Mighty mite balanced SE classA headphone amp

I was going to call this amp 'fleapowered' but after doing some searching online I found fleapowered amps in general were much higher output power than this - single digit Watts typically. Whereas this one is more than a couple of orders of magnitude lower than that (under 10mW) so deserved a different moniker. Seeing as a mite is smaller than a flea, 'mighty mite' seemed to me as good as any descriptor for it.

The mighty mite started life in my imagination after I walked into a local headphones shop and tried out some cans that the shop assistant recommended to me - AT MSR7s. I'd brought my 'Buffalito' headphone buffer with me and some step down transformers for its output - fed from my mobile phone its way too loud for my AKGs and Superluxes. However the MSR7s were too loud even with the 2:1 step down trafo so I told the assistant I'd come back and listen again once I had a more suitable trafo. I figured it needed to be 4:1 step down, so I built a pair of those (PQ32/20) and went back about three days later.

I really liked the MSR7s - particularly their high sensitivity - so I bought them and began dreaming up the ideal amp to drive them. Perhaps classA and battery power could mix, given they're so sensitive? I figured that given these cans needed such a low voltage to go loud, that it would be possible to run classA from a battery and even get decent battery life (by which I mean at least a whole day's listening from a charge).

In fact the projected battery life is quite a bit better than that, given that it runs at 15mA from a LiIon cell rated at 2.8Ah. However seeing I'm not regulating the supply perhaps the full capacity won't be available - still even if only 80% of the charge is usable, there will be about 150hrs of listening from the single cell.

The design is about as simple as I can make it - unbal-bal input trafos which allow the following CCS loaded darlington stages to handle at least 0.5VRMS each without hitting the rails, then bal-unbal again in the output trafo. The OPT is a step down, about 4:1. A single stage LC filter from the LiIon cell attenuates any battery noise there may be, load-induced supply noise is in theory non-existent due to the balanced single-ended approach.

Update - I have found that the lowish shunt inductance from my output trafos causes distortion on some recordings - not classical (which is my main diet) but movie soundtracks with a lot of bass. So I've increased the bias on each of the current sources from 3.5mA to 6mA. This doesn't entirely cure the distortion on my first prototype - I shall entirely cure it with replacement transformers with higher inductance. One thing I've found is that with cheap ferrite cores they don't always fit together very closely along the join and the amount of gap left is crucial. Clamping the two sides together is one solution, using more tape and winding it around more tightly is another - I shall try this second method on my trafos. When I measured the shunt inductance it was about 850mH.

Update2 - I'm so enamoured with the lack of sound of this amp I'm part way through the second build, I want one to loan out on demos while I'm still listening to the first one. On this one I'm experimenting with smaller input trafos, using EP17 for which I have some higher mu cores. Curious to know if the shunt inductance makes a difference to the SQ. I've added a pic of the active elements of this - small enough to fit on a couple of postage stamps. There are 16 BC857Bs in all, plus a few inductors because I'm paranoid about parasitic oscillations in EFs

Update3 - I've been puzzled by what seems to be the very high sensitivity of these MSR7s - trying out the amp with my other cans (AKG 240s, Superlux 668Bs) they just don't go loud enough. But the manufacturer's specifications aren't very different - AT say the MSR7s are 100dB/mW. But then I stumbled upon innerfidelity and on there they've done measurements - the measured sensitivities are starkly different. For most headphones they're quoting around 0.5mW to achieve 90dB SPL, but for the MSR7s, the figure is just 0.01mW So this little mite-power amp is going above 110dB SPL with the MSR7s - but really isn't suitable for any others in its current form.

Attached Thumbnails