Is there such as thing as secondary side regulation based on a simple binary feedback, ie "you're too high" and "you're too low", without indicating by how much? It seems that would be less vulnerable to imprecision in the feedback, and would only require an analog comparator, rather than error amplifier.

 

Certainly a well adjusted on/off feedback system can work quite well, depending on how fast things change and how close control is needed. The oven on the stove and the HVAC system in the house are all on/off systems, at least most of them are. And a lot of people drive that way, as well..Drive a bit too fast and run up on the car ahead, then hit the brakes, then hit the gas again.

So really, a (Bang Bang) servo can work for a lot of things. BUT it is harder to get right.
What do you want to control??
Our Federal reserve system was running that way two years ago, and it sort of worked, sort of. Stomp on the gas, then stomp on the brakes. Certainly it can be done and certainly it can be made to work.
For a DC to DC converter it can work also. BUT how much variation can you live with?? It is all a compromise. But it has to start with a set of requirements: just how close does it need to be.??

 

Is there such as thing as secondary side regulation based on a simple binary feedback

Yes.
Look up "Hysteretic switching regulator".
It has the advantage of the feedback loop being stable without any frequency compensation.

Below is the LTspice sim of a simple 5V hysteretic buck converter:

 

a lot of people drive that way, as well..Drive a bit too fast and run up on the car ahead, then hit the brakes, then hit the gas again.

BUT how much variation can you live with??

My thought is that response time can be made slow enough that there won't be excessive ripple. That is, the comparator should react fast to tiny differences, so when Vout is correct, the comparator would toggle hi/lo very fast. But then filter it to remove ripple.

how close does it need to be.?

Close? If the comparator is hovering around the target voltage, and feedback is smoothed, the output voltage will be steady. There mustn't be any hysteresis in the comparator.

Is a 3-state feedback possible? "low", "high", "locked".

Look up "Hysteretic switching regulator".

Googling "Hysteretic switching regulator" "isolated" "adjustable" gives mostly patents

Below is the LTspice sim of a simple 5V hysteretic buck converter:

Thx for that! But I need an iso converter, Vout adjustable from less than 1V to 5V.

 

The grandfather of all buck regulators the 78S40 works that way, and so do its offspring (MC34063 etc).
So does the LinkSwitch (as I mentioned in your other post on this subject).
Add a clocked D-latch on the output of the error amplifier and you have a Delta-Sigma modulator.
Generally it's a swap for loop stability and simplicity versus accuracy.

 

The grandfather of all buck regulators the 78S40 works that way, and so do its offspring (MC34063 etc).
So does the LinkSwitch (as I mentioned in your other post on this subject).

but not isolated.

Add a clocked D-latch on the output of the error amplifier and you have a Delta-Sigma modulator.

as part of a converter?

Generally it's a swap for loop stability and simplicity versus accuracy.

Constant On-Time (COT) seems to give both.

COT control has become the de facto solution over traditional current and voltage mode control methods due to its faster transient response, higher efficiency, fewer components, and ease of design.
https://www.monolithicpower.com/en/advantages-of-constant-on-time-control-in-dc-dc-converters

Is the "on pulse generator" shown here the Delta-Sigma modulator?

 

Is the "on pulse generator" shown here the Delta-Sigma modulator?

No, it's just a simple hysteretic regulator.

For isolation you would need a fast digital opto isolator between the comparator output and the pulse generator input, and a switching circuit with a transformer or coupled inductor.