The Hardcore Headbangers were designed, like all good devices should be, to purpose. A pair of Lowther Acousta's plus a pair of Altec A7s set the scene: I have had many parties and ratcheted up my tinnitus using the A7s and a 5W triode connected EL84PP amp (started life as a genuine Dynaco SCA35 ended up as a breathed upon ST-35 kinda thing). Since 5W made more than enough of a racket that was the first requirement: low power. I had looked at a couple of designs but copying schematics just gives you all the physical effort and none of the mental exertion. Half the fun is there and all of the performance too. Building amps for me is like building F1 cars or NASA rockets, it got to be the best possible and I want to squeeze out every drop of perfomance that I can. Behringer or Bryston? Dixons or Denon? Crosley or Krell? You can guess what my choices would be?
Next up, PP or SE? Well I built a ~1.5W 6BM8/ECL82...it was not bad with the A7s, was better with the Lowthers but pretty kak on rock music - so definitely not a headbanger. It used crappy OTXs, had some rather questionable bias points and unsurprisingly produced measurements that still make me blush. Hence after a bit of a life as the treble portion of an active set-up, the power amp chassis was assigned to dust gathering whilst it's PSU box was assigned the less mundane duties of powering some experimental pre-amplifiers. But the SE dream is deferred not defunct so watch this space..... I had a stash of 12V6s and I just happened to have four 8k PP OTxs I had living with me rent-free. So PP won that battle.
The intention was to use as much of what was lying around which included a pair of wooden chassis I had had made years ago, some power TXs and, perhaps more importantly it meant no outlay on interstage transformers. There's a lot of dust in the stash room and under that dust a lot of good ideas lying fallow....and then came lockdown and indeed, the time was ripe to resurrect the parts of those now zombie dreams!
I have balanced output DACs, the aforementioned pre-amplifiers (will) also have balanced outputs so balanced inputs where now on the menu. The stage was set...
So:
Low power, PP and 12V6s means: triode operation
Balanced inputs without input transformers (zero budget) means: differential input
Balanced DACs having ~2VRMS output plus that pre-amplifier on the horizon means: minimum 2V input sensitivity
And having two of everything including chassis means monoblocks: woohoo!!
The best way to design power amps is backwards. I already have an OTX, PTX and a pair of tubes. The PTX is a 550V CT which gives about 350V using silicon rectifiers or 330V using a tube rectifier. 330-350V HT would be the voltage into the first capacitor. But this is to be a triode amp not a pentode amp - therefore a clean HT is required. Pentodes can tolerate more ripple on their anodes since it is the screen voltage that defines their current. But a clean HT means filtering and whether active or passive that is going to cost volts. On the other hand it is a PP amp which means it can reject HT ripple to a degree - but what degree that would be I was not sure. If the OTXs turned out to be unbalanced across their halves there would be bugger all I could do about it anyway, and I did have a pair of 15H 100mA chokes with 256R coil resistance. So now HT looked like 330V for silicon and 300V for tube recitifier. I happen to have a couple of GZ34s and since they look much, much nicer than a pair of UF4007s, HT was now 300V. Which isn?t bad because 300V, anode to cathode, just happens to be the max voltage for 12/6V6. Tube rectifiers give the power supply far more impedance than silicon would and this could impact performance?.but if them tubes are cooking in Class A then current demands are a bit more stable and the effects of extra impedance should be mitigated to a degree. The class A box duly got ticked. Max dissipation for these tubes is 12W, the curves show operating points round about 20V bias. This was to be a low power pair of amps but it?s the HT, tubes and OTXs that determine the maximum achievable power. Those devices had already been chosen and HT determined but, I decided I may as well get as much as I could anyways ? a bit like when I am drinking or smoking that zol stuff. Fixed bias gives me all of that 300V to use and so that box was ceremoniously ticked. 12/6V6s demand no more than 100k as grid resistors with fixed bias and, bar a coupling cap or two, that?s the entire power section done and dusted!
The input driver stage will be differential. Each ?side? will graciously accept 2V and gleefully return 20V. Well technically you take the output of a differential stage from between the anodes so you actually want 40V but your balanced DAC output is actually 4V between its legs (snigger, giggle). Either way I needed a gain of x10. Not much gain means driver will double up as the voltage amplifier. What about feedback I thought? Mmmmm. Feedback would mean more gain whch is not really a problem; just have to find the right tube. What is the problem though is whether I also want single-ended inputs and I did, as I have a couple of preamps that don?t have balanced outputs. Adding a single-ended input to a differential stage is easy - just ground one input (look closely at your Mullard front-end). OK but how do I apply feedback? Well with balanced inputs shunt feedback is easiest ? my Altecs are 8R the OTXs have 16, 8 and 4R taps so I could use the 8R tap as a center tap. But it all became a bit of a headache working out how to have that and the ability to switch to single ended inputs. Also feedback required a higher mu tube, I had a GZ34, I had 2x 12V6s and I didn?t really want a tithead to join the party - basically I wanted a GT-shaped tube with high mu. Enter then the 6SL7 - great tube, great reputation, great amount of plate impedance. You now remember them 100k grid resistors?.. So I thought of buffering the 6SL7 using a 6SN7. But if I did that ideally I would want the 6SN7s to be directly coupled to the output stage because this would give me class A2 ability. That would require a negative rail for the buffers, I had the bias rail but it was pretty lowish current due to the transformers I had on hand. And then I realised that I was running out of chassis space because I also had some rather large polypropylene capacitors I wanted to use for the HT. So feedback was dropped to allow SE or balanced inputs plus a non-electrolytic HT supply. No feedback means I needed a tube that could drive 100k from it?s anode and give me a gain of x10. Sounds like that 6SN7 I thought. Well turns out what I had were actually some 12SN7s mmmmm just like them 12V6s. Poindexter uses higher gm tubes in his diff stages and to a degree that is far more sensible because we (me and the Poinz) only have ~300V HT and such tubes are happier than a 12SN7 at lowish anode volatges. But, the 12SN7 is inherently a very linear tube so that will help. The obvious solution would be chokes or constant current sources as anode loads. I did not have those kind of chokes, so I decided on well-matched resistors with the option of trying the CCSs further down the line.
The differential stage had to use a CCS - practically because the negative bias rail wasn?t going to be very large (Poindexter has a dedicated negative rail for his diff stage) meaning I would be forced to use a low value tail resistor. Diff stages need infinite impedance in their tails to give perfect differential operation and a CCS is the easiest way to get close to that ideal. So CCS it was. Morgan Jones reccomends BJT cascode CCSs for lowish volatges and currents over power JFETs, Merlin Blenccoe seems to say nay, the power JFET based sources do way better even at low(ish) voltages and currents. I had lots of BC549s and only 5 appropriate JFETs so Jones won that one. Lo and behold! Them Hardcore Headbangers just got designed!
Next up, PP or SE? Well I built a ~1.5W 6BM8/ECL82...it was not bad with the A7s, was better with the Lowthers but pretty kak on rock music - so definitely not a headbanger. It used crappy OTXs, had some rather questionable bias points and unsurprisingly produced measurements that still make me blush. Hence after a bit of a life as the treble portion of an active set-up, the power amp chassis was assigned to dust gathering whilst it's PSU box was assigned the less mundane duties of powering some experimental pre-amplifiers. But the SE dream is deferred not defunct so watch this space..... I had a stash of 12V6s and I just happened to have four 8k PP OTxs I had living with me rent-free. So PP won that battle.
The intention was to use as much of what was lying around which included a pair of wooden chassis I had had made years ago, some power TXs and, perhaps more importantly it meant no outlay on interstage transformers. There's a lot of dust in the stash room and under that dust a lot of good ideas lying fallow....and then came lockdown and indeed, the time was ripe to resurrect the parts of those now zombie dreams!
I have balanced output DACs, the aforementioned pre-amplifiers (will) also have balanced outputs so balanced inputs where now on the menu. The stage was set...
So:
Low power, PP and 12V6s means: triode operation
Balanced inputs without input transformers (zero budget) means: differential input
Balanced DACs having ~2VRMS output plus that pre-amplifier on the horizon means: minimum 2V input sensitivity
And having two of everything including chassis means monoblocks: woohoo!!
The best way to design power amps is backwards. I already have an OTX, PTX and a pair of tubes. The PTX is a 550V CT which gives about 350V using silicon rectifiers or 330V using a tube rectifier. 330-350V HT would be the voltage into the first capacitor. But this is to be a triode amp not a pentode amp - therefore a clean HT is required. Pentodes can tolerate more ripple on their anodes since it is the screen voltage that defines their current. But a clean HT means filtering and whether active or passive that is going to cost volts. On the other hand it is a PP amp which means it can reject HT ripple to a degree - but what degree that would be I was not sure. If the OTXs turned out to be unbalanced across their halves there would be bugger all I could do about it anyway, and I did have a pair of 15H 100mA chokes with 256R coil resistance. So now HT looked like 330V for silicon and 300V for tube recitifier. I happen to have a couple of GZ34s and since they look much, much nicer than a pair of UF4007s, HT was now 300V. Which isn?t bad because 300V, anode to cathode, just happens to be the max voltage for 12/6V6. Tube rectifiers give the power supply far more impedance than silicon would and this could impact performance?.but if them tubes are cooking in Class A then current demands are a bit more stable and the effects of extra impedance should be mitigated to a degree. The class A box duly got ticked. Max dissipation for these tubes is 12W, the curves show operating points round about 20V bias. This was to be a low power pair of amps but it?s the HT, tubes and OTXs that determine the maximum achievable power. Those devices had already been chosen and HT determined but, I decided I may as well get as much as I could anyways ? a bit like when I am drinking or smoking that zol stuff. Fixed bias gives me all of that 300V to use and so that box was ceremoniously ticked. 12/6V6s demand no more than 100k as grid resistors with fixed bias and, bar a coupling cap or two, that?s the entire power section done and dusted!
The input driver stage will be differential. Each ?side? will graciously accept 2V and gleefully return 20V. Well technically you take the output of a differential stage from between the anodes so you actually want 40V but your balanced DAC output is actually 4V between its legs (snigger, giggle). Either way I needed a gain of x10. Not much gain means driver will double up as the voltage amplifier. What about feedback I thought? Mmmmm. Feedback would mean more gain whch is not really a problem; just have to find the right tube. What is the problem though is whether I also want single-ended inputs and I did, as I have a couple of preamps that don?t have balanced outputs. Adding a single-ended input to a differential stage is easy - just ground one input (look closely at your Mullard front-end). OK but how do I apply feedback? Well with balanced inputs shunt feedback is easiest ? my Altecs are 8R the OTXs have 16, 8 and 4R taps so I could use the 8R tap as a center tap. But it all became a bit of a headache working out how to have that and the ability to switch to single ended inputs. Also feedback required a higher mu tube, I had a GZ34, I had 2x 12V6s and I didn?t really want a tithead to join the party - basically I wanted a GT-shaped tube with high mu. Enter then the 6SL7 - great tube, great reputation, great amount of plate impedance. You now remember them 100k grid resistors?.. So I thought of buffering the 6SL7 using a 6SN7. But if I did that ideally I would want the 6SN7s to be directly coupled to the output stage because this would give me class A2 ability. That would require a negative rail for the buffers, I had the bias rail but it was pretty lowish current due to the transformers I had on hand. And then I realised that I was running out of chassis space because I also had some rather large polypropylene capacitors I wanted to use for the HT. So feedback was dropped to allow SE or balanced inputs plus a non-electrolytic HT supply. No feedback means I needed a tube that could drive 100k from it?s anode and give me a gain of x10. Sounds like that 6SN7 I thought. Well turns out what I had were actually some 12SN7s mmmmm just like them 12V6s. Poindexter uses higher gm tubes in his diff stages and to a degree that is far more sensible because we (me and the Poinz) only have ~300V HT and such tubes are happier than a 12SN7 at lowish anode volatges. But, the 12SN7 is inherently a very linear tube so that will help. The obvious solution would be chokes or constant current sources as anode loads. I did not have those kind of chokes, so I decided on well-matched resistors with the option of trying the CCSs further down the line.
The differential stage had to use a CCS - practically because the negative bias rail wasn?t going to be very large (Poindexter has a dedicated negative rail for his diff stage) meaning I would be forced to use a low value tail resistor. Diff stages need infinite impedance in their tails to give perfect differential operation and a CCS is the easiest way to get close to that ideal. So CCS it was. Morgan Jones reccomends BJT cascode CCSs for lowish volatges and currents over power JFETs, Merlin Blenccoe seems to say nay, the power JFET based sources do way better even at low(ish) voltages and currents. I had lots of BC549s and only 5 appropriate JFETs so Jones won that one. Lo and behold! Them Hardcore Headbangers just got designed!
