[sdiy] Tube amplifier question

[jhaible at debitel.net]

Hi Rene,

wow, very informative post, thank you!

> I'm using a tube amp for monitoring my synths, its a single ended one, 
> based on a russian EL84-like tube. If your speakers are sensitive, then 
> 3W can be plenty....

They aren't sensitive - IQ Mini Lady (I think this was the name) in the
living room and four JBL Control 1's in the studio.

But I listen 99% on headphones, and pretty much Sennheiser HD450s
and HD480s exclusively, so I can design the headphone amp to match these.
The Power amp is just to provide some comfortable listening level when
somebody comes visiting, or when I sit at the bench and solder.


> Sounds like this might be something for you:
> 
> http://www.aleph.co.jp/~takeda/radio/tube/12AU7-muFollow/6N6p-WhitFlw-exp-
schema.gif
> 
> Also very similar to the audio stage of your compressor.


Yes, that's exactly what I was thinking about. What do you think about
the EF86 triode-connected for the voltage gain stage?


> Ouch... I wouldn't recommend a 12AX7 with that. 

Thanks for warning me!


> I'm driving even 32ohms headphones with mine, 

And without a transformer!

>although its actually a mismatch. 

Have you calculated the approx. output impedance?

I guess with a hundred volts of unloaded output swing,
mismatch won't hurt at all - we don't want to vaporize
our headphones after all.
I even wonder about the *optimal* output impedance.
Often when phones are connected to power amps, there's
just a series resistor ar not-so-low-impedance 
voltage divider, and while this certainly changes the
frequency response, I don't know what I'd prefer, now
that I'm accustomed to the rather high driving impedance.


> (Note that the values are optimized to give the lowest output 
> impedance.) 

Is this so? I was surprised to see R4 having a rather
low value in your circuit. I thought higher anode resistor
would decrease output impedance (but decrease large signal
drive capability). So I would have thought your cuircuit
was optimised for drive capability instead of low small
signal output impedance. (?)


> A WCF is a push pull circuit and as such works into an 
> optimum load, where the action is most symmetrical, and the second order 
> distortion dissapears. (I had the maths for the WCF somewhere, but it 
> eludes me right now, but I can dig if necessary.) That means that there 
> will be a compromise between matching and damping.

I have found WCF formula here:
http://www.tubecad.com/october99/
and I have made spice simulations with crude 12BH7 models. The models
are far from accurate, but I hope I can at least trust the trend of
some component value increasing or decreasing a parameter like output
impedance.

> The turn on transients are not very critical, if you provide a discharge 
> path for the cap. (In case you plug in the phone after turn on.) The 
> soft starting feature of the tubes prevents "plopps" on turn on. 
> (Provided you don't use a relay-delay circuit to switch the plate supply.)

I also worry about switching the amp on shortly after switching it off.
I tend to oversize PSU capacitors (avoiding chokes), which will worsen
the problem. I could use a relay to protect the phones, but can I trust the
relay? A transformer might cut the worst spikes by saturating (will it?).
I'm quite anxious about my phones (and even more about my ears).
And I never plug in the phones after power-on. This must be absolutely
foolproof - I tend fall asleep when I hear music at 3:00 AM, and imagine
how a brownout of 20 seconds might wake me, with the phones still on! (;->)

JH.


-------------------------------------------------
debitel.net Webmail