I think most of that diagram relates to the ST alternator.
If you have a dc exciter its dc output brushes will go to the main alternator field brushes.
With its original AVR it may have had the exciter field fed as a shunt winding from the exciter dc brushes, in that case the avr would be a simple regulator controlling that shunt field, the ac line connections would just be for voltage sensing.
The other possibility was that the AVR rectified part of the ac voltage and supplied the exciter field directly, the ac connections then would be voltage sensing and also supplied the exciter field. If it was a fairly accurate AVR there would have been a stabilising signal taken from the main alternator field brushes.
To attempt to use the AVR you have, you need to feed its output to the exciter field, from your comment it may be ZZ but it will not be the main alternator field. ( It will only be one field and no diodes are involved). You will have to flash the field with a battery to get the polarity correct.
If it works you will probably find that it won't stabilise, the other possibility is that you don't have enough residual for it to build up without flashing each time.
If it doesn't stabilise you may be able to use the AVR to feed the main alternator field without the exciter, but I suspect the AVR will not be man enough for a 35kVA alternator main field.
The more likely solution would be to use a Brushless AVR which would have the stabilising network for the exciter and main alternator time constants. The brushless alternator is basically what you have except that the dc commutator and brushes and the main field brushes are replaced by a rotating rectifier, the AVR will not know the difference.
As the main field is not available for stabilising in a brushless machine the stabilising is fudged within the AVR and will still work via a dc exciter.
Flux