I have a couple of questions... I hope someone can help, as I really have no idea about this first one. The second two I can probably muddle through.
There are two pairs of output boards per channel. In each channel an N pair is wired in parallel and a P pair is wired in parallel.
They are NOT wired series as suggested by the schematic. The schematic shows 1 thermistor in the bias circuit per each N/P polarity.
When the two N boards for a single channel are wired in parallel, should there be a biasing thermistor for each board? That would be two per polarity... The build guide shows a V2, but this has twice the boards.... I'm really confused about this.
What value of rail fuse would you use on a V3 that is biased to 80 watts Class A peak, using 4 pairs of transistors per channel? Fast/Slo blow?
Ok, so that's really three questions.
Thanks, and have a great day!
There are two pairs of output boards per channel. In each channel an N pair is wired in parallel and a P pair is wired in parallel.
They are NOT wired series as suggested by the schematic. The schematic shows 1 thermistor in the bias circuit per each N/P polarity.
When the two N boards for a single channel are wired in parallel, should there be a biasing thermistor for each board? That would be two per polarity... The build guide shows a V2, but this has twice the boards.... I'm really confused about this.
What value of rail fuse would you use on a V3 that is biased to 80 watts Class A peak, using 4 pairs of transistors per channel? Fast/Slo blow?
Ok, so that's really three questions.
Thanks, and have a great day!
That's correct, the N and P are not wired in series. The N boards get negative rail, ground and the -drive signal (to the gates), the P boards get positive rail, ground and the +drive signal. If more than one sets of boards then each polarity IS wired in series.
If more than one set of boards are used you only want one thermistor per polarity (one for Ns and one for Ps), best to put it on the MOSFET closest to the middle. You would also use only one R11 & R12 (2.2k) per set also, preferably on the same board that has the thermistor. Same goes for the comp cap (c3 & c4). Only one needed, leave the other empty.
I would advise against using fuses on the rails. If one rail fuse blows it will put the full voltage of the other rail onto the output. That would most likely blow any speakers attached to the output immediately. If you meant the AC input fuse before the TX then it would depend on the voltage/current your amp is capable of. F5Tv3 @ 45v, biased normally, maybe 3-4A slow blow. (Guessing) You want to use the smallest fuse possible that doesn't blow when turning on or running full bore. It's been discussed earlier in the thread. It may also depend on whether you're using a soft start or thermistors (eg.: CL-60) in the AC line. I personally use a 4A slow blow in my F5Tv3 mono blocks @ +/- 47v with a soft start. Because there are so many different versions and options possible there's no set value to use. It depends on your amp's needs and will require some experimenting. Start small and keep increasing until it doesn't blow under normal use.
TJ
Last edited:
Thanks Takitaj!
I should have looked closer at the schematic. To me, it looks like parallel/series does not matter.
The V3 F5T Front end boards have two gate drive outputs per polarity -- these two points are at the same potential, so the bias circuit on from one board will affect the other. With parallel wiring you can use thinner wire for power and output than the series connection.
When using the diyAudio speaker protection circuit, losing one rail fuse should trip the protection on excessive DC offset -- it trips at 1 Volt I believe.
I knew when I started that am very bad at working with electricity. Proof: I've blown this up 3 times with unthinkable mistakes. Thankfully, the diyaudio speaker protection circuit has worked every time so the speakers have never been damaged.
Replacing output MOSFETs is not just time consuming, it is getting expensive! I'd rather be replacing fuses.
I should have looked closer at the schematic. To me, it looks like parallel/series does not matter.
The V3 F5T Front end boards have two gate drive outputs per polarity -- these two points are at the same potential, so the bias circuit on from one board will affect the other. With parallel wiring you can use thinner wire for power and output than the series connection.
When using the diyAudio speaker protection circuit, losing one rail fuse should trip the protection on excessive DC offset -- it trips at 1 Volt I believe.
I knew when I started that am very bad at working with electricity. Proof: I've blown this up 3 times with unthinkable mistakes. Thankfully, the diyaudio speaker protection circuit has worked every time so the speakers have never been damaged.
Replacing output MOSFETs is not just time consuming, it is getting expensive! I'd rather be replacing fuses.
I understand about replacing the MOSFETs. I've had to do it several times for different reasons (all my fault). The Modushop chassis' are nice but you do have to disassemble practically the whole thing to get at them.
Yes, there are 2 pads per polarity for the drive. They're there for flexibility I guess. You can ignore the extra ones or do as I did and put spade connectors there. I put spades on all connections to make it a little easier to take apart if (when) needed.
My protect boards (DIYA) tripped at about 2.5v so I modded it to trip a little earlier (~1.5v) and to increase delay time to allow the amp to settle a little bit more. Same with soft start. I swapped out the resistors for 3 CL-60s in series. That way if the relay fails it won't fry the resistors. The longer delay allows a nice slow soft start for the 240,000 uf I have in there. Not necessary but it's DIY after all.
TJ
Yes, there are 2 pads per polarity for the drive. They're there for flexibility I guess. You can ignore the extra ones or do as I did and put spade connectors there. I put spades on all connections to make it a little easier to take apart if (when) needed.
My protect boards (DIYA) tripped at about 2.5v so I modded it to trip a little earlier (~1.5v) and to increase delay time to allow the amp to settle a little bit more. Same with soft start. I swapped out the resistors for 3 CL-60s in series. That way if the relay fails it won't fry the resistors. The longer delay allows a nice slow soft start for the 240,000 uf I have in there. Not necessary but it's DIY after all.
TJ
I built my own chassis out of heatsinks from Heatsink USA and aluminum plate/angles. Over 150 holes drilled and tapped. It took a whole weekend and then some.
With parallel wiring, I can take out one heatsink and work on one N/P pair at a time. With serial wiring, there is much more disassembly -- serial wiring using house wire was the first attempt.
This is a great idea.
My protect boards (DIYA) tripped at about 2.5v so I modded it to trip a little earlier (~1.5v) and to increase delay time to allow the amp to settle a little bit more. Same with soft start. I swapped out the resistors for 3 CL-60s in series. That way if the relay fails it won't fry the resistors. The longer delay allows a nice slow soft start for the 240,000 uf I have in there. Not necessary but it's DIY after all.
TJ
Nice idea on the soft start!
I put in a relay in series with a thermistor and bypass relay on the secondary. I was concerned that the thermistor would get hot and if the amp was power cycled would not work as expected.
The job of the first relay is to allow power to flow through the next part of the circuit consisting of thermistor/bypass relay and cap bank.
In the event that this is a quick power cycle, like a power outage and restart scenario, the thermistor could be hot while the cap bank was empty. There is a 1 minute power up delay on the first relay to allow the thermistor to cool to working temperature.
The LT spice simulation suggested that the current being passed on relay close was over 15 amps until 1 minute had passed. ( Not sure if that is correct, but I need to be cautious -- if there is a way to break something, I will find it. )
At that point the secondaries are connected to the diode bridge without any thermistors in the way.
Probably overkill, but I have not broken that bit .... yet.
Also, I have used the quasimodo circuit from Mark Johnson to select capacitors/resistors to snubberize the transformer. This showed a significant change on the scope. I have no idea if it sounds better or worse in real life.
I just order them from Mouser 10 or 20 at a time and match them. Most times 10 is enough to get at least a matched quad from. They're usually pretty close. Matching is fairly easy, even easier if you throw together a jig and use a 12 or 15v wall wart. I use a 12v regulated supply and match close to operating current/temp. There's lots of info on it. Search this site for "matching MOSFETs" or "MOSFET matching".
Your soft start sounds way more complicated than it needs to be.
TJ
Your soft start sounds way more complicated than it needs to be.
TJ
6L6 - Thank you for the offer of help, I never would have thought about this project before I joined diy.
I plan on building the power supplies separately and using umbilical cords for the 48vdc rails. I just ordered 2 Pesante 3U chassis for the power supplies.
I watched all the BA 2018 videos, I will be sure to take lots of pics as I go along.
I plan on building the power supplies separately and using umbilical cords for the 48vdc rails. I just ordered 2 Pesante 3U chassis for the power supplies.
I watched all the BA 2018 videos, I will be sure to take lots of pics as I go along.
With the size of the power supplies that people are using, it seems that you can *always* overdrive your amp until it overheats, the diodes conduct and the mosfets blow....
My thinking is that you should be able to choose a fuse on the speaker output that will blow when then amp is pushing too much current for the heatsinks to dissipate.
Does that make sense?
My thinking is that you should be able to choose a fuse on the speaker output that will blow when then amp is pushing too much current for the heatsinks to dissipate.
Does that make sense?
Nope. The heatsinks are always hot, the current is always there, it's class-A which means the power is either going to the heatsinks or the load.
And when the diodes conduct, in this circuit, it's just during musical peaks so the voltage has somewhere to go instead of getting burnt up in the source resistors, letting more voltage swing into the load, and ultimately more power.
You just don't want the standing current (bias) so high that they conduct at idle.
And when the diodes conduct, in this circuit, it's just during musical peaks so the voltage has somewhere to go instead of getting burnt up in the source resistors, letting more voltage swing into the load, and ultimately more power.
You just don't want the standing current (bias) so high that they conduct at idle.
I found out (quite unintentionally) that the F5 does not put rail voltage on the output in case of one missing rail. There is a momentary shift in operating point, after which the output settles back at zero volts.
You can try this with a dummy load. Weak resistors in the CRC are the same thing as fuses. In case of short circuit they act as a fuse (also discovered by accident). No harm befell any of the speakers.
Input JFETs dying due to improper operation, not cascoding, etc - that will kill the speakers for sure and have done so (this was someone else's discovery, thankfully).
I'm assuming that you've biased it so that it won't overheat at idle.
It will pass more current when it switches to Class B operation, which then has to be dissipated. If it cannot do this, then it will overheat with catestrophic results.
So, if my heatsinks are capable of dissipating 100 watts at idle, which may be at 2.5 amps, then would it not make sense to fuse the outputs so that when passing 3 amps, the fuse blows?
Is the strap on the power supply board absolutely necessary? I say this because aren't the grounds from the power supply board connected to the zero volt line already? Zero volt line meaning the diode bridge/thermistor ground loop breaker.