Need substitute for old Siliconix FET to replace Thyratron

I primarily want it to work in these three applications used in this V200 model of Seeburg jukebox: Read-Out, Write-In, and Trip.

Seeburg made a nice little test circuit for the 2D21s as used in their machine - if the tube works in the Read-Out circuit using the adapter, then it will work in the other two as well. Read-Out was always the most critical - reading if those torroids flip state or not. I have a theory document somewhere on my computer, but can't find it because I didn't name it correctly or put into the correct directory - it spoke of how this new model of jukebox (the articlas of 1955) used the torroids as an example of magnetic memory storage in the mid-50s...if I find it I will post a link, it was historically interesting, but not too helpful for solving this particular problem AIR.

Later someone can take that new circuit and use it to replace most if not all 2D21s - would be fine by me!

Thanks,

John :-#)#

<snip>

Have you got a working unit you can use as reference?

RL

Do you have broken semiconductor replacements that you're trying to repair, or are you anticipating constructing a semi replacement based on the literature at hand, assuming that they will work?

RL

I have a working jukebox with a working 2D21 for testing.

I just want to fix this PERMANENTLY! At least, as permanent as solid state items are when run well within their limits that is... Many jukeboxes had trouble with this version of control centre, and my hope is to make a simple replacement part for the 2D21 that can simply be plugged into the tube's socket.

John :-#)#

You have 'one' working 2D21 that you're swapping around to check individual local cct function in a TSR?

What is the selector reciever type number that ~ functions? There are two types in the Seeburg manual. One uses 4x 2D21 - the other uses 2x 2050.

Have you got 7-pin bases to build 2D21 substitutes onto? What's the PN and supplier?

The problem isn't to make the circuit go, so much as to duplicate the innards of the thyratron package. The patent drawings depend on independent heater power to develop 'sky-hook' voltages internally for the substitute cct.

I missed the the -96V bias on the V1, V3 and V7 6V3AC heaters in TSR1-L6. These are the tubes running cathodes at ~ -150VDC.

TSR J5 carriee 150VAC to 'mechanism. Where does this link up in the manual.

You DO know that you can still buy these things, new from distribution? (from Newark FGS )

Not cheap, admittedly, but usefull to make sure that everything else is functional, before diving off the deep end, or starting a science project. The latter would definitely end up costing somebody more than a few weekends of fussing around.

RL

Yes, and a bunch of "new" 2D21s that are too fussy.

I'm using the 2D21 version TSR1-L6

Yes, I found them on eBay, they should work.

From the 0A2 which is the regulator tube on the 6X4 negative supply.

I've bought new 2D21s from Newark, and they don't always work in the machine.

It is possible the machine I have has a problem with the TSR1-L6, but I've rebuilt it, verified the voltages, new resistors, caps, scoped it, poked and prodded, and there is only one 2D21 that works just fine in the Read-Out circuit out of perhaps ten I have that were NOS - five from Newark.

I made a simple solid-state 2D21 that operates the Trip function properly, but won't work in the Read-Out. The Read-Out spikes from the Detent Switch are not getting close enough to Zero volts to trip most of the tubes. The spikes look perfect at the Detent Switch and at the junction of C11 and R12.

Unfortunately I do not have another TSR1-L6 to compare this one too, I suspect I may be overlooking something really obvious...

John :-#(#

Read-out spikes from the detent switch (from J6 pin5, supposedly aimed to fire V3 read-out thyratron) are applied through C11/R11 to a node (C11/R12) that is biased at -150VDC.

To fire the thyratron, this C11/R12 node voltage has to approach the

-118VDC cathode voltage stored on C13 (0.1uF). If the detent switch produces a grounded pulse (150V positive-going impulse), there should DOUBLE sufficient signal still left after filtering at the junction of R12/R13, based on modelling.

If C13 hasn't charged 'up' to -118V, the read-out pulse going into J5 pin5 may be ineffective, but its recharge time is less than 20ms. Recharging the pulse forming input network from the detent switch is longer ~40ms.

Leakage paths for C13 include C1 on J2 pin7 and anything in the 'Service Switch', if J2 is connected. Anything loading J6 pin5 will reduce the amplitude of the detent pulse - it's impedance is very high (at 2M7).

Simulations suggest that with componts currently employed in grid drive and timing, very simple mosfet-buffered SCRs will function in the floating cathode circuit, even with simple back- to-back zeners protecting the gate and a 1K gate-cathode resistor on the scr. There is a delay while the mosfet gate charges, but that's all. No skyhook voltage or heater power required.

(LTcad built-in models for BSC320N20NS3, S4025L, and 1N4746 were used)

Function of V7 Write-In is practically the same.

RL

Sims also suggested that anode loads above 220R seriously affected discharge timing, probably due to high scr 'hold' current. The fet-scr 'darlington' would discharge the 100n cap effectively over a wide range of gate input resistor (1k-100K), if the original input pulse was clean.

RL

Well, thanks for the lesson on the operation of the circuit! I must confess that I never understood it as well as you have figured out with the little information that I have provided.

Much appreciated!

May I ask how (or what program used) you are simulating the operation? If this is an inexpensive simulator I'd consider getting it so I can poke in values to see what I would get...

John :-#)#

<snip>

LTspice is free from (now) Analog Devices.

RL

RL

404 - not found...

I was busy the last week and went to look just now.

John :-#)#

Try again.

The posted link was ok, but the target had a 'space' character mistakenly inserted - and present in my 'checking' links.

I didn't think that a space character could function in an html link,so wasn't looking - appeared as a line wrap on this reader.

RL