1
\$\begingroup\$

I have a soft start circuit that looks like so:

enter image description here

The load will actually be a synchronous boost/buck converter, and ultimately this could also have an AC source instead of just DC and have a rectifier put inline as well.

The issue is that this works fine (and the inverse with an PMOS works fine too but obviously more loss) dropping the surge from 240A initially to 3-4 amps. But this is just a simulation. I'm actually building several circuits that will have voltages anywhere between 40 VDC and 600+ VDC. (1 solar panel at low output versus 10+ in series at high output all of which need to be supported.) This is also an issue for the battery which could be anywhere from 40 VDC to 460 VDC (wide battery support range) so powering this off the battery instead of the solar side isn't really an option either, and ultimately we need to be able to support either battery or solar powering which I think we have solved.

The AC side of things is straight forward because it's a known 240 VAC within a narrow range so calculating R1/R2 isn't a problem. But for the solar string/battery, because of the wide range, calculating values that work. As an example, using the below, if the V1 goes to 460 VDC then the gate would be getting 55 V which of course generates magic smoke. And if I try and adjust for a middle line there, then I'm still at 23 V on the high side with a 1.7 Ohm R2 and 2 V which will either cause magic smoke on the top end, or won't turn on the gate at the bottom.

My initial solution was to use one of these: High Voltage Range Step Down Buck Regulator to just take the wide range and step it down to a nice 12 VDC for my POC and 6ish VDC for the full scale product that will be using GanFETs or SiCFETs (experimentation is required and my be chosen for us if we go up more than 600VDC on the solar series array support) for the synchronous full bridge. That works fine, except the issue there is that I'm worried that it's going to get hammered on inrush, even if I put it in parallel. (i.e. it would go in the place of R1/R2/D1/C2 and I'd use it to power the gate drivers and the inrush control fet in the final design.)

Once I'm past handling the inrush current, the rest works just fine.

So my questions are:

  1. Are my worries founded about the circuit in parallel to power the gate?
  2. If so, how do I power the gate in such a circuit for in rush when the input voltage can be so variable? I'm not opposed to using a good PMOS if I can find one with low Resistance if I can not cook the gate with too high a voltage differential.

I'm really trying to avoid having something like an LTO/LiION battery inside this thing that powers the gate and has to recharge too, just to handle this problem.

\$\endgroup\$
5
  • \$\begingroup\$ Some comments: 1) R1 and R2 have very low values and will waste a lot of power 2) Few MOSFETs can handle the large VGS that M1 will see in this circuit 3) While M1 is conducting but not yet fully on, it will dissipate a huge amount of power. Are you sure it will operate within its SOA range? My guess is that it will not and that it will suffer damage. 5) Why is a softstart needed? \$\endgroup\$ Oct 29 '21 at 16:02
  • \$\begingroup\$ 6) I strongly advise you to take a "step back" and focus on what you're trying to achieve, you now dive into all kinds of solutions and find issues with those while it is unclear what you're after and I have a feeling you're making things more complicated than they should be or need to be. 7) 9 out ot 10 devices that need inrush current limiting use a PTC. That's one component. Done. \$\endgroup\$ Oct 29 '21 at 16:02
  • \$\begingroup\$ @Bimpelrekkie I MUST have a soft start (and I have to add low voltage cut out, reverse polarity etc. too) because the cap to control EMF and the inductor/cap on the synchronous buck will consume 240A easily on charge. As I said, I've just modeled this out in LTspice and it doesn't blow up, but I'm more than open to suggestions. (the VGS can be handled by GaNFETs and SiCFets wtihout issue especially with the NMOS because it's just 12v, PMOS IS V1 - 12), I've already verified and sourced those, but only if I can keep the gate voltage between 6 and 20 Volts (or V1-6-18 Volts obviously if PMOS) \$\endgroup\$ Oct 29 '21 at 16:07
  • \$\begingroup\$ Also note that "high voltage DC to low voltage DC" converters are everywhere. There's no need to re-invent that or use a buck converter. I am talking about AC power adapters. These take mains AC and convert that to DC (so 240 V AC becomes ~400 V DC) and that DC is transformed down using a flyback converter. You can buy ready-made flyback converters as well. For a high input/output voltage ratio, a flyback (using a small RF transformer) is much easier and efficient. \$\endgroup\$ Oct 29 '21 at 16:08
  • \$\begingroup\$ All that you list will be handled by an off-the shelf power supply. Reverse polarity: yes, as these are AC in. Realize that there's a difference between LTSPice and the real world unless you model everything. \$\endgroup\$ Oct 29 '21 at 16:11
1
\$\begingroup\$

I found the answer to this question with an LTC7860 . This chip allows for virtually unlimited high side input using a Zener Diode (see page 10 of the datasheet). It also supports reverse polarity (By adding a second P Channel Mosfet on top of the default N Channel) on top of voltage, current surge and soft start with low voltage recovery and timer safety shut off. (It also will block AC power too!)

This solves the entire problem and doesn't require any flaky PTCs or other solutions that are non-optimal. I've tested here up to 600VDC and it works great in basically every case I could think of (next will be a 1200V solar array using SicFets). I even put a physical fast blow fuse behind it to see if I could create a situation where it would pop and couldn't as long as I sized the Zener properly.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.