I began this about a week ago, and now have the output stage of the first channel working. Thought I'd share what I've done so far.
I wanted to re-use the power devices and obviously the heatsinks. My plan was to find out what was the maximum class A power I could manage with this arrangement.
I decided to use the P channel devices as the current source, very similar to what is shown in Fig 2. of the original Zen article on the Pass DIY website. I'm using a different BJT for Q3, salvaged from the Hafler PCB. I have two MOSFETs in parallel, and a single current sense resistor (R1) of 0.1 Ohm. My R3 is 1K, and I've set the bias to run at 3A total, 1.5 A per device.
I wanted to use the Aleph style current source, so I added a resistor to the base of Q3, and scaled it so that it would modulate the current source appropriately.
For the current feedback signal, I placed a 0.75 Ohm resistor between the load and ground. This develops a ground referenced voltage proportional to the current in the load. I feed this via a capacitor into the base of Q3. I've scaled it so that when 1.5A flows into the load, the current source is sourcing 3.75A. When the load current is -1.5A, the current source generates 2.25A.
Obviously, I needed to replace the power transformer. I have a temporary one installed at the moment while I wait for my toroid kit to arrive. When it arrives, I'll wind it to exactly the voltage I need, to get rails of about +/- 15VDC. Currently, I'm running at about +/- 13.5, but my transformer is maxed out, getting very hot, and I have lots of hum.
I don't have any good way of measuring the temperature. I have the original thermostat installed on the heatsink, and it doesn't trip, but I don't know what temp it is set for. I also have another thermostat set for 85C, and it doesn't trip. I can hold my hand on the heatsink for 5 to 10 seconds. I can hold my finger on the power devices for 3 to five seconds.
For the lower section, I built something like the circuit in Zen variations part 2, figure 1. I tweaked it a little bit, and adjusted the pot until I had no output offset. It wanders around by about 100mV as the temp changes, but seems to stay within +/- 100mV.. I'm running it without feedback at the moment.
By my estimates, I should get about 12W RMS into 8 Ohms. I haven't connected a signal yet. I plan to do that tonight. I have connected a cheap speaker via 10uF, and I hear noticeable hum. If I connect the speaker across either of the supply rails, I hear exactly the same hum, so I think this is my transformer being overworked. I'm not going to worry about it until I get my toroid wound. I don't hear any noise through the speaker, only hum.
Next step is to build an input stage to go with it. I'm thinking I'll do one like the Aleph 30, but haven't decided yet. I have a bunch of BJTs that I originally bought for restoring the Hafler, but now they won't be needed. Maybe I'll use some of those.
So far, so good!
I wanted to re-use the power devices and obviously the heatsinks. My plan was to find out what was the maximum class A power I could manage with this arrangement.
I decided to use the P channel devices as the current source, very similar to what is shown in Fig 2. of the original Zen article on the Pass DIY website. I'm using a different BJT for Q3, salvaged from the Hafler PCB. I have two MOSFETs in parallel, and a single current sense resistor (R1) of 0.1 Ohm. My R3 is 1K, and I've set the bias to run at 3A total, 1.5 A per device.
I wanted to use the Aleph style current source, so I added a resistor to the base of Q3, and scaled it so that it would modulate the current source appropriately.
For the current feedback signal, I placed a 0.75 Ohm resistor between the load and ground. This develops a ground referenced voltage proportional to the current in the load. I feed this via a capacitor into the base of Q3. I've scaled it so that when 1.5A flows into the load, the current source is sourcing 3.75A. When the load current is -1.5A, the current source generates 2.25A.
Obviously, I needed to replace the power transformer. I have a temporary one installed at the moment while I wait for my toroid kit to arrive. When it arrives, I'll wind it to exactly the voltage I need, to get rails of about +/- 15VDC. Currently, I'm running at about +/- 13.5, but my transformer is maxed out, getting very hot, and I have lots of hum.
I don't have any good way of measuring the temperature. I have the original thermostat installed on the heatsink, and it doesn't trip, but I don't know what temp it is set for. I also have another thermostat set for 85C, and it doesn't trip. I can hold my hand on the heatsink for 5 to 10 seconds. I can hold my finger on the power devices for 3 to five seconds.
For the lower section, I built something like the circuit in Zen variations part 2, figure 1. I tweaked it a little bit, and adjusted the pot until I had no output offset. It wanders around by about 100mV as the temp changes, but seems to stay within +/- 100mV.. I'm running it without feedback at the moment.
By my estimates, I should get about 12W RMS into 8 Ohms. I haven't connected a signal yet. I plan to do that tonight. I have connected a cheap speaker via 10uF, and I hear noticeable hum. If I connect the speaker across either of the supply rails, I hear exactly the same hum, so I think this is my transformer being overworked. I'm not going to worry about it until I get my toroid wound. I don't hear any noise through the speaker, only hum.
Next step is to build an input stage to go with it. I'm thinking I'll do one like the Aleph 30, but haven't decided yet. I have a bunch of BJTs that I originally bought for restoring the Hafler, but now they won't be needed. Maybe I'll use some of those.
So far, so good!
Update on the Haf-Nelson..
Last night, I made my first attempt at winding the power toroid.
I'm currently running the amp as a Zen, with 2A bias. I hooked up the finished toroid, and measured.. 28 volts. A bit low, I was expecting more like 32. I thought I'd better measure the current. Ay carrumba! there was 6A flowing.
After correcting a wiring problem in the CCS, I got my 2A back, and now I have the expected 32 Volts. I get about 800mV ripple, but the hum at the output is acceptable (haven't measured it).
As a Zen, it sounds OK, though not much bass, and not enough power. I'm driving it directly from my preamp, and even on max volume, I don't get enough drive to clip the Zen. The gain seems very low.
For reference, I'm comparing it to my Bottlehead SEX amp, which is about 2 Watts, and plays much louder on the same (4 ohm) speaker.
Anyway, it looks like the basic concept is sound. All the power devices seem equally hot. The heatsink is well under 50 Celsius.
Next step: convert back to Aleph-like configuration, and make an imput stage.
Last night, I made my first attempt at winding the power toroid.
I'm currently running the amp as a Zen, with 2A bias. I hooked up the finished toroid, and measured.. 28 volts. A bit low, I was expecting more like 32. I thought I'd better measure the current. Ay carrumba! there was 6A flowing.
After correcting a wiring problem in the CCS, I got my 2A back, and now I have the expected 32 Volts. I get about 800mV ripple, but the hum at the output is acceptable (haven't measured it).
As a Zen, it sounds OK, though not much bass, and not enough power. I'm driving it directly from my preamp, and even on max volume, I don't get enough drive to clip the Zen. The gain seems very low.
For reference, I'm comparing it to my Bottlehead SEX amp, which is about 2 Watts, and plays much louder on the same (4 ohm) speaker.
Anyway, it looks like the basic concept is sound. All the power devices seem equally hot. The heatsink is well under 50 Celsius.
Next step: convert back to Aleph-like configuration, and make an imput stage.
Where did you get the toroid kit from?
I'm interested in what you're doing here. I may adapt a Technics SE-A5 to something else, reusing the transformer, heatsinks, and power meters. The transformer has lower-voltage taps which I will use. Picture:
http://www.diyaudio.com/forums/attachment.php?postid=339089&stamp=1078202143
Nelson advised against a Zen, but I may try a small Aleph.
I'm interested in what you're doing here. I may adapt a Technics SE-A5 to something else, reusing the transformer, heatsinks, and power meters. The transformer has lower-voltage taps which I will use. Picture:
http://www.diyaudio.com/forums/attachment.php?postid=339089&stamp=1078202143
Nelson advised against a Zen, but I may try a small Aleph.
I got the toroid kit from Toroid corp of Maryland
They give you everything you need, including info on how to compute number of turns, etc. Nice folks to deal with. They also stock standard toroidal transformers at reasonable prices.
They give you everything you need, including info on how to compute number of turns, etc. Nice folks to deal with. They also stock standard toroidal transformers at reasonable prices.
Update
OK, I've got the first channel more or less compelte. It's now running with the active current source. I ran out of transistors for matching, so I built my input stage with OPA627 instead.
My target was for 12 V peak into 8 ohms (24V p-p). My power supply is about +- 16.5V. It drives to 12 volts peak with no problem. With heavy clipping, it's about 28V p-p. I don't know what voltage it starts to clip at yet.
So currently, it gives me a solid 8.5 Watts RMS into 8 ohms, and 9 W RMS into 4 Ohms. With an extra Volt on the PSU, it would do 10W RMS into 8 Ohms. I guess I could add a few more turns to my toroid to up the voltage a little.
The heatsink is a tad under 50C after running for a couple of hours. This is on the bench, where the airflow is good. In the chassis it might get somewhat hotter.
Next, I'm going to build the other channel, and then re-install it all into the original chassis. Originally, there were 27000 uF on each supply. I'll pull these out (they are 75 V caps) and install more capacity at lower voltage. I'm adding an IEC inlet socket, a new power switch, and a soft start relay for the toroid inrush current. I also need to rewire the thermal protection and the DC fuses.
Anyway, it's definitely a doable project. I'll have to see later how much I can increase the bias (Currently 2A) while keeping the heatsinks at non-ludicrous temperatures.
Stragne effect I noticed: Increasing the active current source amount reduces 2nd harmonic significantly, but increases higher harmonics. I don't understand this at all.
OK, I've got the first channel more or less compelte. It's now running with the active current source. I ran out of transistors for matching, so I built my input stage with OPA627 instead.
My target was for 12 V peak into 8 ohms (24V p-p). My power supply is about +- 16.5V. It drives to 12 volts peak with no problem. With heavy clipping, it's about 28V p-p. I don't know what voltage it starts to clip at yet.
So currently, it gives me a solid 8.5 Watts RMS into 8 ohms, and 9 W RMS into 4 Ohms. With an extra Volt on the PSU, it would do 10W RMS into 8 Ohms. I guess I could add a few more turns to my toroid to up the voltage a little.
The heatsink is a tad under 50C after running for a couple of hours. This is on the bench, where the airflow is good. In the chassis it might get somewhat hotter.
Next, I'm going to build the other channel, and then re-install it all into the original chassis. Originally, there were 27000 uF on each supply. I'll pull these out (they are 75 V caps) and install more capacity at lower voltage. I'm adding an IEC inlet socket, a new power switch, and a soft start relay for the toroid inrush current. I also need to rewire the thermal protection and the DC fuses.
Anyway, it's definitely a doable project. I'll have to see later how much I can increase the bias (Currently 2A) while keeping the heatsinks at non-ludicrous temperatures.
Stragne effect I noticed: Increasing the active current source amount reduces 2nd harmonic significantly, but increases higher harmonics. I don't understand this at all.
Haf-Nelson: Done.
I'm calling this project more or less complete. I've wired both channels, and packaged it all up for some extended listening.
After a few hours of power, the hottest part of the heatsink I can measure (from the outside) is 124 Farenheit. This is about midway up the heatsink, near the thick wall that mounts the mosfets.
I ended up not adding any turns to the transformer, as I didn't want the heatsinks any hotter than this. This gives me plenty of power for my listening conditions. My bias is running at 1.8 amps per side, so each mosfet is only carrying about 900mA.
I replaced the original PSU filter caps with 2 x 54,000 uF per side. I used Shottky power rectifiers to get a little more voltage. I added an IEC line input connector, and a new power switch that I salvaged from an HP laser printer. I also added a thermistor in series with the transformer primary. I need to add a pilot light, and probably a speaker relay, as I have some annoying power on/off transients.
This has become my reference amp. All I had to compare it with previously was my Bottlehead SET, and a Gainclone. The Haf-Nelson is noticeably superior to the Gainclone in all respects. Compared to the SET, the Haf-Nelson is not as quiet, but has much more extended bass response. This is my first real Class A design. (I don't count the SET, as I built that from a kit.) I'm very pleased with how it went, and learned a lot from the process.
I'm calling this project more or less complete. I've wired both channels, and packaged it all up for some extended listening.
After a few hours of power, the hottest part of the heatsink I can measure (from the outside) is 124 Farenheit. This is about midway up the heatsink, near the thick wall that mounts the mosfets.
I ended up not adding any turns to the transformer, as I didn't want the heatsinks any hotter than this. This gives me plenty of power for my listening conditions. My bias is running at 1.8 amps per side, so each mosfet is only carrying about 900mA.
I replaced the original PSU filter caps with 2 x 54,000 uF per side. I used Shottky power rectifiers to get a little more voltage. I added an IEC line input connector, and a new power switch that I salvaged from an HP laser printer. I also added a thermistor in series with the transformer primary. I need to add a pilot light, and probably a speaker relay, as I have some annoying power on/off transients.
This has become my reference amp. All I had to compare it with previously was my Bottlehead SET, and a Gainclone. The Haf-Nelson is noticeably superior to the Gainclone in all respects. Compared to the SET, the Haf-Nelson is not as quiet, but has much more extended bass response. This is my first real Class A design. (I don't count the SET, as I built that from a kit.) I'm very pleased with how it went, and learned a lot from the process.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.