New Bipolar Boost TX and Front End

[Mdtoday]

@KingJL, Zener used is ZMM16, R25,R30,R35,R41 all at 2.2 Meg and C29,C32,C35,C38 back to original 2.2nF.

[KingJL]

... What values are you currently using for R25 (R30, R35,R41 ) and C29 (C32,C35,C3 )?

@Mdtoday, Can you verify what zener you are using. Every check I have made indicates that with a 16V zener, the driver level should be between 14.4V and 15V with turns ratios of 3.31 to 4. I know that a while ago, I suggested increasing the C29 ( C32,C35,C38 ) to 4.7n to 5.6n. That may have been a faulty calculation for it failed to take into account the input capacitance of the MOSFET, resulting in too large a capacitance. Further checking indicates that 3.3n should be the max for that capacitor. 10 Meg for R25 seems to work well (results in acceptable droop for the damp drive ). Even with C29 at 2.2n and R25, be at 10 Meg, droop is only about 280 mv for 500 usec. Even with a prt of 2000 usec droop on the damp would be acceptable at about 1.2V. I would recommend that C29,C32,C35,C38 be at the original 2.2n and R25,R30,R35,R41 be at 10 Meg.

[KingJL]

... and only getting 9v G-S, so I will have to look at my turns ratio again...

Actually, I think the amplitude may be fine. Remember, the amplitude we are concerned about is between gate and source not gate and GND. To get the true measure put the gate on channel A and the source on channel B and you may see something like the following attachment:

Here green trace is gate and blue trace is source for M1 (coil disconnected). As you see the amplitude of the green trace seems to only increase by 11V, but when compared with the blue trace the increase at the leading edge is 15.7V until the Vth of the MOSFET is reached and then is 13.5V.

I would guess that the output from your pulse transformer gate drive is satisfactory.

What values are you currently using for R25 (R30, R35,R41 ) and C29 (C32,C35,C3 )?

[KingJL]

Here is the test output, please note the damp signals are inverted, I had probes on M6 /M7 (CH3& CH4) on source, however, we can see the amplitude which is all this first test was about.

Actually the damp signals look correct albeit a little low.

>

[Mdtoday]

That is something I would not have expected. It could have been because I did not protect the adequately when I sent them!!!


Nice work on the boards!!!

No not your packing JL, I think the package fell from our kitchen counter with some other mail.

I just finished some testing and only getting 9v G-S, so I will have to look at my turns ratio again but one thing is certain, the stability on these cores is much better than originals

Here is the test output, please note the damp signals are inverted, I had probes on M6 /M7 (CH3& CH4) on source, however, we can see the amplitude which is all this first test was about.

[KingJL]

...It was a hairline fracture and did not take much finger pressure to break it. I am surprised I even got to wind the primary without breaking it..
So a word of caution to those who are using these (and other) ferrite cores, drop them onto a hard surface, they are very likely to be damaged...

That is something I would not have expected. It could have been because I did not protect the adequately when I sent them!!!


Nice work on the boards!!!

[Mdtoday]

strange indeed... almost as if 1 (39uH) or 2 (29uH) turn(s) missing!

Received some more of the same cores today (ZW-41303-TC) and did some more winding.. Also found the reason for one (probably two) of the low primary readings from previous test...broken ferrite.
It was a hairline fracture and did not take much finger pressure to break it. I am surprised I even got to wind the primary without breaking it..
So a word of caution to those who are using these (and other) ferrite cores, drop them onto a hard surface, they are very likely to be damaged...



Wound 2 replacements and the attached shows the results...much better.

Wound another 4 with just the primaries and all measured 74~75uH.

I mounted the new cores onto the little plug-in carrier boards ready for the next test.....firing the board up and checking the pulse transformer outputs..

The attached also shows the size difference between the original (bottom pic) and the ZW cores. They fit nicely.

[KingJL]

... 2 cores were close to the mark of 69uH with 4 3/4 turns, 2 were over half that value...
It's kind of odd to have that variation on these cores.

strange indeed... almost as if 1 (39uH) or 2 (29uH) turn(s) missing!

[Mdtoday]

Got back to some testing tonight thanks to KingJL's timely package of toroid cores, much appreciated JL!

I thought I was on a winning streak winding the primaries but as can be seen with attached snapshot....not to be.



2 cores were close to the mark of 69uH with 4 3/4 turns, 2 were over half that value.
I tested against my standard inductors and meter is fine on all Frequencies and modes. DER-5000 used to measure. Wire gauge 0.63mm

There is another package of cores due in soon so Ill check some of those.
It's kind of odd to have that variation on these cores.
I will check the 2 low value units to make sure they are not cracked.

[Mdtoday]

So, we will stay with the LTC2380-20 as the ADC for the next generation of the interface board. If the price does not come down or the supply becomes more restrictive, we can revisit it in the future. It really does not impact our progress as the FPGA XADC path for RX signal processing will still be available. That being said, I am posting the Kicad project files for the next version of the IF PCB and the TX/RX PCB.

On the TX/RX PCB, GND fencing has been provided for the RX signal from the DG333A (U2C) to the intra-board connector J9 Pin 1. Also a "keep out" was placed in the GND layer in the area of the TX output area to the Coil. This area comprises a lot (relatively) of surface area of the Coil+ and Coil- signals and is meant to reduce capacitance of those pathways.

The TX-RX(ADum version) files:
[ATTACH=CONFIG]48915[/ATTACH]
[ATTACH=CONFIG]48916[/ATTACH]
[ATTACH=CONFIG]48914[/ATTACH]

The I/F (ADuM version) files:
[ATTACH=CONFIG]48920[/ATTACH]
[ATTACH=CONFIG]48919[/ATTACH]
[ATTACH=CONFIG]48918[/ATTACH]
[ATTACH=CONFIG]48917[/ATTACH]

Unless there are any errors found or enhancements/additions to these PCB's, my efforts are now shifting back to proof of concept using the current design. I have finally received my Kapton tape to insulate the pulse transformer/passive components, so I will mount the pulse transformers and continue testing.

If no changes are made to the new version, I will probably have them submitted for fabrication in a few weeks.

Nice work JL, thanks for sharing. As you say we have rev 1 boards to test with and nut out any changes required.

[KingJL]

Yes, that?s not so great!

So, we will stay with the LTC2380-20 as the ADC for the next generation of the interface board. If the price does not come down or the supply becomes more restrictive, we can revisit it in the future. It really does not impact our progress as the FPGA XADC path for RX signal processing will still be available. That being said, I am posting the Kicad project files for the next version of the IF PCB and the TX/RX PCB.

On the TX/RX PCB, GND fencing has been provided for the RX signal from the DG333A (U2C) to the intra-board connector J9 Pin 1. Also a "keep out" was placed in the GND layer in the area of the TX output area to the Coil. This area comprises a lot (relatively) of surface area of the Coil+ and Coil- signals and is meant to reduce capacitance of those pathways.

The TX-RX(ADum version) files:
TX_ADuM.pdf
Kicad_TX_ADUM_drive(12-26-2019).zip
TX(ADuM)-Gerbers(12-26-2019).zip

The I/F (ADuM version) files:
TX(ADuM)_CMOD-A7 IF Board schematic.pdf
TX(ADuM)_CMOD-A7 IF Board BOM.pdf
Kicad_TX(ADuM)_CMOD-A7 IF Board(12-26-2019).zip
TX-IF_Gerber(12-26-1019).zip

Unless there are any errors found or enhancements/additions to these PCB's, my efforts are now shifting back to proof of concept using the current design. I have finally received my Kapton tape to insulate the pulse transformer/passive components, so I will mount the pulse transformers and continue testing.

If no changes are made to the new version, I will probably have them submitted for fabrication in a few weeks.

[Mdtoday]

SINAD down as much as 15 db (depending on which Fin you look at).

Yes, that’s not so great!

[KingJL]

Another contender may be the AD4020 or AD4021 from Analog Devices however it?s SNR is down a little but the device doesn?t look to bad.

SINAD down as much as 15 db (depending on which Fin you look at).

[Mdtoday]

Maybe we could design a solder-on smd module (or plug in) and still keep the option on the IF board for the LTC2378/2380.
I can load the VQFN and send you some modules. Like you say a general purpose module would be useful....food for thought.

Another contender may be the AD4020 or AD4021 from Analog Devices however it?s SNR is down a little but the device doesn?t look to bad.

[Mdtoday]

It would be nice if there were a small module available that could be socketed to our board that used the ADS8900B. In fact a small general purpose module would have many applications. I do like the enhanced SPI interface to reduce clocking speed needed to retrieve the data... but alas... I am not setup to handle it.

Maybe we could design a solder-on smd module (or plug in) and still keep the option on the IF board for the LTC2378/2380.
I can load the VQFN and send you some modules. Like you say a general purpose module would be useful....food for thought.