Driving a 3D coil with the AD9833

Ian0

Joined Aug 7, 2020
9,667
Sure.
It should worth with either, but the MOSFETs should be a lot less trouble!
If you are aiming for a low power system (a few hundred milliwatts) then the MOSFET driver ICs could be used instead of discrete MOSFETs - it would require far fewer components that way!
 

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
Ok, so I designed a small board with both of the transmitter variants. Actually it's just a breakout of the pins of the multiplexer IC, the AD8615 op amp, and the MCP1402 chip. That way I can use whatever caps and resistors I want later, and change them faster.

Now I thought, since Mouser adds 20€ for shipping I rather pack my value up to 50€ and get free shipping. So I wanted to try the other things you suggested @Ian0 a few pages back. The receiver with the SA612 as the mixer and the square waveform generator with the 74HC4060 and a crystal.

But I can't figure out how to wire those two up. I want to leave as much variable components as possible off the board, so I can change them later on the fly. But I think the SA612 needs an oscillator and the 74HC4060 needs a crystal with the right cap sizes. Could someone help me find those right components from Mouser please. I'd rather not buy them a second time.

thanks,
8
 

Ian0

Joined Aug 7, 2020
9,667
I'll draw out the double-balanced mixer for you.
Why do people think that crystal oscillator caps are critical? For everything in the 1MHz-30MHz range, buy the cheapest crystal and load it with 15pF or 22pF either side and it will always work!
You might need a series resistor to get enough phase shift for lower frequencies (32.768kHz, for instance).
The SA612 has a transistor from which you can make a LC oscillator, but it's really for radio frequencies - you can just use the pin as an input.
 

Ian0

Joined Aug 7, 2020
9,667
As it's CMOS the 100k could go as high as 10M and it will still work.
Hypothetically, R2/C2 should be a low-pass filter with a -3dB frequency around the crystal frequency, but
I've use 15pF both sides and no R2 on every crystal oscillator I've built at home and 22pF on all those I built at work, just because of what was in stock, and they all work.
100pF is too big!
 

Ian0

Joined Aug 7, 2020
9,667
Seems like NXP lifted that straight from the much slower HEF4060 datasheet
https://assets.nexperia.com/documents/data-sheet/HEF4060B.pdf
and neither datasheet mentions the frequency of the crystal.
Back in days the days of 4000 series CMOS a 4.433619MHz PAL colour TV crystal was considered a high frequency. Logic that ran at 10MHz was really "pushing it to its limits"!
The crystal generates a phase shift of almost-but-not-quite 180 degrees at its resonant frequency. It needs just a little more to get it to 180 degrees where it will start to oscillate. That's the job of R2/C2 (even when R2 is the output resistance of the gate). Too much, and it reduce the signal so far there won't be enough gain for it to oscillate, and too little there won't get to 180 degrees and won't oscillate.
 

AnalogKid

Joined Aug 1, 2013
10,986
Seems like NXP lifted that straight from the much slower HEF4060 datasheet
and neither datasheet mentions the frequency of the crystal.
Easy enough to drop down to a 2.56 MHz crystal; same price. There is no penalty for using fewer 4060 stages. But you still have to filter the output to have a sinewave for the inductors.

ak
 

Ian0

Joined Aug 7, 2020
9,667
Cd4060 might struggle, buy HC4060 will get to 50MHz.
Here‘s the double balanced mixer. You can use 3 x BCM857( but the pcb layout isn’t easy!) or the SA612.
You will see that the pickup goes between pins 1 and 2, and the 20kHz refer each signal goes to pin6, and the outputs are on 4 and 5.
I‘ve no idea how well it will work, but I bet it will be better than the circuit in the paper.
one thing puzzles me. . . Why time-division multiplex the coils, when the could be all run continuously by using slightly different frequencies?
 

Attachments

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
Ok, I figured I could get away with just breaking out the SA612, since I can feed it a sinewave from the AD9833. It's for sure not the optimal way, but I'm already familiar with it. And I don't want to introduce too many points of failure.

Now about the timer. So I understand that the crystal will shift the phase by 180 degrees. And I guess the 100pF does most of this, and then the additional capacitor should bring it to the full 180 degrees. Just like you said.
But I'm still unsure of what schematic I should use. Is the first counter schematic I sent the right one, or the second ?, or will they both not work ? Is there a calculator to calculate the phase shift and to figure out the right cap sizes ?

thanks,
8
 

Ian0

Joined Aug 7, 2020
9,667
phase is tan^-1(Zc/R) i.e. the inverse tangent of the capacitor impedance divided by the value of the resistor.
don't worry too much about it, you just need a little bit extra phase shift, and 2.2k/100pF belongs to a much lower frequency crystal (at a guess it would be 32.768kHz). Try 15pF with no R2, it always works for me. Or try a low value of R2 say 470 ohms. It's a lot easier to get working than you think!

(The crystal give you MOST of the phase shift, 2.2k/100pF or whatever you decide to use gives you the little bit extra)
 

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
Alright, the parts arrived and I put everything together. Everything works pretty well, even the counter :D. Although I messed up the pin breakout from the counter, so it produces a 10kHz signal now. But I'm using the AD9833 for the square wave for now anyways, so no problem. I love the mixer IC, works a treat.

I tried the mosfet driver(MCP1402) and the opamp(AD8615) variant and attached the waves that were produced under the coil load. In both cases I used a 8.4V battery for the power and the opamp configuration as discussed before. I get much better results now. But there are still some minor problems with the mosfet driver especially. Any ideas to smooth the wave ?

thanks,
8
 

Attachments

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
Well after reading the datasheet, I saw they suggested two 1uF caps on the input. After that the wave improved quite a bit. Can it get better ?
 

Attachments

Ian0

Joined Aug 7, 2020
9,667
I still get the outputs from dividers wrong. Some manufacturer label them starting from Q0 and some start from Q1.
But - if you've got 10kHz instead of 20kHz, then you really need more capacitance on the output. Try doubling it, or even more. You can't really have too much.
Capacitance on the input? - you don't need any for the MCP1402 - connect it straight to the divider output.

You also seem to have more voltage than you originally said - 7.12V. I wonder if the op amp is current limiting.
Anyway, it's not a neat voltage waveform you were trying to achieve, it was a neat current waveform, because the magnetic field is proportional to the current, not the voltage; so put a small value resistor (1 ohm) in series with earth return and have a look at the current waveform.
 

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
I couldn't get the counter work properly, but I found that indeed I used the wrong pin. I attached a thin wire to the right pin and now get 20kHz.

Also, I did the resistor thing you said but used a 4.6 ohms resistor because that was the lowest value I had. I powered the mosfet driver with 5V this time. The wave is attached, and sure enough it's a rather clean triangle wave :)
I'm not so sure about wether or not it's enough current coming through. If my math is right, then 0.2/4.6 = 0.043 A.
Is that expected ? I feel like we anticipated a higher value. But I guess I could just crank up the input voltage.
 

Attachments

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
actually nevermind, I put a capacitor in the wrong place and I also put 5 of those 4.6 ohms resistor in parallel. Now I have roughly 200mV peak to peak with a around 1 ohms resistor. That would equate to around 200mA which is right I think.
 

Thread Starter

8dm7bz

Joined Jul 21, 2020
199
Now that the signal is ok, I need some help to figure the receiver out.

I see that the mixer picks up a signal, which is rather weak though. At the moment I don't use any amplification, but just take the mixer output. How can this be improved ?

thanks,
8
 

Ian0

Joined Aug 7, 2020
9,667
I have no idea what output level to expect from the detector coil.
Either you can amplify the signal from the coil before it gets to the mixer, or you can amplify the low frequency output from the mixer, or both.
First, I'd start by putting the receiver coil into a tuned circuit and see if that gets you anywhere.
C = 1/(4.pi^2.f^2.L)

Just a thought on the driver circuit. If you use two MCP1401s one to drive each end of the coil, and invert the input to one of them (74HC04), then you will get double the drive voltage from the same supply, and won't need the series capacitor.
 
Top