P&B CUF-42-70120 time delay relay

[TimFox]

I got a deal for surplus Potter & Brumfield CUF42-70120 time delay relays a few years back and incorporated them into some vacuum-tube audio equipment to sequence the high-voltage power supply after the heaters had a chance to warm up.  These devices were expensive ($183 at Newark), now seem to be discontinued, and are housed in yellow plastic screw-down shells with the connections made to faston lugs on one face of the insert.  Internally, there is a normal mechanical double-pole relay with a circuit board for the timing circuit. 
The time delay (1 to 120 sec) is set by an external resistance (0 to 2.4 megohm) between two terminals.   This morning, I had to replace one in my current preamplifier power supply, and am now running out of stock, since several had failed previously after months of operation in other amplifiers.  There is no obvious mechanical or contact-welding problem with the DPDT relay.
Of course, there is no documentation on the circuit board, which has through-hole parts:  one TO-92 device (labeled "MSU 12-2"), two polarized capacitors, one 1N400x rectifier diode, and several resistors.  The device is activated with 120 VAC applied to the two input pins, and the six DPDT contact pins are independent.  The external resistor goes to two pins that connect to the board.  The relay is rated for 10 A at 240 VAC, and 1/3 hp at 120 VAC (2 A), and the DPDT contacts switch a transformer primary after a shunt MOV and a series NTC thermistor.  The power transformer is a toroid that feeds two bridge rectifiers with a R-C snubber network across each secondary.  The input current to the rectifiers' primary is < 500 mA at 120 V/60 Hz.
My questions: 
Does anyone have experience with this series of P&B relays? 
Is there any information available about the circuit board? 
Should I suspect the tantalum (?) capacitor (45 uF/25 V) that probably is in the R-C timing circuit (the other capacitor is a Nichicon aluminum 33 uF/25 V electrolytic)?
The manufacturer claims that the unit is "transient protected".  Do I need extra protection for switching a transformer-rectifier load?  There is an R-C network across the two "120 VAC input" terminals, a 100nF/polypropylene/X2 capacitor in series with a 120 ohm resistor.

[TimFox]

I set this aside for a while, but finally fixed the problems.
1.  see attachment for traced circuit.  The active device is a house-numbered unijunction transistor, possibly a 2N4870.
2.  The voltage across the 33uF aluminum electrolytic measured about 30 VDC.  This is limited by the large series resistance and the half-wave rectifier to approximately the mean value of the half-sine-wave, with the finite load of the unijunction, resistor, and relay coil pulling some current through the 3.6k/1W resistor.
2.  The unknown resistor is probably > 2 megohm, to achieve a finite time with an open circuit.
3.  The 45 uF/25 V tantalum capacitor had failed short-circuit.  I replaced it with a 33 uF/35 V tantalum (slightly larger diameter and a tight fit), which reduces the maximum delay time.
4.  The relay coil was unprotected (no shunt diode) so I added an 1N4007 across it.
Now, it seems to work fine, but the original survived many cycles before failure.