W/Repairs

Adapting the Type W plug-in to use 6DJ8 or 12AT7 tubes instead of 8416

It's common to find a Type W plugin missing its three type 8416 electron tubes. The 8416 is the industrial version of the 12DJ8. 8416 and 12DJ8 are crazy-expensive when you can find them at all.

It is possible to modify a Type W to use 6DJ8 (the 6 V version of 12DJ8), which is expensive but readily available.

Studying datasheets, I found that under the conditions in Type W, the 12AT7 has almost the same g_m and r_p as the 6DJ8. The 12AT7 is cheap and abundant.

I have three Type W's. I modded serial number 7465 to use 6DJ8, and 7111 and 2369 to use 12AT7. In each case, I calibrated before the mod using my three precious 8416's, and again after the mod.

Two instruments failed on bandwidth. (2369 spec is 23 MHz, actual is 21 MHz. 7111 spec is 26 MHz, actual is 25 MHz.) But they did no better with 8416's; performance is equal with 6DJ8 and slightly better with 12AT7.

The two mods are about equal in difficulty. The instrument should be calibrated after either mod.

6DJ8/6922/ECC88/E88CC/7DJ8/PCC88

NOTE: These types are often advertised as identical, but their heater current is different. You can use any type but you can't mix them.

Adjust the series resistor and current supply for the type you use (the numbers below are for 6DJ8).

The 6DJ8's heater is half the voltage (6.3 V) and twice the current (365 mA). It requires series resistance to drop the extra 19 V, and a power supply for the extra current.

The supply must be isolated and electrically quiet.

You might think you could just wire them in parallel and run off 6.3 V_AC, but some mainframes elevate that rail to +100 V which will put an uncomfortable 200 V between heater and cathode on V124. Besides, Tek used DC to avoid hum.

A) Add 100 Ω, 5 W in series with R290. I put it on a three-point terminal strip bolted to the top of R290.

B) Add an isolated current source across V124-V214-V114.

Start with either a 6 V - 24 V step-up transformer (details below), rectifier and reservoir, or a shielded switching supply powered by +350 V.

The current regulator is just an LM317 with a resistor between OUT and ADJUST, and output from ADJUST.

Two ¼ W resistors in parallel, one 18 Ω, the other 12 Ω, gives 180 mA, raising the total to 360 mA. (Or use two 15 Ω and a 100 Ω.)

The supply must handle 100 V backfeed if a tube burns out. Just put a diode in series. That's why I regulate current instead of voltage.

12AT7

A) Add 1.2 kΩ, 2 W in parallel with V124-V214-V114. This bypasses 30 mA around the tubes, reducing the heater current from 180 mA to 150 mA.

B) Tek used V214 pin 9 as a tie point. It must float when a 12AT7 is used, because it's the heater center-tap.

C) V114B and V214B grids must be biased 3 V positive for successful operation of the Q154-12AT7 cascode. Details below.

Other types — not tested

12AZ7 requires supplement and grid bias. Curves == 12AT7. Meets V_hk.

12BZ7 requires supplement and grid bias. Meets V_hk.

12DT8 requires shunt R and grid bias. Curves == 12AT7.

6BQ7/6BZ7/6BS8/8BQ7A require series R + supplement + grid bias.

Tube types that do not work

12AU7
12AV7
12AX7
12AY7
12BH7

Heater-Cathode breakdown

Most types exceed their V_hk spec, including the 8416 itself, assuming it's the same as 6DJ8. If Tek can ignore it so can we, I guess.

Effect of a heater-cathode short on V124

R126 will smoke.
Cathode may arc to grid but I don't see it burning anything.
Stuff downstream might get hot.

Transformer power supply for 6DJ8

I used a 120-0252-00 transformer which is used in Type 3T77 and 3T77A. I have several.

The best position is the rear bulkhead upper right corner with the long axis vertical. I put the rectifier/cap/regulator on the rear bulkhead to the left of the transformer, securing the rectifier and cap with hot glue to facilitate short leads.

I strung the diode between the cap and the LM317 to avoid a tie point.

With 200 μF/50 V (two 100's in parallel), I see 4 V peak to peak ripple at IN, with the trough 8 V above OUT. It's a symmetrical triangle wave.

I heat-sinked the LM317 to the rear bulkhead, with a mica insulator.

I can't see any hum at 1 mV/div and line triggering.

Maybe one can use a 120-0481-00 transformer (used in Type 1A4). It has a 20 V winding that puts out 135 mA, should be okay at 180 mA since the other winding won't be used.

A 120-0177-00 (used in Type Z) is 6 V-6 V, so it's no good.

Maybe a 110 V - 36 V transformer on pins 13-14 with primary and secondary reversed, rectifier, filter, and an LM317 would work. The transformer needs to have low magnetic field because it's right next to the tubes.

One AliExpress candidate is YHDC. If we reverse a 110 V primary, we need at least a 20 VA rating to stay within spec for copper loss. That means a big transformer, and hum unless you shield.

Specify 110 V | 110 V primary and 15+15 secondary. Convenient available space in the plugin is 65 mm × 50 mm × 50 mm.\

Here are YHDC 20 VA transformers:

PE5424E-M is PCB mount, 57 mm × 48 mm × 44 mm. Not a toroid.

The following don't fit:

PE5424K-M PCB mount plus holes, 75 mm × 49 mm × 44 mm.
LKB5424-L bolt-down with wire leads, 75 mm × 43 mm × 50 mm.
PU3917B low-profile bolt-down, 68 mm × 55 mm × 33 mm.
PTC25 25 VA PCB with hole toroid – 60 mm × 60 mm × 38 mm.

Switching power supply for 6DJ8

The Sanmim SM-PLG06A-24 is a 6 W model, 51 mm × 24 mm × 19 mm, cost $3. Several stores sell this. Just search for "sanmim" or "sanmin" and grovel through the hundreds of hits for 24 V and 5 W or 6 W. I bought mine at "Your Cee".

>>> SUCCESS!

Mounted in a steel box, with simple filtering on input (1 kΩ) and output lines (0.33 μF), it causes barely perceptible trace thickening at 1 mV/div.

Grid bias increase for 12AT7

Applied to S/N 2369 and 7111.

12AT7 is fine as-is in V124 and V114A/V214A, but V114B/V214B need an additional mod.

Q154's V_CE is the tube's grid bias. The 6DJ8 operates at –2.5V, which gives Q154 2 V. The 12AT7 is only –0.5 V, so Q154 starves. We must give the grids a positive offset.

I tried a resistive divider, 100 kΩ/100 nF to 10 MΩ to +350 V. This fails due to grid current from captured electrons, a normal phenomenon in low-bias tubes like the 12AT7.

Batteries work. One CR2032 in series with R152/R252. Minus side to Q154 emitter, plus side to the resistor. Out of RF paranoia I put a cap across each battery. 100 nF axial-lead MLCC fits in the small space. I_g varies from 10 μA to 70 μA depending on the tube.

Since current is drawn only while the instrument is running, the batteries should last for their shelf life. I put them in insulated wire-lead holders anyway, to make replacement easy and to keep them from shorting.

The batteries dangle below the ceramic strips to minimize stray capacitance. I unsoldered 5 terminals (silver-bearing solder!) and levered parts and wires out of the way to reach R152/R252. It's easier than it sounds.

The grid bias mod is necessary for the 12AT7, but it's compatible with all tube types.