I was looking at the parallel connection of the MOSFET's at the output and was wondering about the exact same thing, while considering Nelson's patent which shows a single MOSFET in the output stage. So, yes, the calculated AC gain for paralleled MOSFET's should probably be 2 X 20.83% which is close to what Nelson's says is the real-life AC gain.
The bit that really confused me is that, according to the patent, the transition from 4 MOSFET's in the output, to 5 MOSFET's in the output, would bring the AC gain of more than 1 (more than 100%).
The bit that really confused me is that, according to the patent, the transition from 4 MOSFET's in the output, to 5 MOSFET's in the output, would bring the AC gain of more than 1 (more than 100%).
Yes.
How SURE are you about this? When I first started looking at the Aleph, I thought these resistors are in between the amplifier output and the speakers, and I thought they would reduce the damping factor. The longer I looked, the more I understood that this is not really the case. The resistors are the output sensing resistors for the Aleph current source. The current source compensates for the "damping that would be lost" in these resistors.
If you replace the current sensing resistors by a wire link, the AC sensing will not sense anything from the output, so it will not convey any AC signal to the current source. You can therefore remove the RC link to the base of the BJT transistor in the Aleph current source. You'll end up with a constant current source. Just take a look at the Zen amps that preceded the Aleph. Also take look at how the Zen (constant current source) and the Aleph (AC variable current source) compare to each other: http://www.firstwatt.com/pdf/art_zv2.pdf
FERD observations
So, I’ve lived with the FERD diodes in my new discrete boards for a few days now and have to say I’m really impressed. Unfortunately, this is also my first experience with a discrete bridge, having used the monolith rectifier bridges until now. So I’m not sure if the improved bottom end, separation, and control are the discrete setup or the FERDs - or a mixture of both. Either way, my AJ has never sounded so good, which I’m sure can also be attributed to the dual mono arrangement. One thing I can say though - the FERDs are surprisingly cool, and my rail voltage increased by 1 volt over the monolith bridges.
Thanks Mark Johnson for the heads up on these!
So, I’ve lived with the FERD diodes in my new discrete boards for a few days now and have to say I’m really impressed. Unfortunately, this is also my first experience with a discrete bridge, having used the monolith rectifier bridges until now. So I’m not sure if the improved bottom end, separation, and control are the discrete setup or the FERDs - or a mixture of both. Either way, my AJ has never sounded so good, which I’m sure can also be attributed to the dual mono arrangement. One thing I can say though - the FERDs are surprisingly cool, and my rail voltage increased by 1 volt over the monolith bridges.
Thanks Mark Johnson for the heads up on these!
^ Yes, I have some of those to try. First amp be my Zen V4, which currently has a set of FEP30 diodes. The old Hafler transformers I'm using are center tapped, so I'm using half of the rectifier PCBs that come with the store PSU boards. The Singing Bush may also get these, if the Zen V4 goes well. I have another set of the 20 Amp FERD diodes to try with a filament power supply for a tube-based preamp which may front my MoFo
My Aleph J has two sets of LVB2560 bridge rectifiers. It is slated to eventually get a set of SMT LT4320 based rectifiers sometime down the road.
My Aleph J has two sets of LVB2560 bridge rectifiers. It is slated to eventually get a set of SMT LT4320 based rectifiers sometime down the road.
Last edited:
It's almost Biblical: Good things come to those who ... read datasheets .
DigiKey parametric search too. Filter out the scrawny parts, then sort on the column "Max Forward Voltage". Aaaah. You'll find that Mouser stocks about 99% of the diodes that DigiKey indexes. So it's not a waste of time to use DigiKey's site to help you shop at Mouser.
_
DigiKey parametric search too. Filter out the scrawny parts, then sort on the column "Max Forward Voltage". Aaaah. You'll find that Mouser stocks about 99% of the diodes that DigiKey indexes. So it's not a waste of time to use DigiKey's site to help you shop at Mouser.
_
Last edited:
Matching the MOSFETs
The last time I went through this exercise it took over 50x MOSFETs to get enough close matches.
I've just bought a tube of 25x from Farnell and the first 5x are all within 1.0% of each other, two pairs within 0.3% of each other.
Has the manufacturing process improved to the extent that they are now much closer when made or have I just been exceptionally lucky ?
The last time I went through this exercise it took over 50x MOSFETs to get enough close matches.
I've just bought a tube of 25x from Farnell and the first 5x are all within 1.0% of each other, two pairs within 0.3% of each other.
Has the manufacturing process improved to the extent that they are now much closer when made or have I just been exceptionally lucky ?
If you order in quantity of 25, that’s almost always going to be a ‘stick’ of TO-247 devices, and also very likely from the same production batch, and the same wafer. The N- devices will be really close in match , and the P-devices will be looser, but still pretty darn good, as the P- devices have more manufacturing variances.
Pass DIY Addict
Joined 2000
Paid Member
I am out of luck with 2K trimpots for R7, the last value in stock I have is 5K. I have used 5K as replacement in most of my circuits without much of hestation, but I am bit concerned here. I also notice there is a parallel 1K as R7 in the circuit which I don't see in many circuits built on this forum. Technically the trimpot would reduce R7 to a fairly low value if I go by the circuit daigram...wanted to check if I could go ahead with the 5K trimpot with 1K resistor is parallel.
This confused me for a while, too, but I think it's there as an artifact of the PCB design process. The PCB permits you to use either a trimpot or a fixed resistor, but to allow both of those components to be used in the exact same place in the circuit, the PCB software treats them as parallel, even though only one is supposed to be installed at a time.