Tank resonance locator

[Vijit Dubey]

any luck with my cct?

I am still waiting for delivery for the RLC meter.
For the circuit you mentioned, I would need a high speed comparator. I tried with the normal one (we both knew it wouldn't work). I would get one soon.

 

[cowboybob]

Let's say I have a capacitor and inductor (used in a tank circuit for induction heating application). It operates at a PLL using CD4046. However, to calibrate the starting frequency I must know the tank's resonant frequency. Using either a micro-controller or any other logic, what would be the best way to identify the tank's resonance frequency by sweeping from a high frequency in such a way that I get a number displayed as the resonance of the circuit.

Perhaps a tad late to suggest this, but...

Have you considered a DGDM (Digital Grid Dip Meter; a convenient, non-contact tank/antenna resonant frequency detecting test device)? See: http://en.wikipedia.org/wiki/Grid_dip_oscillator. An example: http://elm-chan.org/works/ddm/report_e.html

A nominal 200kHz tank frequency is a little more difficult to achieve when designing/building the coupling coil(s), but it is doable.

There is also the "pulling" effect (resonant frequency shift) the coupling coil has on the target tank circuit, but that can be accounted for with a uC based GDM system.

 

[Tony Stewart]

Unfortunately Bob the DGDM is designed for high reactive impedance resonant coils that dip to much lower impedance such that it absorbs energy to dip the grip meter.
This external tank circuit has a reactance of <1 Ohm at resonance so it not not even close to be matched to the meter impedance and thus no variation in grid current can be seen unless it was digitized in parts per million.

 

[cowboybob]

Unfortunately Bob the DGDM is designed for high reactive impedance resonant coils that dip to much lower impedance such that it absorbs energy to dip the grip meter.
This external tank circuit has a reactance of <1 Ohm at resonance so it not not even close to be matched to the meter impedance and thus no variation in grid current can be seen unless it was digitized in parts per million.

You're right, of course. Low Q can result in little or no observable "dip". For sure if using a mechanical meter.

Perhaps some amplification might enhance the detection of the reduction in current at resonance.

 

[BobW]

A dip meter is actually a very good method.
Let's not conflate low Q with low resonant impedance. A dip meter coupled to a high Q tank will still show a distinct dip at resonance regardless of resonant impedance. On the other hand, if the Q is extremely low, then resonance will have a very broad peak and the measurement instrument will have to be very loosely coupled to the tank so as not to influence the resonant frequency. But in that case, there's really no specific circuit that would be superior. It would have more to do with how the measurement circuit is coupled to the tank.

 

[Tony Stewart]

I was thinking that even if the Q of the DUT was 100 ,it might work if the LC tank reactance was < 1 Ohm but it would peak Q times this for parallel and may absorb enough energy to dip a 100 Ohm dip meter a little bit.

But I've never had the pleasure to use one.

Even the Jfet oscillator version would only get about 500 uV with 9mA Idss
But then you can always amplify that.

Unlike below where the DUT on left is high impdance.

 

[Vijit Dubey]

Tony Stewart I was having trouble simulating the following circuit in this datasheet. Link (page 9)

This does not seem to work for me somehow: link

 

[Tony Stewart]

No hysteresis,, negative feedback becomes a voltage follower

https://www.falstad.com/circuit/cir....00009765625+1+-1 o+42+64+0+290+5+0.025+1+-1

Edit Op Amp Voltage to +5, -0V

I tried with Op Amp but found simpler solution. It needs 2 stages to be high impedance and negative feedback with inverting Q to get hysteresis but negative feedback
( confusing but it ends up as Ramp Sweep pulse generator)

 

[Vijit Dubey]

No hysteresis,, negative feedback becomes a voltage follower

https://www.falstad.com/circuit/circuitjs.html?cct=$+2+1e-8+105.18789638808724+1+5+50 r+512+328+512+376+0+33000 c+720+168+720+224+0+1e-7+1.7647368916329036 w+472+136+472+184+0 w+472+136+720+136+0 w+696+72+720+72+0 w+720+224+672+224+0 w+696+104+696+72+0 w+672+224+672+200+0 w+528+184+528+216+0 g+672+224+672+248+0 w+672+168+720+168+0 w+648+360+648+320+0 R+720+72+720+32+0+0+1+5+0+0+0.5 183+528+184+584+184+0+0.5+1.66+3.33 w+720+136+720+168+0 a+648+368+720+368+3+5+0+1000000 t+640+184+672+184+0+1+-1.7647368911329036+4.999999998019801e-10+100 r+640+184+584+184+0+10000 183+472+184+528+184+0+0.5+1.66+3.33 w+584+184+584+216+0 174+720+72+696+136+0+2200+0.41090000000000004+NTC+2.2+CMOS x+739+111+837+114+0+14+NTC+2k2@25'C r+512+416+512+376+0+10000 g+512+416+512+440+0 w+512+376+648+376+0 R+512+328+512+288+0+0+1+5+0+0+0.5 c+456+376+456+416+0+1e-8+1.1627906976732851 w+720+320+848+320+0 w+512+416+456+416+0 w+456+376+512+376+0 x+867+295+965+298+0+14+NTC+2k2@25'C 174+848+256+824+320+0+2200+0.5891000000000001+NTC+2.2k+OPAMP t+768+368+800+368+0+1+-1.7244751883959766+-8.996682471329543e-7+100 w+848+320+848+352+0 R+848+256+848+216+0+0+1+5+0+0+0.5 w+800+352+848+352+0 g+800+408+800+432+0 w+800+408+800+384+0 w+824+288+824+256+0 w+848+408+800+408+0 w+824+256+848+256+0 w+720+368+720+400+0 c+848+352+848+408+0+1e-7+1.7244742887277293 r+768+368+720+368+0+10000 r+648+320+600+320+0+10000000 r+720+320+648+320+0+3300000 g+600+320+600+344+0 c+600+280+648+280+0+1e-10+-1.1537890152019579 w+648+280+648+320+0 w+600+280+600+320+0 w+576+248+944+464+0 w+928+232+568+472+0 o+14+64+0+290+5+0.0015625+0+-1 o+8+64+0+298+5+0.00009765625+0+-1 o+19+64+0+298+5+0.00009765625+0+-1 o+26+64+0+290+2.5+0.00009765625+1+-1 o+41+64+0+290+10+0.00009765625+1+-1 o+42+64+0+290+5+0.025+1+-1

Edit Op Amp Voltage to +5, -0V

I tried with Op Amp but found simpler solution. It needs 2 stages to be high impedance and negative feedback with inverting Q to get hysteresis but negative feedback
( confusing but it ends up as Ramp Sweep pulse generator)

This looks good in the simulation. Let me try to replicate this on my breadboard.
Thanks.

 

[Vijit Dubey]

This looks good in the simulation. Let me try to replicate this on my breadboard.
Thanks.

UPDATE.
It worked great. I checked and found very little noise in the circuit (IGBTs work near resonance). I now am trying to build a circuit monitoring 4 IGBT thermistors triggered by a 555 timer. I would use a CD4017BE to switch between each thermistor every 1.5 second and measure the frequency.
The question is to keep it isolated. An idea comes to mind to use Opto-couplers while switching using CD4017. However, what if one IGBT blows it might send dV/dt to other IGBT and blow that too since the DC source is same.
Using isolated DC-DC converters would be too pricey.
I'll come up with a schematic and get your opinion.
Thanks.

 

[Tony Stewart]

OK.. Just use low impedance drivers shielded with CM beads. Low ESL and low ESR between pre-drivers and output drivers and load. low inductance wiring. Paired wires reduces crosstalk, shield even more and CM chokes as well rated for frequency of transients.

 

[kubeek]

... circuit monitoring 4 IGBT thermistors ...

What kind of alien technology is that?

 

[Vijit Dubey]

What kind of alien technology is that?

ff900r12ip4 IGBTs have thermistors connected to the cooling plate. I am using 4 such thermistors in a timer circuit to monitor temperatures and display on PLC.
Here is the datasheet LINK

 

[Tony Stewart]

Will system response time protect IGBT's?

What is spec?

 

[Vijit Dubey]

Will system response time protect IGBT's?

What is spec?

Specs of what?

 

[Vijit Dubey]

Tony Stewart
I ran it when load was 70% on a 300KW machine. Monitoring the thermistors of inverter IGBTs which are water cooled.
**broken link removed**
when no load is operated. there is no flickering of the frequency.

 

[Tony Stewart]

Tony Stewart
I ran it when load was 70% on a 300KW machine. Monitoring the thermistors of inverter IGBTs which are water cooled.
**broken link removed**
when no load is operated. there is no flickering of the frequency.

Specs for temperature accuracy , thermal resistance of Tjc to estimate junction temp, control system response to load control with dynamic loads (arcs)

 

[Vijit Dubey]

Tony Stewart here are the schematics where we can select upto 24 IGBT thermistors. All you have to do is jump the number of IGBT you want to monitor.
What do you think?
Also, If I want to get a board printed and donot have a account of EAGLEcad (autoroute option). What else do you suggest to design one?

 

[Tony Stewart]

I was wondering if you measured calibration error somehow.

 

[Vijit Dubey]

I was wondering if you measured calibration error somehow.

The thing is that looking at the difference between what simulation tells me and what I found was around 20 degrees but the difference between each temperature was almost similar. So, I am assuming it is due to resistances in the rest of the circuit. I right now used actual resistor(1/4 W) to come up with the green column and I am going to use this as a reference for my thermistor.
I also read the thermistor reading using a multimeter and it did fell on the same range of he green column.

Is there any other way that I could do it?