Discussing just the positive rail of this schematic, R2-C1 and R5-C4 are the filter components. This gives a relatively clean reference to Q1's base. Q2 is the other side of the differential amp and receives feedback from the output. The differential output controls the current through R22, and therefore the voltage to drive U2's gate. Q13 is a CCS which I added back in to make it less sensitive to bulk supply voltage.
I don't think it would be a problem, so long as the chosen pass MOSFETs can take the initial surge. Why would you use such a large output capacitor? The circuit really only needs a relatively small output capacitor just to ensure the output impedance remains low out to higher frequencies. MrEvil's original design only had 100uF on the output. The 4700uF PMI and I chose to use is already overkill IMHO.
Hi donovas,
Use the schematic from post #65 >HERE<. There are enough benefits to having the CCS feed the tail current to the LTP to include it, so I don't recommend the schematic in post #1. The difference in cost for a couple small transistors and LEDs shouldn't be a deal breaker. You can just use the 1µF in the filter if you like at the expense of a little performance, or parallel a couple to keep it performing better.
There is nothing critical about the choice of small signal devices so long as you observe their voltage ratings. Assume any one device could be exposed to the full bulk voltage for a single rail. Therefore the BC550 / BC560 being 45V devices are suitable for about 40V rails. You might consider the 2N5551 / 2N5401 pair for the smaller devices.
For cheap MOSFETs, just use some IRFP240 / IRFP9240 or IRFP140 / IRFP9140, or it will even work with BJT darlington devices like the TIP142 / TIP147. Just remember you need both an N-channel and a P-channel output device if you are building a dual rail supply. Best performance and lowest losses are had when using output devices with the lowest Rds-ON specification you can get.
Use the schematic from post #65 >HERE<. There are enough benefits to having the CCS feed the tail current to the LTP to include it, so I don't recommend the schematic in post #1. The difference in cost for a couple small transistors and LEDs shouldn't be a deal breaker. You can just use the 1µF in the filter if you like at the expense of a little performance, or parallel a couple to keep it performing better.
There is nothing critical about the choice of small signal devices so long as you observe their voltage ratings. Assume any one device could be exposed to the full bulk voltage for a single rail. Therefore the BC550 / BC560 being 45V devices are suitable for about 40V rails. You might consider the 2N5551 / 2N5401 pair for the smaller devices.
For cheap MOSFETs, just use some IRFP240 / IRFP9240 or IRFP140 / IRFP9140, or it will even work with BJT darlington devices like the TIP142 / TIP147. Just remember you need both an N-channel and a P-channel output device if you are building a dual rail supply. Best performance and lowest losses are had when using output devices with the lowest Rds-ON specification you can get.
Hi donovas,
The TIP142 / TIP147 have a lower effective 'threshold' than the MOSFETs, so the resistor values are somewhat dependant on two things, the bulk supply voltage and the output device threshold value.
Referring to the last schematic on the first page of this thread, assuming ten posts per page, R17 and R8 set the tail current and should be calculated to pass about 4-5mA of current. They will see about half of the bulk supply, so a quick calculation using Ohm's Law and select the standard resistance closest to the calculated value that puts the current in the desired range.
Next we need to calculate the appropriate values for R22 and R23 to accommodate the output device threshold voltages. We want to have about half of the tail current through each of these resistors. Assuming a TIP142 / TIP147 are turning on at Vbe of 1.3V and we want about 2-2.5mA through R22 and R23, we would want values of about 680Ω in those places.
You may have to adjust your resistor values experimentally, you have chosen to work with an earlier schematic. Later versions used a simple CCS to reduce some dependance on supply voltage in setting the operating conditions.
The TIP142 / TIP147 have a lower effective 'threshold' than the MOSFETs, so the resistor values are somewhat dependant on two things, the bulk supply voltage and the output device threshold value.
Referring to the last schematic on the first page of this thread, assuming ten posts per page, R17 and R8 set the tail current and should be calculated to pass about 4-5mA of current. They will see about half of the bulk supply, so a quick calculation using Ohm's Law and select the standard resistance closest to the calculated value that puts the current in the desired range.
Next we need to calculate the appropriate values for R22 and R23 to accommodate the output device threshold voltages. We want to have about half of the tail current through each of these resistors. Assuming a TIP142 / TIP147 are turning on at Vbe of 1.3V and we want about 2-2.5mA through R22 and R23, we would want values of about 680Ω in those places.
You may have to adjust your resistor values experimentally, you have chosen to work with an earlier schematic. Later versions used a simple CCS to reduce some dependance on supply voltage in setting the operating conditions.
hello. it works fine. thank you.
sounds also fine
i compare with mr.evil multiplier on 3886 gainclone. yours sound very same. smaller pcb, less component, same sound= win
big but... turn on/off thump with your pcb. mr.evil's quiet. just the circuit change. smoothing cap before and after (10000uf) same. amp same.
sounds also fine
i compare with mr.evil multiplier on 3886 gainclone. yours sound very same. smaller pcb, less component, same sound= win big but... turn on/off thump with your pcb. mr.evil's quiet. just the circuit change. smoothing cap before and after (10000uf) same. amp same.
The capacitor after the actual multiplier does not need to be too big, Mr.Evil used just 100uF and PMI and myself chose 4700uF which is already way more than required. Perhaps there is some difference in your ripple filter sections between the positive side and the negative side which may cause the rails to come up at different rates. That could be the source of the thump.
Jason,
I currently have some Randy Slone Totem Pole LFET amps built with 46v 500va AC transformers ... secondaries setup for dual bridge.
At one point I had a CLC supply with 2.2uH and 12000uF on either side...I eventually pulled this supply and have a traditional cap bank...long story why I did it. Thought I had a problem but the CLC didn't turn out to be an issue. Anyway I loved the sound of the amp with CLC installed, even with the high PSRR of the amp it seemed to benefit from the filtering the inductor provided.
OF course I didn't have enough cotinous current draw to maintain good regulation...regardless of the technical aspects its the best I've ever heard these amps sound.
No looking at these boards would 6A and 67v DC out roughly be something this could support? I'd likely never hit that as my speakers can only take 160 watts sustained..they are 6ohm Tannoy Dimensions towers.
Best Regards,
Theo
I currently have some Randy Slone Totem Pole LFET amps built with 46v 500va AC transformers ... secondaries setup for dual bridge.
At one point I had a CLC supply with 2.2uH and 12000uF on either side...I eventually pulled this supply and have a traditional cap bank...long story why I did it. Thought I had a problem but the CLC didn't turn out to be an issue. Anyway I loved the sound of the amp with CLC installed, even with the high PSRR of the amp it seemed to benefit from the filtering the inductor provided.
OF course I didn't have enough cotinous current draw to maintain good regulation...regardless of the technical aspects its the best I've ever heard these amps sound.
No looking at these boards would 6A and 67v DC out roughly be something this could support? I'd likely never hit that as my speakers can only take 160 watts sustained..they are 6ohm Tannoy Dimensions towers.
Best Regards,
Theo
The capacitance multiplier tries to keep the output a little below the 'trough' of the input ripple, so most of the input ripple can be rejected. Generally we set the drop to be a little greater than the anticipated ripple. I suspect the design will be able to give nice clean power to the amplifier with the appropriate choice of dropout.