier is working properly.
Lateral mosFETs have Source as the middle pin, whereas Vertical mosFETs have the Drain as middle pin. Did you swap pin connections to suit the Laterals?
When the output current increases slightly above zero, the diodes will very effectively bypass the source resistors.
Now you are relying totally on the Lateral mosFETs' negative temperature coefficient to try to maintain close matching of drain current. The diode bypass still has a positive temperature coefficient.
Some will run hotter than others. Some will eventually overheat, even though the heatsink is at what seems like a sensible temperature.
I recommend a lower quiescent Vf until you are sure the amplifier is working properly.
Lateral mosFETs have Source as the middle pin, whereas Vertical mosFETs have the Drain as middle pin. Did you swap pin connections to suit the Laterals?
Last edited:
@AndrewT:
With 0,39V i meant the reading on DVM when checking diodes, not voltage drop in operation. Sorry for not beeing exact.
As far as i know, IRFP240/9140 are vertical power MOSFET, not laterals ?!
So ... having checked the frontend on its own, all resistors and diodes, put in new matched FETs and started it up again this afternoon, everything working perfectly .... at first.
Bias about 0,62A per FET (0,30 - 0,32V on 0,5R source resistor), heatsink could be touched comfortably. Let it run for 30 minutes with sine wave as input signal, connected 15 ohms resistor, offset adjusted to some millivolts, no drift, nice output signal from 0-25VAC with varying input, fine ...
Blew some air across the NTCs, bias current rose slightly (app 0,04A), so NTCs working correctly.
Switched off mains power, connected music signal and test-speaker, switched on mains again and had 1,6V offset instantly. No smoke, no bang, no fuse blown.
Result: one IRFP9140 defective (Rgs=1,9k, Rds=5k), all other FETs, diodes and resistors measured OK.
So if anybody has some idea or thoughts about this, i´d be delighted to hear about. I can´t see any mistake, so maybe someone else does.
The only thing that comes to my mind is to get IRFP9240 instead of 9140 and try again. Maybe someone who also uses 240/9140 can share his experience ?
Time for a beer now ...
With 0,39V i meant the reading on DVM when checking diodes, not voltage drop in operation. Sorry for not beeing exact.
As far as i know, IRFP240/9140 are vertical power MOSFET, not laterals ?!
So ... having checked the frontend on its own, all resistors and diodes, put in new matched FETs and started it up again this afternoon, everything working perfectly .... at first.
Bias about 0,62A per FET (0,30 - 0,32V on 0,5R source resistor), heatsink could be touched comfortably. Let it run for 30 minutes with sine wave as input signal, connected 15 ohms resistor, offset adjusted to some millivolts, no drift, nice output signal from 0-25VAC with varying input, fine ...
Blew some air across the NTCs, bias current rose slightly (app 0,04A), so NTCs working correctly.
Switched off mains power, connected music signal and test-speaker, switched on mains again and had 1,6V offset instantly. No smoke, no bang, no fuse blown.
Result: one IRFP9140 defective (Rgs=1,9k, Rds=5k), all other FETs, diodes and resistors measured OK.
So if anybody has some idea or thoughts about this, i´d be delighted to hear about. I can´t see any mistake, so maybe someone else does.
The only thing that comes to my mind is to get IRFP9240 instead of 9140 and try again. Maybe someone who also uses 240/9140 can share his experience ?
Time for a beer now ...
No such issues on the build that I've used this pairing.
However it didn't use the diodes.
And the thermistor was connected 'vanilla' style, to adjust Vgs of all devices.
I also used the current limiter with the same values as the vanilla F5. To prevent exactly this kind of situation. With three pairs the total limit is high enough that I need not worry about the amp running out of juice even into very low impedance loads.
My theory for your failure? The startup bias is high enough (with cool thermistors) that the diodes start conducting, possibly creating some unbalance and the weakest device/s get taken out. Simple enough to test this, and prevent it too.
Do note that this thread is about the original F5, not the Turbo. Maybe you'll get more opinions on that thread.
Thank you AndrewT and sangram for your information, have to search for "vanilla", did not find posts related to this problem so far.
sangram, your explanation seems right, though NTCs are mounted to 6 kg of heatsink. Got to do some tests on startup-behaviour.
At least i now know that it is right basically, so maybe omit diodes, put in current limiters, change mounting of NTCs, or IRF and Vishay-devices behave differently, or something else ...
Somehow mixed up the threads ...
p.s. second channel OK during tests. Identical twin, maybe walking a fine line.
sangram, your explanation seems right, though NTCs are mounted to 6 kg of heatsink. Got to do some tests on startup-behaviour.
At least i now know that it is right basically, so maybe omit diodes, put in current limiters, change mounting of NTCs, or IRF and Vishay-devices behave differently, or something else ...
Somehow mixed up the threads ...
p.s. second channel OK during tests. Identical twin, maybe walking a fine line.
'Vanilla' refers to the original F5 schematic posted in the first post of this thread. It has the schematic for the current limiter and the thermistor is connected to the rail instead of the device source.
I have asked before, and it has been indicated that there is no issue with the way you have connected it. I still went with my gut. Never had a failure except when I did something foolish like forget to bolt down a device to the sink, or connect the power supply backwards.
Good luck!
I have asked before, and it has been indicated that there is no issue with the way you have connected it. I still went with my gut. Never had a failure except when I did something foolish like forget to bolt down a device to the sink, or connect the power supply backwards.
Good luck!
F5 failure.
I haven't been following this thread for some time so this suggestion may have been made already. I have built several F5 amps for friends and generally they go OK for years. However one I kept for some time blew up one channel when I used some exotic high capacitance speaker cables. Initially I hot swapped them and it ran OK all night. When I switched them on the next day one channel expired. After a lot of mucking about it seems that the F5 doesn't like high capacitance speaker leads or maybe high capacitance in the speakers themselves due to maybe a crossover issue. I believe the amp can go into oscillation with too much capacitance on the output. The other problem could be DC on the input signal. I never used input capacitors on my amps and always made sure there was no DC from the signal source. Good luck with fixing the amps. David.
I haven't been following this thread for some time so this suggestion may have been made already. I have built several F5 amps for friends and generally they go OK for years. However one I kept for some time blew up one channel when I used some exotic high capacitance speaker cables. Initially I hot swapped them and it ran OK all night. When I switched them on the next day one channel expired. After a lot of mucking about it seems that the F5 doesn't like high capacitance speaker leads or maybe high capacitance in the speakers themselves due to maybe a crossover issue. I believe the amp can go into oscillation with too much capacitance on the output. The other problem could be DC on the input signal. I never used input capacitors on my amps and always made sure there was no DC from the signal source. Good luck with fixing the amps. David.
Many amplifiers have this problem. It's actually a region of instability due to the phase shift.
With enough additional load capacitance it will be stable again, but with reduced performance.
Last edited:
When I got my F5 I got it as a kit from Jims Audio on Ebay and I noticed that it had a capacitor in the circuit which obviously isn't meant be there! I presume he added it for this reason to stop oscillation.
This amp also seemed to have a harsh Trebble sound to me so Im guessing it was that ?
Im building another F5 now and I think is it best I buy the Diyaudio boards this time ?
can someone recommend a specific Kit or PCB ? I want to get building asap for our campervan. which will run the amp from 4x 12v 150ah batteries and SMPS
This amp also seemed to have a harsh Trebble sound to me so Im guessing it was that ?
Im building another F5 now and I think is it best I buy the Diyaudio boards this time ?
can someone recommend a specific Kit or PCB ? I want to get building asap for our campervan. which will run the amp from 4x 12v 150ah batteries and SMPS
I can recommend the boards from the diyAudio shop here.
They are simple clear and work well.
Make sure that you are working from the V3 circuit (not the BOM which is for an early version.)
Components all (apart from jfets ) from Digikey makes life easy.
Farnell or RS also have them in the uk.
As do Cricklewood.
jfets from Spencer, or possibly Little Diode in UK. Integrated Circuit items in 2sj74 store on eBay!
They are simple clear and work well.
Make sure that you are working from the V3 circuit (not the BOM which is for an early version.)
Components all (apart from jfets ) from Digikey makes life easy.
Farnell or RS also have them in the uk.
As do Cricklewood.
jfets from Spencer, or possibly Little Diode in UK. Integrated Circuit items in 2sj74 store on eBay!
My F5 has a warm, but still detailed, rich sound with plenty of well controlled bass.
A much better sound (IMHO) than the 2A3 amp that it is replacing.
IRFP 240 / 9240 output devices.
I seem to have 5x IRFP240 and 8x LSK170 but thats all, and I guess if I try to find others matching them would be a pain for someone like me who does not know what they are doing.
If I order the Mosfets from Diyaudio I guess they will be better matched ? LSJ74 cost a lot on Diyauadio.
Illl have a look at the links you sent cheers.
If you like the F5 the F5t is a monster!
If I order the Mosfets from Diyaudio I guess they will be better matched ? LSJ74 cost a lot on Diyauadio.
Illl have a look at the links you sent cheers.
If you like the F5 the F5t is a monster!
This would be a case where an output Zobel (maybe 5-10 ohms in series
with .047 - .1 uF in series) would be appropriate.
Take a look at my article on Speaker Cables from 1980 (articles section at
FirstWatt.com) for some examples of the problems with low inductance /
high capacitance cables).
I'm going to order all the parts from Diyaudio does anyone think it's the best choice for buying in the UK?
Have checked out the other links here and some good options have used RS electronics before too.
I'm going to be powering the F5 in our Van with solar setup
660w panels
4x 155 ah batteries.
Should be good!
Sent from my SM-G900F using Tapatalk
Have checked out the other links here and some good options have used RS electronics before too.
I'm going to be powering the F5 in our Van with solar setup
660w panels
4x 155 ah batteries.
Should be good!
Sent from my SM-G900F using Tapatalk
Last edited: