Moskido - Hybrid Aikido/MOSFET amplifier

[DarthBubba]

Greetings!

I'm just joining this thread and have not yet had the time to read all 159 pages, so I apologize if this was just asked a day ago: Would someone please direct me to the most recent version of the schematics for this project. The "Official Version" is preferred if there is one, and then any of your personally designed ones too. I have an extra Hafler DH-200 chassis w/good MOSFETs that needs a front-end transplant.

Until later...

[koldby]

Look at post 1547.
Koldby

[beau2317]

Here's a link. But I'm about to change the "official" version soon, so it uses aikido-push/pull version. But it's only a couple of resistors different. When I do I will publish the whole lot again.

[DarthBubba]

Thank you both, koldby and beau2317! I look forward to the revised push-pull version.

What kind of output (total wattage to the speakers) does this current design generate?

BTW, has anyone experimented w/two CCDA's in series, the first a 12AX7 (sort of the voltage amp), the second a 6SN7 (sort of the voltage and current amp) that then drives the MOSFETs?

[beau2317]

Hi Darth - welcome to moskido-land!

Wattage : you can design the output stage to produce various wattages. But because they are class A watts, they somehow seem louder than regular watts (Actually true for some reason). The one I made, which is what is documented in the drawings, produces 50W class A int0 4 ohms. I run Magnepan 3.6R speakers, which people normally say need 200W plus, but the moskido drives them just great, to loud volumes.

If you want to drive 8 ohm speakers, or make more or less power then you can adapt the output stage - and we can help with that. But remember that with class A amps, you need BIG heatsinks as they make a lot of heat. My one makes about 150W per channel in heat!

CCDAs - there's an interesting article by John Broskie (the designer of the Aikido on which the moskido is based) where he talks about CCDAs. Here's the link.

But he says:

How did it sound? I found the sonic presentation familiar, as I have built many CCDA line-stage amplifiers over the years. In fact, I was surprised by how good it sounded, no doubt due to the high-quality power supply components, heater regulator, and high-end coupling capacitors. Still, it was no Aikido, sounding only as good as your average $3,000 tube line-stage amplifier ;} I missed the sonic ease and wider sound stage that the Aikido offered. (By the way, I used the exact same 12AU7s and power transformers and chassis that I had used in the Aikido All-in-One line-stage amplifier; so it was oranges versus oranges.)

If you want to learn more about tubes, John's site is a treasure trove!

[DarthBubba]

Originally Posted by beau2317

Hi Darth - welcome to moskido-land!

Wattage : ...you can design the output stage to produce various wattages. But because they are class A watts, they somehow seem louder than regular watts (Actually true for some reason).

... But remember that with class A amps, you need BIG heatsinks as they make a lot of heat. My one makes about 150W per channel in heat!

What level of output do you think that the heat sinks on a Hafler DH-200 would support? Has anyone already adapted the driver card to do this. I'd like to use the MOSFETs that came w/the Hafler, as they already are matched.

Originally Posted by beau2317

CCDAs - there's an interesting article by John Broskie (the designer of the Aikido on which the moskido is based) where he talks about CCDAs. Here's the link.

Okay, language problem. When I was more heavily into Hollowstate equipment CCDA referred to Constant Current Differential Amplifier, A.K.A. "the long-tailed pair" with a current source. It's a self-balancing input/driver circuit w/two diff amps more-or-less cascaded, the first acting mostly as a voltage gain stage (a 12AX7) and the second acting mostly as a current gain stage (6SN7). The 6SN7 was much better at driving 6550/KT88 output tubes due to its higher current output. Both gain stages had constant current sources in the "tail" which helped significantly with making the input stage immune to the ravages of the output tubes on the power supply.

Anybody done anything like that?

[beau2317]

Not with mosfets. They don't require much in the way of current to drive. The idea of the moskido is to use the aikido as voltage amp and mosfets as current amp. And the aikido has a high PSRR along with it's great sound.

I used a CCDA in my DAC, where the DAC's differential output goes to a CCDA which goes into a broskie cathode follower to do balanced to unbalanced. See here.

[parb]

im planning to drive 8R speakers and i think i need more power than you published. the speaker designer recommends no less than 60W and no more than 110W at the upper end.

to me this suggests that i increase the voltage to the mosfet's as well as add in more outputs.

i am obviously concerned about cooling, and i'm spending quite a bit of time thinking through how to cool this. i'm even trying to think through how to build a copper block with water channels to mount the mosfets on and transport the heat outside of the device to some form of heat exchanger/radiator.

i digress though, how would you recommend me to go about to get to 60w into 8R loads from the moskido. I'd like to keep the Voltages from the PSU so that i can use a whole pile of 63V capacitors that i already own.

any thoughts?

[beau2317]

parb : I doubt that you need more power, but you will need more volts.

Let me explain more...

Let's forget about AC for the moment and just consider simple resistors and DC.

Power = V*V/R or V= root(power*R) = root (50*4) = 14.1 volts

What this means is that a 4R resistor needs 14 volts to dissipate 50W

But for an 8R resistor, the equation is root(50*8) = 20 volts.

So you need 50% more volts.

Now the moskido as I have it runs 30V rails. This is actually plenty for 4ohms. You see, for DC, 14 volts is enough, but for AC you need to multiply by root(2), so the output needs to swing about 20V for 4R or about 30V for 8R

So you need about 36-40V rails for 8R

BUT, you also need less current. So where I run 2.5A standing current for 50W (the equation for that is (2.5*2)*(2.5*2) * 4 /2 = 50)

For 8R, you can use only 1.8A standing current.

======================
Back to what the speaker maker recommends - basically there is no difference between 60W and 50W. Even if you double the power (eg go from 50W to 100W) you can barely hear the difference as it's a logarithmic scale. You get a 3dB increase with a doubling in power, and 1dB is the minimum sound level difference that a human can hear. Going from 50 to 60 would produce well less than 1dB in difference.



My Magnepans are normally recommended to have 200W plus, but with my moskido they can make you jump. There's no lack of volume.

========================
Heat : this is the big problem with class A. At least you are thinking about the problem. The usual solution is to use big heatsinks. But you also use smaller heatsinks with a fan (or fans). You can get very low noise fans now with magnetic bearings for PCs. They are very quiet and long life. I use one in my DAC and I can't hear it inside the case, so that's one possibility.

I'm note sure about water channels. Water and high voltages and hot tubes don't mix too well..

[parb]

thanks for reminding me of some fundamentals. Its been 25 years since i last worked on analog circuits and while i have studied it, its been a long while and clearly i got derailed in my thinking.

i spent the evening talking to my brother who is a designer of cooling and heating technologies for scientific instruments (he uses analog circuit sensors to measure nano degree temperature differences as well as create very perfect heaters that are programmable).

he has used the water cooling mechanisms for some things, to move heat away from the object being cooled and then using peltier elements to cool th water.

for what it is worth, my brother is actually in your camp, he suggested i go with forced air and a larger sized (120mm) fan in a high efficiency heatsink arrangment. the one he thinks should be good enough is a 0.07C/W heatsink that he has used for similar arrangements (they use mosfet controllers to control their peltier elements and drives 50V 6A RMS through those mosfet devices). he told me he will go and look at the datasheets tomorrow (we are 12 timezones away from each other and it got late).


i really think i should think through how i want to tackle heat as that will settle a lot of other things for me.