Automatic Tap Changer (or AC Voltage regulator) - Ideas please!!

[oldway]

Back in 1988-89 I used to work at Sola Basic Australia. We used to make UPSs, ferro resonant transformers and tap switchers. I used to work on them all day every day.

Did anyone actually read my post #19?

Yes, your post #19 is all right.
With inductive load, we must ensure 2 triacs will never conduct together or use a low value power resistor to limit short circuit current between taps.
On low end triac voltage regulators used for computers, I often had to repair them (all triacs faulty) only because user had connected his fan at the output (inductive load).

Slow acting voltage regulator (tap's switching by relays) is dangerous for the load: let's think mains voltage rapidly vary from 90 to 220V, the load voltage will then go to 537V for a short time before switching from 90V tap to 220V tap.

[xibalban]

The cheap commercial units on eBay use simple relay tap switching, so that's what I would try.  Here are some photos of the guts of one.

Thanks for the pictures  ! I see that the unit uses relays, but apart from a bunch of transistors where is the intelligence of the electronics to change between taps?
Coming back to the pertinent question of inductive loads and the challenges associated with switching those, if I could live with the manual knob system - I ought to be fairly content with the use of triacs. The caveat would be to feed the MCU with information about "completely off" state of the preceding triac, thus compromising on speed of operation.

Slow acting voltage regulator (tap's switching by relays) is dangerous for the load: let's think mains voltage rapidly vary from 90 to 220V, the load voltage will then go to 537V for a short time before switching from 90V tap to 220V tap.

Do you remember my stating that I would retain the analog over-voltage relay (that senses the output voltage), which originally came with the manual regulator which simply disconnected the load if the output voltage exceed some pre-set value?

[oldway]

The caveat would be to feed the MCU with information about "completely off" state of the preceding triac, thus compromising on speed of operation.

Absolutely not. It does not compromise speed of operation. If you are sure previous triac does not conduct any more, you can trigger any other triac without risk, it has not to be sequential.
In my project, I was integrating every positive half cycle of input voltage and choosing by comparators what was the right tap. This is possible because such a voltage regulator has a low internal impedance and voltage drop is also low.

[xibalban]

The caveat would be to feed the MCU with information about "completely off" state of the preceding triac, thus compromising on speed of operation.

Absolutely not. It does not compromise speed of operation. If you are sure previous triac does not conduct any more, you can trigger any other triac without risk, it has not to be sequential.
In my project, I was integrating every positive half cycle of input voltage and choosing by comparators what was the right tap. This is possible because such a voltage regulator has a low internal impedance and voltage drop is also low.

Do you have a sample schematic to share, if that's not much of a problem?

[oldway]

Do you have a sample schematic to share, if that's not much of a problem?

I make this project in 1983, I don't have this schematic any more, and I can't have it because I am now living 10.000 Km's far from where I lived in this time and the company where I worked does not exist any more.
I can only give you informations that I can remenber.
That was not a simple tape changer because in such case, triacs had to be high current triacs and this was too expensive.
In my project, it added or substracted voltages with 9 levels, using 6 triacs TIC226.
Nominal input voltage was 80 to 140V, output voltage was 110V or 127Vac +/-5% (two models) 1Kw
Two auxiliary transformers where used, adding or substrating voltages in proportions of 1 and 3.
The steps where:
-4 : triacs of -3 and -1 on
-3: triacs of -3 and 0 on
-2: triacs of -3 and +1 on
-1 ; triacs of 0 and -1 on
0 : both triacs 0 on
+1 : triacs of 0 and +1 on
+2 : triacs of +3 and -1 on
+3: triacs of +3 and 0 on
+4: triacs of +3 and +1 on
To avoid changing of taps continually, comparators had an hysterese.
There also was a delay avoiding changing taps before to stay a complete cycle on the same tap so DC voltage can't be applied to the load.
As the auxiliary transformers had their secundaries in serie with the load, they could not stay with open primary, as they would act as current transformers and develop high primary voltages, I had to accept the occurrence of short circuits during steps changes with inductive loads and protect the triacs with low value power resistors.

Here is an example of a voltage regulator with 115Vac nominal output and 8 taps.

[oldway]

Do you remember my stating that I would retain the analog over-voltage relay (that senses the output voltage), which originally came with the manual regulator which simply disconnected the load if the output voltage exceed some pre-set value?

You need 30 or 40ms for actuating the relay, more some ms or tenth's of ms for the over-voltage sensor. This could be too slow to protect electronic instruments.

What seems to be the best and more secure solution would be:
- a SMPS with PFC (input 90 to 230Vac), output 48V 20A (for telecomunications)
- an inverter 48V / 220V 1KVA (seems to be enough if you don't use all your equipment at the same time)

[xibalban]

Do you remember my stating that I would retain the analog over-voltage relay (that senses the output voltage), which originally came with the manual regulator which simply disconnected the load if the output voltage exceed some pre-set value?

You need 30 or 40ms for actuating the relay, more some ms or tenth's of ms for the over-voltage sensor. This could be too slow to protect electronic instruments.

What seems to be the best and more secure solution would be:
- a SMPS with PFC (input 90 to 230Vac), output 48V 20A (for telecomunications)
- an inverter 48V / 220V 1KVA (seems to be enough if you don't use all your equipment at the same time)

I get what you imply, and I am seriously going to consider it. Here, my multitap-transformer wouldn't be much useful though.
Your advice and feedback are highly appreciated. Thank you  !

[edavid]

Do you remember my stating that I would retain the analog over-voltage relay (that senses the output voltage), which originally came with the manual regulator which simply disconnected the load if the output voltage exceed some pre-set value?

You need 30 or 40ms for actuating the relay, more some ms or tenth's of ms for the over-voltage sensor. This could be too slow to protect electronic instruments.

A small relay will actuate in under 10ms.  How long does the generator take to ramp up?  Isn't it at least a few cycles?