Totem-pole gate driver VS Active pull down VS Fast Fet turn off

[green]

Still looking at the current waveform by scoping a 0.1ohm resistor in the ground return from the coil, except that the coil is replaced by a 6.6 ohm resistor (3x 2R2), with the coax still in circuit. All the hash has disappeared and we have a clean couple of cycles of decaying ringing. Next is for the resistor to be straight on the board i.e. no coax. Still a ring but all decayed before 1uS. I measured the coax with the resistor termination and it gives a reading of 0.475uH which together with a capacitance of about 80pF may account for the extended ringing in the first plot. Perhaps my wrap around shielding contributes some additional ringing frequency that explains the additional 'noise' on yesterday's plot.

[ATTACH]54465[/ATTACH] [ATTACH]54466[/ATTACH]



Eric.

Hi Eric
Wondering if you could try a couple things the next time you are testing. Repeat of the tests with coil and 6.6ohm resistor you have done + repeat tests with scope ground lead and scope probe connected to same point(ground side of .1ohm resistor. I get the high frequency both cases with coil connected, wondering if you do or is my scope the problem. Need to get a .1ohm resistor and try with a resistor replacing the coil. I have been using twisted pair instead of coax so I need to try coax also to compare UCC27518 with what you are using.

[green]

Hi Eric
Wondering if you could try a couple things the next time you are testing. Repeat of the tests with coil and 6.6ohm resistor you have done + repeat tests with scope ground lead and scope probe connected to same point(ground side of .1ohm resistor. I get the high frequency both cases with coil connected, wondering if you do or is my scope the problem. Need to get a .1ohm resistor and try with a resistor replacing the coil. I have been using twisted pair instead of coax so I need to try coax also to compare UCC27518 with what you are using.

Tried test with what I have. Wasn't expecting the high current reading with the .01ohm resistor, calculates over 60A peak at turn off. Reads about 10mV(1A)when Tx on. Need a better current measuring resistor or something else causing the problem?

[Ferric Toes]

Hi Eric
Wondering if you could try a couple things the next time you are testing. Repeat of the tests with coil and 6.6ohm resistor you have done + repeat tests with scope ground lead and scope probe connected to same point(ground side of .1ohm resistor. I get the high frequency both cases with coil connected, wondering if you do or is my scope the problem. Need to get a .1ohm resistor and try with a resistor replacing the coil. I have been using twisted pair instead of coax so I need to try coax also to compare UCC27518 with what you are using.

Hi Green, I tried what you suggested and there is no difference in the ringing. I have tried shielded and unshielded coils and no difference. The flyback is 750V and is not avalanching. Spec gives 850V for Vdss. Scoping the gate shows a very similar ringing waveform to that on the 0.1ohm resistor, so maybe that is the cause. Further investigation needed as to why this is happening. The drive to the capacitor coupling is a TLC555. I tried an LMC555 and the result was the same. If I can find one, I will try a bipolar 555 and see if that makes a difference. The other thing to look at is whether a suitable zener in series with the gate/source diode will improve things. this will increase the hold off volts at the expense of drive volts as per the Texas paper. From the Mosfet spec I have about 3V in hand to do this. I can also try using direct coupling without the RC arrangement.

This is really fine tuning things though. Even with the ringing it appears I could sample at 2uS but my present Rx will only let me sample at 5.5uS. I'm thinking of trying a series blocking circuit so that there are no large transients saturating the preamp.

Eric.

[green]

Hi Green, I tried what you suggested and there is no difference in the ringing.


Eric.

Thanks, What do you mean by no difference?
I get an oscillation signal on the scope when I connect scope probe to scope ground lead and touch probe to +battery(noise signal). About 100mV p-p, with .1ohm resistor 1A p-p. To repeat your test with my Tx circuit I would have to eliminate the noise signal first and purchase a .1ohm resistor. Which I would try if the result would be different than doing the test with a 1ohm resistor which I have. Does anyone think result would be different? Think decay signal looks like it should with very little oscillation. Small drop in current while charging capacitance like I see with spice.

Maybe to eliminate noise. Solder scope lead coax direct to .1ohm resistor?

scope ground leads were connected to +batt
MUR460 diode between mosfet and coil

[Ferric Toes]

Thanks, What do you mean by no difference?
I get an oscillation signal on the scope when I connect scope probe to scope ground lead and touch probe to +battery(noise signal). About 100mV p-p, with .1ohm resistor 1A p-p. To repeat your test with my Tx circuit I would have to eliminate the noise signal first and purchase a .1ohm resistor. Which I would try if the result would be different than doing the test with a 1ohm resistor which I have. Does anyone think result would be different? Think decay signal looks like it should with very little oscillation. Small drop in current while charging capacitance like I see with spice.

Maybe to eliminate noise. Solder scope lead coax direct to .1ohm resistor?

scope ground leads were connected to +batt
MUR460 diode between mosfet and coil

I always use 0.1 ohm because the coil current is not materially affected and 1A gives 100mV. My coils are usually 5ohms or 3ohms depending on which wire I use. By no difference, I meant that the amplitude, duration and frequency of the ringing was not discernably different. Attached is a plot of the voltage across the 0.1 resistor in blue and the gate switchoff edge of the TLC555 as trigger and the ringing seen in red. . Which one causes the other?
The next plot shows on a slower timebase, two successive 250uS Tx pulses, peaking at about 1.6A in blue. The red waveform is the 8V drive pulses on the gate drive.

Eric.

[Altra]

The ground lead on the scope can add ringing and other noise. This video demonstrates the problem. The second video shows a nice solution for the type of testing you guys are doing.


https://www.youtube.com/watch?v=zodp..._channel=w2aew


https://www.youtube.com/watch?v=-4q8..._channel=w2aew

[Ferric Toes]

Eric,

Based on my research, the energy in the oscillations which need to be damped before sampling can occur comes from any capacitance source as seen by the coil and comes from these sources.
1. Coil capacitance
2. MOSFET COSS
3. Coax cable capacitance due to cable type and length
4. Coil to shield capacitance
5. Circuit capacitance

I found that using Scotch 24 as a shield was not very detectable, had less surface area than a solid shield material and was pretty easy to lay around a homemade coil to keep the self resonance higher than other solid shield materiel.

I came to visualize that the oscillations that need to be damped comes from unwanted capacitance from all its sources.

I can see a new coil design where the coil wires continue up the plastic shaft connected to the coil where they directly connect to the first electronic active circuits either TX or RX to eliminate the coax capacitance and thus make for a potentially faster coil. This also eliminates the coax solder joints inside the coil housing that may become detectable small targets.

Thanks,

Joseph J. Rogowski

Hi Joe, Thanks for the list of things to check. When I have wound my coil, I wrap it with thin cable wrap, and then another heavier polyethylene cable wrap so that it is a tight bundle, and then measure the inductance and resistance. My present ones are 330uH and 5 ohms. I then measure the resonant frequency which comes out at near 1.25mHz. I then wrap it with woven fabric emc tape with gap and a drain wire toward one end. Drain wire and one end of coil are joined. Resonant freq. is measured again and usually drops to about 1.0mHz. This drops again when cable is added according to length required.

Coss is minimised due to series diode.

I have Scotch 24 but rarely use it because of the difficulty of holding it in place. It is basically a flattened tube of mesh and I have tried splitting it down one edge and laying the coil between the V, then stitching the open edge. I soon gave up on that and returned to the conductive adhesive backed woven nickel plated copper fabric tape.

I'd be interested how you shield with the Scotch 24.

Eric.

[Mr.Jaick]

I read somewhere, that a good "rule of thumb" for the Mosfet gate resistor is about 10 times the Mosfet's Source Drain Resistance. Have not really tried it yet as I was still trying different Mosfets.

I calculate the gate series resistor this way, I hope it helps:

Ig = Qg / tr
Rg = Vdrv / Ic

where:
Ig : max gate current when gets hit by the rising edge of the drive voltage
Qg : Total gate charge in nC (nano coulomb)
tr : Rise Time (in nano seconds typically)

Rg : minimum gate resistor value to avoid oscillation
Vdrv : gate drive voltage

(you can find Qg and tr in datasheet)

[Mr.Jaick]

I always use 0.1 ohm because the coil current is not materially affected and 1A gives 100mV. My coils are usually 5ohms or 3ohms depending on which wire I use. By no difference, I meant that the amplitude, duration and frequency of the ringing was not discernably different. Attached is a plot of the voltage across the 0.1 resistor in blue and the gate switchoff edge of the TLC555 as trigger and the ringing seen in red. [ATTACH]54476[/ATTACH]. Which one causes the other?
The next plot shows on a slower timebase, two successive 250uS Tx pulses, peaking at about 1.6A in blue. The red waveform is the 8V drive pulses on the gate drive. [ATTACH]54477[/ATTACH]

Eric.

hi Eric

I know this sounds a bit odd but have you used bypass capacitors for 555 ic?
i would probably solder a 6.8 or 10nF directly on the power pins of the chip and maybe a 2.2uF tantalum in parallel with that
and have you used a minimum gate resistor or only bare wire?
just wondering

[bbsailor]

Hi Joe, Thanks for the list of things to check. When I have wound my coil, I wrap it with thin cable wrap, and then another heavier polyethylene cable wrap so that it is a tight bundle, and then measure the inductance and resistance. My present ones are 330uH and 5 ohms. I then measure the resonant frequency which comes out at near 1.25mHz. I then wrap it with woven fabric emc tape with gap and a drain wire toward one end. Drain wire and one end of coil are joined. Resonant freq. is measured again and usually drops to about 1.0mHz. This drops again when cable is added according to length required.

Coss is minimised due to series diode.

I have Scotch 24 but rarely use it because of the difficulty of holding it in place. It is basically a flattened tube of mesh and I have tried splitting it down one edge and laying the coil between the V, then stitching the open edge. I soon gave up on that and returned to the conductive adhesive backed woven nickel plated copper fabric tape.

I'd be interested how you shield with the Scotch 24.

Eric.

Eric, I see from your coil inductance and resonant frequency that your 330uH coil with a 1.25mHz self resonance represents 51pf of capacitance. Then, when you added your shield your 330uH coil now has a self resonance of 1.0mHz which represents 76pf of capacitance. This is only a 25pf of difference as what is called distributed capacitance but only represents about one quarter of what the actual coil to shield capacitance would be if you measured either coil wire to the shield for capacitance. In this case the space between the shield and dielectric constant of the spacer materiel makes the biggest difference with a thicker spacer and lower dielectric constant materiel causing lower capacitance.

Then I thought about the area of the coil shield and found Scotch24 and found that I had a lower coil to shield capacitance compared to using a solid shield materiel such a thin aluminum foil. The Scotch24 when waved under the coil on wood stick did not respond. The best way that I have found to mount on to the coil is to put it over about 3mm of polyethylene spiral wrap covering the coil. I cut the Scotch24 just wide enough to not overlap too much when I lay electrical tape half on the Scotch24 and half to secure it to the inner diameter of the coil. I do this to not form a conductive loop around the cross-section of the coil. Then I leave the circumference gap where the wires enter and exit the coil winding. I found that the Scotch24 has about half the capacitance as a solid shield would have in exactly the same place.

It all gets down to finding all creative ways to minimize the coil capacitance so you can use a higher damping resistor value. This does two things.
1. Higher damping values stimulate smaller targets better with a faster and steeper discharge pulse
2. Higher damping values allow for earlier sampling

I have been a big fan of you pulse induction metal detecting forum. I really liked how you planned to detect broken needles in fabric rolls. You needed some very low delays to detect these small broken needle parts.

My latest thinking is to eliminate the coax cable capacitance by looking for a way to miniaturize the necessary active circuits that both drive and then receive the target signal at the end of the aluminum pole that connects to the plastic coil shaft. I am sure that many Geotech1 forum members could begin prototyping something like this and see an immediate improvement in minimizing the delay even more.

Eric, thanks for sharing your knowledge.

Joseph J. Rogowski

[green]

I calculate the gate series resistor this way, I hope it helps:

Ig = Qg / tr
Rg = Vdrv / Ic

where:
Ig : max gate current when gets hit by the rising edge of the drive voltage
Qg : Total gate charge in nC (nano coulomb)
tr : Rise Time (in nano seconds typically)

Rg : minimum gate resistor value to avoid oscillation
Vdrv : gate drive voltage

(you can find Qg and tr in datasheet)

Still trying to understand how to determine required gate resistor. Including a schematic(mosfet driver4) of my test(mosfet driver3). Only mention of gate resistor I found in the UCC27518 literature The 2.4x overdrive capability provides an extra margin against part-to-part variations in the QGD parameter of the power MOSFET, along with additional flexibility to insert external gate resistors and fine tune the switching speed for efficiency versus EMI optimizations. However, in practical designs the parasitic trace inductance in the gate drive circuit of the PCB will have a definitive role to play on the power MOSFET switching speed. The effect of this trace inductance is to limit the dI/dt of the output current pulse of the gate driver. R1 my test is a piece of AWG30 wire about 1/2 inch long. What should I want the scoped decay curve to look like? What would you suggest for a value for R1 to try?

Forgot to include a 1u capacitor on pin 5 UCC27518 on the schematic. When looking in Digikey for information on STF11N50 I see it's obsolete, any suggestions for a replacement?

[Mr.Jaick]

Still trying to understand how to determine required gate resistor. Including a schematic(mosfet driver4) of my test(mosfet driver3). Only mention of gate resistor I found in the UCC27518 literature The 2.4x overdrive capability provides an extra margin against part-to-part variations in the QGD parameter of the power MOSFET, along with additional flexibility to insert external gate resistors and fine tune the switching speed for efficiency versus EMI optimizations. However, in practical designs the parasitic trace inductance in the gate drive circuit of the PCB will have a definitive role to play on the power MOSFET switching speed. The effect of this trace inductance is to limit the dI/dt of the output current pulse of the gate driver. R1 my test is a piece of AWG30 wire about 1/2 inch long. What should I want the scoped decay curve to look like? What would you suggest for a value for R1 to try?

Forgot to include a 1u capacitor on pin 5 UCC27518 on the schematic. When looking in Digikey for information on STF11N50 I see it's obsolete, any suggestions for a replacement?

1_ I can't recognize any bad behavior with your decay curve, so not really sure to be honest.

2_for 11N50 for the worst case minimum is 32 ohms and for typical it's 20 ohms
maybe try 27 ohms

3_I found these for replacement with even better rise and Fall time, take a look
2SK3468
2SK3695
13N50 (not as good and it needs a bit more current than your driver can supply)

how can I send pdf?

[green]

Still trying to understand how to determine required gate resistor. Including a schematic(mosfet driver4) of my test(mosfet driver3). Only mention of gate resistor I found in the UCC27518 literature The 2.4x overdrive capability provides an extra margin against part-to-part variations in the QGD parameter of the power MOSFET, along with additional flexibility to insert external gate resistors and fine tune the switching speed for efficiency versus EMI optimizations. However, in practical designs the parasitic trace inductance in the gate drive circuit of the PCB will have a definitive role to play on the power MOSFET switching speed. The effect of this trace inductance is to limit the dI/dt of the output current pulse of the gate driver. R1 my test is a piece of AWG30 wire about 1/2 inch long. What should I want the scoped decay curve to look like? What would you suggest for a value for R1 to try?

Forgot to include a 1u capacitor on pin 5 UCC27518 on the schematic. When looking in Digikey for information on STF11N50 I see it's obsolete, any suggestions for a replacement?

Was trying another test when Mr.Jaick replied. Tried different resistors(1/4W are thru hole, leads cut short)in series with wire between UCC27518 out and mosfet gate. Could try some different values if I have them.

Added coil volts with 200 ohm in series. Not seeing much difference with added gate resistance except for delay.

[Mr.Jaick]

good test green
so i think it's safe to say the problem is not the gate ringing
maybe some sort of complex impedance happening at the first turn on moment in that RLC rabbit hole(coil, damping, diode and mosfet capacitance, connections and packages,....)
by the way why your gate voltage on the scope is reverse?
it's an N mosfet and current starts when gate voltage falls in your pics, why is that?

[Ferric Toes]

I made a fitting for the scope probe exactly as portrayed in the YouTube video that Altra recommended in post 36. This was directly across the 0.1 ohm resistor.

The TLC timer already had a 22uF tantalum bead across the pins to the 8V supply, but I added a 4.7uF thin film ceramic capacitor. The ringing on the tail of the switchoff edge of the drive pulse seems much less.

Ringing still apparent on the switchoff of the Mosfet as seen using the better probe connection, but does seem less.The large initial ring seems to coincide with the flyback peak at 0.4uS

I think I am going to accept the result as all seems clean after 1.5 - 2.0uS and I don't need to sample that early anyway. Instead I am going to look into ways of eliminating the slow recovery characteristics of the preamp.

One thing I might quickly look at is the same Mosfet but driven by a complementary emitter follower drive.



Eric.