This inverter is identical externally to the 50-127 described here.

The ratings are the same as the 50-127, with an output of 110V 60c/s at 75W continuous and 100W intermittent. It is not clear why there are two Terado models that look the same, with identical ratings. However, an internal look shows things to be quite different.

The most obvious difference is the use of a dual interrupter vibrator; something I was expecting to find in the 50-127 but was noticeably absent. Instead what appears to be a standard single contact type was used in that model. It could be they upgraded to a dual interrupter type for longer vibrator life, and hence produced the model described here.
In the 50-138-3, the vibrator is a 1006 dual interrupter type with a 6V driving coil. For 12V use, a 22R series resistor is used.
The transformer primary has two separate windings which is the correct method when using a dual interrupter vibrator. Doing so provides more equal current sharing between the two contact sets. Buffer capacitance is 10uF on the secondary only. There is no other RF output filtering. Input filtering is provided by a choke and condenser in the same way as the other models. RFI suppression is also provided by .01uF ceramic condensers across one set of contacts. Why the other set is not bypassed is not clear.

Construction is the usual compact layout as per the other Terado inverters. All that is above the chassis is the transformer, vibrator, and fuse.

Under the chassis shows the new 10uF buffer condenser and other components. The 1uF paper bypass can stay as leakage is harmless.

It seems that Terado wasn't interested in servicing their own products.

Indeed, it is undesirable to run the inverter with no load, hence this warning. Experiments showed that 2uF of buffer capacitance was preferable for very light loads, but on full load the specified 10uF is actually required. Evidently, Terado gave up on the high voltage overwind and smaller buffer capacitor as used in the 50-127.
The specification for using a "Portable TV Only" is interesting. Why not any other load? A portable TV for 110V  will have a series heater string using 450mA or 600mA valves, so that in itself is a purely resistive load. The B+ will come from either a half wave rectifier, or a half wave voltage doubling rectifier. The current draw will therefore be asymmetrical, causing more loading than a full wave rectifier load would. It appears that this inverter is therefore beefed up to take care of this, so such a TV set would be better used with the 50-183-3 and not the 50-127. This would imply that purely resistive loads of more than 75W could be used on this inverter.
However, it does seem the ratings are being pushed as they are. If the TV set uses 450mA valves, the power to heat them is 50W, leaving only 25W for the B+, which equates to 220mA, about what a small set would require. For a set with 600mA valves, things are not looking good. Here, 66W is drawn by the heaters, leaving only 9W for the B+. That's only 81mA - hardly enough to run a TV set, even a small electrostatic one.
Obviously, the "Portable TV" must be one using 450mA valves. There is no way a non technical person would know if this was the case. One possible explanation is that this inverter was actually supplied with a specific TV set as a package. In fact, there is some mention of certain GE portables being sold this way in the 1960's. As the Compactron valves used in these sets have 450mA heaters, it would allow just enough current for the B+.

The output waveform is as it should be. Because there's a possibility of one set of vibrator contacts not working and the user not being aware of this, all four were checked with a light bulb in series with 240V to ensure they were clean and working. A new 12V supply cable with two pin polarised plug was fitted, and the paper buffer capacitor replaced with a new polypropylene type.