PI never gets old - constant power drive with added bonuses.

[Teleno]

Thanks for the fresh ideas Teleno.

How do you look at this, what if we exclude lm317 constant current source (as badass energy eater) and instead use analog feadback loop to regulate NE555 TX pulse width.
This may help to keep current constant during battery discharge.
But, i see, with lm317 positive voltage raill will be less spiky.

240mW at 12V, 180mW at 9V and that includes the power applied to the pulses.

The thing is I don't want to regulate the pulse width but the energy in the coil. That's why I charge the capacitor to a predetermined voltage and discharge it to a second, lower predet. Voltage. At the same time I provide a smooth analog ground. No spikes would be present on the Vcc line. Only the common mode rejection needs to be high, but most op amps provide that.

The pulse width will vary according to the inductance, which can eventually be used for ferrous discrimination.

This time I'm building the project, it's in an advanced stage. I've written and tested AVR code for ADC and EF acquisition, integrator, period measurement and delay control via a rotary encoder. I expect the PCB to be ready in 1-2 weeks, then it'll be soldering time.

[green]

240mW at 12V, 180mW at 9V and that includes the power applied to the pulses.

240mW or 240mA?

[Teleno]

60mA constant current for charging the cap during 1msec, for a 1A pulse of 30us at the coil.

240mW at 9V.

Run the simulation, you'll see.

[green]

60mA constant current for charging the cap during 1msec, for a 1A pulse of 30us at the coil.

240mW at 9V.

Run the simulation, you'll see.

Thanks, need to figure out how to run the sim. I have problems with spice when I hit run and it doesn't(my problem not yours). Coil current peaks at 2A, maybe 90us in reply #1 not 1A at 30us.

[Teleno]

Thanks, need to figure out how to run the sim. I have problems with spice when I hit run and it doesn't(my problem not yours). Coil current peaks at 2A, maybe 90us in reply #1 not 1A at 30us.

The timing varies with L, C, the value of the constant current and the lower and upper thresholds of the capacitor voltage.

For example, charging a 240 uF cap from 5V to 5.25V stores an energy of 1/2 (27.625 - 25) x 240e-6 = 307 microjoules. When discharged on a 300uH coil the peak current is sqrt( 2 * 307e-6 / 300e-6) = 1.4A (not accounting for losses in the damping resistor and internal resistance). As a rule of thumb, the actual current is about 85% the calculated value.

At 60mA current, the time to charge the cap is 0.25 x 240e-6 / 60e-3 = 1ms, which is roughly the Tx period of such a PI.

[Vadim18]

240mW at 12V, 180mW at 9V and that includes the power applied to the pulses.

The thing is I don't want to regulate the pulse width but the energy in the coil. That's why I charge the capacitor to a predetermined voltage and discharge it to a second, lower predet. Voltage. At the same time I provide a smooth analog ground. No spikes would be present on the Vcc line. Only the common mode rejection needs to be high, but most op amps provide that.

The pulse width will vary according to the inductance, which can eventually be used for ferrous discrimination.

This time I'm building the project, it's in an advanced stage. I've written and tested AVR code for ADC and EF acquisition, integrator, period measurement and delay control via a rotary encoder. I expect the PCB to be ready in 1-2 weeks, then it'll be soldering time.

Oh, thats looks interesting.
Not much time ago i take some fun with relaxation oscillator based on db3 dinistor and capacitor. I charged 10uf capacitor up to 30v and than drive coil thrue hi-voltage NPN transistor, dinistor acts like trigger.
Keep going.

[Teleno]

Well I've built the circuit only to find out that the ESR of C3 is too high and severely limits the peak current in the coil.

For the circuit to work acceptably the ESR has to be 0.1 ohm or less, my build with a cheap electrolythic is showing over 2 ohms. It went over my head to add this parameter to the simulation.

Can anyone point to some good readily available electrolythics?

[eclipse]

You should try Aluminium Organic Polymer Capacitors with ESR usually lower than 0.05Ohm , for example A759MS477M1CAAE015 by Kemet 470uF/ 16V / 15 mOhms

[Teleno]

I had a better capacitor at hand, not really great with an ESR of 0.15 Ohm but the results have vastly improved.

It's only 150uF of the 220uF that I need. That's why the flyback is only 300V (Coil is 188uH / 210pF / 2.1 Ohm. Rdamp = 390 Ohm).

Still it looks like this PI circuit is feasible. The signals now are very close to the simulation.

Vbatt = 12V.

The gate signal (blue) output by the 555 is sharp and clean.






Yellow is the floating ground (the voltage at cap C3).
Blue is the outpout of U6 that controls the 555.
The period can be neatly adjusted to 1KHz.

[Teleno]

You should try Aluminium Organic Polymer Capacitors with ESR usually lower than 0.05Ohm , for example A759MS477M1CAAE015 by Kemet 470uF/ 16V / 15 mOhms

Looks good. Thanks for the tip!

[Carl-NC]

Also use multiple parallel caps.

[Teleno]

Also use multiple parallel caps.

Added another 150u cap in parallel for a total of 300u.

The flyback now is 430V, which for a damping resistor of 390 R it's about 1.1 A in the coil.

I'm satisfied with the transmitter, next I'll be soldering the amplifier and see what the common mode looks like.

[Vadim18]

Well I've built the circuit only to find out that the ESR of C3 is too high and severely limits the peak current in the coil.

For the circuit to work acceptably the ESR has to be 0.1 ohm or less, my build with a cheap electrolythic is showing over 2 ohms. It went over my head to add this parameter to the simulation.

Can anyone point to some good readily available electrolythics?

From my practice samwha had less ESR, next was yageo and jamicon.

[Teleno]

The amplifier output is at odds with the simulation and can't figure out an explanation.

In the sim, the transient decays to 1mV within 6 usec, but the real circuit takes over 80usec. Why the huge difference?

The following is the output of a a two-stage amplifier with gain 2 x 25 based on a fast LM4562.



Coil parameters:: L=188uH; C=220pF, Rser=2.1R

The damping resistor is the theoretical, 450R. It's implemented on the PCB as a switch-configurable bank of metal film resistors connected in series and scaled as multiples of two: 400, 200, 100, 50, 25 and 12.5. Miniature bridges on the PCB short sections of the bank to obtain the desired value. See top right on the PCB:



My first thoughts about what's going on:

- parasitic inductances/capacitance of the resistor bank and its PCB traces.

- lower actual gain of the opamp the mV range of the input stage. Notice the feedback loop includes the damping resistor, see schematic.

-fake op amps, though I've tried an TL072 and an NE5532 with exactly the same results. Surprisingly, the lowest noise was measured in the TL072.

I would appreciate any insights.

[waltr]

I have found that simulations do not take into account how long the op-amp take to recover from saturation.
Both the 072 & 5532 take 10's of usec to recover when their outputs are driven into the rail Voltage.