partzmans board ATL

Author Topic: partzmans board ATL (Read 36146 times)

Verpies	Re: partzmans board ATL « Reply #700 on: 2025-03-14, 22:28:57 »
Group: Administrator Hero Member Posts: 4513	Quote from: partzman on 2025-03-14, 15:28:42 The schematic is the first pix. Untwisted partzmans board ATL CS Ver2d1 Schematic.png (27.64 kB, 880x496 - viewed 285 times.)

Verpies	Re: partzmans board ATL « Reply #701 on: 2025-03-15, 05:55:29 »
Group: Administrator Hero Member Posts: 4513	Does this circuit behave differently ? partzmans board ATL CS Ver2d2.png (28.46 kB, 880x496 - viewed 265 times.)

partzman

Re: partzmans board ATL « Reply #702 on: 2025-03-15, 13:43:20 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: verpies on 2025-03-14, 22:28:57

Untwisted

Verpies,

The reason I drew the circuit as I did and not the "untwisted" way, is I wanted the viewer to see the physical positioning of the capacitor in the toroid relative to the primary or L1. IOW, the dot end of L1 is the primary start where the wire starts at the top side of the toroid, enters the hole, comes out the bottom of the toroid, proceeds up the outside of the toroid and returns to the top of the toroid for the next turn. Cs1 has the negative end positioned at the top of the hole in the toroid so the schematic gives a picture of this physical relationship.

I agree that your version is a little easy to read!

Regards,
Pm

partzman

Re: partzmans board ATL « Reply #703 on: 2025-03-15, 14:11:38 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: verpies on 2025-03-15, 05:55:29

Does this circuit behave differently ?

It's operation is different but the overall results are nearly the same. Although I find it difficult to admit to the fact that charge separation is not needed for this circuit to produce OU, but it is true.

The attached sim schematic shows the circuit without any charge separation which yields OU on the bench but not in simulation. IOW, the simple circuit using two caps with dissimilar biasing voltages requires more investigation.

CS does enhance the performance in some ways however as is seen when using different recovery techniques on the caps.

regards,
Pm

partzmans board ATL

NoCS Schematic.png (17.38 kB, 831x519 - viewed 235 times.)

Verpies	Re: partzmans board ATL « Reply #704 on: 2025-03-15, 14:49:43 »
Group: Administrator Hero Member Posts: 4513	This is worth watching before accounting for energy stored in biased caps. https://youtu.be/Tfatk7wmnhs?t=292

web000x

Re: partzmans board ATL « Reply #705 on: 2025-03-15, 18:25:39 »

Sr. Member

Posts: 441

Quote from: partzman on 2025-03-15, 14:11:38

It's operation is different but the overall results are nearly the same. Although I find it difficult to admit to the fact that charge separation is not needed for this circuit to produce OU, but it is true.

The attached sim schematic shows the circuit without any charge separation which yields OU on the bench but not in simulation. IOW, the simple circuit using two caps with dissimilar biasing voltages requires more investigation.

CS does enhance the performance in some ways however as is seen when using different recovery techniques on the caps.

regards,
Pm

Very interesting Partzman, thank you. I had a sneaking suspicion that the charge separation wasn't the source of what you were seeing. Most of your earlier calculations were from calculating the energy from .5*C*V^2 which were a little ambiguous because no real power seemed to be able to be siphoned off before the caps returned to zero. A single loop of wire showed almost identical waveforms in my preliminary experiments. Dollard spoke about the difference in two types of volts 1)e = Psi/C and 2)E = Phi/t. It appears the latter was appearing on the wires but didn't actually charge the cap to that measured voltage. It wasn't until you started biasing the caps with a precharge that you started to see measurable gains by integrating the power on the scope shots.

I decided to not go into work this evening due to impending weather so hopefully I can get to some experiments and fingers crossed that my utility power stays on uninterrupted.

@verpies,

I guess if we want to know the energy in a charged capacitor that we'll just have to remove it from the circuit and see what kind of energy density it can dump into a load from the bias levels.

Regards,

Dave

partzman

Re: partzmans board ATL « Reply #706 on: 2025-03-15, 19:54:25 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: web000x on 2025-03-15, 18:25:39

Very interesting Partzman, thank you. I had a sneaking suspicion that the charge separation wasn't the source of what you were seeing. Most of your earlier calculations were from calculating the energy from .5*C*V^2 which were a little ambiguous because no real power seemed to be able to be siphoned off before the caps returned to zero. A single loop of wire showed almost identical waveforms in my preliminary experiments. Dollard spoke about the difference in two types of volts 1)e = Psi/C and 2)E = Phi/t. It appears the latter was appearing on the wires but didn't actually charge the cap to that measured voltage. It wasn't until you started biasing the caps with a precharge that you started to see measurable gains by integrating the power on the scope shots.

I decided to not go into work this evening due to impending weather so hopefully I can get to some experiments and fingers crossed that my utility power stays on uninterrupted.

@verpies,

I guess if we want to know the energy in a charged capacitor that we'll just have to remove it from the circuit and see what kind of energy density it can dump into a load from the bias levels.

Regards,

Dave

The value of an electrolytic capacitor when being charge separated is always less than the measured capacitance. I never did come to any conclusion for this but suspect the outer aluminum case to have something to do with it. With film caps, there is still a small difference however.

So far I have not built the remaining bench circuitry to reset the caps starting voltages for the final phase so the jury is still out. My tests so far for this final phase have been done in LtSpice with the models as accurate as possible IMO.

Pm

partzman

Re: partzmans board ATL « Reply #707 on: 2025-03-15, 20:09:30 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: verpies on 2025-03-15, 14:49:43

This is worth watching before accounting for energy stored in biased caps.
https://youtu.be/Tfatk7wmnhs?t=292

After some consideration, this could be the reason for the differences between the sim and bench for the circuit with just the biased caps. The answers will come with a completed bench circuit which I hope to have results from over the next few days.

Pm

web000x

Re: partzmans board ATL « Reply #708 on: 2025-03-17, 01:31:43 »

Sr. Member

Posts: 441

Quote from: partzman on 2025-03-15, 20:09:30

After some consideration, this could be the reason for the differences between the sim and bench for the circuit with just the biased caps. The answers will come with a completed bench circuit which I hope to have results from over the next few days.

Pm

How much do your results change using these caps without the toroid?

I used a 5600uF 400V cap for Cs and a 1400uF 400V for Cx while using an air core 1 mH coil for L1 and was not able to see any gains into Cx and the battery vs the loss in Cs using just the caps and no charge separation. I also tried a 390uF 450V for Cs and 120uF 450V for Cx and got similar results as before. I can gather details on these if needed but haven't since there is seemingly nothing to call home about.

I would have replicated the charge separation circuit variation but didn't have any higher value axial caps that would fit in any of the toroids I have. It would seem that for those small toroids that you've been using that in order to fit a 1000uF cap inside of one that the voltage rating might be rather low. I'm assuming it is listed as 63V electrolytic. This could be why you're seeing the gains with just the caps as verpies pointed out in the video with regards to derating upon charge saturation. I might not be seeing anything from this circuit because my caps are rated for a higher voltage and I was only biasing with about 55V, not to mention my ESR's measure 100 - 200 mOhm.

I was going to try some film caps that I have that are 10uF 250V (20uF for Cs and 10uF for Cx) with the charge separation circuit and a toroid but they're a little more difficult to maintain a steady charge before throwing the pulse trigger. The mosfet gate pulse cycle is a little awkward and I've not come up with a way using my cheap Chinese function generator to hold G2 on indefinitely to charge balance Cs and Cx before pulsing G1 with a burst pulse. I may need to write an Arduino program for it.

Dave

Edit: Capacitors measure in the 100 - 200 mOhm range, not 1 -2 mOhm.

« Last Edit: 2025-03-17, 14:53:40 by web000x »

partzman

Re: partzmans board ATL « Reply #709 on: 2025-03-17, 13:37:27 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: web000x on 2025-03-17, 01:31:43

How much do your results change using these caps without the toroid?

IF one considers only the energy generated in V+, there is not much difference. However, it is the voltages remaining on Cs and Cx after the first phase that are important and I'm presently documenting those differences.

Quote

I used a 5600uF 400V cap for Cs and a 1400uF 400V for Cx while using an air core 1 mH coil for L1 and was not able to see any gains into Cx and the battery vs the loss in Cs using just the caps and no charge separation. I also tried a 390uF 450V for Cs and 120uF 450V for Cx and got similar results as before. I can gather details on these if needed but haven't since there is seemingly nothing to call home about.

I would have replicated the charge separation circuit variation but didn't have any higher value axial caps that would fit in any of the toroids I have. It would seem that for those small toroids that you've been using that in order to fit a 1000uF cap inside of one that the voltage rating might be rather low. I'm assuming it is listed as 63V electrolytic. This could be why you're seeing the gains with just the caps as verpies pointed out in the video with regards to derating upon charge saturation. I might not be seeing anything from this circuit because my caps are rated for a higher voltage and I was only biasing with about 55V, not to mention my ESR's measure 1 - 2 mOhm.

My Cs and Cx are rated at 50v and Cs is operating at that level. In regards to the video that Verpies pointed out, it is reasonable to assume that with the small dE involved with both Cs and Cx that any resultant dC would also be small however, it is these small dE's that are creating the not so small energy levels so the jury is still out!

Your esr's is an order of magnitude less than mine and this could be the reason I'm seeing anomalous energy comparisons in Cs and Cx.

Quote

I was going to try some film caps that I have that are 10uF 250V (20uF for Cs and 10uF for Cx) with the charge separation circuit and a toroid but they're a little more difficult to maintain a steady charge before throwing the pulse trigger. The mosfet gate pulse cycle is a little awkward and I've not come up with a way using my cheap Chinese function generator to hold G2 on indefinitely to charge balance Cs and Cx before pulsing G1 with a burst pulse. I may need to write an Arduino program for it.

Dave

One way I have created the G1 and G2 pulses is to use a half bridge to drive g1 and G2 connected together. In this way, G2 is "on" before the cycle start thus charging Cs to the full Vbias voltage. The problem with this is that at the end of the cycle, G2 remains on and the current in L1 continues to conduct in the negative direction.

Pm

Magluvin

Re: partzmans board ATL « Reply #710 on: 2025-03-17, 14:34:02 »

Group: Mad Scientist
Hero Member

Posts: 1105

caps that are used in high end car amplifier power supplies are usually very low esr and when in use can handle a lot of in/out currents.

also, if you use smaller value low esr caps, and parallel them for the value needed, the esr is lowered even more. in my soundstream reference 705, you can see the yellow cap bank. each 1000uf and specify low esr on the label.

mags

partzmans board ATL

UQAAAgGXzeA-960.jpg (662.55 kB, 960x1280 - viewed 208 times.)

web000x	Re: partzmans board ATL « Reply #711 on: 2025-03-17, 14:38:42 »
Sr. Member Posts: 441	My ESR for my capacitors was actually in the 100 -200 mOhm range. I accidently shifted the decimal the wrong way in my head as I was making the post. I'd found that the half bridge works well too for initializing the conditions for the pulse but then it becomes difficult to turn it off in time without oscillations causing energy discharge from V+. Dave

Magluvin	Re: partzmans board ATL « Reply #712 on: 2025-03-17, 15:45:16 »
Group: Mad Scientist Hero Member Posts: 1105	may be wrong on the 1000uf. the pic is from online. looks like 2200uf 16v. will take a pic of my amp mags

partzman

Re: partzmans board ATL « Reply #713 on: 2025-03-25, 21:30:30 »

Group: Moderator
Hero Member

Posts: 2232

These are the test results of a single cycle bench recovery of the Cs1 and Cx1 capacitor voltages back to their original starting values as seen in the energy generating cycle in my previous post #698.

The schematic is shown first of the circuitry used. Essentially, L1 is first charged to an energy level from Cx1 which reduces the energy in Cx1 then, the energy level in L1 is discharged into Cs1 which raises it's energy level and Cx1 which reduces it's energy level even further. L1 must be capable of high currents without saturation plus have low resistance.

SP1 shows the beginning voltage levels on Vc12 CH3(pnk) and Vc1 CH2(blu) of 50.09v and 429.1mv respectively. CH4(grn) shows the current ramping in L1.

SP2 shows the ending voltage levels on Vc12 and Vc1 of 49.73v and -17.61mv respectively.

SP3 shows the peak current reached in L1 on CH4(grn) of 4.703A.

Although these recovered voltages vary slightly from the voltages on the energy generating phase on post #698, it can be seen that there is a slight gain in both Cs1 and Cx1 during this recovery process. The voltage across Vc2 remains relatively constant from cycle to cylce but increasing over time. In continuous operation, these voltage levels will vary up and down slightly but overall the resulting gain in the LAB will be infinite.

I might add that the vertical on CH3(pnk) is set such that it is operating at 15-16 bit resolution.

Regards,
Pm

partzmans board ATL

Cs1_Cx1 Recovery Schematic.png (21.12 kB, 769x670 - viewed 192 times.)

partzmans board ATL

SP1.png (23.06 kB, 800x480 - viewed 57 times.)

partzmans board ATL

SP2.png (23.13 kB, 800x480 - viewed 51 times.)

partzmans board ATL

SP3.png (23.24 kB, 800x480 - viewed 48 times.)

web000x

Re: partzmans board ATL « Reply #714 on: 2025-03-25, 23:05:03 »

Sr. Member

Posts: 441

Hi Partzman,

I am failing to see the gain you speak of in this circuit. From my calculations, you have a loss of 7.43 mJ in Cx1 and a gain of 3.25 mJ in Cs1 so the gain in the latter is less than what is drawn from the former. I'm also unclear how such a circuit is feeding anything back into the LAB as there doesn't appear to be a current path through the battery. Maybe you can elaborate on the numbers you've crunched or maybe I'm not seeing a crucial detail in circuit operation?

Thanks,

Dave

partzman

Re: partzmans board ATL « Reply #715 on: 2025-03-26, 14:21:54 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: web000x on 2025-03-25, 23:05:03

Hi Partzman,

I am failing to see the gain you speak of in this circuit. From my calculations, you have a loss of 7.43 mJ in Cx1 and a gain of 3.25 mJ in Cs1 so the gain in the latter is less than what is drawn from the former. I'm also unclear how such a circuit is feeding anything back into the LAB as there doesn't appear to be a current path through the battery. Maybe you can elaborate on the numbers you've crunched or maybe I'm not seeing a crucial detail in circuit operation?

Thanks,

Dave

Hi Dave,

I can see that some confusion might exist because of the way that I've presented this concept. My bad! Anyway, take a look at my post #698 with the circuit and scope traces. This post discloses the energy generation phase to the LAB of the whole concept. The circuit is not intuitive and must be studied closely because it's operation is somewhat complex.

L1 is initially charged by the differential voltage between Cs1 and Cx1 that is created by V+. The instant that V+ is applied to L1, the voltage on the negative end of Cs1 (which initially is at zero volts) goes to ~+3v due to charge separation which is the volts/turn of the primary circuit. IOW, 12.5/4=3.125v . Also during the charging of L1, the voltage across Cs1 decreases and the voltage across Cx1 increases theoretically but in actual measurement may not appear to be so. This is due to the individual esr's of Cs1 and Cx1. This is evidenced by the sudden rise in the voltage on VL1a at the switching of M1 and M2 and at the end of the completed cycle.

After L1 reaches a predetermined current peak, M1 switched off and M2 is switched on and again due to charge separation, the voltage on VL1a is reduced by the volts/turn to ~0v . This allows the current in L1 to collapse again thru Cs1 and Cx1 and also thru V+. During this collapse, energy is generated in V+ plus the voltage across Vs1 is reduced more along with the voltage across Cx1 is increased more. When the current in L1 reaches zero, M2 is switched off resulting in the increase in VL1a due to the esr.

At the end of this cycle, we now have new voltage levels from beginning to end. These ending voltage levels are the beginning voltage levels used in the recovery circuit shown in post #713 . So yes, there is energy lost in Cs1 and energy gained in Cs1 as they are restored to their starting voltage levels. This is done at no cost to V+ . The charging of L1 also takes no energy from V+ . So, the gain of 1.928mJ seen in post #698 comes at no cost and is therefore infinite.

I might add that the 160uH inductor labeled L1 in the recovery circuit should be labeled Lr as it is separate from L1. Also, L1 is shorted in the recovery mode to prevent unwanted interaction from it's open inductance as the current thru Ls1 creates induction in L1!

Regards,
Pm

Magluvin	Re: partzmans board ATL « Reply #716 on: 2025-03-31, 23:32:27 »
Group: Mad Scientist Hero Member Posts: 1105	hey pm i see the circuit in the post you refered to. the winding and cap in the box.... have you shown in pics exactly what that entails? if not, then we are stuck with experimentaion of infinite possibilities and maybe no success. i personally cant afford that time with guesses. i see you have presented the circuit, but for me there needs to be more. mags

partzman

Re: partzmans board ATL « Reply #717 on: 2025-04-01, 15:55:45 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: Magluvin on 2025-03-31, 23:32:27

hey pm

i see the circuit in the post you refered to. the winding and cap in the box.... have you shown in pics exactly what that entails? if not, then we are stuck with experimentaion of infinite possibilities and maybe no success. i personally cant afford that time with guesses.

i see you have presented the circuit, but for me there needs to be more.

mags

Mags,

There are pix of examples on posts #500, 505, and 512 plus others. You are correct in the many possibilities with the circuit and for that reason, I have decided to refrain from posting more info and rather concentrate on building a final working proof of concept. This way others like yourself who wish to replicate won't be wasting their time during the development phase!

Regards,
Pm

Verpies	Re: partzmans board ATL « Reply #718 on: 2025-05-12, 18:32:17 »
Group: Administrator Hero Member Posts: 4513	Quote from: partzman on 2025-04-01, 15:55:45 I have decided to refrain from posting more info and rather concentrate on building a final working proof of concept. How's it going ?

partzman	Re: partzmans board ATL « Reply #719 on: 2025-06-03, 14:40:56 »
Group: Moderator Hero Member Posts: 2232	Quote from: verpies on 2025-05-12, 18:32:17 How's it going ? Work in Progress. Pm

Verpies	Re: partzmans board ATL « Reply #720 on: 2025-06-03, 15:44:03 »
Group: Administrator Hero Member Posts: 4513	Did you account for the dependence of capacitance on voltage in some capacitors, when calculating E=½CV ?

partzman

Re: partzmans board ATL « Reply #721 on: 2025-06-05, 15:40:30 »

Group: Moderator
Hero Member

Posts: 2232

Quote from: verpies on 2025-06-03, 15:44:03

Did you account for the dependence of capacitance on voltage in some capacitors, when calculating E=½CV ?

Yes, but this effect doesn't seem to cause any appreciable change in the capacitors I'm using. What is appreciable, is the apparent lower value of the charge separated capacitance as compared to the normally measured capacitance as I've explained in this thread.

Smudge	Re: partzmans board ATL « Reply #722 on: 2025-06-30, 07:31:11 »
Group: Professor Hero Member Posts: 2332	I have just come across something that I downloaded in 2017 that applies to Partzman's work here. An Experimental Evidence of Energy Non-Conservation_1005.0078v1.pdf (88.53 kB - downloaded 44 times.) partzmans board ATL Extract.png (42.34 kB, 599x407 - viewed 55 times.)

F6FLT

Re: partzmans board ATL « Reply #723 on: 2025-07-01, 15:08:23 »

Group: Professor
Hero Member

Posts: 2435

Quote from: partzman on 2025-03-14, 15:28:42

...
So, with a reset scheme for Cs1 and Cx1 that costs zero energy, we have an apparent infinite gain of 1.928mJ generated in V+.
...

That's not what we can conclude from the diagram and the measurement. I see a ‘Vbias’ generator whose energy input (or consumption) is not measured. I see a rectangular signal generator which switches the G1/G2 gates and whose energy input is not monitored either, whereas the G/S or G/D capacities can inject a signal into the rest of the circuit.

---------------------------

"Open your mind, but not like a trash bin"

Verpies	Re: partzmans board ATL « Reply #724 on: 2025-07-01, 17:04:06 »
Group: Administrator Hero Member Posts: 4513	Quote from: F6FLT on 2025-07-01, 15:08:23 ..., whereas the G/S or G/D capacities can inject a signal into the rest of the circuit. Yes, gate power is not negligible at high frequency. It should be accounted for.