Skip to main content ↑ ↓
Recent Posts
1
Wind Power Machines / Re: Interfacing wind turbines to the grid
Last post by Bruce S -
I look forward to seeing how the programming goes, never got very good at it.
One of my instructors told me my brain was too many steps ahead of what my fingers would do.
I understand the fun part, I still figure out ways to rebuild even newer laptops, it gets kinda ZEN like when looking at how stuff is assembled .

Cheers!!
Bruce S
2
Wind Power Machines / Re: Interfacing wind turbines to the grid
Last post by SparWeb -
I'm getting more and more excited about this project.
My first try is still on the wall in the garage, exporting energy every time it's windy. 
I haven't given it a meter to track how much has been exported. 
For the first try, functional was the winning goal.  Anything is better than zero!

I've designed a new version.  Actually I discarded version 2 and moved on to a version 3 already.  
It will more fully exploit the capability of the micro-inverter that it interfaces to.

Today, the box of parts for V3 came in from Digikey.  I'm really looking forward to this one!
  • Compact relay! 
  • Controlled by a FET! 
  • Controlled by a PIC microcontroller! 
  • Yes, I'm programming in microcode again!
  • Every memory bit on those chips is precious!
Things like this are fun.  To me. Hard to explain to others.  My wife says "uh-huh, okay"
4
Wind Power Machines / Re: Interfacing wind turbines to the grid
Last post by SparWeb -
Watching it functioning for a day, and there are several improvements to be made before a permanent device can be built and trusted to run long-term.

1) Microcontroller is too wimpy to run its built-in relay.  The "Feather" microcontroller is compact, but has a very limited power supply.  Trying to drive a relay (even one designed to be compatible with the Feather) is pushing its ability.  Every fourth power cycle or so, the relay drops out immediately after it pulls in.  The system has to reset and try again.

2) The timing is set up to vary the length of its cycle.  This isn't necessary.  I don't need the potentiometer / adjustment knob.

3) Using a timer to control the time connected is not ideal.  Whatever amount of time is set, there's a situation where the time won't be appropriate.  On a very windy day like this, there is no need to cycle the timer every 20 minutes.  On a gentle wind day, maybe 20 minutes is good.  On a day with no wind, there's still no need for a timer.

4) Orient the control solely to the detection of the diversion load activity.  I've already built this in, and it's a factor in activating the buck controller.  I should exploit this more fully.  Instead of stopping the export with a timer, I can just make the microcontroller monitor the diversion load's activity, and stop the export when the diversion load stops.

5) Plainly obvious cable clean-up!

6) Need a series resistance in the DC cables to the microinverter's MC4 connectors.  I think about 1 Ohm will help the microinverter settle down it's MPPT algorithm into an operational phase.  I see it always stuck in a testing phase where it randomly varies the current.  It's trying to compute a pattern to guide its MPPT program, but with a steadily regulated 48.00v coming out of the buck controller there's no pattern to see.  A 1-Ohm resistor will give it something to lock in on.  Some energy will be wasted as heat, but what's the diversion load doing anyway?
6
Wind Power Machines / Re: Interfacing wind turbines to the grid
Last post by SparWeb -
Success!
My first prototype* did its thing without any magic smoke escaping.

I haven't posted about this for several months because I spent February building the kit and deciding to drastically simplify it compared to the system I described and sketched for you all before.  In late Feb I had parts of it working while I refined the code.  In March the WT started making funny noises and then my charge controller quit. So it's been out of service until this week.  That paused the development of this idea for 3 months. Also, instead of using discrete devices like FETs or SSRs to regulate the battery current, I've chosen to use a buck controller instead.  It got me to the testing phase quickly.

The buck controller converts the 50-60V from the battery into a flat regulated 48.0V.  Feeding a microinverter with that has some limitations (I'll describe later) but for now I'm satisfied that the microinverter won't run away to infinity or go berzerk.  When plugged into 240V AC the microinverter's output went happily to the AC feed in my garage.  I happened to have my car plugged in at the time, meaning that most of the 300W being exported actually went into the car.  :) First time my car was charged from WIND.

The buck converter didn't have any trouble either.  It was windy and as long as wind power was coming in, the battery was at float about 54V feeding the bucker enough that it maintained 48.00V without a twitch.  This sounds great at first but there's a detail that makes this a small problem.  With 6 Amps going through the buck controller it's little fan was running on the heatsink, but its little MOSFETs were below 35C (95F).

Later, while still running the buck converter + microinverter combination, I shut down the wind turbine for a while.  The charge controller stopped, too, and while the microinverter was still running the battery voltage dropped to about 50V.  At that point, the buck converter's output started to sag below 48v.  The microinverter actually responded to this, interestingly.

Microinverters have a start-up procedure that they do every time they start up.  The one I'm using has a 5-minute start where it watches the incoming voltage before it connect any current flow.  It's probably making sure it's not responding to a spurious voltage pulse.  Once current does start to flow, the microinverter varies the current randomly.  It's strange to watch because the current swings from 2 to 6 to 4 to 5 to 3 to 4 to 6 and basically wanders all over the place.  Eventually I figured out that it's building some kind of "performance table" in its memory.  Once it's built the performance table then it knows how the "solar panel" responds to change in current.  This makes sense because solar panels have a varying power curve. Full sun, partial sun, clouds, changing angle during the day, all affect them.  So it seems the microinverter will keep trying to calculate what this performance table looks like until it has one that makes sense and can predict where the MPPT point is.  While it doesn't have a solution to this problem, it keeps shifting current up and down randomly. 

For all the time the buck converter was feeding the microinverter a tightly-regulated 48.00V, the microinverter couldn't do anything with the current that would affect the voltage.  This gave it nothing for the MPPT algorithm to use and it doesn't stop randomly testing the current/voltage changes.  During the time the WT was continuing to charge the battery and it was held at float, the buck converter could maintain its 48.0V, preventing the microinverter to settle into its full power MPPT operation.  Then, when I did shut off the WT, and the battery voltage sagged, this is what allowed the battery voltage, and thus the output of the buck controller to finally start varying with current.  Then the microinverter got the answer it was looking for.  Once that happened, it's output went from fluctuating between 100W to 300W to suddenly rise to a nearly steady 600-630W.

* AKA "box of random bits"
8
Other Electronics / Re: Tristar TS60 Repair attempt
Last post by SparWeb -
Great idea! I can see breaking the DIP sw block away a piece at a time.  Once the big pieces are cleared off the board I can finish the cleaning.  Then the re-soldering will be the mot effective.
9
Other Electronics / Re: Tristar TS60 Repair attempt
Last post by MaryB -
Quote from: SparWeb – on
The little wire shunt didn't work.  Shrug, worth a try, and not much too lose.

Yes with some light and a magnifying glass I can see that under the DIP switch panel there's still a lot more crap under there probably corroding the rest away, maybe other pins are about to go next.  A proper fix is only to de-solder the whole DIP switch block (16 pins) and put a new one on.  That wouldn't be too hard or expensive, so I think I'll try it.  First just for the challenge and second for the chance to restore a devise that has otherwise worked well for many years.

I tried prying the heat-sink off but the FET's are stuck on tightly and I didn't want to break anything.  Nonetheless Morningstar tells me it's possible and provides instructions to do it in a manual for ambitious folks who want to de-solder and replace burned out MOSTFETs.  That doesn't seem to be my problem but if the heat sink can be removed that would help me access and replace the DIP switch panel.

A little 90% isopropyl alcohol dribbled on the transistor tab/heatsink interface in a way that lets it run down down behind the tab will help loosen them. Heat sink compound gets old and dried up and replacing it wouldn't hurt either!

Unless you are skilled in desoldered something on a double sided board I wouldn't try it. Better yo break the dip switch case apart and nibble it down to just pins sticking thru then remove one pin at a time. At the casino we had boards with them that were in the path of drink spills and the boards weren't high quality. So we destroyed the switch and removed a pin at a time then use solder wick(Chem Wik brand for best results!) to clean each hole being very careful how much heat is applied. If you lose a bad don't panic, you can look on the schematic/follow the traces and run a tiny wire between points trying to follow the original trace path. Did a lot of that at the casino when drink spills ate traces off boards. A slot machine CPU board was a $10,000 replacement so we patched until it was impossible to patch any more  :o