You probably want to bump it up to 18 VAC. With 15 VAC if you use a standard canned bridge you'll get ~15*1.4 - 1.2V rectifier loss = 19.8V at the CRC input, or 17V rails. That's pushing the lower limit. If you use Patrick's rectifier recommendations you'll lose another volt or two. Go to 18 VAC and you'll end up with 18.5 - 20V rails depending on rectifier choice, a better operating point for the output mosfets.
Be sure to provide heat sinking for the rectifiers and power resistors. The resistors can take high temperatures, but will be dissipating close to 12W. At a minimum they should be bolted to a thick case panel.
Be sure to provide heat sinking for the rectifiers and power resistors. The resistors can take high temperatures, but will be dissipating close to 12W. At a minimum they should be bolted to a thick case panel.
Bob,
I think Patrick was shooting for 16V rails in his design. I'm trying to follow that lead.
BTW, I'm sure this resides somewhere in the F5 thread (a search is not revealing), but what are NPs 2.2K resistors for on his F5 PS? I'm beginning to understand the CRC Pi filter, but the 2.2K resistors don't seem to fit the filter.
I think Patrick was shooting for 16V rails in his design. I'm trying to follow that lead.
BTW, I'm sure this resides somewhere in the F5 thread (a search is not revealing), but what are NPs 2.2K resistors for on his F5 PS? I'm beginning to understand the CRC Pi filter, but the 2.2K resistors don't seem to fit the filter.
Last edited:
The transformer in post #587 :
http://www.diyaudio.com/forums/pass-labs/172770-balanced-f5-question-12.html#post2371245
will apply equally to regulators using 2SK3497 / 2SJ618 and CRC using 0R68.
Using STTH2002 diodes and one full bridge per secondary, you will get 19.6V output at full load after the rectifiers.
Drop another 2.7V over the 0R68, you still get almost 17V.
So you can even go to 0R9 before you have to worry about insufficient voltage.
And even then you can start considering Schottky diodes like MBR40250.
And the circuit does not cease to work at +/-15V rails.
It just have less headroom for impedance variations of your speaker.
Patrick
.
http://www.diyaudio.com/forums/pass-labs/172770-balanced-f5-question-12.html#post2371245
will apply equally to regulators using 2SK3497 / 2SJ618 and CRC using 0R68.
Using STTH2002 diodes and one full bridge per secondary, you will get 19.6V output at full load after the rectifiers.
Drop another 2.7V over the 0R68, you still get almost 17V.
So you can even go to 0R9 before you have to worry about insufficient voltage.
And even then you can start considering Schottky diodes like MBR40250.
And the circuit does not cease to work at +/-15V rails.
It just have less headroom for impedance variations of your speaker.
Patrick
.
Last edited:
To use CRC at R=0R68, you should use at least a resistor rated at 40W.
The easiest solution is to use 4x 10W 0R18 vertical ceramic wire wound resistors in series, e.g. :
http://www.vitrohm.com/download/series/15/
As long as you have proper thermal insulation from the PCB, they will do fine at 30% of the rating.
You want to connect in series and not in parallel in this case as inductance help to provide extra filtering.
Patrick
The easiest solution is to use 4x 10W 0R18 vertical ceramic wire wound resistors in series, e.g. :
http://www.vitrohm.com/download/series/15/
As long as you have proper thermal insulation from the PCB, they will do fine at 30% of the rating.
You want to connect in series and not in parallel in this case as inductance help to provide extra filtering.
Patrick
My concern was not that the circuit would cease to work, but that the output devices become more nonlinear as Vds drops. I cannot provide a link, but I remember Papa it would be better to keep the rails closer to 20V+ for IRFP240.
That depends on what is the output level, which is halved in voltage terms per side in a balanced circuit. So if play your music at say 6V pk to pk, which for me is already every loud with 4 ohm speakers (4.5W rms), the MOSFET still has (16-6/2=) 13V Vds, which is quite OK in terms of "pentode" region & non-linear capacitance.
Want more voltage ? Need to add more FETs and reduce bias on each FET. Then you increase the capacitance, and change the characteristics of the amplifier (maybe for the better, maybe not).
Patrick
Want more voltage ? Need to add more FETs and reduce bias on each FET. Then you increase the capacitance, and change the characteristics of the amplifier (maybe for the better, maybe not).
Patrick
Beside the voltage drop, is there a better Pi filter outcome to 0R68 vs. a lower value? I've never done the Pi filter calculations and looking at them, I'm poorly equiped to know the "best" outcome.
In other words, is the 0R68 a choice because it gets you the "right" voltage drop, or the best Pi filter?
In other words, is the 0R68 a choice because it gets you the "right" voltage drop, or the best Pi filter?
You can split the R into 2x 0R33 and use CRCRC at 10Hz. This will then become a second order filter with approx. 25x attenuation at 50Hz.
This compares to 10x attenuation for first order using 0R68 for CRC, but cost more space (and money) because of extra caps required.
If you want to drop more voltage, you can use CRCRCRCRC....... No fixed rules.
Patrick
This compares to 10x attenuation for first order using 0R68 for CRC, but cost more space (and money) because of extra caps required.
If you want to drop more voltage, you can use CRCRCRCRC....... No fixed rules.
Patrick
When all the layout work is finished, we shall first order a small batch of PCB to test and debug (by at least 2 members of the team).
Only when that work is completed, we shall publish all the schematics and bills of material as used in the tests.
Realistically around Easter 2011, I think. I still have not got all the PCBs in Gerber as yet.
People are busy around Christmas, and this is voluntary hobby work.
So I can only beg for patience and understanding.
Patrick
Only when that work is completed, we shall publish all the schematics and bills of material as used in the tests.
Realistically around Easter 2011, I think. I still have not got all the PCBs in Gerber as yet.
People are busy around Christmas, and this is voluntary hobby work.
So I can only beg for patience and understanding.
Patrick
I have not been able to find the CRC equation for a Pi filter. I found a few for CLC. Can you point me in the right direction for CRC calculations? Some have recommended Duncan's PS program, but I was too inexperienced to understand the different options.
I'm not sure what the right answer is. I was going to use an Antek 500VA 2x15V toroid for each channel.
It puts out 14.6 under load which goes to 20.6V nominal. With MUR1560s I lose ~1V, with the Pi filter, another 2.72V gets me to 16.8V - a little higher than desired.
I'm not sure what the right answer is. I was going to use an Antek 500VA 2x15V toroid for each channel.
It puts out 14.6 under load which goes to 20.6V nominal. With MUR1560s I lose ~1V, with the Pi filter, another 2.72V gets me to 16.8V - a little higher than desired.
In Rush Limiters
NP uses the CL-60 for inrush limiting on the F5. In looking at the datasheet, the CL-60 has a maximum steady state current rating of 4A.
I'm interested in keeping the thermistor as part of the build. Would the CL-30 be more appropriate? The Max Steady State goes up to 8A, but the R70 drops to 2.5ohms from 10 ohms of the CL-60.
NP uses the CL-60 for inrush limiting on the F5. In looking at the datasheet, the CL-60 has a maximum steady state current rating of 4A.
I'm interested in keeping the thermistor as part of the build. Would the CL-30 be more appropriate? The Max Steady State goes up to 8A, but the R70 drops to 2.5ohms from 10 ohms of the CL-60.
> but the R70 drops to 2.5ohms from 10 ohms of the CL-60.
So what's the problem ?
Using 15V secondaries, the rectified voltage is about 20V peak after rectifiers. So your peak current is 8A for 2.5R. The transformer is 500VA for 2x15V, so max output current is about 12A DC. When the NTC is hot, the ripple current under normal operation is a lot higher.
There are many threads and many websites dealing with the subject of using NTC for inrush current. There is also a section on the subject in Bob Cordell's book.
Patrick
So what's the problem ?
Using 15V secondaries, the rectified voltage is about 20V peak after rectifiers. So your peak current is 8A for 2.5R. The transformer is 500VA for 2x15V, so max output current is about 12A DC. When the NTC is hot, the ripple current under normal operation is a lot higher.
There are many threads and many websites dealing with the subject of using NTC for inrush current. There is also a section on the subject in Bob Cordell's book.
Patrick
I guess the reason I keep checking with the thread is this is the first time I've done this. Yes, the formulas and issues are trival, but the money involved (for me) is not. As such, I like to get confirmation of my first time assessments. I kind of thought that's what this website was about.
I am not that far as yet, but I probably will use a single NTC at the primary side of the toroid to avoid tripping the mains fuse box.
I do not use GE NTCs but use RTI (just availability), and I follow their recommended procedure to the word :
Surge-Gard Circuit Protection Devices - NTC Thermistors - Inrush Current Limiting Devices
And an alternative one from EPCOS :
http://www.epcos.com/web/generator/...ta__en.pdf;/PDF_InrushCurrentLimiting_AN2.pdf
If you could wait, then we shall eventually publish a complete solution built, tested and documented by at least two members of the team. If you want to be able to make your own choices, and there are many solutions to inrush current, I can again only suggest understanding the subject more by reading books and technical documents from manufacturers. No one else can have more expertise than those who design, manufacture and sell those products.
Hope this is a satisfactory answer for you.
Regards,
Patrick
.
I do not use GE NTCs but use RTI (just availability), and I follow their recommended procedure to the word :
Surge-Gard Circuit Protection Devices - NTC Thermistors - Inrush Current Limiting Devices
And an alternative one from EPCOS :
http://www.epcos.com/web/generator/...ta__en.pdf;/PDF_InrushCurrentLimiting_AN2.pdf
If you could wait, then we shall eventually publish a complete solution built, tested and documented by at least two members of the team. If you want to be able to make your own choices, and there are many solutions to inrush current, I can again only suggest understanding the subject more by reading books and technical documents from manufacturers. No one else can have more expertise than those who design, manufacture and sell those products.
Hope this is a satisfactory answer for you.
Regards,
Patrick
.
Last edited:
Sorry, guys newb question here.
Can someone explain why there is matching fets etc, as well as circuit redesign for a balanced f5?
From my limited understanding, wouldn't making the f5 balanced be a simple matter of jumping wires on two f5 boards?
I'm referring to the explanation Mr.Pass gives on the F4 manual pg 7 where he explains balanced mono, and parallel mono.
Also doesn't this also quadruple the power output of the f5?
Sorry my electronics is quite rusty and I need some caveman explanation.
Can someone explain why there is matching fets etc, as well as circuit redesign for a balanced f5?
From my limited understanding, wouldn't making the f5 balanced be a simple matter of jumping wires on two f5 boards?
I'm referring to the explanation Mr.Pass gives on the F4 manual pg 7 where he explains balanced mono, and parallel mono.
Also doesn't this also quadruple the power output of the f5?
Sorry my electronics is quite rusty and I need some caveman explanation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.