"Shrine" SE Amp using 6AH6 and 1625

[wrenchone]

Some exchages with Tubelab have inspired me to build an SE amp similar to his Simple SE design, but using a 6AH6 pentode in triode mode at the input and a 1625 output tube (both extremely cheap and plentiful). I'm using the budget Edcor 5K single ended transformers at the outputs. The unit will be powered with an SMPS of my own design, providing power for both B+ and filaments. the whole shooting match will go in a 6" cube with the tubes poking out on top and the output iron and power supply inside. Finish will be black crackle, with a blue LED pilot light to complement the plate wires. Attached is a picture of the top deck. Next step is to power up the amp with a couple of bench supplies to look for screw-ups and check bias points. After that it's wring out the power supply (built and putting out voltage, but still needing a real going-over and EMI scan), integrate supply and amp, and throw it in the box. Schematic for the amp will follow when I'm sure it's all working.

[wrenchone]

Here is a pic of the messy bottom...

[wrenchone]

Here is the preliminary amp schematic. I don't expect it to change too much except perhaps some tweaks for bias.

[Brian Beck]

Are you asking for feedback? If not, please ignore the following. If you are, let me suggest that the load CCS on the first stage might be improved for real sonic benefit. The CCS, whatever it is, will leave a sonic footprint in this critical location. I think that the non-linear capacitance behavior of CCSs explains much of the sonic differences between them. Differences in CCS capacitance can be surprisingly audible, even when the RC time constant suggests a pole frequency in the megahertz region, due to minute phase intermodulation effects in the audio band. All this assumes that the low-frequency dynamic resistance of the CCS is very high to begin with, which should be the case here. Specifically, you might consider something to reduce the non-linear capacitive load on the 6AH6’s plate created by Q3’s Ccb through C2 to the B+ supply. If Q3 were a FET, the gate resistance would be so high that you could then bias the gate with very large-value resistors which would provide a series isolating resistance to Cdg. If you insist on using the MJE350, choose a high-hfe MJE350 and try inserting some resistance into the base circuit. Just my two cents…

[wrenchone]

Noted - nothing comes for free. The MJE350 is there to protect the JFET and stiffen the current source, hopefully increasing its output impedance. Putting a substantial resistance in series with the transistor base would muck up the cascode effect. I may try putting a few kohms in series with the drain, but not until I fire the amp up and see how much voltage swing I have available with the loop closed.

[motherone]

Nice work!

[Brian Beck]

Originally posted by wrenchone
Noted - nothing comes for free. The MJE350 is there to protect the JFET and stiffen the current source, hopefully increasing its output impedance. Putting a substantial resistance in series with the transistor base would muck up the cascode effect. I may try putting a few kohms in series with the drain, but not until I fire the amp up and see how much voltage swing I have available with the loop closed.

I don't think putting some resistance in series with the drain would help in this circuit. The MJE350's Ccb is still run straight to ground through C2 and Q2 (which is used as a zener I surmise). A HV FET in the place of the MJE350 would allow those large gate resistors with no degradation in cascode performance. Or consider a double (cascode) Supertex DN2540 See the link for performance, which indicates that it presents only 0.177pF to the load.

http://www.pacifier.com/~gpimm/ccs_performance.htm

[astouffer]

Good to see the 6AH6 being used. Its a fine tube.

[frank754]

Once you have this completely ironed out & tweaked nicely, I'd love to see a detailed schematic.

A single-ended 1625 design would be really cool, with the proper
power supply and both pushing it high yet aiming for good full-range audio performance, especially music.
The 1625's with plate caps really look cool, and may do well
if designed for SE, parallel SE, or PP, keeping the components limited and effective (lean).

There are a few 1625 / 807 amps out there but not too common.
I had posted a thread a few days ago about the same subject.
I saw a few last night on a schematics site, like these for example:

http://www.nijnkonijn.nl/machmat/sch...p/gromm215.gif

http://www.nijnkonijn.nl/machmat/sch...pp/a515amp.gif

http://www.nijnkonijn.nl/machmat/sch...p/807ab2a1.gif

http://www.nijnkonijn.nl/machmat/sch.../ppp/807-s.gif

[zigzagflux]

Originally posted by Brian Beck
If you insist on using the MJE350, choose a high-hfe MJE350 and try inserting some resistance into the base circuit.

Interesting analysis, and easy to overlook these stray capacitances.

What about the possibility of the attached current source? I have even considered replacing the LED with a few series schottky's, which have extremely low capacitance and excellent high frequency behavior.

It would appear to me that the capacitance path you have identified would be mitigated/minimized, yet still keeping the same basic design. I would even think you could reduce the 1uF cap by a factor of 10 and not really affect the performance all that much. LM285 has very low dynamic impedance.