New Bipolar Boost TX and Front End

[Mdtoday]

I had the same problems with the Chinese toroids. I ended up revisiting the Vallon design and they are using over 20 turns on the secondary. So my current design is using 6 turns #24 on the primary and 23 turns of #30 for the secondary. My primary has a 1 ohm in series. The secondary wire gauge didn't seem to matter. I am getting over 20 volts of gate drive which is more than enough. Good idea sourcing a spec'ed toriod. Too much going on in these circuits to wonder if your Chinese parts are the problem. Can't wait until you start the signal processing stage.

Attached photo - no coil, gate voltage 21 volts for 100uS.

Thanks for the info Altra, are you still using the Chinese toroids?

[Mdtoday]

I have determined the specifications for the toroid cores we have:
measured OD = 11.9 mm
measured ID = 7.417 mm
measured H = 3.048 mm
measured inductance @13 turns = 1185

using the calculator @ https://coil32.net/online-calculator...meability.html
ui = 24782
AL=7011.83

using the calculator @ https://coil32.net/online-calculator...alculator.html
4t = 112 uH ( 14.228 cm AWG 22 (w/5cm x 2 leads)
15t = 1578 uH ( 25.87 cm AWG28 (w/ 5cm x 2 leads)
16t = 2351 uH ( 29.4 cm AWG28 (w/5cm x 2 leads)

15t : 4t is a 3.75:1 ratio which should be more than enough since the 5V supply will always be 5V or greater (high enough to generate 1 amp TX current. If TX current is too high at 5V, need to adjust R6, R7, R8, R10 values to maintain >=5V).
16t : 4t is a 16:1 ratio if we need it.

Thanks Jl, much appreciated. I have quite a few of these toroids left so will rewind using your posted specs, see how we go.

Currently I have only fitted R6 btw

[Mdtoday]

If there were screen shots of M6_G and M7_G then we have a problem... extremely easily corrected but still a problem. The M6 & M7 gates shold be 0 during the TX time and go high shortly after TX turns off and stay high until slightly before TX (A or B) turns on again. In other words it will be high for ~398 usec and low for ~60 usec.

As I recheck the schematic, I see that I took DAMP_S to pin 4 of U14 and U15... I think this is wrong... my rationale was wrong. If this is a fact, then I can correct it by changing the constraint file for the Cmod-A7. All I will have to do then is regenerate the project and post it.

My bad, they were gate drives for M2,M4 not M6/M7.

I need to reset my brain.

I will take a fresh set of measurements of all gate drives and label the files correctly, sanity check them and then post..

[KingJL]

My bad, they were gate drives for M2,M4 not M6/M7.

I need to reset my brain.

I will take a fresh set of measurements of all gate drives and label the files correctly, sanity check them and then post..

Thanks @Mdtoday... I am fairly certain that I have the DAMP_S and DAMP_E going to the wrong side of the EL7202's. I am going to change the constraints file and regenerate the Vivado project. Will post when you confirm my foul-up.

[KingJL]

Thanks Jl, much appreciated. I have quite a few of these toroids left so will rewind using your posted specs, see how we go.

Currently I have only fitted R6 btw

I do not think we want to go over 4:1 on the turns ratio as that may allow us to start bumping up against the possibilty of exceeding Vgs limits.

Rewinding with AWG 28 enameled magnet wire, the windings wont be as cramped... should only take up a 3rd of the toroid as opposed to over half.

[Mdtoday]

In the next version of the TX board, I think I will include an isolated test point to be able to check the MOSFET gate drive and route the pad for the test point to the top side.
[ATTACH=CONFIG]48496[/ATTACH]



It does not show the actual voltage of the drive pulse, but the ratio can be determined. It is at a 0V reference and testing with it will not change the electrical charactaristics of the circuit under test (like dissapating any capacitively stored voltage).
[ATTACH=CONFIG]48497[/ATTACH]

I was thinking, for testing and connecting a logic analyser we could have a little pass through spreader board that fits in between the board to board connector J9 and fans out to header pins.
Or, just have header pin capability on the IF board for say, 0.1" pitch SIL header.
It would be great for testing, just don't load for normal operational boards.

[KingJL]

I was thinking, for testing and connecting a logic analyser we could have a little pass through spreader board that fits in between the board to board connector J9 and fans out to header pins.
Or, just have header pin capability on the IF board for say, 0.1" pitch SIL header.
It would be great for testing, just don't load for normal operational boards.

Either one would be a great a great help...

[Mdtoday]

I do not think we want to go over 4:1 on the turns ratio as that may allow us to start bumping up against the possibilty of exceeding Vgs limits.

Rewinding with AWG 28 enameled magnet wire, the windings wont be as cramped... should only take up a 3rd of the toroid as opposed to over half.

Yes, I agree, that is the way to proceed. I will wind some later today and post results.

[Mdtoday]

Thanks @Mdtoday... I am fairly certain that I have the DAMP_S and DAMP_E going to the wrong side of the EL7202's. I am going to change the constraints file and regenerate the Vivado project. Will post when you confirm my foul-up.

Ok, thanks JL, I will post later today.

[Mdtoday]

Ok, thanks JL, I will post later today.

@KingJL, here are some screen shots of M1,M2, M6 gates.






I spent some time making up 4 new transformers but they failed to produce more than 5v. I used 0.36mm x 16 turns on secondary and 0.63 x 4 turns on primary. Both were enamelled wire. Bit disappointing but I will recheck again in the morning to see where it went wrong.
Installed the original transformer modules and back to last night's voltages.

[KingJL]

@KingJL, here are some screen shots of M1,M2, M6 gates.


[ATTACH]48504[/ATTACH]
[ATTACH]48503[/ATTACH]

As I suspected, M6 is wrong... I fouled up an swaped the inputs to the EL7202,s for the M6_G and M7_G. I have a fix r4eady for the Vivado project that will fix rhis. It involved switching the ports around in the FPGA constraints file. Will put up the link to it in another post.

[KingJL]

New Vivado project archive.

There is an error in the schematic for the New Bipolar Boost TX and Front End that results in the DAMP signal sent to the DAMP MOSFET's being inverted. But the problem can be fixed by swapping the DAMP_s and DAMP_e signals set by the FPGA to the TX/RX board. The swap is accomplished by editing the constraints file to reflect the swap of signals.

This problem is now fixed in the new project archive, Vivado Bipolar_PI _Nextion(11-24-2019).zip

I need to revisit the schematics and change the Labels on the nets goint to the EL7202's for M6 and M7. It really does not physically change anything on the PCB layout... it is just changing signal names.

[Mdtoday]

There is an error in the schematic for the New Bipolar Boost TX and Front End that results in the DAMP signal sent to the DAMP MOSFET's being inverted. But the problem can be fixed by swapping the DAMP_s and DAMP_e signals set by the FPGA to the TX/RX board. The swap is accomplished by editing the constraints file to reflect the swap of signals.

This problem is now fixed in the new project archive, Vivado Bipolar_PI _Nextion(11-24-2019).zip

I need to revisit the schematics and change the Labels on the nets goint to the EL7202's for M6 and M7. It really does not physically change anything on the PCB layout... it is just changing signal names.

Thanks for posting JL, an easy fix for sure. I will try this one later today and post screen shots.

[waltr]

There is an error in the schematic for the New Bipolar Boost TX and Front End that results in the DAMP signal sent to the DAMP MOSFET's being inverted. But the problem can be fixed by swapping the DAMP_s and DAMP_e signals set by the FPGA to the TX/RX board. The swap is accomplished by editing the constraints file to reflect the swap of signals.

This problem is now fixed in the new project archive, Vivado Bipolar_PI _Nextion(11-24-2019).zip

I need to revisit the schematics and change the Labels on the nets goint to the EL7202's for M6 and M7. It really does not physically change anything on the PCB layout... it is just changing signal names.

That is the beauty of programmable devices for control. No cutting or soldering just build a new image and load.

Great progress guys.

[Mdtoday]

I spent some time making up 4 new transformers but they failed to produce more than 5v. I used 0.36mm x 16 turns on secondary and 0.63 x 4 turns on primary. Both were enamelled wire. Bit disappointing but I will recheck again in the morning to see where it went wrong.
Installed the original transformer modules and back to last night's voltages.

I was wondering why, if I wound a transformer with same primary turns, same gauge wires and just increased the amount secondary turns with near identical gauge wires, would I measure only 5v Pk-Pk pulses.
Well, the answer is the inductance was decreased by using super-glue to hold the windings. Rewound a couple more without using super glue to hold the windings in place and all was good again.
Even completely drying the superglue did not help. Won't be doing that again. When not cured I can see the problem but not dried.
Normally I use a nail-polish type glue that I use on trim pots but I was in a hurry to finish the 4 transformers and thought I'd be smart and dowse them in superglue...not a great idea.

The 2 new ones I did rewind, were 4 turns primary 22g magnet wire, 16 turns of teflon coated 28g wire for the secondary on one and 18 turns on the second transformer.
See attached for waveforms.




I think I will wind another 4 units with 19 turns on the secondary and test again, there is room for 20 turns comfortably but as JL has stated, we do not want to test the limits of Vgs