Tír Na HiFi

Step up transformers are working now, its sounding great on 'audiophile' stuff (Suzanne Vega for example, her voice startled me on first appearance) but less than stellar with my normal fayre of classical. Gets a bit muddled at the top when there's a lot happening, so I'm going for the MOSFET buffer after the trafos as the next step. With a K339 between the TKs and the trafos the input capacitance will go down by an order of magnitude.

<later> Buffer engaged... goosebumps!

Great stuff

Got any more pictures

What did you get with the buffer in
Resolution, clarity, tautness ?

Still having trouble getting my pics down small enough - what's the limit for a .jpg?

The buffer made classical music listenable. On less complex stuff it increased the emotion quotient very nicely. Highly addictive now, can't wait to build the next power level version... :D

I find file hosting easier
I use Flickr is that blocked ?
Hosting allows the use of 1000x1000 pixels or under

So it was sorta smooth sounding but a little muddy through the complex before

A flavor some might prefer?

I think all the photo sharing sites are blocked. I had been using Pinterest but that got annihilated in the past 3 months. Sent you a pic on LinkedIn messaging.

the hexagon PCB is a bank of caps just for an experiment,
not sure they're needed.
Buffer's got a 0603 red led on it ,
trafos are under the PCB next to it.

Thanks nige - forgot to add the small board in the middle has 4 output caps, two for each channel. That's where the individual buffers are paralleled. I've not done measurements to prove that each side of the parallel buffer is pulling its weight, but I'll get to that eventually. For now its too much fun to play music...

are u gonna pull that cap bank out ?
see what changes?

hows the sound developing?

I'm not doing really serious listening as I don't have a set up where I have a listening seat facing a pair of speakers. I guess I need to package up the DAC, give it its own PSU then I can set up a system away from my workbench. So there won't be serious 'sweetspot' listening for a while, just off-axis which is still jolly satisfying as the speakers are effectively invisible.
Next up to get the active gain stage finished also I'm working on a DAC which directly outputs a 20V+ swing which will be an interesting comparison.

What impresses me most about the sound of this amp are vocals. Guess that means that its got excellent mid-range. Even before I fitted the buffer Suzanne Vega's voice sent shivers down my spine. I've also fired up Leonard Cohen's last album, the palpability is awe-inspiring. I also went back to an old reference, Rebecca Pidgeon 'The Raven' but that's no longer a reference for me, its clear through this set-up that her voice is smeared, washed-out - I suspect its the S-D ADC Chesky used (and have now gotten rid of).

Nige - since you were planning on going active sometime in the future, you might like the next idea I have for a tweeter amp. (The ramblings that follow are thinking on ways to get more voltage swing out of the step-up trafo (more powa!)).

The step-up transformer idea for voltage gain looks to be a good one judging by the resulting SQ - however with a 1:10 step up we're reaching the practical limit of step-up ratio. Any higher and the leakage inductance together with the self-capacitance (even assuming no load) of the secondary starts to encroach on the audio bandwidth. Put another way - the self-resonance freq of the trafo gets too low to be practical.

Ways to address this include reducing the leakage inductance through different winding geometry - interleaving the windings. But that'll probably increase self-capacitance so no net gain. Going to a toroidal will certainly help lower the leakage L but I don't know a way to DIY a toroid with a few thousand turns on its secondary. Commercial toroids are a possibility though but the smallest is 15VA and with a steel core its likely the self-capacitance on that secondary will increase (ferrite being an insulator). I do have some 15VA toroids here I got made for step-down so I could have a play with the LCR meter. <Did that - got 40nF for the input capacitance (12V in, 110V out) and the leakage L is 10X lower (even lower at HF). So probably there are dividends to be paid in interleaved windings>

Then I got to thinking about the limitations on the output voltage of the trafo and realized its limited by the maximum input voltage (1VRMS) which in turn is determined by the low-frequency bandwidth requirement - we want to go down to 20Hz. But what if we give up that requirement? Then with a restricted LF bandwidth we can stick higher input voltages into the trafo without worrying about saturating the core. If we designed a system that needed an external powered sub then we could set the LF limit at 80Hz without too much trouble, buying us a factor of 4X in output voltage. More than plenty for any practical SE classA amp (40VRMS) and with the same small and cheap trafos.

But instead of 80Hz what could be achieve if we built a much more restricted LF amp - i.e. one which goes down only to 2kHz? In other words, a tweeter amp. The extra voltage swing available at the input means we could use a step down trafo at the output which might well help with tweeter dynamics as your toroidal step-downs helped on your battery-fed amp. Running the TK13E25Ds at (say) 200V instead of 24V makes their feedback capacitance (Crss) much more linear and also bootstraps away their input capacitance to negligible (I estimate 2pF). The output transformer only needs to be a small ferrite cored one, not a big steel toroidal as the working freq is so much higher.