Dr. Stiffler returns with a new device: SFM

[TinselKoala]

TK,

We are driving the LED arrays directly with our Rigol sig gens so we can't try your suggestion above.  We really are pushing the accuracy of our scopes at these frequencies with stray  pickup, etc, so the possibility of measurement error is pretty high.

Regards,
Pm

Oh, I thought several people had already progressed to stand-alone oscillators at the desired frequencies. I thought I saw that Lidmotor, Slider, and others reporting on OU forum seem to have dispensed with the need for the signal generator. Do these people have negative input power results? If so they should be able to do the suggested supercap test.

[partzman]

Itsu,

I have concluded at this point in time (in my mind anyway) that the differential probe measurements on the csr are beyond our scopes capability of giving accurate consistent results.  So, the attached test was taken with my TCP0020 current probe that I set the deskew on manually.  I used a 50 ohm non-inductive resistor with short leads at 15MHz to get a zero voltage/current phase angle and this was done following a scope recal.  I believe this is the most accurate measurement I can muster on this device at this point in time.

TK- Yes, I believe those fellows have built various drivers but I'm not sure if they have attempted accurate input measurements.

Regards,
Pm 

[Itsu]

PM,

Using severall browsers (Chrome, FireFox, IE, Edge), they all give audio on my latest video.


I will do the same calibration with a 50 Ohm resistor on my scope.

Looking at your screenshot i see you still have a negative value for input power (-40.02mW).
But is this with the leds full (as full as possible) on, so in resonance or somewhere inbetween (chasing
the negative value so to speak)?

Itsu

[ion]

I've stated earlier that the stand alone low power oscillator unit should reveal the truth quickly by enabling easy test of input power at DC. (post #48)

I agree with TK and the proposed stand alone unit and supercap test, with one change: I would not use a supercap as you probably don't need anywhere near that level of storage capacity to see negative power, and may get in trouble with things like dielectric absorption fooling you.

If you don't have  accurate DC power measuring devices on hand (couple of DMM's and a CSR) or don't trust your computations or DMMs, I think a good quality lower value, low dielectric absorption cap should reveal the truth and quickly by noting the charge decay over a short time period as the circuit is tuned for alleged negative power. Use a push switch to power supply or battery to replenish while tuning. Note the decay during each discharge cycle. You've hit paydirt when you no longer need to recharge the cap and the circuit is still running.

Regards

[partzman]

PM,

Using severall browsers (Chrome, FireFox, IE, Edge), they all give audio on my latest video.


I will do the same calibration with a 50 Ohm resistor on my scope.

Looking at your screenshot i see you still have a negative value for input power (-40.02mW).
But is this with the leds full (as full as possible) on, so in resonance or somewhere inbetween (chasing
the negative value so to speak)?

Itsu

Itsu,

My audio problem disappeared with a computer re-boot.  I have also run some light tests but I was not using the SG for the input source.  In those tests, I seemed to get equal power levels or close to it comparing DC to high frequency.

The negative input I manage to get is at a lower frequency than the resonance frequency that gives the most light from the led array.  There is generally no light output from the array and the frequency is sensitive to adjust for the maximum negative input.  With the leds full on at resonance, the input power is around 130mw or so.

Regards,
Pm

[TinMan]

Itsu,

My audio problem disappeared with a computer re-boot.  I have also run some light tests but I was not using the SG for the input source.  In those tests, I seemed to get equal power levels or close to it comparing DC to high frequency.

The negative input I manage to get is at a lower frequency than the resonance frequency that gives the most light from the led array.  There is generally no light output from the array and the frequency is sensitive to adjust for the maximum negative input.  With the leds full on at resonance, the input power is around 130mw or so.

Regards,
Pm

Pm

While you have the DUT in the negative value range,could you try a small test--turn all your lights off,and see if that changes your P/in value on the scope.


Brad.

[partzman]

Itsu,

This is the result when using a square wave to drive the circuit.  I know Dr. Stiffler recommends sine wave only but really, with a series resonant circuit as it is, the only difference would be increased output with the sq wave IMO.  But I may be somehow missing the 'magic'.

The change in the normally flat top and bottom of the sq wave is revealing.  One can see that as the current reverses thru zero, the sq wave voltage increases in the positive or negative direction depending on the polarity.  This is evidence of power being returned to the signal source.  This is neat because it gives us a way to utilize this negative energy due to the fact that a pulsed sq wave source is far easier to build to recover the negative energy verses a sine source.

I'm curious to see what results you get with your led array driven with a sq wave.

Regards,
Pm

[partzman]

Pm

While you have the DUT in the negative value range,could you try a small test--turn all your lights off,and see if that changes your P/in value on the scope.


Brad.

Brad,

OK, I just did as you suggested and I see no change in the negative input value.  I might add that my lab gets quite dark with no lights on.

Regards,
Pm

[Itsu]

Itsu,

My audio problem disappeared with a computer re-boot.  I have also run some light tests but I was not using the SG for the input source.  In those tests, I seemed to get equal power levels or close to it comparing DC to high frequency.

The negative input I manage to get is at a lower frequency than the resonance frequency that gives the most light from the led array.  There is generally no light output from the array and the frequency is sensitive to adjust for the maximum negative input.  With the leds full on at resonance, the input power is around 130mw or so.

Regards,
Pm

PM,

in also calibrated my current probe with an inductionfree 50 Ohm resistor at 15Mhz to have a phase difference of 0° (or very near) between voltage and current.
I had to add 6ns deskew to the current probe channel.

Then measured again the leds scanning for most negative input, which seems to be near your -40mW as its -45.32mW at 11.2Mhz (below resonance).
phase difference current -> voltage is 130°, see schreenshot.

The leds where not on, and covering them with something did no change the value.


What equipment would be needed to accuratly measure the input power?  Would it be something like a Vector Network Analyzer (HP 8753D or so)?


Itsu

 

[Itsu]

Itsu,

This is the result when using a square wave to drive the circuit.  I know Dr. Stiffler recommends sine wave only but really, with a series resonant circuit as it is, the only difference would be increased output with the sq wave IMO.  But I may be somehow missing the 'magic'.

The change in the normally flat top and bottom of the sq wave is revealing.  One can see that as the current reverses thru zero, the sq wave voltage increases in the positive or negative direction depending on the polarity.  This is evidence of power being returned to the signal source.  This is neat because it gives us a way to utilize this negative energy due to the fact that a pulsed sq wave source is far easier to build to recover the negative energy verses a sine source.

I'm curious to see what results you get with your led array driven with a sq wave.

Regards,
Pm

OK,  here with the setting as earlier (below resonance etc.), just changed from sine wave to square wave:


Itsu

[partzman]

OK,  here with the setting as earlier (below resonance etc.), just changed from sine wave to square wave:


Itsu

Itsu,

OK, we are in reasonably close agreement with our measurements even though we are using different led arrays.   

Maybe someone who has used the HP VNA could answer your question on measurement accuracy but I have no experience with those so I'm no help.

In the meantime, I ran a test using the same frequency with an Ixys mosfet driver which does not like to run at 15MHz with negative energy being fed back to it.  I mean it only runs for a few seconds before overheating and I have it heat sinked!  Anyway I finally captured the scope shot below and with the supply at 25vdc, the p-p drive is higher than can be achieved with the Rigol SG.

Maybe the next step would be to raise the series inductance to lowering the operating frequencies and see if the effect goes away.

I also have ripped apart a 60w/9w led flood lamp that has 15 leds in series and if I can get the array off the heat sink I'll give it a try.

Regards,
Pm

[Itsu]

Hmmm,  the IXDD609 is a very sturdy driver (9A source/sink), so i am amazed it gets that hot that fast.

So now you should add the pulled power from the driver into the mix to get an accurate input calculation.

Itsu

[partzman]

Hmmm,  the IXDD609 is a very sturdy driver (9A source/sink), so i am amazed it gets that hot that fast.

So now you should add the pulled power from the driver into the mix to get an accurate input calculation.

Itsu

It's the frequency that's the killer.  In the data sheet it shows a typical rise and fall time of 22ns and 15ns respectively but with the the period of 15MHz being 67ns, there will be considerable power generated in the device especially if the rise and fall times are slower.

I have found that the 609 is not that efficient even at several MHz so, the power drawn from the supply at 15MHZ will be enormous.  I'm only considering the power from the waveform feeding the device with the idea of engineering an efficient driver later on.

Pm   

[Centraflow]

OK I have been thinking about this and what I have come up with is:_

A diode has a significant capacitance and can be deemed as a capacitor in certain circumstances. The LED's form a loop, an energy consuming loop as it is made up of light emitters. The loop is completed by two HF diodes in series, we now have a HF current loop.

Yes for some I have talked about this before.

Now how is this loop driven, well it is a capacitive drive on the back plate of the loop from a HF source, in this case the coil with either a square or sine wave. As PM say's it can be square wave as the coil will generate the sine wave from the square wave.

So what do we have? We have a HF magnetic loop which does not need a ground plane as they use the natural Earth as the ground plane (it is a magnetic and not an electric near field). The field is radiated directly into the LED's which form the loop. The two or more diodes are the tuning capacitance of the loop, I think if you know the series capacitance of the two looping diodes and replaced them with a capacitor of the same value, the LED's will light.

Regards

Mike

[Centraflow]

I said radiated in my above post, it is not radiated, it is probably a standing wave lighting the diodes and when you tune and they don't light the current goes back to source, hence your negative input.

Regards

Mike

[Itsu]

It's the frequency that's the killer.  In the data sheet it shows a typical rise and fall time of 22ns and 15ns respectively but with the the period of 15MHz being 67ns, there will be considerable power generated in the device especially if the rise and fall times are slower.

I have found that the 609 is not that efficient even at several MHz so, the power drawn from the supply at 15MHZ will be enormous.  I'm only considering the power from the waveform feeding the device with the idea of engineering an efficient driver later on.

Pm   

I tried an IXRFD630 RF MOSFET driver (30A up till 40Mhz), but it also does not like this setup.
Its pulling 2A (@ 24V) and heats up rapidly using 5Vpp square wave as input so i stopped using it.

I think it needs a solid return line, and not the floating return through the air/capacitance.

Itsu

[Itsu]

OK I have been thinking about this and what I have come up with is:_

A diode has a significant capacitance and can be deemed as a capacitor in certain circumstances. The LED's form a loop, an energy consuming loop as it is made up of light emitters. The loop is completed by two HF diodes in series, we now have a HF current loop.

Yes for some I have talked about this before.

Now how is this loop driven, well it is a capacitive drive on the back plate of the loop from a HF source, in this case the coil with either a square or sine wave. As PM say's it can be square wave as the coil will generate the sine wave from the square wave.

So what do we have? We have a HF magnetic loop which does not need a ground plane as they use the natural Earth as the ground plane (it is a magnetic and not an electric near field). The field is radiated directly into the LED's which form the loop. The two or more diodes are the tuning capacitance of the loop, I think if you know the series capacitance of the two looping diodes and replaced them with a capacitor of the same value, the LED's will light.

Regards

Mike

Mike,

nice theory, so i have replaced the 2 diodes (1n4148) with a capacitor, but i cannot light the leds that way.

According to the data sheet, the 1n4148 has a capacitance of 4pF, so 2 in series makes 2pF
I replaced them by 3 x 10pF caps in series, so 3.3pF, but as said, no leds turn on.

The signals look similar as with the diodes, including the resonance frequency around 15Mhz (voltage dips) and
the negative power value below resonance.

Regards Itsu

[partzman]

Mike and Itsu,

I also had no light at resonance with a 5pf mica cap but the negative input was nearly the same level as with the 2-1N4148s.  I also seem to have over double the negative input when being driven from the SG this afternoon as we have a heavy cloud cover with slight rain!

I have also tried lower frequencies but don't see any appreciable negative input thus far.

Pm

[partzman]

Out of curiosity, I decided to connect just a foil to the inductor instead of the led array.  The scope pix below shows the results.

The frequency is 13.1MHz and the series inductor is 20.51uH.  The resonance calculation results in an equivalent C of 7.2pf.  The foil is 13mm x 25mm x 0.2mm thick (w/paper backing).

There is no light however! 

Pm

[Itsu]

So this measured/calculated negative input power has nothing to do with the leds and/or the diodes.

Replacing the coil with something else would probably show that its coming from there, but what could
be an equivalent to a coil?

Could the air humidity influence the capacitive return to ground, my first guess says no, but need to
check on that.

Anyway, it seems that input power measurements with this setup are almost impossible.

I could go the route mentioned by TK and ION as i have a cheap chinese dual CH FG up till 8MHz which is powered
by 5V via a 12V battery.

Using a bigger coil to resonate around 7MHz or so and measure the DC input when in resonance etc.
The problem i have is that its only capable of about 10Vpp.

Itsu

[Centraflow]

There is no light however! 

Pm

I nearly fell off my seat and choke on my first coffee this morning.

Now a question, does the forward current and the reverse current add up to more than the front end input current?

It can be calculated that it does on paper, so why is main stream shying away from OU?

I'm going to hopefully show this with a little help from my friends


Regards

Mike

[TinMan]

So this measured/calculated negative input power has nothing to do with the leds and/or the diodes.

Replacing the coil with something else would probably show that its coming from there, but what could
be an equivalent to a coil?

Could the air humidity influence the capacitive return to ground, my first guess says no, but need to
check on that.

Anyway, it seems that input power measurements with this setup are almost impossible.

I could go the route mentioned by TK and ION as i have a cheap chinese dual CH FG up till 8MHz which is powered
by 5V via a 12V battery.

Using a bigger coil to resonate around 7MHz or so and measure the DC input when in resonance etc.
The problem i have is that its only capable of about 10Vpp.

Itsu

Would this little circuit be of any use Itsu?

4.5 - 13.7 MHz

http://www.vk2zay.net/article/93


Brad

[TinMan]

Out of curiosity, I decided to connect just a foil to the inductor instead of the led array.  The scope pix below shows the results.

The frequency is 13.1MHz and the series inductor is 20.51uH.  The resonance calculation results in an equivalent C of 7.2pf.  The foil is 13mm x 25mm x 0.2mm thick (w/paper backing).

There is no light however! 

Pm

Looks like the same sort of track my wife takes around the shopping center,when we go grocery shopping 


Brad

[partzman]

So this measured/calculated negative input power has nothing to do with the leds and/or the diodes.

It would appear so.  However, try a test with the 2-1N4148 diodes removed from the led array.  It will still produce negative input with some lower frequency adjustment but is weaker than with the 2 diodes in place. 

Replacing the coil with something else would probably show that its coming from there, but what could
be an equivalent to a coil?

Could the air humidity influence the capacitive return to ground, my first guess says no, but need to
check on that.

Anyway, it seems that input power measurements with this setup are almost impossible.

Yes and can be appreciated if one makes the attempt.  I understand that we are capacitively coupling back to ground thru anything in proximity of the test, but why the current phase which is apparently producing the negative input.  Perhaps we are coupling back to our current probes.  Maybe a better calibration would be to do the 50 ohm deskew cal with the inductor connected without the led array.  Hmmm, will have to try that.

I could go the route mentioned by TK and ION as i have a cheap chinese dual CH FG up till 8MHz which is powered
by 5V via a 12V battery.

Yes, that is a good idea to account for the input energy but, I think that until the power level is raised to a point of overcoming the generator (whatever means) losses, the results will be disappointing.

Pm

Using a bigger coil to resonate around 7MHz or so and measure the DC input when in resonance etc.
The problem i have is that its only capable of about 10Vpp.

Itsu

[partzman]

I nearly fell off my seat and choke on my first coffee this morning.

Now a question, does the forward current and the reverse current add up to more than the front end input current?

If I understand correctly, the input current at the point of measurement with the probe will or should average to zero over complete cycles.  Portions of this current will at times be supplied by the generator and for the remainder of time will be returned to the generator.  Are you meaning to measure the magnitudes of these two current conditions and compare to see which is greater?

Edit-

Pm

It can be calculated that it does on paper, so why is main stream shying away from OU?

I'm going to hopefully show this with a little help from my friends


Regards

Mike