Tank resonance locator

[Tony Stewart]

How will you calibrate frequency? Is it analog scope or DSO?
It should only take a few seconds to measure .
No generator? or will you calibrate VCO V to f and then use analog meter or scope.

BTW your LC values are way too big to resonate near 50kHz

 

[Vijit Dubey]

How will you calibrate frequency? Is it analog or DSO?
It should only take a few seconds to measure .

How accurate? Is a freq counter avail, or just for calibration?

I am planning to use the output of the VCO and send it to a freq meter. The output will be in form of a digital pulse, which will be calibrated to some value and will be updated on a screen (digital output using a HMI). Do you think its a right approach.

I am thinking to ring it using a 9v-24v battery, so that the response is quick but I would want to hold it for some time.
My inductors and caps look humongously big around 2 feet x 2 feet x 3 feet.

If I short my resistance, it will behave as a parallel tank, will that help?

 

[Tony Stewart]

This is a fairly simple problem made hard by communication here.

the values you chose resonate at 2.91 kHz (1uH 7 mF)
maybe you meant microFarad (7uF !!) in which case that makes more sense 60.1kHz
The impedance Zo at resonance is 500mΩ.
Q= Ratio of Zo/Rs ( including coil) = resonance factor and for Q= 1, 10, 100, Rs= 500mΩ, 50mΩ, 5mΩ ( perhaps the coil Resistance) which results in a step response of <1 , 5 and 50 cycles

You will need 50 cycles for < 1% accuracy unless very precise threshold adjustment on counter.

This also means your connections must use braided wire or Litz wire or be very short.

Once way is to use direct COAX connection to an amplified shunt R of 5mΩ directly across coax and terminated by counter and scope.

Since this now means 5 mV/'per Amp, you can now choose your voltage and amplifier gain at 150KHz BW
I might suggest a pulse circuit that shorts out the coil after energizing it, however the counter AC couple may not response quickly enough and DC couple will saturate counter front end with battery voltage.

1st answer this
What accuracy do you need?

Using the pulse method is not low current.

Using my method with swept F is low current with gain on signal input from shunt in CD4046

 

[Tony Stewart]

Yes parallel test is easier.. But if you dont use a series R you can generate several hundred volts

Now R parallel / Rs is the Q or voltage amplification unless attenuated. Now use a 10: probe to counter.

 

[Vijit Dubey]

This is a fairly simple problem made hard by communication here.

the values you chose resonate at 2.91 kHz (1uH 7 mF)
maybe you meant microFarad (7uF !!) in which case that makes more sense 60.1kHz
The impedance Zo at resonance is 500mΩ.
Q= Ratio of Zo/Rs ( including coil) = resonance factor and for Q= 1, 10, 100, Rs= 500mΩ, 50mΩ, 5mΩ ( perhaps the coil Resistance) which results in a step response of <1 , 5 and 50 cycles

You will need 50 cycles for < 1% accuracy unless very precise threshold adjustment on counter.

This also means your connections must use braided wire or Litz wire or be very short.

Once way is to use direct COAX connection to an amplified shunt R of 5mΩ directly across coax and terminated by counter and scope.

Since this now means 5 mV/'per Amp, you can now choose your voltage and amplifier gain at 150KHz BW
I might suggest a pulse circuit that shorts out the coil after energizing it, however the counter AC couple may not response quickly enough and DC couple will saturate counter front end with battery voltage.

1st answer this
What accuracy do you need?

Using the pulse method is not low current.

Using my method with swept F is low current with gain on signal input from shunt in CD4046

View attachment 94289

I was just being approximate with the values, hence I mentioned rough values for resonant frequency.
Talking about accuracy, let's say the resonant frequency is 30KHz, the maximum deviation can be 31KHz, but on the higher side. I do not want it to sweep further and give me frequency below the resonance freq.

Could you explain more on the method which uses low current. I would prefer a test system which is low power and safe.

In 4046 are you still using phase compactor 2 or we do not need it. I remember that you previously told me that I only need a VCO and not the entire PLL. Also, in 4046 VCO pin 6, 7, 11 and 12 are used to define frequency sweeping range (min and max). Is there a way to regulate the sweeping rate too (is it important, since the circuit will automatically run in natural frequency).

Also, if we are taking the voltage feedback what are we comparing it to and does the VCO output affect the IGBT firing to regulate the frequency.


Vijit.

 

[Tony Stewart]

If you can isolate coil and cap in parallel, then try this direct to counter.
5W but only for as long as it takes to momentarily charge up the resonator RC= 70us.
Discharge is AC coupled with series cap and probe / counter impedance 1M or 10M with sensitive input on counter.

Check sensitivity of counter for zero crossing error and max gain or 1:1 or 10:1 probe,

If you can't get this to work. then we can get fancy with VCO , peak detector m scan ramp and peak and hold circuit with Control Voltage pause on resonance,

Counter must be in gated trigger mode for >100us then tryup to 1ms .
and report accuracy using 3.7V LiPo. If you want it safer, add 1A fuse.The 10 Ohm resistor will draw a few Watts but only for <1ms for each contact bounce. So use a good switch ( momentary ) that is low in contact bounce. If cap is not connected it may arc on switch, so TVS diode may be used for additional safety ( user fault)

This is low voltage, low energy... few mJ (3W x 1ms)

This does not trigger IGBT.

Test time after connected 1 second. Push and read.

 

[Tony Stewart]

When you specify accuracy , use a ratio then it does not change with frequency. i.e. %

 

[Vijit Dubey]

If you can isolate coil and cap in parallel, then try this direct to counter.
5W but only for as long as it takes to momentarily charge up the resonator RC= 70us.
Discharge is AC coupled with series cap and probe / counter impedance 1M or 10M with sensitive input on counter.

View attachment 94292

Check sensitivity of counter for zero crossing error and max gain or 1:1 or 10:1 probe,

If you can't get this to work. then we can get fancy with VCO , peak detector m scan ramp and peak and hold circuit with Control Voltage pause on resonance,

Counter must be in gated trigger mode for >100us then tryup to 1ms .
and report accuracy using 3.7V LiPo. If you want it safer, add 1A fuse.The 10 Ohm resistor will draw a few Watts but only for <1ms for each contact bounce. So use a good switch ( momentary ) that is low in contact bounce. If cap is not connected it may arc on switch, so TVS diode may be used for additional safety ( user fault)

This is low voltage, low energy... few mJ (3W x 1ms)

This does not trigger IGBT.

Test time after connected 1 second. Push and read.

Tony Stewart I understand and I am sure that this would work. But I realized that this process would always need a scope. My eventual aim is to replace everything with a circuit which is used to plug-in to a tank and give me a number. I would not want to use a function generator or scope during the test (just a circuit which works on its own). Do you have any idea about that. My error range can be upto 5% and I would prefer it in the higher side (more than resonance).

One other idea was to use a function generator and put signal across the tank, at resonance the feedback voltage across tank will be minimum and maximum across resistance. I could use a function generator to sweep the frequency (manually) and indicate the resonance using LED. However, this would also need a function generator, which I do not want.

What do you suggest?

 

[Tony Stewart]

what scope?

I only described the setting for a good low cost frequency counter with a sensitive input. ( in my last couple comments)

By careful selection of passive values and counter trigger thresholds, the test takes 1 second and reads frequency of the burst directly, instantly.

Make sense yet? WHich part did you not understand? The probe? I was suggesting using a scope probe for the counter. But as R in raises Q and duration of ringing, it also attenuates input so amplification may be needed. The counter usually has a good limiter inside.

 

[Vijit Dubey]

When you specify accuracy , use a ratio then it does not change with frequency. i.e. %

On that note, can I create a frequency sweep using a VCO, measure voltage across the series tank. The moment the voltage drops near 0V, it stops sweeping and gives me the value of the frequency.

 

[Vijit Dubey]

what scope?

I only described the setting for a good low cost frequency counter with a sensitive input. ( in my last couple comments)

By careful selection of passive values and counter trigger thresholds, the test takes 1 second.

Make sense yet? WHich part did you not understand? The probe? I was suggesting using a scope probe for the counter. But as R in raises Q and duration of ringing, it also attenuates input so amplification may be needed. The counter usually has a good limiter inside.

Let me think it over. I'll get back to you.

 

[Tony Stewart]

"can I create a frequency sweep using a VCO"

you can but you dont seem to be following my better advice.

 

[Vijit Dubey]

"can I create a frequency sweep using a VCO"

you can but you dont seem to be following my better advice.

Tony Stewart I read your comments over and over again. Are you suggesting to ring the tank circuit momentarily, for 1 second using a 3.7V LiPo and then measure the frequency using a frequency counter? Will it work?

If yes,
I didn't know that a frequency counter even existed. Is making a frequency counter cheap, let's say I buy a microcrontoller, can I get the code online. Or should I buy a frequency generator online (is it cheap). Again, my idea is to come up with one circuit with probes which you put across a tank and it gives you a number.

Vijit.

 

[Tony Stewart]

What is the Q of all resonators to be tested?
if 1uH 7uF then f= 60kHz and Zo~ 500mΩ
if Rs of coil and wire is 1mΩ then Q= Zo/Rs = 500
if Rs=10mΩ then Q=10
This determines feasibility of the pulse method into a gated counter by the number of cycles available.
Since Zo=0.5Ω for this example, voltage out is determined by excitation current, thus for low energy it must be pulsed.

Using a sweep VCO , same problem with Zo=0.5Ω at series resonance demands high current or parallel resonance demands high impedance with clamp to suppress flyback.

So what are all the range of values for L, Rs C? without this a precise solution is impossible.
An industrial frequency counter is essential. Don't waste time trying to make one. Only consider brand names like used HP for best triggers, best value.(Agilent)

 

[Vijit Dubey]

Tony Stewart
Can I use This frequency counter : **broken link removed**

I was thinking if frequency counter could be this cheap, I can integrate this into a box and come up with just one product.

I'll get back with the appx. values of L,C,R. Give me a moment.

 

[Tony Stewart]

If you know how to design a limiter and trigger circuit to exclude noise and transients and generate a gated pulse maybe...but you dont.

That circuit is for testing Crystals with a smaller series motional capacitance ( 0.01pF) with a 10 to 20 pF parallel load, yours is 7uF in series and in parallel, both arrangements wont work due to Zo=0.5 Ohm

 

[Vijit Dubey]

If you know how to design a limiter and trigger circuit to exclude noise and transients and generate a gated pulse maybe...but you dont.

That circuit is for testing Crystals with a smaller series motional capacitance ( 0.01pF) with a 10 to 20 pF parallel load, yours is 7uF in series and in parallel, both arrangements wont work due to Zo=0.5 Ohm

So for testing out here will be my values,
L- 1uH
C- 7uF
I can remove the induction circuit and introduce a resistance (whichever you would recommend) and I can make it series or parallel, whatever you recommend.

Now using a frequency counter of ~$300, would limit it. I am looking for something cheap and mobile.

Is there any other way I could do it, under $50 and something mobile, independent and I can make a lot of them.

 

[Vijit Dubey]

Tony Stewart
I went through the following document
https://www.ece.utah.edu/~ece3510/Lab_PLL_Basic_S09.pdf
They are sending the VCO output to function generator. I can't understand what is the feedback coming to pin 14 coming from.

Can I use 4046. I really can't see buying a frequency counter.

 

[Tony Stewart]

Tony Stewart
I went through the following document
https://www.ece.utah.edu/~ece3510/Lab_PLL_Basic_S09.pdf
They are sending the VCO output to function generator. I can't understand what is the feedback coming to pin 14 coming from.

Can I use 4046. I really can't see buying a frequency counter.

I thought you knew how the 4046 works... pin 14 is for a low level AC input signal
if you dont understand how to design this, how are you expected to find a solution?

IF you dont have a counter do you know how to calibrate VCO precisely to 1%?

To make the automated circuit you imagine... think of block diagrams then and specify all inputs and outputs.

1. VCO: 0-Vcc input, 5:1 freq range output, 300R output (internal ) impedance
2. Low impedance Driver: for 5 Ohm load.
3. driver to 0.5Ohm parallel resonator with 5 Ohm current sense to ground.
4. Precision rectifier Op Amp for resonant peak signal. Vp
   
5. Negative sweep integrator sample and hold circuit for V_ctrl
   
6. 1 shot cct to dump Cap voltage to Vcc with MOSFET switch to restart sweep.
   
7. Sample and hold cct #2 for peak signal. (low leakage switch & low leakage Film cap)
   
8. Comparator between V(t) - V(t-1) between S&H#2 cap and rectifier out , +Ve out if signal > Hold voltage to resample during pulse .. logic level out.
9. ! shot timer to stop test if resample pulse has not occurred within 1 ms and hold V_Ctrl
10. then use control limits or look up table for Vctrl before it drifts. for V vs F
11. Or use cheap counter circuit with VCO direct to Crystal input pin with small series cap.

So you need a precision OpAmp rectifier, two S&H, one with 1 minute leakage decay time constant for ramp gen. and one for comparing signal to previous sample.
With 10:1 frequency range in 0~5V range and 1% freq accuracy, V vs f is about 28kHz/V for a VCO from 13k to 130kHz, so for 0.5kHz error ~1% , voltage error must be < 18mV on V_control. This also means your Vcc or V ref must be more accurate than this.

 

[Vijit Dubey]

I thought you knew how the 4046 works... pin 14 is for a low level AC input signal
if you dont understand how to design this, how are you expected to find a solution?

IF you dont have a counter do you know how to calibrate VCO precisely to 1%?

To make the automated circuit you imagine... think of block diagrams then and specify all inputs and outputs.

1. VCO: 0-Vcc input, 5:1 freq range output, 300R output (internal ) impedance
2. Low impedance Driver: for 5 Ohm load.
3. driver to 0.5Ohm parallel resonator with 5 Ohm current sense to ground.
4. Precision rectifier Op Amp for resonant peak signal. Vp
   
5. Negative sweep integrator sample and hold circuit for V_ctrl
   
6. 1 shot cct to dump Cap voltage to Vcc with MOSFET switch to restart sweep.
   
7. Sample and hold cct #2 for peak signal. (low leakage switch & low leakage Film cap)
   
8. Comparator between V(t) - V(t-1) between S&H#2 cap and rectifier out , +Ve out if signal > Hold voltage to resample during pulse .. logic level out.
9. ! shot timer to stop test if resample pulse has not occurred within 1 ms and hold V_Ctrl
10. then use control limits or look up table for Vctrl before it drifts. for V vs F

Sorry, I have dealt more with high power electrical engineering. So, my questions might be pretty amateur.
I previously have used 14pin for a current feedback which is then compared to vco and sent to gate signals in my previous project. My question was that if I am going to use low power, I cannot use a current feedback. So can I use a voltage feedback and use that. Would it work, how do do you suggest going ahead with it.

Vijit.