I did that to get unity gain off the the differential pair for a 1K base resistor. I was running with the assumption of a 1V input and unity gain so that R1 and R2 only needed to drop 1V from the +rail to avoid clipping. That and I was trying to keep things to nice simple round numbers as a starting point.
Feel free to point out obvious errors, as I am a metallurgical engineer fascinated with an audio hobby trying to do electrical design, learning via the trial-by-fire method.
Feel free to point out obvious errors, as I am a metallurgical engineer fascinated with an audio hobby trying to do electrical design, learning via the trial-by-fire method.
As I already posted, the post I made is not about environment and the like.grataku said:
And don't come at me with power wasting is not that bad. It is bad for the very fact alone that it proves an extremely inefficient design.
BTW, your sh|t talking has anything to do with your avatar or your underpaid situation? Because is is strange, underpaid but happy to pay for class A energy bills?
Metalman, I see that you anticipate me.
The first schematic is now available in pdf at
www.passlabs.com/np/GC-SS-1a.pdf
This was intended to get some response while giving the
geniuses a chance to start filling in their own values.
Last night I filled in the values that I would start with - keep in
mind that none of this has been tested. This is found in
www.passlabs.com/np/GC-SS-1b.pdf
The first schematic is now available in pdf at
www.passlabs.com/np/GC-SS-1a.pdf
This was intended to get some response while giving the
geniuses a chance to start filling in their own values.
Last night I filled in the values that I would start with - keep in
mind that none of this has been tested. This is found in
www.passlabs.com/np/GC-SS-1b.pdf
Nelson Pass said:Metalman, I see that you anticipate me.
The first schematic is now available in pdf at
www.passlabs.com/np/GC-SS-1a.pdf
This was intended to get some response while giving the
geniuses a chance to start filling in their own values.
Last night I filled in the values that I would start with - keep in
mind that none of this has been tested. This is found in
www.passlabs.com/np/GC-SS-1b.pdf
Looks very interesting. I like added zobel network.
--
Brian
Grataku,
No worries, I was just acknowledging you pointing out the obvious. Constructive feedback is always welcome
Nelson,
Not sure whether that means you thought I was anticipating hints from you, or if I actually guessed something right, but I like the schematic. Very interesting. I see some home theatre monoblacks in my future.
No worries, I was just acknowledging you pointing out the obvious. Constructive feedback is always welcome
Nelson,
Not sure whether that means you thought I was anticipating hints from you, or if I actually guessed something right, but I like the schematic. Very interesting. I see some home theatre monoblacks in my future.
The blank sch from Mr.Pass always differs from the one he fills in. There is R17 and C5 , somekind of bootstrapping, or R17 is somekind of folded cascode that ensures the total current in R5 and R6 always the same all the time (no..., that has been ensured by CCS, didn't it?)
Mr.Pass, I impressed by the article of ZEN V-7. In the listening test report, even though you heard ZEN V-7T not for long, you find it more musical and lively. Why don't we use common CT inductances in place of R5 and R6 here?
For advance option, maybe put 2 secondary winding, each connected parrarel to R9, R10, to get galvanic generator for feeding signal to GC. In this advance option, we dont have to use C1-R7 and C2-R8
Mr.Pass, I impressed by the article of ZEN V-7. In the listening test report, even though you heard ZEN V-7T not for long, you find it more musical and lively. Why don't we use common CT inductances in place of R5 and R6 here?
For advance option, maybe put 2 secondary winding, each connected parrarel to R9, R10, to get galvanic generator for feeding signal to GC. In this advance option, we dont have to use C1-R7 and C2-R8
Roibm,
This design is efficient. Chipamps are efficient power amp (just don't compare it with classD, compare it with classA), some said with very good result. So dont worry about energy preservation.
(I also concered about nature abuse. Mankind cannot make crude oil himself, cannot grow trees in a minute, but why we spend those greedily like we can replace them in the next minute? Whats left for your grandson? Nuclear waste? My brother works in Canada, developing fuel-cells. They wanted to make efficient nature friendly power transfer. They produces hydrogen to replace batteries)
But here in diyaudio, I can say that hot mosfets are GOOD... This is an exception design from the master. He turns from hot mosfets (...eh, there is no mosfet at all here).
The schematic that Mr.Pass filled in is having the same gain in the GC (R11-R13) and the one in the differential (R1-R3). Usually the one in the GC should have more than in differential?
This design is efficient. Chipamps are efficient power amp (just don't compare it with classD, compare it with classA), some said with very good result. So dont worry about energy preservation.
(I also concered about nature abuse. Mankind cannot make crude oil himself, cannot grow trees in a minute, but why we spend those greedily like we can replace them in the next minute? Whats left for your grandson? Nuclear waste? My brother works in Canada, developing fuel-cells. They wanted to make efficient nature friendly power transfer. They produces hydrogen to replace batteries)
But here in diyaudio, I can say that hot mosfets are GOOD... This is an exception design from the master. He turns from hot mosfets (...eh, there is no mosfet at all here).
The schematic that Mr.Pass filled in is having the same gain in the GC (R11-R13) and the one in the differential (R1-R3). Usually the one in the GC should have more than in differential?
nowater said:
None. Keep in mind that these are starting point values.
He who builds it first will probably discover others that give
better distortion / sound / or are more frequency stable.
Also keep in mind that this may not be all the parts for an
optimized circuit. For example, in 1b we added output RC
networks to help with frequency stability. There may be other
compensation elements needed before it's all done.
Who knows? Sometimes a good looking paper design is
crappy in real life.
One of the key issues is the enormous open loop gain of the
chips, which can give rise to some difficulty. There are ways
of throwing some of that open loop away, and I'll post a
couple of general examples for chips tomorrow, when I have
working ftp again.
Here are some images of a quick layout that I made last night for this circuit. I will post the gerber files in an attachment following this post if anyone wants to try it out.
I used the J510 for the current source, which roughly gives 4mA. The values in the schematic aren't properly labeled, since I ran no simulations for this circuit. Does anyone have a spice model for the LM3875 to run simulations?
The layout is a bit unconventional, since I mounted one of the LM3875 to the bottom of the pcb and a few other components to the bottom of the pcb to achieve better symmetry, but I am fairly happy with the result. The board size is 2.7" wide and 2.1" deep. I made it for using the compact 0.5w phoenix SFR16S resistors from Digikey, or Caddock MK132 resistors (the resistors I am using for my other boards). For the 10uF capacitor, I am planning on trying a 10uF BlackGate non-polar capacitor, which should work fine. I initially had the 10uF panasonic polyester caps, but they were huge. The zobel components are 3w panasonic resistors, and BC polyprop. or Panasonic polyester cap (5mm or 7.5mm spacing).
I have no intentions of selling kits with these boards, as they use Nelson's patent, so please don't ask. The board files are free to use for non-commercial use.
Keep in mind that I just started the layout last night, and only spend a few hours on it, so it might still have issues that need to be addressed. Any comments are appreciated.
--
Brian
I used the J510 for the current source, which roughly gives 4mA. The values in the schematic aren't properly labeled, since I ran no simulations for this circuit. Does anyone have a spice model for the LM3875 to run simulations?
The layout is a bit unconventional, since I mounted one of the LM3875 to the bottom of the pcb and a few other components to the bottom of the pcb to achieve better symmetry, but I am fairly happy with the result. The board size is 2.7" wide and 2.1" deep. I made it for using the compact 0.5w phoenix SFR16S resistors from Digikey, or Caddock MK132 resistors (the resistors I am using for my other boards). For the 10uF capacitor, I am planning on trying a 10uF BlackGate non-polar capacitor, which should work fine. I initially had the 10uF panasonic polyester caps, but they were huge. The zobel components are 3w panasonic resistors, and BC polyprop. or Panasonic polyester cap (5mm or 7.5mm spacing).
I have no intentions of selling kits with these boards, as they use Nelson's patent, so please don't ask. The board files are free to use for non-commercial use.
Keep in mind that I just started the layout last night, and only spend a few hours on it, so it might still have issues that need to be addressed. Any comments are appreciated.
--
Brian
Here are the gerber files for the design (along with NC drill file)
Keep in mind that nothing has been tested yet. I am planning on making regulated power supply board with a bigger capacitance bank along with the regulators, so don't assume that those 2 small caps are all that is needed.
A few more things:
- I am wondering if the LM3886 would be better, since it is better with lower impedence loads.
- If I was to add an output resistor on the output of the LM3875, parallel 2 of them off each leg to get more power out of this circuit?
--
Brian
Keep in mind that nothing has been tested yet. I am planning on making regulated power supply board with a bigger capacitance bank along with the regulators, so don't assume that those 2 small caps are all that is needed.
A few more things:
- I am wondering if the LM3886 would be better, since it is better with lower impedence loads.
- If I was to add an output resistor on the output of the LM3875, parallel 2 of them off each leg to get more power out of this circuit?
--
Brian
I have a couple more for you today, Brian. I believe they
will address some of these issues.
I don't personally know that the 3886 is better sounding than
the 3875, but I would would bet that it is based on the data
sheets: Lower open loop gain, higher bias, smoother spectral
distortion plots. I'll let you know when I've had the time to
play with them properly.
will address some of these issues.
I don't personally know that the 3886 is better sounding than
the 3875, but I would would bet that it is based on the data
sheets: Lower open loop gain, higher bias, smoother spectral
distortion plots. I'll let you know when I've had the time to
play with them properly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.