My EL84 SE

[jerluwoo]

I just wanted to share my little EL84 SE amplifier I built a few months ago. I purchased a Motorola phonograph console at a local flea market, which was horrible sounding. So I decided to gut it and make my own circuit using the parts. What is a little different about it is that the output tube (EL84/6BQ5) has its cathode grounded through the output transformer secondary, giving 100% feedback so that the EL84 is running as a power follower. There is an article by John Broskie which sparked the idea. At first it sounded a little dark and I didnt think I was going to keep it this way, but, after a week or so of use it settled in and sounds quite nice. Attached is the schematic as it is.

[Jeb-D.]

That is an interesting feedback method which I want to experiment with when I get time. C5 and R8 probably aren't necessary since the grid is pulled down from R2 through the transformer secondary. Also, have you tried it with feedback only to the output stage?

[Boris_The_Blade]

It's not exactly a power follower, but a small % of cathode feedback, according to the transformer ratio. Also gives a bit of ultralinear effect since vg2 is fixed.

[Jeb-D.]

Also gives a bit of ultralinear effect since vg2 is fixed.

Good catch. In the TCJ artical g2's decoupling cap is tied to the cathode, so Vg2 follows the cathode voltage which makes it still Pentode. This is not exactly the case in the above schematic.

[jrevillug]

I haven't seen that topology since I traced the circuit of my rescued Stereogram (console) amplifier, er, today.

This amp dates back to the 50s - there is nothing new under the sun.

It does have the cathode resistor and capacitor in it too. The first stage is a 12AX7 with gridleak biasing.

Sounds pretty horrendous - loads of magnetic couling between the power and output transformers, all those 50-year-old components and the lack of power supply filtering conspire to make it worse than it should be.

I am going to use the iron and tubes in a DIY amplifier - not sure whether to use this schematic (if I may, that is) or Alex Kitic's RH84. I am leaning toward the RH84 at the moment.

James

EDIT - do you have a link to the TCJ article? I can't seem to find it.

EDIT 2 - My console amp doesn't have any feedback around the 12AX7, though.

[Jeb-D.]

do you have a link to the TCJ article? I can't seem to find it.

http://www.tubecad.com/2006/06/blog0068.htm

[jrevillug]

Thank you

[Jeb-D.]

Mmm. after reading that article, the claims seem a little delusional. Using the (1/gm)/winding ratio equation given, I've gotten some pretty low output impedance calculations with only 6-9dB of feedback. I'll be curious to simulate some when I get a chance. If you get a chance to do some simulations let me know.

[jerluwoo]

Thanks also for posting the link Jeb, it seems a bit difficult to find exactly what your looking for on JB's website. True enough that r8 can be omitted but c5 is there because the winding resistance of the OT secondary produces a voltage drop. In my case it is 30mV. I could also use considerably less nfb to the 12ax7,or none at all, but my source produces 2Vrms and the amp just begins to audibly distort at max volume, so this setup works well for me. The low output impendance is quite noticeable in the bass region as it is quite tight and smooth more like a SS amp would be, but it still has the nice warm midrange we expect of a tube amp. Imaging is quite nice as well. The power supply uses a 5y3gt rectifier and basic rc filtering, Supply noise is hard to hear even with an ear right against the speaker.

[Jeb-D.]

but c5 is there because the winding resistance of the OT secondary produces a voltage drop. In my case it is 30mV

That is cautious thinking, which isn't necessarily bad. However, if you were to remove the cap, your source will be separated from that 30mV by that 470k+47k ohm resistance.
With the capacitor in there the source will see 0mV of that 30mV and the grid of the tube will see all 30mV.

If you remove the cap and you’re driving it with a direct coupled source with a Zout of example 1k you would have

((1k / (1k + 470k + 47k)) * 30mV = .058mV on the output of the source

((1k + 47k)/(1k + 47k + 470k)) * 30mV = 2.7mV on the grid of the tube.

These numbers are in similar range of offset errors from op-amps in sources or voltage produced by grid-leak of your input tube.
If the amp is driven by a capacitive coupled source then it will just behave as it does now with the coupling cap. I'm not saying your wrong by having it in there, but just that there aren't really any dangers by removing it.