partzmans board ATL

[partzman]

Jon,

OK,  thanks.

Now, concerning the voltage across L1 using the 3/4 Bridge.

I now drive the circuit using my Rigol FG in burst mode (1 pulse per 100ms).
As the Rigol has a grounded lead, i cannot directly measure the voltage across L1 using a single probe as its ground lead will short out the MOSFET, so i use 2 probes in differential mode and subtract the result using the math function.

Here is my result of the voltage across L1 (40V supply on the MOSFET) in red and the current through L1 in green.



So i have an AC square wave across L1 (+/- 40V).
The current trace shows saturation where the green trace curls up.

Can you confirm that you also have such an AC square wave voltage signal across L1?

Itsu

Itsu,

In your schematic attached below I have indicated where to connect the scope ground and then the probe positions to measure the square wave across L1.

Your measurements should look like my attached measurements in the scope pix below.  My voltage levels are different from yours because this setup was operating at 12v.

Regards,
Pm

[Itsu]

PM,

OK, now things get clearer for me.

So you are not directly measuring (scoping) across L1, but instead measure (scope) each leg of L1 compared to ground.
Are the "probe" points in the diagram the yellow and blue traces? (yellow being the P- and N- channel MOSFETs junction and blue the other junction?).

What is the purple trace in your scope shot? (the voltage across C1 (load cap) which i do not have yet?).

I understood you were running the MOSFETs at 64V DC, so i would expect the voltages (yellow and blue) to be at 64V, (mine are at 40V) but instead they are at 12V (use to be 10V in your earlier screenshots), so did you lower the 64V DC voltage to 12V?

I also lowered my voltage on the MOSFETs to 12V (from 40V) and measured the scope points you pointed out (yellow the P- and N-channel MOSFET junction, Blue the other junction) compared to ground and this is the result:



Pulse frequency is 13kHz at 100ms repetition rate, and we now see similar voltage traces like in your screenshot, only my current is shown at 50mA/div. (probably due to my CH4 being at 50 Ohm).
Induction of L1 = 4.7mH (10 turns).

Itsu

[partzman]

PM,

OK, now things get clearer for me.

So you are not directly measuring (scoping) across L1, but instead measure (scope) each leg of L1 compared to ground.
Are the "probe" points in the diagram the yellow and blue traces? (yellow being the P- and N- channel MOSFETs junction and blue the other junction?).

The yellow is the signal generator input while pink is the P and N channel mosfets and blue the other.

What is the purple trace in your scope shot? (the voltage across C1 (load cap) which i do not have yet?).

I don't have a purple trace but maybe you could identify which scope pix you are referring to on which post.

I understood you were running the MOSFETs at 64V DC, so i would expect the voltages (yellow and blue) to be at 64V, (mine are at 40V) but instead they are at 12V (use to be 10V in your earlier screenshots), so did you lower the 64V DC voltage to 12V?

Yes, I was using 12v DC for that test.

I also lowered my voltage on the MOSFETs to 12V (from 40V) and measured the scope points you pointed out (yellow the P- and N-channel MOSFET junction, Blue the other junction) compared to ground and this is the result:



Pulse frequency is 13kHz at 100ms repetition rate, and we now see similar voltage traces like in your screenshot, only my current is shown at 50mA/div. (probably due to my CH4 being at 50 Ohm).
Induction of L1 = 4.7mH (10 turns).

Itsu

Now you are ready to place a cap in the core!

Regards,
Pm

[Itsu]

Thanks PM,

my purple is your pink 

Here my corrected screenshot using your probe positions:





Yellow is the FG input signal
Pink is the L1 coil leg (P-channel side)
Blue is the L1 coil leg (schottky diode side)
Green is current through L1
All referenced to ground.

Itsu

[Verpies]

my purple is your pink 

[Itsu]

As i did not have any useful 2500uF axial electrolytic capacitor available, i had to order some, which arrived today.


So first i replicated the diagram from Post #516 (see below) where PM only used a 2500uF cap parallel with a 48uH inductor to do some measurements.

I used only 35V on the MOSFETs (so not the 64V PM used) but my L1 inductance is higher 4.7mH instead of 2.5mH.

My results shown here are similar as PM shows in his 1st screenshot in his post #516:



Continue to set up the Vload circuit now....

Itsu

[PhysicsProf]

  Thank you, Itsu, for replication - and thank you, Jon, for the invention!
I look forward to learning more, hoping for the best.

[partzman]

Itsu,

Thanks you for the replication in your post #530.  I would like to point out an anomaly that I have not been able to completely analyze that is visible in your scope pix.  I thought previously that it was due to the onset of core saturation but your Finemet core is without any indication of saturation. 

Notice the rapid discharge of the capacitor just prior to the falling edge of your input pulse.  Also notice the drop in your 35v DC supply that occurs at the same time even though the primary current remains linear!  The drop in your supply indicates a current draw that exceeds the ability of the power supply to hold voltage regulation.  The primary peak current is ~2.2A.  Is this over the limit of your power supply?  If yes, then what is happening is that the primary voltage is decreasing with the drop in power supply voltage thus rapidly lowering the E-Field in the core.

Regards,
Pm

[Itsu]

PM,

i did notice the difference in my screenshot compared to yours in post #516 1st screenshot regarding the pink and blue traces.

The PS i use is a 40V 10A PS so it should be able to maintain the current through the primary L1.

But the green trace is not measuring the current through the primary (when you say: "though the primary current remains linear!) but is the current through C1 / L2.

Anyway, i could strengthen up the PS using a 10mF / 250V capacitor in parallel to see if the 35V stabilizes then.

Itsu

[partzman]

PM,

i did notice the difference in my screenshot compared to yours in post #516 1st screenshot regarding the pink and blue traces.

The PS i use is a 40V 10A PS so it should be able to maintain the current through the primary L1.

But the green trace is not measuring the current through the primary (when you say: "though the primary current remains linear!) but is the current through C1 / L2.

Anyway, i could strengthen up the PS using a 10mF / 250V capacitor in parallel to see if the 35V stabilizes then.

Itsu

Itsu,

OK, I would recommend you check the current in the primary as it is probably showing some core saturation.

Regards,
Pm

[Itsu]

PM,

Despite what i am saying in the video, am not sure about saturation.

I made a short video showing the result of increasing the voltage on the MOSFETs (from 35V to 40V) both in the C1 L2 circuit as in the primary input (35V) circuit:

https://youtu.be/neqN-QiLaYI

Does it not more looks like there is some runaway in the MOSFETs (P-channel and N-channel both active at the switch over point)?

Itsu

[partzman]

PM,

Despite what i am saying in the video, am not sure about saturation.

I made a short video showing the result of increasing the voltage on the MOSFETs (from 35V to 40V) both in the C1 L2 circuit as in the primary input (35V) circuit:

https://youtu.be/neqN-QiLaYI

Does it not more looks like there is some runaway in the MOSFETs (P-channel and N-channel both active at the switch over point)?

Itsu

Itsu,

When your green channel measuring current in the primary begins to increase from a linear ramp even a little bit, the core is beginning to saturate.  When your PS is 40v DC, your core is greatly saturated.  As the core saturates, the current in the primary will increase and no longer be linear, the voltage across the secondary (in this case the capacitor) will decrease, and there will be flux outside the core depending on how heavy the core is saturated.  In your case, you can see the voltage across the cap decrease to zero and below plus your 40v DC supply is also decreasing in voltage.

I'm not sure what you mean by "runaway" in your last question?  Could you be more specific?

Regards,
Pm

[Itsu]

PM,

thanks, so the saturation indeed seems to be the most logical after all.

What i meant with runaway is the lack of "dead time" between the MOSFETs being driven (top left P-channel and bottom left N-channel) causing a momentary short (both active), but that would be the case all the time, not only when increasing the voltage above a certain point.

Why you can run with 64V on the MOSFETs without saturating the core (L1 = 2.5mH) while my core (L1=4.7mH) gets saturated at 35V (and saturation at 24V when my L1 is also 2.5mH) is not clear with me, but probably has to do with the makeup of the used core.

Itsu

[Verpies]

When your green channel measuring current in the primary begins to increase from a linear ramp even a little bit, the core is beginning to saturate. 

Because approaching the V/R limits curves the inductor current in the opposite direction to saturation - it can mask the saturation.

I'm not sure what you mean by "runaway" in your last question?  Could you be more specific?

Shoot-through current.  An extreme form of crossover distortion.

[partzman]

Itsu,

The saturation of our inductors will depend on the applied voltage, inductance, and dt or the amount of time the voltage is applied to the inductor.  Up to now, you have been discharging the "C" with a known inductor.  When you begin to use "C" to charge a Vload, you will notice the required dt will be short due to the small lead inductances creating a relatively high resonant frequency.  Then you will find your core will normally not enter saturation.

Verpies,

Ah yes 'shoot-thru'!  My simple circuit was not designed to eliminate shoot-thru so it will be present.  A final design would use any number of the quality full bridge drivers that have the ability to control shoot-thru.

Regards,
Pm

[Itsu]

Thanks guys.

verpies,

Because approaching the V/R limits curves the inductor current in the opposite direction to saturation - it can mask the saturation.

as my 10 turn coil has only 0.2 Ohms of resistance, the V/R limit will be very high and thus won't be playing a significant role here, right?

Itsu

[Verpies]

as my 10 turn coil has only 0.2 Ohms of resistance, the V/R limit will be very high and thus won't be playing a significant role here, right?

At 35V your V/R limit is 175A, but maybe partzman's limit is lower.

[Itsu]

OK,  thanks

[Itsu]

PM,

in your Post #519 (7 pulses demo), where did you have the current probe and where is the pink probe?

I have my current probe (green) where in your diagram the word "Vload" is en the pink probe across the Vload source.
I now have my 2.5mH L1 core active, have 40V on the MOSFETs and 15.5V as Vload, pulse frequency is 30kHz.

Here is my screenshot of those 7 pulses:



Itsu

[partzman]

PM,

in your Post #519 (7 pulses demo), where did you have the current probe and where is the pink probe?

I have my current probe (green) where in your diagram the word "Vload" is en the pink probe across the Vload source.
I now have my 2.5mH L1 core active, have 40V on the MOSFETs and 15.5V as Vload, pulse frequency is 30kHz.

Here is my screenshot of those 7 pulses:



Itsu

Itsu,

In post #519, the CH4(grn) and CH3(pnk) are located as follows-

In SP1, CH4 is measuring the current into Vload and CH3 is measuring the voltage across Vload.  The Math(red) channel shows the mean power of CH3*Ch4.

In SP2, CH4 is measuring the current in R2 and CH3 is same as above.

In Sp3, CH4 is measuring the current into the 3/4 bridge on line Vs, CH2(blu) is measuring the voltage across Vs, and the Math(red) channel is calculating the mean power of CH2*Ch4.  CH3 is same as above.

In SP4, CH4 is measuring the current into Vload and CH3 is measuring the voltage across C1 at Vc1.  This CH3 measurement is taken to see the voltage loss (if any) in the bulk capacitance of C1 which would result in an energy loss.

Regards,
Pm

[Itsu]

Thanks,

so my last screenshot should be similar as your SP1 screenshot, as i had the current probe and the pink probe in the same positions as you had.

But the results are not similar, looking into that.....

Itsu

[partzman]

Thanks,

so my last screenshot should be similar as your SP1 screenshot, as i had the current probe and the pink probe in the same positions as you had.

But the results are not similar, looking into that.....

Itsu

Itsu,

One thing that is not clear in that post is that the M1 mosfet is switched on to connect R2 to ground continuously for the completion of the seven cycles only and is then shut off.  I can tell by your voltage and current waveforms that this is not the case with your circuit.  Also, the input from the signal generator to the 3/4 bridge is running at 45% duty cycle.

Regards,
Pm

[Itsu]

PM,

well, i put up the blue trace being the drain of the M1 MOSFET, and it shows in the below screenshot it's at ground level during the 7 pulses and at 15.5V before and after.

I also put up the duty cycle of the yellow trace which is the input from the signal generator to the 3/4 bridge, and it shows 50%.



What i did change in this screenshot is the position of the current probe as i understood it should be where in your diagram the word "Vload" is (between Vload input point and R1/D2), but it actually is in the Vload input line, see the below diagram for the probe positions i have now.

Now i have the same negative going current in between the pulses.

The shape of the current pulses (and power) though is not really the same as yours yet.


Itsu

[partzman]

PM,

well, i put up the blue trace being the drain of the M1 MOSFET, and it shows in the below screenshot it's at ground level during the 7 pulses and at 15.5V before and after.

I also put up the duty cycle of the yellow trace which is the input from the signal generator to the 3/4 bridge, and it shows 50%.



What i did change in this screenshot is the position of the current probe as i understood it should be where in your diagram the word "Vload" is (between Vload input point and R1/D2), but it actually is in the Vload input line, see the below diagram for the probe positions i have now.

Now i have the same negative going current in between the pulses.

The shape of the current pulses (and power) though is not really the same as yours yet.


Itsu

Itsu,

You now have the current and voltage probes in the correct positions as my explanation in my last post was incorrect.  The original description in post #519 is correct.

However, the voltage should not be squared as you have.  It should be a decreasing ramp as in SP1 during that time period.  What diode are you using for D2?  It seems as though there is a reverse connection back to Vc1 rather than having a disconnect via D2 during that time period!

Regards,
Pm

[Itsu]

PM,

A diode blocks current, no voltage, so it seems logical to me.

But i have a 1N5819 Schottky diode, both for D1 as D2, but i see what you mean (the pink voltage trace is different as yours).

I will measure the diode and resistor tomorrow.

Itsu