Re: Phase accurate primary current sensing
hen918, Mon Dec 31 2018, 11:53AM

There is one series of IC that will do it: the allegro acs732 series. Bandwidth of 1 MHz, only problem is the current rating, which is 40 A bidirectionally; a little low.
A current transformer will do fine though, it fits every requirement and can even be made pretty easily, just don't forget a burden resistor to convert the current to a voltage!

Re: Phase accurate primary current sensing
klugesmith, Mon Dec 31 2018, 04:58PM

Hen's current transformer suggestion sounds good to me. There are ferrite core materials with high permeability and low loss up to hundreds of MHz, e.g. in baluns. One might explore Rogowski coils, if no suitable ferrite core is on hand, but I think they are not trivial at high frequencies.

Since you only care about zero crossings, the burden on CT secondary could be two diodes in "antiparallel". How about a single Zener diode instead, to simplify digital logic level shifting?

That Allegro series current sensor could probably have its range extended with appropriate external shunt resistor (mind the inductances). Or use several identical devices in parallel, and combine their outputs. I bet it's based on Hall effect. What's the time delay corresponding to its 1 MHz bandwidth rating, and does the bandwidth vary widely with process and temperature?

If both ends of DRSSTC primary are "hot", voltage wise, consider electrostatic shielding in or around the current sensor.

Why is zero current a time of interest in a DRSSTC? Just curious.

Re: Phase accurate primary current sensing
Avalanche, Mon Dec 31 2018, 08:06PM

I did some work on this a while ago. I've dug up the old thread because it had some excellent info from Richie Burnett -

Re: Phase accurate primary current sensing
Uspring, Tue Jan 01 2019, 05:09PM

Why is zero current a time of interest in a DRSSTC? Just curious.

Usually a transistor bridge drives the primary. Switching at zero current minimises switching losses and transients of the inductive load.

Re: Phase accurate primary current sensing
Avi, Wed Jan 02 2019, 12:16AM

as Uspring said, it is best (for the semiconductors) for the interrupt turn off to be delayed until there is 0 current going through the primary

here is a comparison of a non isolated resistive bridge (top waveform), vs a current transformer (bottom waveform)
you can see the ~90 degree phase shift


Interesting!
unfortunately without a DSO I have no idea what sort of peak currents are involved to calculate the shunt (when run from 330v rectified mains)

im thinking if i use a shunt bar (similar to the one attached), it'd be ok

Re: Phase accurate primary current sensing
Weston, Wed Jan 02 2019, 05:15AM

Current transformers can work reliably up to many MHz. They do use some patented distributed resistive termination scheme, but you can buy Pearson current transformers that provide accurate phase measurement to 100MHz+. Most of the DRSSTC controllers use a current transformer for phase accurate current sensing.

Unless you used a completely unsuitable core (something like powdered iron) the phase shift you are seeing is probably in the resistive shunt. Big shunt resistors like that typically have a very low resistance but a comparatively large inductance due to the large physical size, becoming predominately inductive around a few tens or hundreds of KHz.

Re: Phase accurate primary current sensing
klugesmith, Wed Jan 02 2019, 05:17AM

Are those nice pictures yours? What's the frequency, or the scope horizontal scale?
An ideal current transformer with resistive load has no phase shift. What if some or all of the apparent phase shift is in the signal from resistive shunt, due to parasitic inductance, and/or from induced voltage in its scope-probe-and-ground loop? One way to check the latter is to connect probe signal and ground to the same end of current shunt.

To make a point about inductance in low-ohm circuits, here's a thought experiment using ACS732-ish values.
Make a current divider by connecting two 1-milliohm resistors in parallel.
Disturb the symmetry by adding, in just one branch, 1 nanohenry of equivalent series inductance.
What is the current ratio between branches A and B, at 1 MHz?
What is the phase difference between currents A and B at 100 kHz?

[edit] Written before but had to be re-typed. Nanohenries can be sneaky. Consider power supply bypass capacitors, on a multilayer circuit board with dedicated power and ground planes. Place leadless ceramic chip capacitors with vias right next to the pads. Each one comes with equivalent series inductance on the order of a nanohenry, when you include the connection loop.

Re: Phase accurate primary current sensing
Uspring, Wed Jan 02 2019, 10:45AM

Current transformer phase accuracy depends on the inductance reactance to shunt resistance ratio. Ideally it becomes a 100% accurate, if the shunt resistance is zero. In practice one has to add the copper resistance of the CT to the shunt resistance, so it never is really a 100% accurate. Usually the copper resistance can be neglected. Care hast to be taken, though, not to make the shunt resistance too large.

Re: Phase accurate primary current sensing
Avi, Wed Jan 02 2019, 10:49AM

ohhhhhhhhh!

not that those prices :(

given that everything is connected with clipleads, a shunt-bar is probably the least inductive part of the setup!

"nice"?
yes they are mine

probably about 320Khz, however it must work with different coils too without manually tuning some delay parameter

if i was to use a CT, seems i have to chose between a Flame Emitting Resistor


or an out of phase signal

Re: Phase accurate primary current sensing
hen918, Wed Jan 02 2019, 11:48AM

a turns ratio of 6 is very, very low. A turns ratio of 1000 solves all your problems. Burden resistor of 10 ohms, 1 A maps to 10 mV, no nasty explodyness, negligible phase shift.

Re: Phase accurate primary current sensing
Uspring, Wed Jan 02 2019, 12:34PM

if i was to use a CT, seems i have to chose between a Flame Emitting Resistor

You will be running your DRSSTC at 100% duty cycle? Or just use a smaller resistance.

Re: Phase accurate primary current sensing
klugesmith, Wed Jan 02 2019, 04:24PM

As Hen and Uspring pointed out, something in those Falstad animated simulations led Avi astray (no pun intended).
How can we see the invisible details, like voltage source amplitude or transformer parameters?

If CT burden were a pair of antiparallel diodes, as mentioned above, then one could be replaced with the LED of an optoisolator. Then optical receiver speed becomes the challenge. Many fiber-coupled receivers work at many gigabits per second.

Those ACS732 hall-effect current sensors come with linear full-scale ranges up to +/- 75 amperes, and aren't damaged by out-of range current. As long as the RMS current (figured over a whole DRSSTC burst-to-burst interval) doesn't overheat the device. It was a surprise to see "1 MHz" devices spec'd with 220 pF capacitance on the output! Risetime and prop delay: 0.7 and 0.14 microseconds typical, unspecified max and min.

Let's wait and see if anyone takes up my challenge about effect of 1 nanohenry
on a 1 milliohm traditional current shunt or ACS732 primary-side leadframe.