Nah, Ryssen,
use leds in a string (stack up in series) as a much quieter replacement for the 8volt zener - the j511 current source has a lot of equivalents but suggest use a jfet (like a 2sk170, etc) connected as a current source but with a variable resistor (pot) across the current setting resistor to adjust the current thru the led string for simple adjustment amd more precise output offset (near zero, if possible).
This sort of mucking about isn't really necessary but is good to get a good grasp on just what is happenong in the cct and so, can adjust sound a bit to suit own taste.
The fets DO sound better when quite hot - my first rough assembled unit runs over 60*C and hasn't complained once - remarkable design.
I did find it's can be a bit picky about what you feed it with (preamp, vol control, etc) and added a simple JC buffer (K170/j74) and this was then used to isolate the Lightspeed volume control, rather than a pot on the input - made a huge improvement in clarity and dynamics.
use leds in a string (stack up in series) as a much quieter replacement for the 8volt zener - the j511 current source has a lot of equivalents but suggest use a jfet (like a 2sk170, etc) connected as a current source but with a variable resistor (pot) across the current setting resistor to adjust the current thru the led string for simple adjustment amd more precise output offset (near zero, if possible).
This sort of mucking about isn't really necessary but is good to get a good grasp on just what is happenong in the cct and so, can adjust sound a bit to suit own taste.
The fets DO sound better when quite hot - my first rough assembled unit runs over 60*C and hasn't complained once - remarkable design.
I did find it's can be a bit picky about what you feed it with (preamp, vol control, etc) and added a simple JC buffer (K170/j74) and this was then used to isolate the Lightspeed volume control, rather than a pot on the input - made a huge improvement in clarity and dynamics.
I think you'll find this is very beneficial - considering that you have both +/- rail, suggest a simple discrete reg off the main rails for a B1 that uses the +/- 9 - 10 volts.
There's a "Symetrical B1 buffer" thread (sorry about the spelling) and there's a group buy for a pcb going on now, I think.
Incidently, the caps that are added across the zeners (or led string) have quite an effect on the sound - on the first one I did, I tried Panas, then Silmics, Phillips, Nichis, and ended up some 470uF/35v Siemens (S&M B41590) with RC network across it (1.5R +100uF Rifa peg124). These are all axial caps (leads out each end) and take up a lot of space, so the new pcb is about 4" long now and I still need to reduce some tracks lengths - and nothing like Patrick's super neat design, unfortunately.
If you particularly want to use the j511s and other bits as per Patrick original design, perhaps you could email Steenoe - he's up in your area. For the k170s, perhaps ask AndrewT (in Scotland) as he had quite a few of the Lovaltaec ones ready sorted - the led string will generally draw about 5mA, so ask for fets in that range - if you're looking towards a Salas shunt Symetrical B1, you will need some more k170 fets.
There is a guy called "Audiojoy" that advocates the benefits of the Russian K73-16 (PETP) caps - I got some from the Ukraine for a B1 tryout and they are a very cheap good cap - I got a variety and found the lower voltage ones work best (63volt) - I wouldn't say they're the best cap ever made, but extremely good value, despite the size.
Curiously, I've added one of their giant sized K75-10 (POI) caps across the big 22,000uF Siemens Electrolytic on the output of my F3 amp, and is excellent, first class.
Regards ...
There's a "Symetrical B1 buffer" thread (sorry about the spelling) and there's a group buy for a pcb going on now, I think.
Incidently, the caps that are added across the zeners (or led string) have quite an effect on the sound - on the first one I did, I tried Panas, then Silmics, Phillips, Nichis, and ended up some 470uF/35v Siemens (S&M B41590) with RC network across it (1.5R +100uF Rifa peg124). These are all axial caps (leads out each end) and take up a lot of space, so the new pcb is about 4" long now and I still need to reduce some tracks lengths - and nothing like Patrick's super neat design, unfortunately.
If you particularly want to use the j511s and other bits as per Patrick original design, perhaps you could email Steenoe - he's up in your area. For the k170s, perhaps ask AndrewT (in Scotland) as he had quite a few of the Lovaltaec ones ready sorted - the led string will generally draw about 5mA, so ask for fets in that range - if you're looking towards a Salas shunt Symetrical B1, you will need some more k170 fets.
There is a guy called "Audiojoy" that advocates the benefits of the Russian K73-16 (PETP) caps - I got some from the Ukraine for a B1 tryout and they are a very cheap good cap - I got a variety and found the lower voltage ones work best (63volt) - I wouldn't say they're the best cap ever made, but extremely good value, despite the size.
Curiously, I've added one of their giant sized K75-10 (POI) caps across the big 22,000uF Siemens Electrolytic on the output of my F3 amp, and is excellent, first class.
Regards ...
Ah Yes, saw your name there - sorry.
Current for the leds? From memory, the fets need about a 100R pot, not 1kR. I think I set the current then pulled the pot and dropped in a resistor when I sorted the voltage - it's only now that I'm seeing what I missed out on playing with!!
Unfortunately, there is a bit of variations here - I aimed for Patrick's original 8.2 volts and if you push the little green leds hard enough, they'll give you just over 2 volts each and hence close to the 8V2 mark - I wasn't too keen on this, so used 5 series infra red SMD leds that were just under 1.7 volts at 5mA to give me just over 8v3 - I did try 4 greens at 5mA plus a single diode that summed up to about the 8.2v mark, but found the IR leds were quieter (or that my story, anyway!)
I don't actually think it's all that critical for the cct to work well, but then have a bit of fun playing with EVERYTHING. It's not very often that you'll get a cct that is quite so simple, so responsive, sounds so good and also teaches you all about "learning your ears" as the saying goes. Also, I'm using the AKG 170 with a 4 way multistrand rewire, so not standard. The Beyer 880s and the old Senn 424s are still original, but the AKGs bring out the magic thru the DAO amp.
I'm just about to go over to a computer source with the Wolfson dac chip (wavelength), so this might change things a bit.
The B1 current is also set somewhere about the 6 - 7 mA area, I think but here the k and j fets will alter as the current and the applied voltage is changed - Nelson calls this optomising the pair to get the "sweet spot" according to you
Current for the leds? From memory, the fets need about a 100R pot, not 1kR. I think I set the current then pulled the pot and dropped in a resistor when I sorted the voltage - it's only now that I'm seeing what I missed out on playing with!!
Unfortunately, there is a bit of variations here - I aimed for Patrick's original 8.2 volts and if you push the little green leds hard enough, they'll give you just over 2 volts each and hence close to the 8V2 mark - I wasn't too keen on this, so used 5 series infra red SMD leds that were just under 1.7 volts at 5mA to give me just over 8v3 - I did try 4 greens at 5mA plus a single diode that summed up to about the 8.2v mark, but found the IR leds were quieter (or that my story, anyway!)
I don't actually think it's all that critical for the cct to work well, but then have a bit of fun playing with EVERYTHING. It's not very often that you'll get a cct that is quite so simple, so responsive, sounds so good and also teaches you all about "learning your ears" as the saying goes. Also, I'm using the AKG 170 with a 4 way multistrand rewire, so not standard. The Beyer 880s and the old Senn 424s are still original, but the AKGs bring out the magic thru the DAO amp.
I'm just about to go over to a computer source with the Wolfson dac chip (wavelength), so this might change things a bit.
The B1 current is also set somewhere about the 6 - 7 mA area, I think but here the k and j fets will alter as the current and the applied voltage is changed - Nelson calls this optomising the pair to get the "sweet spot" according to you
not generally true - you need to use sensitivity, impedance and source level to calculate required gain and Ipeak
consumer source can be 2 Vrms for desktop CD players/DACs to as low as 0.5 Vrms for sources such as portables
not uncommon dynamic headphone sensitivity and impedance vary hugely 90 to 135 dB Spl re 1 Vrms and 16-600 Ohms
as mentioned HD600 is a popular audiophile headphone that really requires V gain
for the 300 Ohm, 102 dB/V HD600 Amp gain is often set to 10+
I think that for dynamic source material the amp should be able to drive the headphone to 120 dB for instantaneous peaks like snare drum hits, cymbal crashes
http://headwize.com/articles/hearing_art.htm
for the HD600 120 dB spl requires ~ +/-11 Vswing
re pwr: the HD600 would need 420 mA pk or ~2W amplifier pwr dissipation in hopelessly inefficient SE Class A drive
parts selection from current production would lead me to much smaller devices - the nonlinear parasitic C of 10A transistors and transition to exponential (not sq law) behavior at low "sub threshold" bias make these big speaker amp parts the wrong choice for headphone amps
even TO-92 plastic pkg transistors can handle headphone pwr levels if paralleled:
http://headwize.com/projects/showfile.php?file=gilmore3_prj.htm
I wouldn't count paralleling of identical devices against topological "simplicity", especially when it opens up a much larger universe of more appropriate parts
Last edited:
I think both channels sink about 20 watts total, and with a couple lost in the resistors will leave about 15 or so thru the fets.
I personally have no problem with running these devices at quite high temps and simply divided the power by the temp rise I wanted (increase of 35*C) and went looking for a small flat sided heatsink with approx 2*C/W. (about 8" long and 3" high with 24 fins about 2" long for both channels, then blocked some of the fins to get it up to 60*C)
So your sink will run quite cool with both channels on it - could use half of it and get pretty warm - the fets do sound better when hot, about 50*C is the target point.
I've been testing a couple of the big ones (FP240s) at over 100*C for at least 4 months now in a 5 amp power supply and not a problem - unfortunately, it's just the opposite for the Shottky diodes and have to keep them as cool as possible (they leak noise when hot!) and the 2 x 4 matched TO220 diodes have the same sized sink!
The only thing you have to keep in mind is the dc offset after the amp warms up - this is why to match-up the led strings quite closely, to each other. Perhaps it is simpler to just use the original zeners - they still work okay and the low noise 8v2 ones aren't so easy to find.
I personally have no problem with running these devices at quite high temps and simply divided the power by the temp rise I wanted (increase of 35*C) and went looking for a small flat sided heatsink with approx 2*C/W. (about 8" long and 3" high with 24 fins about 2" long for both channels, then blocked some of the fins to get it up to 60*C)
So your sink will run quite cool with both channels on it - could use half of it and get pretty warm - the fets do sound better when hot, about 50*C is the target point.
I've been testing a couple of the big ones (FP240s) at over 100*C for at least 4 months now in a 5 amp power supply and not a problem - unfortunately, it's just the opposite for the Shottky diodes and have to keep them as cool as possible (they leak noise when hot!) and the 2 x 4 matched TO220 diodes have the same sized sink!
The only thing you have to keep in mind is the dc offset after the amp warms up - this is why to match-up the led strings quite closely, to each other. Perhaps it is simpler to just use the original zeners - they still work okay and the low noise 8v2 ones aren't so easy to find.
You first need to decide how much SPL you need from the phones.
Then the datasheet of the phone will tell you the sensiticity, which allows you to convert SPL to voltage. A typical CD player without volume control will output 2V rms max. If 2Vrms is not sufficient to give you the SPL you want, then you need voltage amplification.
The DAO is a follower with (slightly less than) unity gain. It gives you lots and lots of current with very little distortion, but it will only follow the voltage at the input about 1:1.
The easiest way to determine whether you need additional voltage amplification is to connect a 10k log pot and a simple opamp circuit of say +10x gain using e.g. an OPA2604 or OPA2132 (for stereo) supplied by 2x 12V batteries. If this is too loud, then lower the gain of the opamp circuit until it reaches the right level for you, including the amount of volume control you need.
The you know for sure how much (if any) voltage gain you need, and only then you choose the voltage gain block circuit. The output (current gain) stage is the DAO, already done for you.
Patrick
Then the datasheet of the phone will tell you the sensiticity, which allows you to convert SPL to voltage. A typical CD player without volume control will output 2V rms max. If 2Vrms is not sufficient to give you the SPL you want, then you need voltage amplification.
The DAO is a follower with (slightly less than) unity gain. It gives you lots and lots of current with very little distortion, but it will only follow the voltage at the input about 1:1.
The easiest way to determine whether you need additional voltage amplification is to connect a 10k log pot and a simple opamp circuit of say +10x gain using e.g. an OPA2604 or OPA2132 (for stereo) supplied by 2x 12V batteries. If this is too loud, then lower the gain of the opamp circuit until it reaches the right level for you, including the amount of volume control you need.
The you know for sure how much (if any) voltage gain you need, and only then you choose the voltage gain block circuit. The output (current gain) stage is the DAO, already done for you.
Patrick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.