An energy harvesting project

[pearldragon]

Hi Julian,

Thanks for the update/info to solve the spike issue with the amp meters. Will implement those caps (at some point) in my device as well (currently blew up both my meters ).
I want to respond a bit more elaborate on your post, also on some previous ones, but still looking for the time to so, so for now I just share some of my initial ‘quick and dirty’ results.

COP Tests
Quick and dirty, because I wanted some ball park COP (Ah in versus Ah out), numbers and as long as they are far below 1 and/or, not a noteworthy change compared to a previous measurement I don’t care too much about the accuracy yet. So to explain what I mean with that/how I performed my measurement so far:
-Discharge 1Ah out of my output battery
-Next day charge with the Solid State Charger (SSC)
---Noting every 5 tot 10min the amp reading on the display of my power supply (see image below). (this measurement is probably a bit inaccurate, still have to start using my recording multimeter)
---Record the rising voltage in my output battery with the CBA
---Stop charging at 14.30V (since that is where my conventional charger charged to as well
-Wait for an hour (while recording with CBA)
-Then start at step 1 again Discharge 1Ah.

For the calculation of the Ah the PSU put in I took the amp reading @ +/-60% of the charge time, and multiplied that by the total time of the run; as mentioned this is very rough measurement.

Generally I would do one charge + discharge a day, except for the first tests (ID 230528 CM & 230530 CM), there I did two full charge/discharge runs in 1 day.

Test IDs 230528 CM / 230530 CM / 230532 CM / 230534 CM are done with IRF840.
Test IDs 230601 CM / 230603 CM / 230605 CM are done with STP20N95K5.
Rest of the settings can be found in the attachments.

Conclusions/follow up actions:
-For test ID 230528 I did not discharge 1Ah out of the output battery the previous day, but 4 days ago. The COP is therefore +/-10% higher then when I discharge the previous day. This I noticed as well with the Bedini SG, sometimes this difference could be even far more than 10%.
-With currently selected settings no significant difference between IRF840 and STP20N95K5.
-I was planning to continue testing with the STP12N120K5, then STW12N150K5, finally STW12N170K5, but reasoned that finding the right frequency might be more important to figure out first, so will probably focus on that as a next step.

Peak HV pulse of active devices
I re-did the measurements, this time a bit more in a reproducible manner so I could track the increase of peak voltage over a couple of minutes and comparing the different active devices. I included my measurements for the IRF840 / STP20N95K5 / STP12N120K5.

Best regards,
Rodolphe

[JulesP]

I'm sorry I forgot to include the issue of spike effects on the meter. Yes, initially I had a similar issue with the meters being unable to cope with the HV spikes being added to the DC voltage, although the main battery should normally act as a sink for them to effectively dampen them, so theoretically they should not reach the meters.

What I did was put two small capacitors (100uF, 50V) across the negatives and the meter voltage feeds as shown in the attached graphic and the photo, and that seemed to solve the problem.





The graphic shows the wiring for the panel that holds the PWM module, with its own On-Off switch, and the two small (not very accurate) panel meters.

Your spike voltages are good for the device used (IRF840) although seem to be switching off slower, hence the curved slopes. Off course the time base you are using will change the appearance a lot.

J

PS. Please note that my inserting the ammeter into the negative line via the plug panel is not in fact the best place for it to get an accurate reading. I found that going from the supply battery directly to the meter and then out to the negative line was best. I can give more info on that if you need. Of course, with battery swapping, you can't really use a digital meter since when the meter is in battery line during the charging stage, it won't like the spikes and may damage it so I used the configuration shown with swapping off for the CoP tests, and also similarly when substituting the supply battery with a PSU.

[pearldragon]

Hi Julian,

Some quick HV pulse measurement results with the following setup:

-100:1 probe between H19 and H10/H14:
-IRF840 as active device
-100Hz / 50% duty cycle
-Input battery voltage under load: 12.07V
(-Input battery voltage @ rest: 12.13V)

-Air coils (no core material) 62.31mH/coil: 656V (see image below)

-Welding rod as core material, 267mH/coil: 632V

The max voltage of the peaks seemed to be climbing slightly when leaving the machine run. E.g. with the welding rods it started out at +/-596 and climbed to 632V. It might have climbed a bit further if I would have left it on longer.

Regards,
Rodolphe

[pearldragon]

Hi Julian,

Just a quick update:

-Replaced the IR2121 and the last signal of your post #85 did appear (with SW5 switched ON). See attachment. Post 88 - measurements.pdf

-After finishing the chapter ‘SWITCH ON & DIAGNOSTIC CHECKS’, I connected the coils and with a battery on my input and my output, I switched ON SW5 (FET Driver). I had a separate multimeter connected to my output battery to double check that the voltage would increase which it did.
However, when I looked back to the panel meters, I saw that the panel meter for my output battery was not functioning anymore. I had one spare, replaced it. Switched on the machine again and the moment I switched ON SW5 again I toasted my spare panel meter as well .
I double checked your panel meters* and although the PCB components look slightly different than mine**, for the rest they look very similar.
Just to double check: your output panel meter did not blew up?

*
https://www.ebay.co.uk/itm/154574694810?
**
https://www.ebay.nl/itm/394215176060...3ABFBM5Nf9sbth

-I have some minor updates for the manual, but will e-mail you about it.

Best regards,
Rodolphe

[JulesP]

I will check when I get back off holiday.

J

[pearldragon]

Hi Julian,

I’ve been/am snowed under by other stuff currently, so time spent on the machine has been very little unfortunately.

Before I redo my measurements/check and compare with yours, just one thing to double check:
You write before showing your measurements:

“Regarding your PWM measurements, you should see the following traces on the various TP points going from the PWM input on H37-1 through to the Driver output/ FET Gate input on H22 (yes that should be H22 in the manual) and with the Driver off (SW5 off).”

In your last image/screenshot, from the H22-1/2 TP (from output pin 7 of the Driver) shows a square wave, giving me the impression SW5 was ON there. Or do I misinterpreted?

Best regards,
Rodolphe

[JulesP]

Hi Rodolphe,

Bad connections have been a problem with me on occasions when a circuit error occurred and only by going through every possible option step by step in a logical way, and including the most unlikely things, was the fault found. In one case a biasing resistor for a transistor, that is involved in switching on one of the cap dump routing FETs was not in fact 10k but around 5ohms! Is it Murphy’s Law or another one? So now I check each component before I solder it in place.

When I ‘hot’ recharged a battery after a test involving a controlled discharge and then HV pulse charging (as in the regular CoP test run), the battery would be left till the charger naturally stopped and then usually left overnight. So long as you know the voltage when you start a controlled discharge, and then leave it for an hour or more before you start the pulse charging, then it shouldn’t really matter what voltage is reached. You’re not expected to have a ‘globally’ consistent value and each test I treated as its own independent one.

I expect batteries and capacitors will both display an ‘electret’ effect that will boost the voltage a little but the degree and significance are worthy of another whole study!


Regarding your PWM measurements, you should see the following traces on the various TP points going from the PWM input on H37-1 through to the Driver output/ FET Gate input on H22 (yes that should be H22 in the manual) and with the Driver off (SW5 off).


H37-1 and H17-1/2 where you should see the PWM square wave input as here, so about 7V max:




H13-1/2 where you should see the inverse of the above but it will probably look the same timing-wise on the scope. If not then replace the 2N3904 and repeat




H16-1/2 where you should see a similar square wave but I expected a lower voltage as it enters pin 2 of the Driver. It reads 100V since I noticed later that I moved the x10 switch on the probe, so really it's 10.0V.




H22-1/2 there should be a similar square wave of increased voltage to 'hard drive' the FET gate for rapid switch on and off. Again this is really 12.6V max.



If not then switch out the IR2121 chip and repeat.

These checks should allow you to zoom in on the problem.

Regards,

Julian

[pearldragon]

Hi Julian,

Swap LED 3
I disconnected SW 2 completely (Swap On/Off), but LED 3 still lit very dimly as before when switching on the main switch. Maybe it relates to the topic below?:

FET driver chip / U2 / IR2121 issues/signal
For the measurements below the following things/settings were used:
-Jumper 4 and jumper 5 capped (connected)
-None of the 5 coils was connected to the PCB
-PWM module “V-“ and “Ground” connection connected to each other (and together connected to H34 (PWM-)
-C7 is indicated on the electrical schematic as a polarized capacitor, I used a non polarized:
https://www.reichelt.com/de/en/folie...NGUAGE=EN&&r=1

1) H13 & H16, SW5 OFF -> approx. 10V square-ish signal
When I disconnect SW5 completely, the signal stays the same (=no leak through switch).
2) H13 & H16, SW5 ON -> approx. 12V Square signal
3) Pin 1 IR2121, SW5 OFF -> approx. 8V Sawtooth-ish signal
4) Pin 1 IR2121, SW5 OFF, PWM module OFF -> approx. 8V signal
5) Pin 1 IR2121, SW5 ON -> approx. 12V signal
6) H22/Pin7 IR2121 SW5 ON -or- OFF -> no signal

I assume 1) 3) 4) are signal readings that are not supposed to be there with SW5 switched off. Do you have an idea what could be wrong/what I should check?
Post 84.pdf

Manual update, page 47:
Where it reads: “…then an output should be measured at TP-H17…”
I think it should read: “…then an output should be measured at TP-H22…”

Best regards,
Rodolphe

[pearldragon]

Hi Julian,

The manual was/is more than helpful, did not get further than the battery swapper section yet; spent the day before quite some time on double checking the PWM generator and the frequency counter. Got some really random results, just before I decided to tear the freq. counter apart I decided to test my test leads and found out that one of them had a bad solder connection . And so there were another couple of silly things that took me a while to figure out, but had nothing to do with the proper/improper functioning with the device itself.

Swap LED 3
It certainly hints in the direction towards C6 (or C4); it lights up when I switch ON the main switch and then goes really dim. I soldered the connections to my switches (also SW2) so cannot disconnect it easily, unless I unsolder it temporarily, which I might try in the weekend. When I switch the main switch off (to short cut C6), the LED 3 goes out as well with it. But if you say it is not of much consequence, I probably not worry about it/leave it as is.

Voltage divider
I tried to use the adjuster on the probe as well to get the signal square, but although it had an effect, I could not get it square. I’ll have a look this weekend if one of my scopes has the option to generate an impulse signal.

Battery baseline
At the moment I’m trying to get steady baseline reading with my ‘output’ battery I want to use for my first tests, meaning that I try to figure out where the battery’s resting voltage is after it is charged.
I was assuming that if I hook it up to my charger, it would at some point hit a ‘max’ voltage where the charger keeps it stable. Apart for the fact that it fairly long (10h of charging) to reach this point, if I compare he graph of the day before to today, the voltage where it flattens out sits a bit higher than yesterday… 13.90V versus 13.82V.
Also the 1h rest period seemed not enough to stabilize the voltage complete in this particular battery (24Ah, Flooded Lead acid VMF 52805 12N24-4). I charge currently with the ‘small’ setting, 0.9A. The standard setting is 3.5Ah, but which is actually intended for batteries from 35Ah onward. But I assume the standard setting will get to the ‘max’ voltage easier/quicker, but potentially damage the battery… Charger is a Noco G3500. I’ll continue playing around with this charger/battery setup, try a couple of different things, might at some point also try that standard setting.
Update 2023-05-04: today the voltage where the charge lines flattened after 10h was 19.05V; the difference gets smaller. It seems the battery needed to be ‘woken’ up again to get to its max after not being in use for a while. (I did charge it with a cap/pulse charge of TeslaGenX a month ago or so to keep it alive).
I’ll probably do another run tomorrow and then also let the battery rest for longer than an hour to get a better ‘flat line’ value of the battery’s resting voltage.
Post 83 - 2023-05-03 - 08.0x - 230504.0 CM - LA1.pdf
Post 83 - CBA measrements Noco G3500 - LA1 V1.0.pdf

Best regards,
Rodolphe

[JulesP]

Hi Rodolphe,

Good to see your progress on the build. I hope the diagnostics section I presented in the manual was helpful in getting things set up and checked in a straightforward manner.

Regarding your swapper LED, there is not enough power to operate it and start the chip counting so it is probably not going to be a problem. If SW2 is not isolating properly then some leftover charge, stored in C6 (1,000uF) and which, if you look at the attached schematic, feeds the input to SW2, and could be getting through. Disconnect the center wire on the switch and see if that stops it. Or you can try shorting out C6 after switching off the whole circuit and seeing what happens. If not then, as you say, it must be some leakage within the chip, but it is not likely to be of any consequence. When using my cap dump circuit, I often get the power on LED dimly lit for a while as some small power leakage is finding its way through to it when the PSU is on.



From memory, my little cap adjustment inside the voltage divider didn’t do much either but the adjustment on the cable feed to the scope channel did so I was able to get nice square waves using that.

However, if you are going to use the voltage divider then you will need to calibrate it against an ‘impulse’ signal from a signal generator and not a square wave. The 5V impulse signal I used is far more like the spikes you will be measuring. Super accuracy here is not required since the divider is just to give confirmation of the peak HV and which should be similar to the avalanche rating of the device.

There is no reason why you can’t use your 2000V probe. The conventional power in the spikes is small as there is hardly any ‘hot current, a fact demonstrated by the fact that the small relays are passing the HV spikes through to the correct battery with no problem or damage.

J

[pearldragon]

Hi Julian,

Testing/Startup
This weekend I started with testing components and sub-systems; went pretty ok, see image below. Swapper works. One thing that I noted was that when SW2 is OFF (swap on/off), LED3 (swap on) was still lit very faintly… I measured the voltage over R11 (10k), this was 0.12V. (9.6V when SW 2 is in the ON position). So either that switch is not completely disconnecting in the off position, or U1 (CD4060 unit) is leaking from somewhere a bit.
(White tape indications will be replaced with stickers further down the road)


Voltage divider
Furthermore I put a +/-2V 1kHz square wave signal from my scope on the voltage divider and used the probe from the same scope to compensate/tune the radius-ed edge of the square wave output with the trimmer, but to no avail: in the screenshot below you see the signals with the trimmer in both extreme positions, 8.9pF & 57pF. Any idea what might be the cause? Before using the voltage divider I did calibrate the probe itself with the same square wave signal (but without the voltage divider in between).
I disconnected the trimmer to double check if it didn't burn out in the soldering process (very tiny and sensitive thing ), but it was working properly and connected it again.

Another thing I realized is that I do have an additional set of HV probes, with a ratio of 100:1. My question is; could I use on of those probes directly, or would the energy in the spike be too much and damage it? https://eleshop.eu/owon-oscilloscope-probes-2kv.html

Best regards,
Rodolphe

[JulesP]

Hi Rodolphe,

The highest inductance will be with Ferrite rods.

I had reason to contact Devan so he is on the look out for an email from you so try again. He might find your previous ones but write anyway saying what you need.

Tests so far are indicating the best voltage for cap dumps in actually much lower around 20V but tests are not complete. In a few weeks I might have a more complete picture.

J

[pearldragon]

Hi Julian,

I find it 273mH versus 370mH i find still quite a difference. I will machine the rods I have here to size when I have time, but it won't hold me back of continuing finishing the setup and move in the direction of starting the test (subcomponents of the) system.
If anything, having the option of testing with different inductances might be an other interesting variable to test with: dV/dT of the HV spike. With the rods machines to size I will then have 3 options: air coil, welding rods, ferrite rods.

I haven't tried to contact the supplier again yet since currently there are no components I still neede although in saying that; i do not have the STW12N170K5 Mosfet yet... I might shoot them another mail and let you know if I get a response.

Regards,
Rodolphe

[JulesP]

Rodolphe, Your inductance and resistance readings are similar to what I got (14Ohms, 368mH) Of course when the coils are in parallel then the total values will change.

Did you manage to try the quote form for the ‘hard to get’ component place? If all else fails I can suggest another place.

Julian

[pearldragon]

Hi Julian,

Quick update regarding the coils:
-Resistance (individual): +/-14 Ohm
-Inductance (individual) without any core material: 62.5mH
-Inductance (individual) with core material*: 273.5mH

*The Ferrite rods I ordered from Aliexpress are slightly too big, so stacked 1 coil with welding rod material, see picture (material: R45-G / DIN: 8554: G1 / w.nr 1.0324)
Will put the ferrite rods on a lathe and turn them down to fit and remeasure.

Measured with: LC100-A Digital LCD High Precision Inductance capacitance L/C Meter (aliexpress)

Regards,
Rodolphe