I'm just catching up on yesterday's development.
Joe, I'm so sorry this has happened to your left channel. Please take some time to
regroup.
I really don't know what might have happened. However, since this happened with
the left channel which had a previous issue, perhaps there was some
residual damage (say the diodes) that we didn't consider.
I wonder if you would consider skipping the diodes all together, since they are only
there to allow higher current. The F5t article mentioned 23A without the diodes and
38A with the diodes. Perhaps 23A is already enough for your application?
Joe, I'm so sorry this has happened to your left channel. Please take some time to
regroup.
I really don't know what might have happened. However, since this happened with
the left channel which had a previous issue, perhaps there was some
residual damage (say the diodes) that we didn't consider.
I wonder if you would consider skipping the diodes all together, since they are only
there to allow higher current. The F5t article mentioned 23A without the diodes and
38A with the diodes. Perhaps 23A is already enough for your application?
I agree with Dennis. Skip the diodes. I have seen too many F5 turbo's go up in smoke becouse of this devices. I just can't se the benefit overweighing the risk.
I for one would never EVER trust this amp with the diodes. It can smoke anytime. on day one, after a year, or anywere in between.
I for one would never EVER trust this amp with the diodes. It can smoke anytime. on day one, after a year, or anywere in between.
Last edited:
I wonder if an F5T without the diodes is worth the trouble. Why not make a vanilla F5 instead?
Anyways: luvrockin where you adequately warned before commencing this build? I was, and dropped it.
An F5T is by far the most complex available pass DIY amp. A dual mono one, complicates matters further.
That said, do you have separate fuses for each tranny, and each fuse as small as possible? If you have / or had that, the diodes wouldn’t be allowed to conduct without the fuse blowing. Nelson speaks about this in his article.
Anyways, I suggest there be even clearer warnings against building this amp in the store.
Something like «Do not try this at home, unless your name is Nelson, Dirk, Dennis, Jim, Mighty, Mark, Itsall, Audiosan or something like that, or you have built many amps before». It is stated it is not a beginners project, but could possibly have been even clearer wrt consequences.
Regards,
Andy
Anyways: luvrockin where you adequately warned before commencing this build? I was, and dropped it.
An F5T is by far the most complex available pass DIY amp. A dual mono one, complicates matters further.
That said, do you have separate fuses for each tranny, and each fuse as small as possible? If you have / or had that, the diodes wouldn’t be allowed to conduct without the fuse blowing. Nelson speaks about this in his article.
Anyways, I suggest there be even clearer warnings against building this amp in the store.
Something like «Do not try this at home, unless your name is Nelson, Dirk, Dennis, Jim, Mighty, Mark, Itsall, Audiosan or something like that, or you have built many amps before». It is stated it is not a beginners project, but could possibly have been even clearer wrt consequences.
Regards,
Andy
Hard to argue with that. But once a build is as complete as Lucrockins it is probably tempting to also squeeze those extra amps out. But if the risk is fire and the problem is not easily solved, the F5T V1 is probably a good route.
Btw, 6L6, how would you describe the audible differences between vanilla, V1 and v2?
Btw, 6L6, how would you describe the audible differences between vanilla, V1 and v2?
Last edited:
The diode conduction point varies with temperature, which means it can be fully stable one moment and run away if the diodes start to conduct when they're not meant to.
Remember the diodes have no way of knowing anything except their own temperature and the voltage across them. They have no control input and cannot wait for anything else in the circuit to trigger. When they conduct they're basically a dead short which is perfect if it syncs up with the output current. If it doesn't, then it's havoc.
The diode idea is great for low-bias Class AB amps. When you are on the edge of the conduction voltage most of the time in a Class A amp, it becomes a little tricky. It also sets a limit on the overall standing bias achievable. Not having the diodes, IMO, loses nothing from the circuit and in fact renders some extremely useful gains, such as the lower distortion resulting from higher bias.
The F5T is an extremely flexible topology that scales well without the diodes, all the way to 100W/ch given enough heatsinking. The idea of the diodes is to minimise the voltage drop across the source resistors, which is also achieved by running more pairs in parallel. The source resistors are a thermal feedback mechanism for the output device, and in hot conditions - especially in hot conditions, should not be bypassed.
Keeping the diodes off the main sink is a possible way of mitigating the issue. There are others, but none of them can be hacked into the existing circuit easily. As technology improves, manufacturers will move in direction of higher efficiency, not lower. This means Vf will trend downwards in a given part number.
I’ve been weighing in on all the valuable input about losing the diodes and I have a couple questions. How does removing the diodes affect the audible sound characteristics of the F5T amp, if any? I will still have 2 output stages, will I still have 50wpc output? If not, what would I need to do to achieve at least the 50wpc output, My speakers are only 86db and I’d rather build it now than rip it apart later when it doesn’t do what I’d like it to do. Lastly we came to discussing the diodes that burn up but it’s my mosfets that burned up. Is there thought the diodes possibly caused this? I wasn’t sure if that came through in the pics. Thanks
I don't recall anyone stating the diodes burning up. They don't.
The diodes have a current rating of 30A. The MOSFETs can only do 12A, so they are the devices that fail when faced with full rail voltage, a small input signal and zero source resistance (because the diode effectively short out the resistors).
If you drop the diodes you can bias to a full 1.6-1.7A (about 400-425mV across 0.5 ohms) and this does nothing but improve the sonics (IMO), because of the higher bias of the output stage. What the diodes do is allow a little more current into complex reactive loads at the expense of a bit of standing bias.
If you look at things a little differently, Nelson's XA series amps use lots and lots of output devices, with very large source resistors, and not a diode in sight. There's a very good reason for all of that. The F5T (with diodes) should be considered a tinkerer's circuit and while it's definitely possible to make it work well, using three to four pairs of output devices and more optimally sized source resistance should be the way to go for builders looking for more power.
Edit: I might add that to get the full benefit of dropping the diodes, you should look at slightly lower source resistance. About 0.33 ohms is good, lower is not advisable. You'll need to compute the correct voltage reading from the bias and the resistance value.
The diodes have a current rating of 30A. The MOSFETs can only do 12A, so they are the devices that fail when faced with full rail voltage, a small input signal and zero source resistance (because the diode effectively short out the resistors).
If you drop the diodes you can bias to a full 1.6-1.7A (about 400-425mV across 0.5 ohms) and this does nothing but improve the sonics (IMO), because of the higher bias of the output stage. What the diodes do is allow a little more current into complex reactive loads at the expense of a bit of standing bias.
If you look at things a little differently, Nelson's XA series amps use lots and lots of output devices, with very large source resistors, and not a diode in sight. There's a very good reason for all of that. The F5T (with diodes) should be considered a tinkerer's circuit and while it's definitely possible to make it work well, using three to four pairs of output devices and more optimally sized source resistance should be the way to go for builders looking for more power.
Edit: I might add that to get the full benefit of dropping the diodes, you should look at slightly lower source resistance. About 0.33 ohms is good, lower is not advisable. You'll need to compute the correct voltage reading from the bias and the resistance value.
I think Sangrams post is fantastic, and a great learning opportunity for me at least.
You keep the 50 watts without diodes, but won’t «gain» the extra and high risk diodes that Sangrams describes so wonderfully. Less source resistance gives you more amps.
If you post details about your speakers here, brand and type, impedance and maybe links to impedance measurements, the gurus here will be able to enlighten you wrt how well it might work without diodes.
Regards,
Andy
Hello Nikwatt,
you always need good cooling in a F5T - my opinion.
What I saw - take care with the thermistor in direct contact with the aluminum heatsink! Those thermistors coating have failed in a few builds and made a short!
This is the reason why I glued my thermistors to the plastic body of the Mosfet.
Dear Cubinchincer
You suggest me replace my thermistor with other type?
It's better no contact between thermistor ad heatsink?
thank You
nik
Greets
Dirk
to Nikwatt #934
Hello Nikwatt,
Andreas explained in post #935 exactly what I wanted to say.
You do not have to change the thermistors. It was reported by two F5T-
builders, that the surface coating of the thermistors got damaged (by the heat?
by the glue/compound?) and those who attached them to metal surfaces in contact to the heatsink (=ground) seemed to get a short ( I read from 2 cases).
Although anodised aluminum parts are no good electrical conductors.
But there also could have been other issues in those builds/amps?
Greets
Dirk
Hello Nikwatt,
Andreas explained in post #935 exactly what I wanted to say.
You do not have to change the thermistors. It was reported by two F5T-
builders, that the surface coating of the thermistors got damaged (by the heat?
by the glue/compound?) and those who attached them to metal surfaces in contact to the heatsink (=ground) seemed to get a short ( I read from 2 cases).
Although anodised aluminum parts are no good electrical conductors.
But there also could have been other issues in those builds/amps?
Greets
Dirk
Hi Guys, I’m back at it. I got the new MOSFET kit today from the store, removed all of the diodes from both channels. Taking advantage from all the advice above, I decided i don’t want the risk of failure. Thank you all for the input.
Sangram or anyone who can offer, as a newbie just learning, can you please explain how to come up with the biasing number 1.6-1.7A 400-425mv (like in the above quote) and if I were to change the 1 ohm resistors to .33 ohm as you suggested, I would have to compute the correct voltage rating from the bias and resistance value. Can you tell me how that is done? Thank you in advance
You don't change the 1 ohm to 0.33 ohm. Two 1 ohm resistors in parallel are 0.5 ohm, this is the existing arrangement. A third resistor in parallel will get you to the 0.33 ohm number.
To work out the voltage drop, first you need to know how much heat your heatsinks can take. 1.7A at 32V rails is (1.7 x 64) 108 watts. With a 25C ambient, the heatsink should not exceed 55C after an hour of running. It is safer to go up in steps. At 300mV voltage, the actual bias is 1.2A and you should increment in 100mA steps.
The voltage drop across a known source resistance Rs is the product of Rs and the current through it. There are two pairs, so each pair has half the current of the total bias. So for 1.7A and 0.33 ohm the voltage reading will be ((1.7x0.5) x 0.33) = 0.28V. I've used down to 0.22 ohms with zero issues, as long as the MOSFETs are very tightly matched (and you have a nice big heatsink, or some airflow over it) it's not an issue.
What does happen with higher bias is that the overall loop gain goes up. This increases the actual feedback and if you don't tweak the feedback resistors, you could have some oscillation. I've used three pairs up to 2A and needed a 220pF cap across the feedback resistors to completely kill the slight oscillation that resulted. Different layouts will have different stability implications, so the 1nF suggested in the article is a good place to be on average and should be installed anytime you change any of the overall parameters of the amplifier.
To work out the voltage drop, first you need to know how much heat your heatsinks can take. 1.7A at 32V rails is (1.7 x 64) 108 watts. With a 25C ambient, the heatsink should not exceed 55C after an hour of running. It is safer to go up in steps. At 300mV voltage, the actual bias is 1.2A and you should increment in 100mA steps.
The voltage drop across a known source resistance Rs is the product of Rs and the current through it. There are two pairs, so each pair has half the current of the total bias. So for 1.7A and 0.33 ohm the voltage reading will be ((1.7x0.5) x 0.33) = 0.28V. I've used down to 0.22 ohms with zero issues, as long as the MOSFETs are very tightly matched (and you have a nice big heatsink, or some airflow over it) it's not an issue.
What does happen with higher bias is that the overall loop gain goes up. This increases the actual feedback and if you don't tweak the feedback resistors, you could have some oscillation. I've used three pairs up to 2A and needed a 220pF cap across the feedback resistors to completely kill the slight oscillation that resulted. Different layouts will have different stability implications, so the 1nF suggested in the article is a good place to be on average and should be installed anytime you change any of the overall parameters of the amplifier.
to luvrockin #937
Good morning luvrockin,
nice to see that you didn't give up!
If your new Mosfets are mounted I hope all will be fine.
At first start up I wouldn't bias to the full amount. Stay alittle bit lower for the first few days. You can increase the bias later.
Perhaps a silly comparison: if I repair a motorcycle engine and it is running
the first time, then I don't go on the highway and give it full throttle....
And I wouldn't send 'nasty' test signals through the new amp at high input levels.
Greets
Dirk
p.s.: when I test/listen to a new amp, I have a pair of pretty cheap JBL Control1 speakers. They are running the new amp in. If anything goes wrong...
My ones are real 'survivors'
Good morning luvrockin,
nice to see that you didn't give up!
If your new Mosfets are mounted I hope all will be fine.
At first start up I wouldn't bias to the full amount. Stay alittle bit lower for the first few days. You can increase the bias later.
Perhaps a silly comparison: if I repair a motorcycle engine and it is running
the first time, then I don't go on the highway and give it full throttle....
And I wouldn't send 'nasty' test signals through the new amp at high input levels.
Greets
Dirk
p.s.: when I test/listen to a new amp, I have a pair of pretty cheap JBL Control1 speakers. They are running the new amp in. If anything goes wrong...
My ones are real 'survivors'