Hi Shaan, a question:
I am interested in separating the power rails into just FET's and rest of circuit. I think I can do this without much effort. However, I have two 30-0-30 volt toroids, 160VA, which I want to use as the 'rest of the circuit' and buy some new toroids with a liitle more current for the output FET's. But my FET's are limited to just 8 amps each, these are the ALF of the original VSSA design, which I bought at the time. They are unused. Would the sensible thing to do be to use a 300VA toroid at 25-0-25v (35v rails) at a high current or 30-0-30v (42v rails) at a slightly lower current?
Please note, I am using two amplifiers to feed one loudspeaker (lower mid range to super tweeters, not bass <100Hz)) so the total power output will not be more than 100 watts into 4 Ohms for the two amplifiers, I guess. So each 'channel' will contain two amplifier boards, one 160VA toroid, one 300 VA toroid (and a 12V toroid for switch on and output monitoring).
ps I have used a polypropylene capacitor across the inputs to the FETs as I think polyprop sounds much better than polyesters, but I haven't finished the amplifiers to test this, yet!
I am interested in separating the power rails into just FET's and rest of circuit. I think I can do this without much effort. However, I have two 30-0-30 volt toroids, 160VA, which I want to use as the 'rest of the circuit' and buy some new toroids with a liitle more current for the output FET's. But my FET's are limited to just 8 amps each, these are the ALF of the original VSSA design, which I bought at the time. They are unused. Would the sensible thing to do be to use a 300VA toroid at 25-0-25v (35v rails) at a high current or 30-0-30v (42v rails) at a slightly lower current?
Please note, I am using two amplifiers to feed one loudspeaker (lower mid range to super tweeters, not bass <100Hz)) so the total power output will not be more than 100 watts into 4 Ohms for the two amplifiers, I guess. So each 'channel' will contain two amplifier boards, one 160VA toroid, one 300 VA toroid (and a 12V toroid for switch on and output monitoring).
ps I have used a polypropylene capacitor across the inputs to the FETs as I think polyprop sounds much better than polyesters, but I haven't finished the amplifiers to test this, yet!
Last edited:
You can run the MOSFETs from 35V while the previous stages run from 42V (with maximum power available being 15-20W more than if running the whole amp from 35V alone i.e. about 70W/8R with the option to run it into 4R load if needed). So I'd suggest a 25-0-25 300VA toroid.
It's gonna get loud.
Yes a better quality capacitor is always welcome anywhere. However, this capacitor (C2) doesn't affect the sonics as much as C3 and C4. So if space allows make these polypropylene too, or if space doesn't, the best polyester you can get.
thanks, Shaan. Polyprop 100nF ordered for all those on board. Toroids to get!
As regards loudness. I have pondered this, and have come to the conclusion that two amplifiers covering different parts of the audio spectrum won't sound any louder than one covering all of it. I'm doing it to ease the burden, especially as I have a low impedance for the tweeters (a soft dome and a true ribbon above 4.5kHz) in parallel! Impedance may dip below 4 Ohms, but as classical music slopes off quite rapidly above 1kHz, I think I should be OK.
Again, many thanks.
As regards loudness. I have pondered this, and have come to the conclusion that two amplifiers covering different parts of the audio spectrum won't sound any louder than one covering all of it. I'm doing it to ease the burden, especially as I have a low impedance for the tweeters (a soft dome and a true ribbon above 4.5kHz) in parallel! Impedance may dip below 4 Ohms, but as classical music slopes off quite rapidly above 1kHz, I think I should be OK.
Again, many thanks.
C3 to C6 are feedback caps. They provide a path for AC signal to be attenuated accordingly and reach the error correction node, But their own non-linearities are not corrected by an amplifier, instead they are amplified. So the best quality (i.e. most linear) components should be used in here.
A capacitor being marked "for audio" doesn't automatically make it superior. All the test THD measurements of V4 were done with industrial grade capacitors and the sonics never seemed to complain.
Really good news Shaan, it looks like my Mosfets survived. I managed to adjust both channels on DMM, next is testing with cheap speakers.
I noticed the total consumption (aka bias) is increasing bit by bit in 10-15 minutes, when does it stopes?
How long i need to wait before i can say my bias is stable?
Thanks alot for your help!
PS. Practical advice: for me it was much easier to use two DMM's in order to achieve VAS biasing and 0mV DC out in the same time. So one DMM on R29/30 and another one on SPK out and target 450mV on one and 0mv one the other one
I noticed the total consumption (aka bias) is increasing bit by bit in 10-15 minutes, when does it stopes?
How long i need to wait before i can say my bias is stable?
Thanks alot for your help!
PS. Practical advice: for me it was much easier to use two DMM's in order to achieve VAS biasing and 0mV DC out in the same time. So one DMM on R29/30 and another one on SPK out and target 450mV on one and 0mv one the other one
Last edited:
It depends a bit on cooling of the VAS+MOSFET stage and thermal coefficient of the MOSFETs at a particular bias.
Usually the process of optimum biasing is to let the amp warm up for half an hour and then re-adjusting the bias to normal value. After cooling when you start the amp again it will start with lower than set bias and after half hour of warming up (or less time if loaded) it will reach the previously set bias.
Yup two is better than one.
J1,J2 open (during VAS trimming) = VAS current limited to 15mA so no chance of burning up Q9/Q10 while trimming VR1/VR2.
J1,J2 closed (after trimming is done) = VAS overcurrent protection off.
It would seem unnecessary to close J1,J2 since an always on protection seems better than one that's not. But this protection pulls down slew rate from 100V/uS to about 1V/uS. (very very slow for a CFB amplifier
, 1960's LTP based amps were faster than that!)So signal performance is hampered when VAS protection is on (but we don't need to play music during trimming). After VAS bias is set properly to ~10mA it doesn't need the protection anymore until any future trimming if necessary. That's why we close J1/J2, turning off Q5,Q6 and letting the amp run at its full speed.
Dual CRC psu's for the V4 boards stacked one over the other
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
PSU Sizing
I am starting to size up the PSU requirements for a stereo build and, from comments here and in the main design thread, I think I have arrived at the following specs. I would appreciate for some confirmation of what I'm thinking below and that I'm not hanging my hat too much on Ohm's law where it does not apply
At 35VDC into 8R, there will be 35/8 = 4.375A per channel of current at a full voltage output. This lines up with other recommendations of 150VA per channel transformer recommendations: 4.375 * 25VAC = 110VA.
Using these numbers, I am hoping to use a 300VA, 28VAC transformer in conjunction with Prasi's CRC PSU board to power a pair of PeeCeeBees. The specific transformer specs I plan on using are at: http://www.antekinc.com/content/AS-3428.pdf. This shows 26.6VAC output at a load of 10.7A, close to what I'll be targeting.
Given this, 26.6*1.414 = 37.6VDC rectified output, plus a diode voltage drop of 2*1.25V = 2.5V (using a MUR1540 as an example), gives me a total rectified voltage of 35.1VDC.
The next voltage drop would be from the resistor in the RC filter. In calculating this component's specs, I am looking at voltage drop and power rating for the resistor. At 4.375A across a 0.2R resistor, I would get a voltage drop of 4.375*0.2 or 0.875V and a required power rating of 4.375* 0.875 or about 3.8W. I could then plan on using a 5W+ resistor at 0.2R and would arrive at a final voltage (under full load) to the PeeCeeBee of 35.1-0.875 = 34.23V. I am assuming this would be sufficient for the PeeCeeBee?
I am starting to size up the PSU requirements for a stereo build and, from comments here and in the main design thread, I think I have arrived at the following specs. I would appreciate for some confirmation of what I'm thinking below and that I'm not hanging my hat too much on Ohm's law where it does not apply
At 35VDC into 8R, there will be 35/8 = 4.375A per channel of current at a full voltage output. This lines up with other recommendations of 150VA per channel transformer recommendations: 4.375 * 25VAC = 110VA.
Using these numbers, I am hoping to use a 300VA, 28VAC transformer in conjunction with Prasi's CRC PSU board to power a pair of PeeCeeBees. The specific transformer specs I plan on using are at: http://www.antekinc.com/content/AS-3428.pdf. This shows 26.6VAC output at a load of 10.7A, close to what I'll be targeting.
Given this, 26.6*1.414 = 37.6VDC rectified output, plus a diode voltage drop of 2*1.25V = 2.5V (using a MUR1540 as an example), gives me a total rectified voltage of 35.1VDC.
The next voltage drop would be from the resistor in the RC filter. In calculating this component's specs, I am looking at voltage drop and power rating for the resistor. At 4.375A across a 0.2R resistor, I would get a voltage drop of 4.375*0.2 or 0.875V and a required power rating of 4.375* 0.875 or about 3.8W. I could then plan on using a 5W+ resistor at 0.2R and would arrive at a final voltage (under full load) to the PeeCeeBee of 35.1-0.875 = 34.23V. I am assuming this would be sufficient for the PeeCeeBee?
I am starting to size up the PSU requirements for a stereo build and, from comments here and in the main design thread, I think I have arrived at the following specs. I would appreciate for some confirmation of what I'm thinking below and that I'm not hanging my hat too much on Ohm's law where it does not apply
At 35VDC into 8R, there will be 35/8 = 4.375A per channel of current at a full voltage output. This lines up with other recommendations of 150VA per channel transformer recommendations: 4.375 * 25VAC = 110VA.
Using these numbers, I am hoping to use a 300VA, 28VAC transformer in conjunction with Prasi's CRC PSU board to power a pair of PeeCeeBees. The specific transformer specs I plan on using are at: http://www.antekinc.com/content/AS-3428.pdf. This shows 26.6VAC output at a load of 10.7A, close to what I'll be targeting.
Given this, 26.6*1.414 = 37.6VDC rectified output, plus a diode voltage drop of 2*1.25V = 2.5V (using a MUR1540 as an example), gives me a total rectified voltage of 35.1VDC.
The next voltage drop would be from the resistor in the RC filter. In calculating this component's specs, I am looking at voltage drop and power rating for the resistor. At 4.375A across a 0.2R resistor, I would get a voltage drop of 4.375*0.2 or 0.875V and a required power rating of 4.375* 0.875 or about 3.8W. I could then plan on using a 5W+ resistor at 0.2R and would arrive at a final voltage (under full load) to the PeeCeeBee of 35.1-0.875 = 34.23V. I am assuming this would be sufficient for the PeeCeeBee?
You're on the right track.
Sure! Welcome to the party.
GB2 Update:
Mdpae - 2 PCBS - DELIVERED
kartk - 2 PCBs - DELIVERED
Manoranjan - 2 PCBs - DELIVERED
lifewater - 2 PCBs - DELIVERED
NishantS - 4 PCBs - DELIVERED
muthumuthiah - 2 PCBs - DELIVERED
cganeshprabhu - 4 PCBs - DELIVERED
airkrishna2015 - 4 PCBs - DELIVERED
patriz - 4 PCBs - SHIPPED
cgraf - 2 PCBs - SHIPPED
potepuh - 4 PCBs - SHIPPED
andrewboyp - 10 boards - SHIPPED
mj777 - 2 PCBs - SHIPPED
arthur - 6 PCBs - DELIVERED
zebulo - 2 PCBs - SHIPPED
electrix - 2 PCBs - SHIPPED
HH - 2 PCBs - SHIPPED
phill094 - 4 PCBs - SHIPPED
ben123 - 4 PCBs - SHIPPED
vasillis - 4 PCBs - SHIPPED
rdobbis - 4 PCBs - SHIPPED
nikosokey - 4 PCBs - SHIPPED
JaredC79 - 4 PCBs - SHIPPED
CitizenKane - 2 PCBs - SHIPPED
tome - 2 PCBs - SHIPPED
dersucher - 4 PCBs - SHIPPED
brookhart995 - 2 PCBs - SHIPPED
Mdpae - 2 PCBS - DELIVERED
kartk - 2 PCBs - DELIVERED
Manoranjan - 2 PCBs - DELIVERED
lifewater - 2 PCBs - DELIVERED
NishantS - 4 PCBs - DELIVERED
muthumuthiah - 2 PCBs - DELIVERED
cganeshprabhu - 4 PCBs - DELIVERED
airkrishna2015 - 4 PCBs - DELIVERED
patriz - 4 PCBs - SHIPPED
cgraf - 2 PCBs - SHIPPED
potepuh - 4 PCBs - SHIPPED
andrewboyp - 10 boards - SHIPPED
mj777 - 2 PCBs - SHIPPED
arthur - 6 PCBs - DELIVERED
zebulo - 2 PCBs - SHIPPED
electrix - 2 PCBs - SHIPPED
HH - 2 PCBs - SHIPPED
phill094 - 4 PCBs - SHIPPED
ben123 - 4 PCBs - SHIPPED
vasillis - 4 PCBs - SHIPPED
rdobbis - 4 PCBs - SHIPPED
nikosokey - 4 PCBs - SHIPPED
JaredC79 - 4 PCBs - SHIPPED
CitizenKane - 2 PCBs - SHIPPED
tome - 2 PCBs - SHIPPED
dersucher - 4 PCBs - SHIPPED
brookhart995 - 2 PCBs - SHIPPED