Preregulation of a linear bench PSU

[Atom]

thanks for the link...
I'll have a look later, but as far as I remember HP uses triacs not scr, and they are switched on zero-crossing so a simple mosfet cannot replace them.
You need back-to-back mosfets, zero-crossing detector and insulated drivers.

Yes my bad, hp uses triacs and not scr since they are switching the ac portion and not a dv voltage. By the way beginner question here why spikes of current are harmful for mosfets? Mosfet rated for 60A continous current are for 200A spikes so even a large current spike of 60 amps or more shouldn't damage them.

For example the irf has a continuos current of 110amps and a peak current of 390A , i know that those are values that you can't even approach  but we aren't going anywere near them.

[Kleinstein]

If the MOSFET is turn on hard with capacitors on both sides, there can be quite some peak current spike. The thread started with the problem of high current spikes that looked like up to 100 A.
The current spike may be OK for the MOSFET if it does not get much lager. However with so large a current there is noise and also possible voltage spikes.  I would be more worried about loss at the capacitors and also some of those peaks coming through to the output. The audible noise is a second problem. On fast load changes (short to open, CC to CV) there could be also quite some current spike - though usually only at a low frequency.

For the switching there are several options: Triac before the transformer, relays before the rectifier, SCRs as part of the rectifier and a MOSFET behind the rectifier. The switching can be hard between taps or in some cases also with a continuous output. The circuit from the beginning is turning on in zero crossing and doing some continuous adjustment. The circuit does not look so bad - I would expect even more current spikes from the continuous adjusting  SCR version.

[Atom]

If the MOSFET is turn on hard with capacitors on both sides, there can be quite some peak current spike. The thread started with the problem of high current spikes that looked like up to 100 A.
The current spike may be OK for the MOSFET if it does not get much lager. However with so large a current there is noise and also possible voltage spikes.  I would be more worried about loss at the capacitors and also some of those peaks coming through to the output. The audible noise is a second problem. On fast load changes (short to open, CC to CV) there could be also quite some current spike - though usually only at a low frequency.

For the switching there are several options: Triac before the transformer, relays before the rectifier, SCRs as part of the rectifier and a MOSFET behind the rectifier. The switching can be hard between taps or in some cases also with a continuous output. The circuit from the beginning is turning on in zero crossing and doing some continuous adjustment. The circuit does not look so bad - I would expect even more current spikes from the continuous adjusting  SCR version.

i know that you are a really good EE and your knowledge is far greater than mine..i wish my professors could teach the things that i learn here. yes the preregulator proposed at the start has the problem that makes the transformer buzz and has a continuos adjustment creating high current spikes, but the circuit of not1xor1 switches only once in a time when i need to switch between taps so there would be a single current spikes at the mosfet turn on, what i'm trying to say it's that if the mosfet takes a current spike once there shouldn't be a problem since this circuit doesn't switches continuosly.

[Kleinstein]

For the circuit from  not1xor1 it still depends on how the switching is controlled. It could be switching with hysteresis and thus not so often or it could still be fast control within a line period.

If going from lop tap to high tap there could be quite some current spike, as the filter cap is charged fast. This could be more than the 60-100 A seen at start. Here switching near zero would still help, as the charging is no faster than the sine wave. Still some 10 V/ms and 10mF of filter capacitor would be 100 A in theory, if there is no other limit.

The other point would be rare cases where the switches are active both at the same time, if 2 Fets are used (using a diode instead of the 2nd FET (at COM) could avoid this). These peaks could be damaging.

Another point that could help a little against high current spikes could be a fuse before the rectifier - it's resistance also limits the current a little. As rectifiers sometimes fail short, it is usually a good idea to have a fuse to protect the transformer.

[not1xor1]

thanks for the link...
I'll have a look later, but as far as I remember HP uses triacs not scr, and they are switched on zero-crossing so a simple mosfet cannot replace them.
You need back-to-back mosfets, zero-crossing detector and insulated drivers.

Yes my bad, hp uses triacs and not scr since they are switching the ac portion and not a dv voltage. By the way beginner question here why spikes of current are harmful for mosfets? Mosfet rated for 60A continous current are for 200A spikes so even a large current spike of 60 amps or more shouldn't damage them.

For example the irf has a continuos current of 110amps and a peak current of 390A , i know that those are values that you can't even approach  but we aren't going anywere near them.

The problem of the previous circuit I showed is that in some instances both mosfet are on and that should be avoided.
The current spikes through the capacitor are less relevant as there are plenty of mosfets able to withstand that.

[not1xor1]

If the MOSFET is turn on hard with capacitors on both sides, there can be quite some peak current spike. The thread started with the problem of high current spikes that looked like up to 100 A.
The current spike may be OK for the MOSFET if it does not get much lager. However with so large a current there is noise and also possible voltage spikes.  I would be more worried about loss at the capacitors and also some of those peaks coming through to the output. The audible noise is a second problem. On fast load changes (short to open, CC to CV) there could be also quite some current spike - though usually only at a low frequency.

For the switching there are several options: Triac before the transformer, relays before the rectifier, SCRs as part of the rectifier and a MOSFET behind the rectifier. The switching can be hard between taps or in some cases also with a continuous output. The circuit from the beginning is turning on in zero crossing and doing some continuous adjustment. The circuit does not look so bad - I would expect even more current spikes from the continuous adjusting  SCR version.

The single tap switch uses just one capacitor. The current spikes are the usual current spikes you get when you switch on a transformer or when you switch transformer windings via relais. Nothing different, no noise, no overshots.
The only problem with my first draft was that in some cases both mosfets were on. That is easily solved by switching on zero crossing.
In the circuit I'll discuss tomorrow I also enforced fast switch off on both mosfets, just to make that even safer.

You get instead huge voltage spikes when you switch the mosfet off as soon as the capacitor gets to the required voltage, because the mosfet and the inductance of the transformer work like a stepup switching circuit.
You can only mitigate a bit the problem by slowly switching the mosfet off, but at the expense of more power dissipation.
And yet some more power has to be dissipated by the transformer and a bleeding resistor and or a TVS.

[not1xor1]

For the circuit from  not1xor1 it still depends on how the switching is controlled. It could be switching with hysteresis and thus not so often or it could still be fast control within a line period.

hysteresis plays no useful purpose here because even if the comparator would "misfire" at 50 Hz (i.e. every other cycle) the capacitor voltage would be higher than the center tap voltage and that alone would prevent the center tap mosfet conduction.
I'll try to show that tomorrow...

[Atom]

hysteresis plays no useful purpose here because even if the comparator would "misfire" at 50 Hz (i.e. every other cycle) the capacitor voltage would be higher than the center tap voltage and that alone would prevent the center tap mosfet conduction.
I'll try to show that tomorrow...

i'll look forward to it.

[not1xor1]

hysteresis plays no useful purpose here because even if the comparator would "misfire" at 50 Hz (i.e. every other cycle) the capacitor voltage would be higher than the center tap voltage and that alone would prevent the center tap mosfet conduction.
I'll try to show that tomorrow...

i'll look forward to it.

sorry... I've been too busy yesterday... anyway here it is



The circuit uses a couple of BJTs to detect when the 100Hz (or 120Hz) rectified AC gets to 0V. Then on the collector of Q6 you get a short (compared to the AC line period) positive pulse.

Each positive pulse then gets to one input of the two NAND gates U5 and U6.
Depending on the status of the voltage comparator U2 output (and inverted U1 one) the positive pulse becomes a negative one at the output of either U5 or U6.

If the flip-flop is not yet in the required status, the pulse changes that, switching off one of the mosfets and leaving the other get to the on state.

The circuit has been simulated with LT1013, LM358, LM393 (with pull-up resistors on the output) but should work with any rail-to-rail I/O opamp and probably with most single supply one. I also added some hysteresis as suggested by Kleinstein to reduce unneeded tap switching.

It looks like the peak current on switching from the low tap to the high one gets at once about 4 times higher than the full load peaks and then gets back to normal in 3-4 cycles. The output voltage ramps up at a speed of few milliVolts per µs so shouldn't cause too much trouble for the voltage regulator stage.

Possible improvements might include usage of N-MOSFET (greater choice, low cost and low Rds-on) and switching them off via an optocoupler.

With few modifications, including back-to-back MOSFETs switches and a bit more complex logic, two transformers and bridges (or a single one with dual split winding secondaries) might be stacked in series to get 4 voltage levels.

I'm also attaching the .asc file (see below). Apart from CD4000_v.lib and logic gates symbols it uses just LTspice stock components.

The following screenshots show the logic signals:





These show the MOSFETs current and source-gate voltage:











These last two screenshots show the peak current and voltage spike when switching from low to high voltage.
The simulation has been run with a huge 47.000µF capacitor and a load of 10Amps.

[blackdog]

Hi,

If you only need one switchpoint than take a look at this schematic, it uses a Schottky diode and a PowerMOSfet.
www.bramcam.nl/Diversen/CO-2016-PSU.pdf

The whole Power Supply Page on Circuitsonline: https://www.circuitsonline.net/forum/view/130041/1#highlight=voeding+co+0

And this is a page on Circuitsonline with my customized version of this power supply: https://www.circuitsonline.net/forum/view/131554/1#highlight=voeding+co%200

I don't think it's too difficult to adjust the circuit around the transformer, the extra Schottky diode and the PowerMOSfet for 5 to 10 Amps.

But beware, the peak currents at large electrolytics can rise considerably, which is normal with large capacitances and large output currents.

Kind regards,
Bram

[Atom]

These last two screenshots show the peak current and voltage spike when switching from low to high voltage.
The simulation has been run with a huge 47.000µF capacitor and a load of 10Amps.

Really nice circuit i'll try it, i have already on hand some low rdson Pmos
I was a little worried about the peak current 200A but then i saw that the measurments were done with 47000uf @ 10 amps

[Atom]

If you only need one switchpoint than take a look at this schematic, it uses a Schottky diode and a PowerMOSfet.
www.bramcam.nl/Diversen/CO-2016-PSU.pdf

thanks i'll try also this out

[Kleinstein]

Hi,

If you only need one switchpoint than take a look at this schematic, it uses a Schottky diode and a PowerMOSfet.
www.bramcam.nl/Diversen/CO-2016-PSU.pdf
...

That version of tap switching looks nice and simple.

The nice point with the diode is that, there can be no cross conduction. The voltage loss only happens at in the lower tap case, where is does not hurt much. The relatively slow turn on of the MOSFET limits peak currents, when going from the center tap to the full voltage. There would be some heat loss at the MOSFET, but not that much (e.g. comparable to 1/4 the energy in the filter caps) and more of a transient mode.
Moving the switching from low to high tap close to the zero crossing could further reduce the current peak a little.

[not1xor1]

That version of tap switching looks nice and simple.

The nice point with the diode is that, there can be no cross conduction. The voltage loss only happens at in the lower tap case, where is does not hurt much. The relatively slow turn on of the MOSFET limits peak currents, when going from the center tap to the full voltage. There would be some heat loss at the MOSFET, but not that much (e.g. comparable to 1/4 the energy in the filter caps) and more of a transient mode.
Moving the switching from low to high tap close to the zero crossing could further reduce the current peak a little.

I agree that Blackdog's circuit is much simpler (I made something even simpler as it was for regulation on the negative rail).

When converting from AC to DC you have a fixed voltage loss (max peak-to-peak ripple voltage + max dropout of regulator) plus diode bridge and transformer loss.

In the case of center tap transformer the diode bridge and transformer losses are halved, but the other aren't so saving that fraction of volt wouldn't be so bad. 

In any case such a complicated circuit is obviously an overkill for a single tap. I designed that just for the puzzle solving fun.
But in future I would start from there to replace the triacs in the HP 3 taps switching design.

[Kevin.D]

A full wave voltage doubler could be added on to that to give two extra extended ranges. There is one disadvantage I can see
though as it's halving the output capacitance (since 2 caps in series) on the lower voltage (higher current) ranges where its needed most.
Notice on the third range how the top caps gets charged upto double the voltage of the bottom cap to get us the third range (sw1 closed sw2 open).
The available outputs currents and voltages I give on the schematic are what should be theoretically available from the
transformer and don't include rectification losses.

Happy New Year . 

[Mechatrommer]

why dont use one chip solution? like 50¢ 34063? i once try to build my own discrete pre regulator from some comparator and/or schmitt trigger and/or mcu pwm. but with all the hassle, on another circuit, i just use the one chip solution, all i have to take care of is the noises spike and ground loop. in fact i'm building another one right now. ymmv.

[Kleinstein]

A switched mode converter is in deed a real option. However not an easy one, as it takes care to keep higher frequency noise out. The slow tap switching has a chance to produce less higher frequency noise.

However the MC34063 is not a good choice for higher power - if at all it is acceptable when the internal switch is sufficient and efficiency is not important. There are more modern chips available and also older standard ones drive an external MOSFET switch. A cheap old one would be UC2842 and similar.

It is in dead a little odd to look at the low noise tap switching at high power (e.g. 8-10 A current). Especially the slow phase angle like switching versions to get a continuous regulation can also produce quite some ripple / noise problems. Even with just a normal rectifier, at 10 A load current the peak currents at the filter cap can be significant.

[Mechatrommer]

However the MC34063 is not a good choice for higher power - if at all it is acceptable when the internal switch is sufficient and efficiency is not important.

it can drive external mosfet, pnp, npn etc. there are circuits examples in the datasheet for buck, boost, invert etc with or without external switcher (power element). the nice thing is comparator, hysterisis, PWM, logic are all taken cared of in single chip, there is also Timing Capacitor (TC) pin i believe to control switching frequency (but i dont care i just use recommended value) and current limit (Ipk sense) pin as well (which i also disabled in my current circuit). the output is open collector-open emiter bjt (much like optocoupler) so the limit is your imagination.

currently i'm upgrading my earlier version long serving but now dead dual tracking psu to add few features, one of it is smps pre regulation. in the attached picture (bottom part) you can see i have to arrange 34063 on -ve rail to mirror the 34063 on +ve rail (top part), its Vcc is PSU's GND, and its GND is PSU's -ve input voltage (VCC-, -20V in my case), feedback is uA741 full swing +ve to -ve rail supply, i know what fanboys are going to say, but nevermind i made them to work for my need, and dont mind the Vrfs (voltage reference for smps, slightly higher but adjustable than  Vref for linear regulation output) its not good enough i should've add buffer to it since i use it in voltage division, i overlooked that until i populate them on pcb.

i dont know what specificity of the OP to use diy version, sorry if its not suitable, just thought its related. if say we want to limit current spike maybe we can slow down the mosfet driver to managable level, in my circuit attached, the driver is maxed to about an amp in and out of the gate, maybe i overstress the 3906/3904 pair but o well. for power element you can put anything you like (AO3400/3401 rated 5A in my case although my psu only rated to 0.5 - 1A) fwiw...

[Atom]

it seems we have new people that have joined the discussion

I agree that Blackdog's circuit is much simpler (I made something even simpler as it was for regulation on the negative rail).

When converting from AC to DC you have a fixed voltage loss (max peak-to-peak ripple voltage + max dropout of regulator) plus diode bridge and transformer loss.

In the case of center tap transformer the diode bridge and transformer losses are halved, but the other aren't so saving that fraction of volt wouldn't be so bad. 

In any case such a complicated circuit is obviously an overkill for a single tap. I designed that just for the puzzle solving fun.
But in future I would start from there to replace the triacs in the HP 3 taps switching design.

today i had time to tes blackdog circuit so i used a 12mohm nmos in a to252 package the current of the load was 6/7amps at 25V ....the mosfet died after the first switch, i think the peak current killed it, no big deal it was a chinese mosfet

2nd try .....i don't have low voltage high current mos on hand....i used a 14nk50z to 247 on heatsink .....0.380ohm of rdson 500vds 14A continuos

https://www.st.com/resource/en/datasheet/stb14nk50z.pdf

i know that's not the right mosfet for the job but i wanted to try it, it works, but there is a problem the voltage measured from drain to source is 11V pkpk so 5.5v. @ 6.6A load

power disspiantion is Rdson = 5.5/6.6 = 0.83ohm not good  power disspiation is 0.83*36= 30W (the mos was on a heatsink).

yes that's not the right mosfet but it works, i'm waiting for low voltage high current mos.

for anyone intrested the diode used was byw99w200 : https://www.st.com/resource/en/datasheet/CD00000692.pdf

Happy new year to everyone!!

[Atom]

A full wave voltage doubler could be added on to that to give two extra extended ranges. There is one disadvantage I can see
though as it's halving the output capacitance (since 2 caps in series) on the lower voltage (higher current) ranges where its needed most.
Notice on the third range how the top caps gets charged upto double the voltage of the bottom cap to get us the third range (sw1 closed sw2 open).
The available outputs currents and voltages I give on the schematic are what should be theoretically available from the
transformer and don't include rectification losses.

Happy New Year . 

really intresting i was taught that voltage doublers are used in low current application and for generating supply rails if you have an ac signal nearby withouth using overcoplex circuits...never thought to use it in a power supply, it could be used in a more hv / low current psu.

why dont use one chip solution? like 50¢ 34063? i once try to build my own discrete pre regulator from some comparator and/or schmitt trigger and/or mcu pwm. but with all the hassle, on another circuit, i just use the one chip solution, all i have to take care of is the noises spike and ground loop. in fact i'm building another one right now. ymmv.

yes the mc34063 i have some on hand..but still i'm trying to keep the noise as down as possible so a switchmode prereg is excluded...btw that doesn't mean that i'm not intrested in switching prereg...maybe in a smaller and more compact unit than the one that i'm trying right now.
i've now ordered boards from jlcpcb the 100*100 mm pcb incudes a tpa3116 class d amp (but that's irrelevant, props to the author https://github.com/kurozz/kamp_tpa3116d2)
 
the other 2 boards are one with the LM2679 simple switcher, main candidate to a smaller psu prereg because it's really simple and can deliver quite a bit of power
the second one is based around the LM5116 Synchronous Buck Controller, the board were generated by webench designer and tweaked to make them smaller and better by me.
planning to play around with them and see what i can do, another option would be the xlsemi 4005 or similar but the 32v max input is a bottle neck

LM2679 : http://www.ti.com/lit/ds/symlink/lm2679.pdf
LM5116 : http://www.ti.com/lit/ds/symlink/lm5116.pdf
xl4005   : http://www.xlsemi.com/datasheet/xl4005%20datasheet.pdf

then i discovered that some seller on aliexpress is selling those dcdc industrial modules (3.4mm thick pcb) made from DELTA and MURATA, i'm going to take some of those since they are extremely cheap (4E). maybe the feedback of those one can me modified so we can control the voltage of the output.

MURATA  https://it.aliexpress.com/item/Lungo-staffa-con-la-dissipazione-di-calore-piatto-Murata-RBQ-isolamento-ad-alta-potenza-DC-DC/32957694449.html?spm=a2g0s.13010208.99999999.262.42653c00ZZeXhp

DELTA     https://it.aliexpress.com/item/Con-dissipazione-di-calore-Delta-Isolato-500-w-ad-alta-potenza-DC-DC-modulo-di-conversione/32955585915.html?spm=a2g0s.13010208.99999999.259.42653c00ZZeXhp

[prasimix]

Hi Atom, seems that you are on the similar journey as myself, searching for power pre-regulator solution. I've started with "blackdog" type of stuff, and gave up (possibly too early, since at least @tombi from this forum reported good results independently from @blackdog). I spent some time after that with LM5118, LM5088, and finally end up with LTC3864 for EEZ H24005 power supply. I didn't stop on DC-DC converter as the final solution since it still require a) huge mains transformer or b) an AC-DC coverter to get "DCbus" voltage for pre-regulator (48 Vdc in my case). Therefore I finally decide to make an AC-DC converter that can works with fixed or widely adjustable output voltage as pre-regulation. Progress is reported here.
I'd like to mention once again what was said here, if you are in the phase to try DC-DC converter as pre-regulator first.

Finally not completely on-topic (sorry for that): It's interesting that people when someone mention bench power supply start to talk about low noise and ripple. That could be indeed beneficial, but question is for what kind of applications? Does everyone is working with sensitive RF circuits, have ultra-precise voltage reference, or whatever? I'm still didn't find a good article or post about noise levels/categories with explanation when low PARD become of paramount importance and present a possible show stopper or not.

Keep up with good work, and all the best in New (Western counting) Year to you, and all members of the forum.

[Atom]

Hi Atom, seems that you are on the similar journey as myself, searching for power pre-regulator solution. I've started with "blackdog" type of stuff, and gave up (possibly too early, since at least @tombi from this forum reported good results independently from @blackdog). I spent some time after that with LM5118, LM5088, and finally end up with LTC3864 for EEZ H24005 power supply. I didn't stop on DC-DC converter as the final solution since it still require a) huge mains transformer or b) an AC-DC coverter to get "DCbus" voltage for pre-regulator (48 Vdc in my case). Therefore I finally decide to make an AC-DC converter that can works with fixed or widely adjustable output voltage as pre-regulation. Progress is reported here.
I'd like to mention once again what was said here, if you are in the phase to try DC-DC converter as pre-regulator first.

Finally not completely on-topic (sorry for that): It's interesting that people when someone mention bench power supply start to talk about low noise and ripple. That could be indeed beneficial, but question is for what kind of applications? Does everyone is working with sensitive RF circuits, have ultra-precise voltage reference, or whatever? I'm still didn't find a good article or post about noise levels/categories with explanation when low PARD become of paramount importance and present a possible show stopper or not.

Keep up with good work, and all the best in New (Western counting) Year to you, and all members of the forum.

the project of the power supply that you linked (the one on the diy audio forum is really intresting), and i was already following the development of your smps preregulator. that said i think a full smps prereg is far out of my capabilities (for now )

and for the noise thing personally (form a student stanpoint) i want to exclude any kind of noise(of the psu) from the pcb under test so i already know that the supply isn't the problem, so if you have to troubleshoot something you have already the certainty that the psu isn't the problem, and then when you're moving into more complicated stuff you have a nice low noise psu to help you.

i know that's not the right mosfet for the job but i wanted to try it, it works, but there is a problem the voltage measured from drain to source is 11V pkpk so 5.5v. @ 6.6A load

power disspiantion is Rdson = 5.5/6.6 = 0.83ohm not good  power disspiation is 0.83*36= 30W (the mos was on a heatsink).

yes that's not the right mosfet but it works, i'm waiting for low voltage high current mos.

mesuring the output of the psu there is a 100mvpkpk 1mhz wave (ac copuling).. i think is the mosfet oscillating ? probably because it isn't the right type of mos for this application, that would explain the large power dissipation and the fact that i wasn't able to go to the full 30vdc on the output, anyway i have to wait or the better mos to arrive.

[Mechatrommer]

the project of the power supply that you linked (the one on the diy audio forum is really intresting)

maybe not really hitting it but audio amps that i usually fixed last time are using TL494 running full banana (50% PWM 30+ KHz) into 1:1 & 1:-1 toroidal transformer to create dual rail. fwiw...

that said i think a full smps prereg is far out of my capabilities (for now )

if you want to learn offline smps, there are good people explaining it out there https://www.eevblog.com/forum/projects/offline-smps-psu-explained/msg1459572/#msg1459572 i myself dont have time for such project since no urgent need and yes its much more involved (and dangerous) project.

i know that's not the right mosfet for the job but i wanted to try it, it works, but there is a problem the voltage measured from drain to source is 11V pkpk so 5.5v. @ 6.6A load

power disspiantion is Rdson = 5.5/6.6 = 0.83ohm not good  power disspiation is 0.83*36= 30W (the mos was on a heatsink).

yes that's not the right mosfet but it works, i'm waiting for low voltage high current mos.

mesuring the output of the psu there is a 100mvpkpk 1mhz wave (ac copuling).. i think is the mosfet oscillating ? probably because it isn't the right type of mos for this application, that would explain the large power dissipation and the fact that i wasn't able to go to the full 30vdc on the output, anyway i have to wait or the better mos to arrive.

from the way you explain it i think your mosfet is running in saturation (current limit) mode (linear region in bjt term) its not fully off and not fully on, not enough hysterisis? the datasheet you linked is a beefy mosfet capable of 14A (48A pulsed). the fact that it only run at 6A and burn is that, running at current limit mode, its not about its a not good china knock off mosfet.

[not1xor1]

Hi Atom, seems that you are on the similar journey as myself, searching for power pre-regulator solution. I've started with "blackdog" type of stuff, and gave up (possibly too early, since at least @tombi from this forum reported good results independently from @blackdog).

So far I've not yet seen any data regarding power efficiency at various output voltages and currents.

In that kind of pre-regulator the MOSFET and the transformer secondary behave like a powerful step-up circuit.

And the problem is not just the huge voltage spikes, but all the energy stored in the transformer secondary which has to be wasted in some way.

Blackdog reduced a bit both the spikes and the energy by reducing the speed of the switch-off at the expenses of more dissipation in the MOSFET(s) switch.

But the doubt I got from multiple simulations of lot of variations of that kind of pre-regulator is that the efficiency might be beaten even by a simpe center tapped transformer.
I can't be sure of that as the transformer model I use is just too rough.

Besides that the screenshots posted in the blog you linked tell us nothing about the spikes.
It would be much more interesting to see in finer details what happens when the MOSFET goes off and that would require different time-base settings on the scope.

[tombi]

Hello,

Sorry - I've been away - I didn't see great spikes at switch-off of the pre-regulator and I also didn't have issues with the transformers getting hot or humming.

Could be the transformers I used were over-kill. I used  160VA ones from RS but never draw more than 4.8A from the output of the supply (at up to 15V 3A at 30V)
https://au.rs-online.com/web/p/toroidal-transformers/6719012/?relevancy-data=636F3D3126696E3D4931384E525353746F636B4E756D626572266C753D656E266D6D3D6D61746368616C6C26706D3D5E2828282872737C5253295B205D3F293F285C647B337D5B5C2D5C735D3F5C647B332C347D5B705061415D3F29297C283235285C647B387D7C5C647B317D5C2D5C647B377D2929292426706F3D3126736E3D592673723D2673743D52535F53544F434B5F4E554D4245522677633D4E4F4E45267573743D36373139303132267374613D3637313930313226&searchHistory=%7B%22enabled%22%3Atrue%7D

I slightly modified the classic circuit and used a comparator. I added a little hysterisis and I made the circuit slow to reduce the pre-regulator voltage but quick to ramp it up. This stops the PSU getting caught out if it briefly goes into current limit.

The biggest issue I had was the bounce in the ground line when the pre-regulator MOSFET turns on. I used a differential amplifier to measure the output voltage fed back to the analog control loop though so this minimized the impact.

I don't have a current probe but if you want me to capture any voltage waveforms I'm happy to crack the unit open and poke a probe into it.

Tom