If only positive rail from cap multiplier is used (for single rail amp) with fully populated pcb, should unused rail be loaded in some way? Is it safe to leave unused rail output unloaded?
I have not tested that.
If you mean with one of my boards, I think I would just pull the fuse on the other side, because the two circuits are independent. In the original version the jfet CCS crosses the ground to the other rail. I did not test the original circuit with un-balanced load when I had that on perf-board.
The feedback resistors provide enough load by themselves, so you can safely leave one or both rails unloaded.
Kind of along this thinking, can this cap multiplier work in series just using the + and - outs for a single rail amp?
Although you could do that, by grounding the negative rail instead of the normal ground, I would definitely not recommend it. It would be setting someone up for a tragic Magic Smoke Release Event™ due to all the bits on the PCB marked "GND" that aren't actually connected to ground.
LED and LED Series Resistor
I have been reminded to post a note on the series resistor value...
As it is now, on the schematic and in the BOM, the LED series resistor (R13-R14) value and 1/4 watt rating are a bit small for a typical supply voltage over 30V. At the time I thought that adjusting the value (or power) for the desired supply voltage was sort-of obvious, but better safe than sorry, so...:
The LED is rated at 20ma, but it typically needs much less than that current to get more than sufficient brightness for a power-on indication. However, the LED and its series resistor are also used to drain the large reservoir caps when no load is connected to the power supply. This can take several minutes, during which time the LED will dim gradually as the capacitor voltage drops. So, the resistor should not be arbitrarily large, either. Also, if the LED is omitted completely, the pads should be shorted on the bottom side of the board, to maintain a closed path to the bleed resistor.
Most modern resistors can take a slightly larger than rated load, but to be on the safe side, here is a small table of power supply voltages and 1/4 watt resistor values:
<30V 3.3K~4.7K
<40V 6.8K~10K
<50V 10K~15K
I have been reminded to post a note on the series resistor value...
As it is now, on the schematic and in the BOM, the LED series resistor (R13-R14) value and 1/4 watt rating are a bit small for a typical supply voltage over 30V. At the time I thought that adjusting the value (or power) for the desired supply voltage was sort-of obvious, but better safe than sorry, so...:
The LED is rated at 20ma, but it typically needs much less than that current to get more than sufficient brightness for a power-on indication. However, the LED and its series resistor are also used to drain the large reservoir caps when no load is connected to the power supply. This can take several minutes, during which time the LED will dim gradually as the capacitor voltage drops. So, the resistor should not be arbitrarily large, either. Also, if the LED is omitted completely, the pads should be shorted on the bottom side of the board, to maintain a closed path to the bleed resistor.
Most modern resistors can take a slightly larger than rated load, but to be on the safe side, here is a small table of power supply voltages and 1/4 watt resistor values:
<30V 3.3K~4.7K
<40V 6.8K~10K
<50V 10K~15K
Attachments
Will that LED stand its own dissipation in the long term when constrained under the fuse socket? Makes a nice diffused illumination effect nonetheless.
It would seem to me if you were building it solely for this purpose, you would leave off any "ground" lugs and just treat the - rail as ground. Am I missing something?
Yes. To the best of my knowledge, at 1/2 rated current, a 3-mm LED will do fine even hermetically sealed in a piece of plastic.
Not sure, but this is not ideal, because the two rails will not drop to zero at the same rate. So, for some time after you turn off power, you may have different rail voltages on the amp boards. In case of VSSA, this is not a big factor, because the bias current is much higher than the ~10mA LED current, but it might be an issue with other circuits.
MOSFETs used
Passively following this thread with big interest since August 2013, I decided to jump in 🙂
I intend to build this circuit as well, but only the Cap multiplier part. The reasons are that I have already a board for Power supply and Loudspeaker protection.
For VSSA as it is everything is clear and I intend to use one circuit per channel.
However, for the balanced VSSA I will power both branches with one module and I wonder how it will behave? I was thinking of using lower Rds transistors:
N channel: IRF3205, Rds=8mOhm, ID=110A Vds=55V Datasheet
P channel: IRF4905, Rds(on)=20mOhm, ID=-74A, Vdss=-55V Datasheet
Allowing a Drop-off voltage of 0,8 V I hope they can make it, being mounted on the main heatsink. What is the designer's opinion?
A bit off-topic:
I think I can use a "Coldamp power supply" for both balanced channels (four VSSA modules), these cap multipliers being used between the common PSU and the modules, providing a nice separation as well. Being able to supply 800W continuously (and 1300W peak) I guess it's up to the task. I have used another one before to successfully power 2 channels of 180W each. I have another one as well that I intend to use here.
In my (future) case I hope to get more than 200W from balanced VSSAs. If this will be a success, I will stop building power stages for a while
😀
Cheers!
Passively following this thread with big interest since August 2013, I decided to jump in 🙂
I intend to build this circuit as well, but only the Cap multiplier part. The reasons are that I have already a board for Power supply and Loudspeaker protection.
For VSSA as it is everything is clear and I intend to use one circuit per channel.
However, for the balanced VSSA I will power both branches with one module and I wonder how it will behave? I was thinking of using lower Rds transistors:
N channel: IRF3205, Rds=8mOhm, ID=110A Vds=55V Datasheet
P channel: IRF4905, Rds(on)=20mOhm, ID=-74A, Vdss=-55V Datasheet
Allowing a Drop-off voltage of 0,8 V I hope they can make it, being mounted on the main heatsink. What is the designer's opinion?
A bit off-topic:
I think I can use a "Coldamp power supply" for both balanced channels (four VSSA modules), these cap multipliers being used between the common PSU and the modules, providing a nice separation as well. Being able to supply 800W continuously (and 1300W peak) I guess it's up to the task. I have used another one before to successfully power 2 channels of 180W each. I have another one as well that I intend to use here.
In my (future) case I hope to get more than 200W from balanced VSSAs. If this will be a success, I will stop building power stages for a while
😀 Cheers!
Is there a complementary for BS170 (which is cheap and available here)? I just paid one dollar for each ZVP2106A and wonder if there is cheaper substitutes for ZVN/ZVP pair?
I still think it's too easy to come back to it in the future and forget that it's configured in that way (I know I've done that sort of thing before).
It should work fine.
Those look excellent. They will allow lower dropout voltage and higher output current.
BS250 is close enough for this.
I prefer the two bridge rectifiers and voltage divider instead of a reference voltage source, according to this scheme.
Last edited:
Following this recommendation I placed an order with Tayda.
I got 550 resistors, 20 sets of croc clipped cables, 100 transistors, 50 varistors and 50 of those T0220 heatsinks for ~£20 ($31.525) delivered by registered post in 10days.
Everything arrived today in good order.
There is a burr around the tapped hole. This must be sanded off, or deburred, to allow a "flat" contact Thermal Interface.
Last edited:
It looks like a nice circuit, but do you still intend to use it as a filtered supply, whose output rises and falls with line an load variations, or something closer to a supply with an adjustable output?
With the original circuit, the trim pot adjusts the voltage drop across the pass transistor to whatever value is desired to filter the amount of mains ripple and power supply noise. The Vdrop stays the same over normal line and load variations, and for class AB amplifiers, it can be adjusted to a fairly small value (a little over 1/2 idle ripple voltage at intended load).
Using a voltage reference or a zener diode makes the circuit into something close to an adjustable regulated linear supply. To keep the output clean, Vdrop has to be set much higher to cover not just the power supply noise and ripple, but the full range of the expected line and load variation. This can easily double or tripple the power dissipation in the pass transistor, and its heatsink has to be proportionally bigger.
It is still a nice circuit, but it gives up some of the original benefits, which was a fairly high efficiency relative to the amount of noise filtering achieved, low voltage loss, and low heat.
No, that is a good question. I do not take it for granted that I will be able to hear improvements just because something looks better on a scope, or in a simulation.
Compared to a conventional rectifier/filter, I can hear the difference. The differences are significant. Not like a fraction of a percent of THD. You can hear it.
Compared to another highly regulated or filtered supply? Not personally, no.
I originally compared a conventional power supply to an old linear Tektronix bench supply, both using my ears and a scope, and it eventually led me to this circuit.
My goal was a linear supply which could be used instead of a Switch-Mode PSU for the original VSSA modules from LazyCat, and for my Through-Hole version adapted from Shaan's circuit.
I also wanted a supply with a soft turn-on to avoid power-on transients in the amplifier circuit (I can hear the difference on startup, too).
- Status
- Not open for further replies.