The isolation transformer output would feed the bridge rectifier input, to give the 90V DC supply for the motor.
However, from the original schematic, it appeared you were using a smoothed 90V supply, which it turns out is not the case.
As you do not have any smoothing on the motor supply, it's not actually working as a true PWM controller, and may give some odd resonances as the PWM pulses drift through the AC half cycle timing.
You may be better off with a conventional "Drill speed controller" type circuit? Those interact with the motor back EMF so actually give pretty good control, rather better than an open-loop PWM system.
This is a typical circuit of that style - for a DC motor, add a bridge rectifier before the power input, but no smoothing! It relies on the zero crossing of the AC waveform for the thyristor to turn off at each half cycle.
And what looks to be a similar unit in ebay:
Find many great new & used options and get the best deals for AC 110V-220V 4000W Thyristor SCR Adjustable Motor Speed Controller Module at the best online prices at eBay! Free delivery for many products.
www.ebay.co.uk
The use a suitable thyristor and an adjustable supply voltage to the gate (set by P1 in that circuit). Any time the motor back EMF drops below the gate voltage, the thyristor fires and maintains the motor speed.
Very simple, very reliable and that version even has a kind of "soft start" effect if you switch the input power to it, as C2 charges up to the set voltage over a second or two.
The TIC106D is rated 5A, you would need a higher rated one for a bigger motor, if you intended building one.