[R-390] 26Z5W replacement

[John Wendler]

Rationale for replacing the 26Z5W:
1) Cost
2) Lifetime
3) Excess heat (filament + plate resistance)

Rationale for getting rid of the 1N561 diodes:
1) The power supply on hand had the solid-state mod done with a butane
torch instead of a soldering iron and the diodes were damaged.  The 1N561
is no longer manufactured.
2) The voltage drop across the diode is small, leading to greater output
voltage and potentially stressing the other components in the receiver.

A pair of vernier calipers and interpolation were used to pull points off
the IV-curve on the Tung-Sol 26Z5W data sheet.  The model was fitted using
Matlab.  The best fit was:

Ip = (1.1e-3)*Vp^1.47 where Ip and Vp are the plate current (A) and voltage
(V), respectively.  (Plates in parallel, as in the R-390A)  The residual
error was 9.7e-6 over Vp = 0-40 V.

A SPICE-like simulator (Agilent Advanced Design System, ADS) was used to
evaluate various options to match the curve in the 100 mA - 200 mA range
that in which the R-390A runs.  (Standby / Operational)  I looked at the
variation over temperature, as well.

The first option was to replace the tube with a 1N4007G diode.  The 1N4007G
is probably a good modern replacement for the 1N561.  Like the 1N561, it
has an extended temperature range.  This voltage drop is somewhere under 1
V and the equivalent series resistance is negligible.  The consequence of
any single diode replacement is that the output voltage is greater than the
circuitry was originally designed for.

The second option was to cascade 8 1N5007G diodes in series with an 80 Ohm
resistor.  This moved the knee voltage out to the 4-6 V range (over
temperature) so that the linear portion of the solid-state and tube I-V
curves would coincide.  The problem was that the voltage drop is somewhat
sensitive to temperature, changing 1.3 V over the range -40 to +85 C.  Many
people might be fine with this as most are not running over an extended
temperature range.

The third option was to cascade a 1N4007G with a 1N5337B Zener and an 82
Ohm resistor.  The drop across the zener moved the knee out to the right
region so that the series resistance would overlay the linear part of the
curve.  This simulation gave very close agreement with the tube curve over
the critical 100 mA to 200 mA range, with minimal variation over
temperature.  This was the option that was built.

Construction was on perfboard.  The resistors were 82 Ohm Dale RH-10 10 W
1% resistors.  I used 1 mm diameter brass rod for the tube pins; the
nearest english-unit rods will be too thick or too thin.

A Keithly pulsed Stimulus-Measurement unit was used to evaluate the I-V
curves.  Unfortunately, this is a 100 mA maximum compliance instrument.
 The knee ends up outside the tube's knee, as expected.  At 100 mA, the
solid-state version is approaching the tube curve (within 0.1 V or so) and
is on a slope to intersect.  Note that the series resistance will likely
increase under continuous duty, because the series resistance of diodes
will increase with temperature.

Overall, this is a pretty good replacement to a 26Z5W, and a much better
replacement that a single solid-state diode, regardless of part number.
 The filament heat was removed but the heat due to the plate resistance
still remains - this is what drops the voltage to the original design.

Modeling or measuring the zener in series with the 1N561 and 82 Ohm
resistor was not tried; it would probably to do a reasonable job, as well.

No attempt was made to do a gradual turn on to emulate the warming of the
heater.  The 82 ohm resistor will provide some current limiting at turn on.


References and notes:

1) Any such project must reference Dr. Kurt Schmid's (Sigmapert) beautiful
replacement diodes. His results are works of art and are designed to fit in
the same physical envelope as a tube.  I have never touched one and so
cannot comment on the circuit.
2) Perugini, Stefano, "Vacuum diode Models & PSpice Simulations.  (Article
originally appeared in Glass Audio, V. 10, No 4., but can be found on web)
3) Leach, W. Marshall, "SPICE Models for Vacuum Tube Amplifiers," J. Audioo
Eng. Soc., V 43, No. 3, March 1995.
4) http://www.excem.fr/download/usergui5.pdf (Excem Vacuum Tube Modeling
Package User's Guide)
5) Koren, Norman, "Improved Vacuum Tube Models for Spice Simulations,"
http://www.normankoren.com/Audio/Tubemodspice_article.html
6) Standby and Operational current estimates from The R-390 - Y2K Manual:
http://www.r-390a.net/Y2K-R3/index.htm

As always, thanks to the members of the R-390 list for encouragement and
answers to questions!

73 de N5CQU

John (JP) Wendler