PopularFX
Home Help Search
Advanced search 
Login Register
Welcome,Guest. Please login or register.
2025-03-31, 22:19:47
News: Check out the Benches; a place for people to moderate their own thread and document their builds and data.
If you would like your own Bench, please PM an Admin.
Most Benches are visible only to members.

Pages: 1 2 3 [4] 5 6 7 8 9
Author Topic: Tetra Replication  (Read 16367 times)
Group: Moderator
Sr. Member
*****

Posts: 308
No, first you multiply the i&v samples and then you average their products.  Averaging first and multiplying later is a different mathematical operation that generates different results (results that are not indicative of the input power).
Also, 100ks/s is usually insufficient.  Fortunately even the cheap modern scopes such as the DS1054Z can sample at 1Gs/s.
The 8-bit vertical resolution of the cheap ADCs introduces significant quantization errors, especially when the ADC is underdriven or overdriven.  For this reason a 12-bit or 14-bit ADC are preferred, e.g. as in the DHO900.  See the difference below:


If you can filter the input so it results in pure DC then you can get away with using even a slow voltmeter and an ammeter.

Makes perfect sense to me, I think filtering the input to enable accurate measurement with a simple meter, then add an error factor, would be the simplest approach.

If I get non-null results, I plan on releasing all of the data in spreadsheet form. Try to quantify all the variables that could effect the result, etc...
   
Group: Moderator
Sr. Member
*****

Posts: 308
Here is the pulse driver, one per coil, except coil A, which will have three of these in parallel.
   
Group: Experimentalist
Hero Member
*****

Posts: 2148
@bte

We can also use a very simple device to measure RF voltages (not currents): a simple diode followed by a capacitor, which rectifies the RF voltage. But this only works in one particular case: the signal must be periodic, unmodulated, have a level of at least several volts, and have no DC offset or offset to compensate for. Since what we're measuring is the peak voltage, we also need to know the shape of the signal to obtain the RMS value. If the signal is sine or rectangular, this is easy.
I've often used it to measure RF output voltages from radio transmitters. It's accurate. For better precision, don't forget to add the 0.6v knee voltage to the measurement if the diode is Si.

An oscilloscope measurement doesn't work well at high frequencies, especially if you're using a high-impedance output and a high-impedance ordinary probe. The divider in your diagram with resistors R4/R5, which are very large and probably more or less inductive, will form a low-pass filter with the input capacitance of the probe. In addition, ambient radiation can induce currents directly in the probe, greater than the signal connected to the probe. This will work approximately for a few tens of KHz, but if we're talking MHz, the measurement will be false.


---------------------------
"Open your mind, but not like a trash bin"
   
Group: Moderator
Sr. Member
*****

Posts: 308
I guess it makes it easier that anything higher than switching noise at the output would be significant and unusual. Shouldn't be any real coupling I would guess.
   
Group: Moderator
Sr. Member
*****

Posts: 308
Note to look into helmholts coil for magnetic field.
   

Group: Professor
Hero Member
*****

Posts: 3670
Note to look into helmholts coil for magnetic field.
While you're at it, look at Maxwell coils, too.
   
Group: Moderator
Sr. Member
*****

Posts: 308
Need a gauss meter recommendation if possible. Want plot magnetic field strength as a parameter. Is there a way to detect the earth's magnetic field? Like a meter that can go to zero absolute field strength?
   

Group: Professor
Hero Member
*****

Posts: 3670
Need a gauss meter recommendation if possible. Want plot magnetic field strength as a parameter. Is there a way to detect the earth's magnetic field? Like a meter that can go to zero absolute field strength?
Yes, look at the posts of the member "Itsu".  There should be some photos and demonstrations of his gauss meter in his posts or videos.
IMHO that meter has a good sensitivity and quality to price ratio.
   

Group: Professor
Hero Member
*****

Posts: 3670
Pulse width 25ns, 1mhz repetition rate, 300v rail.
21.966ns would be the ideal pulse width to match pure metallic iron's NMR frequency in room temperature and in absence of an external magnetic field.  If you apply such field then the required pulse width will increase, albeit only by 0.033% up to 0.6T (more afterwards).  Temperature will have much more impact.

The pulse repetition frequency (PRF) is of secondary importance.  In fact lower PRFs increase the density of the signal's spectrum - it's counterintuitive but that's Mr. Fourier for you. 
Denser spectra are more likely to contain the desired frequency (if the fundamental one is off a bit).

« Last Edit: 2025-02-22, 16:11:30 by verpies »
   
Group: Moderator
Sr. Member
*****

Posts: 308
21.966ns would be the ideal pulse width to match pure metallic iron's NMR frequency in room temperature and in absence of an external magnetic field.  If you apply such field then the required pulse width will increase, albeit only by 0.033% up to 0.6T (more afterwards).  Temperature will have a much more impact.

The pulse repetition frequency (PRF) is of secondary importance.  In fact lower PRFs increase the bandwidth of the signal - it's counterintuitive but that's Mr. Fourier for you.

Incredible insight. Thank you.
   

Group: Professor
Hero Member
*****

Posts: 3670
Have you read this patent yet ?
« Last Edit: 2025-02-22, 16:44:44 by verpies »
   
Group: Moderator
Sr. Member
*****

Posts: 308
Variables to plot:
Pulse voltage
Repetition Rate
Pulse width
Magnetic field strength
Coil copper mass
Coil inductance
Distance from center
Humidity
Air pressure
Time of day
Location
Output power (voltage and current and their product)
Total input power (same as above)
Temperature

Hopefully most of these have no impact on output power
« Last Edit: 2025-02-22, 21:38:42 by bte »
   

Group: Professor
Hero Member
*****

Posts: 3670
Variables to plot:
Pulse voltage
Repetition Rate
Pulse width
Magnetic field strength
Coil copper mass
Coil inductance
Distance from center
Humidity
Air pressure
Time of day
Location
Output power

Hopefully most of these have no impact on output power
No temperature and current ?
   
Group: Moderator
Sr. Member
*****

Posts: 308
No temperature and current ?

Added temperature
   
Group: Moderator
Sr. Member
*****

Posts: 308
Going to try thermal via for these tiny small components. I'm going to overestimate the power dissipation of the ganfet at 10W and try to overengineer the thermal design of the board. Idea is thermal vias to ground plane on bottom half of board, which can be sandwiched with a cold plate.

I watched some videos on thermal mistakes with smd and the ground plane being contiguous is vital.

I have a thermal camera I'm going to use to test each board, temperature vs switching frequency.

Board design for the pulser almost done, just waiting on components to arrive so I can double check dimensions.

I use filled planes where I can for strong connections. Traces for minor components. I have high hopes for this board and hopefully it works without issue first try, fingers crossed.

Once I begin experimenting with the unit I'm going to try to isolate each variable as best I can. Some variables like time of day and location won't be focused on, rather the important ones like coil distance which should greatly impact operation of the device.

Edit: board will be 4 layer
   

Group: Professor
Hero Member
*****

Posts: 3670
Once I begin experimenting with the unit ...
I hope you minimized the loop area of the high di/dt current paths.
What model of oscilloscope will you use to monitor the high-speed signals on this board ?
   
Sr. Member
****

Posts: 280
There's really no reason to not use a 6 layer board at this point. The price difference between a 4 and 6 layer board is marginal at places like JLCPCB, and they offer higher spec 6 layer boards with added value things such as controlled impedance, free via in pad, free epoxy filled and capped vias and tighter tolerances.

The extra 2 layers will also give you a bit more freedom for the layout.

SMD components generally use the ground plane as a heatsink too, so I would recommend spending extra for 2oz copper.
   

Group: Professor
Hero Member
*****

Posts: 3670
There's really no reason to not use a 6 layer board at this point.
IMO that's all premature.
The high-speed pulsing circuits are so finicky that it makes sense to build them in the air first and test with low PRF so the FET power dissipation is negligible and no heat sinking is required.

Because air is such a fantastic dielectric, it outperforms PCB designs with respect to undesirable capacitive coupling.
Also, the 3D freedom allows the designer to minimize the loop area of high di/dt current paths and undesirable inductances and inductive coupling.  Soldering the driver right on top of the FET with minimum lead lengths being the best example of it. 

This allows the designer to empirically test whether the components perform as expected (especially the driver and the FET+snubber) before committing to a PCB design.
If it does not work in the air - it will never work on the PCB.
   
Sr. Member
****

Posts: 280
IMO that's all premature.

Don't disagree with breadboarding it first. My point was when faced with the choice between 4 layer and 6 layer, there's no reason to choose 4 layer.
   
Group: Moderator
Sr. Member
*****

Posts: 308
I hope you minimized the loop area of the high di/dt current paths.
What model of oscilloscope will you use to monitor the high-speed signals on this board ?

This is another area I need to upgrade I think. My current scope is 100Mhz bandwidth, 1Ghz sample rate. I'm looking at this scope:

https://www.tequipment.net/Rigol/MSO5354/Digital-Oscilloscopes/?Source=googleshopping&msclkid=413dfc3c457811f3c7af9b86c3d8b084&utm_source=bing&utm_medium=cpc&utm_campaign=Shopping%20Campaign(BSC)&utm_term=1100101189689&utm_content=Shopping%20Adgroup
   

Group: Professor
Hero Member
*****

Posts: 3670
This is another area I need to upgrade I think. My current scope is 100Mhz bandwidth, 1Ghz sample rate.
That's not so bad.  The 100MHz front end will attenuate the amplitude of the pulse but you still should be able to see it with 1Gs/s sample rate. Give it a chance...

I'm looking at this scope:
That's a good choice if you have the $2k to spend.  The only objection I have is the low vertical resolution of the ADC.
What country are you in ?

What about a signal generator?
If you don't have one I recommend the cheapest model of this series because it can be hacked for free to become the highest model of that series.
   
Group: Moderator
Sr. Member
*****

Posts: 308
I have two SDG 1032X signal generators, I was hoping to link them to have it generate a 3 phase signal but it's too limited. Works well as a function generator though.

Edit: I'm in the US
   
Group: Moderator
Sr. Member
*****

Posts: 308
work in progress on routing this board.

the 57mm is the dimensions of the cold plate, since these thru hole components will touch the plate I had to offset the board off of the cold plate, probably going to use a shallow wooden structure to fasten the board to, and fasten the cold plate in place.

i hate routing boards, it's a never ending puzzle that has no correct solution.

note that i simplified the connectors to avoid lengthening the board too much

going to add 5v power plane and a few ground planes

layer facing cold plate will be copper fill ground plane probably. hv traces will stay on the top layer.

need to add fill zones for all the signal traces and do some research on how to best handle the 50ohm line coming in for that signal trace at the bottom (J1).

edit: missing attachment
   
Group: Moderator
Sr. Member
*****

Posts: 308
Came up with this board edge cut pattern, will allow 4 boards for every cold plate, so I'll only need 2 cold plates in total.

Looks funky but it should work fine.

Edit: I keep forgetting attachments
   
Group: Moderator
Sr. Member
*****

Posts: 308
Here is the board routed as first pass. Ground plane hidden, 5V plane in brown/orange. Top layer in red. Ideally the 5V source would be closer to where it's used. I don't know. Tradeoffs. I'd rather have the 5V source be a bit farther away than move the high speed pulse signal loop further away.
   
Pages: 1 2 3 [4] 5 6 7 8 9
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2025-03-31, 22:19:47
Loading...