Yup, the circuit is taken from the DCB1 pre. The relay keeps the output shorted to ground a couple of seconds during startup and quickly shorts it to ground again during shutdown.
jwb is using a similar circuit in http://www.diyaudio.com/forums/headphone-systems/22817-jwbs-ultimate-aleph-headphone-amp.html
jwb is using a similar circuit in http://www.diyaudio.com/forums/headphone-systems/22817-jwbs-ultimate-aleph-headphone-amp.html
Attachments
Last edited:
The amp has now logged several hours of operation with a dummy load (35R resitance, see third pic on the left). I biased the output FETs @ 0.5A and set the Aleph current source AC gain to 60%. Output offset is stable at about 0.1V on both sides, which means the headphones have to sink 0.2mW in idle. I will adress this issue at some later point though when I switch to JFET inputs. The rails are @ +17.50 and -17.49 VDC, a bit higher as planned. The heatsinks are properly sized and get comfortably warm.
The current limiter circuitry is in place, so far no smoke has escaped from the FETs.
The amp is driving my HE-400i with great ease and really makes them sing, I'm really satisfied.
The current limiter circuitry is in place, so far no smoke has escaped from the FETs.
The amp is driving my HE-400i with great ease and really makes them sing, I'm really satisfied.
Attachments
> The current limiter circuitry is in place, so far no smoke has escaped from the FETs.
I don't think you got the message.
Aleph-H has negative feedback, probably around 20~30dB.
For any feedback amp, if there is input signal and the output is short circuited, the NFB will keep driving the output devices to deliver more current,
in an attempt to achieve the required output voltage.
The limit in this case is 2x nominal (Vgs + I*Rs), as this is the maximum drive voltage supplied by the input diff pair to the MOSFET gate.
And this "excess" current will be sunk to ground by the relay via the output mosfets.
Not that it is necessarily an issue, as long as your MOSFETs and relay can take it.
But you should be aware what happens when you short circuit outputs.
Patrick
I don't think you got the message.
Aleph-H has negative feedback, probably around 20~30dB.
For any feedback amp, if there is input signal and the output is short circuited, the NFB will keep driving the output devices to deliver more current,
in an attempt to achieve the required output voltage.
The limit in this case is 2x nominal (Vgs + I*Rs), as this is the maximum drive voltage supplied by the input diff pair to the MOSFET gate.
And this "excess" current will be sunk to ground by the relay via the output mosfets.
Not that it is necessarily an issue, as long as your MOSFETs and relay can take it.
But you should be aware what happens when you short circuit outputs.
Patrick
EUVL is right that shorting the output will drive it to the rails. It doesn't matter in my original design; the available current isn't enough to break anything in ten seconds or whatever the startup time is. Think about the voltages and bias current in your design and the FETs and relays you used to check if it's safe.
Measurement revealed that the two parallel 0R47 resistors at the output drop 110mV during a cold startup with a shorted output. This means the current through them equals 0.11/(0.47/2)=0.468A. Should be well within the capabilities of the output stage FETs. The RY12-RK relay can handle 1A@24VDC. Am I missing something?
I am getting back to my old Aleph H. It was packed away for a few years and I feel it is time to revisit and see if I can learn a bit more.
From memory, it sounded best when it was running biased rather high but had an annoying psu-related hum. After reducing the bias and playing around with the psu I got rid of the hum, but from memory some of the magic was gone.
Lately I have been thinking about the Aleph circuit. If I understand correctly it sounds best when the Aleph current source delivers about 50% under load. I assume that (at a given, fixed configuration) the Aleph contribution will be different for a 32R headphone than say a 300R headphone. If that is indeed the case, it could be expected that it sounds better with the load it is optimised for.
Does it make sense to use a fixed resistor (parallel to the load) as a kind of shunt to ensure the amp sees a much smaller change in load when changing headphones?
From memory, it sounded best when it was running biased rather high but had an annoying psu-related hum. After reducing the bias and playing around with the psu I got rid of the hum, but from memory some of the magic was gone.
Lately I have been thinking about the Aleph circuit. If I understand correctly it sounds best when the Aleph current source delivers about 50% under load. I assume that (at a given, fixed configuration) the Aleph contribution will be different for a 32R headphone than say a 300R headphone. If that is indeed the case, it could be expected that it sounds better with the load it is optimised for.
Does it make sense to use a fixed resistor (parallel to the load) as a kind of shunt to ensure the amp sees a much smaller change in load when changing headphones?
name of the game is having upper part of amplifier sort of current follower (following lower art but antiphase) , without adding own flavor
same applies to Aleph , ACA , or any of last Papa's SE thingies
so , type of device (being driven from pretty low impedance ) isn't so important as choice of lower one
one can try LU as lower OS device , recipe given eons ago by EUVL
(in that instance Mighty Moi being Public Fool , for oomphtienth time
)
same applies to Aleph , ACA , or any of last Papa's SE thingies
so , type of device (being driven from pretty low impedance ) isn't so important as choice of lower one
one can try LU as lower OS device , recipe given eons ago by EUVL
(in that instance Mighty Moi being Public Fool , for oomphtienth time
)
http://www.diyaudio.com/forums/pass-labs/111197-aleph-schematic-6.html#post1358353
http://www.diyaudio.com/forums/pass-labs/111197-aleph-schematic-6.html#post1361918
And you should use the 200mA triode cell of the DAO for headphones.
Patrick
