ATmega128 old age

Go To Last Post
28 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not only am I getting old, but the ATmega128 in my cat feeder is getting old.  The mega128 and the RFID circuit I built have been running continuously for about 5 years.  I have a push button on the feeder that opens or closes the sliding door when I push it.  The button is connected to ground and a pin in PORTA so pushing the button pulls the pin in PORTA low.  The internal pullup resistor on the pin is enabled.  I had no external pullup resistor or bypass capacitor on that pin.  A couple times recently the door on the feeder suddenly started opening and closing and opening and closing all on it's own as if the push button pin was being held low.  Is it possible the internal pullup resistor could fail intermittently from old age and allow that pin to be pulled low by external noise or whatever?  I replaced the push button a couple of months ago when it did the same thing, thinking it was worn out and needed replacing.  This time I put a 10k pull up resistor on the pin and a 0.1uF bypass cap.  It is very strange.  I am wondering if I have a ghost in the house.  It just started doing it all on its own when nobody was near it, except the ghost.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not likely a "wear-out". That pull-up resistor is really a soft-turn-on FET. It appears as durable as any other FET on the chip.

 

Since you replaced the switch, which would be my first guess, and it  persists, you COULD have EMI problems with some local radio transmitter. Might be police or taxi or ham or garage door opener or whatever. Make sure you have the leads of that bypass cap as short as possible and connected as directly to ground as possible. What kind of enclosure is it in (metal vs plastic)?

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I doubt if there is an wear out inside a chip.
Measure the resistance across the push button with a DMM. Dirt, grime, flux residue can make a difference especially in humid weather.
Alternatively measure the voltage with a DMM across the button when kit is powered.
.
Or your cat knows how to operate the push button.
.
David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If, even with 10K and 0.1uF, the ghost enjoys playing with it in its free time, you may like trying 1K and 10uF to see if it will be scared smiley But, if the random close/open persists, we may say that a very strong ghost visits your environment one a while lately surprise (for instance, did you install something new in your house?)   

Last Edited: Sat. Jul 6, 2019 - 10:03 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A few random thoughts.

1) How are you detecting button press in software eg. are you doing something like read button every 10ms, need 5 consecutive low readings to consider the button pressed, or just polling for the button pin to go low.

2) Maybe problem isn't with the button input. What about the circuit the microcontroller is driving to open/close door. Is that circuitry held in a defined state when microcontroller is not driving it eg. if micrcontroller was resetting for some reason (what's the power supply?)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In my experience the component most likely to fail (esp in a "hot" environment) are electrolytic because the dielectric dries out. I doubt you'd have much worries with resistors unless you pass astronomical current through them.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is the board in a place where it could have slowly collected moisture along with grime, creating a leakage path in the switch circuit?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How is the "board" constructed? Perf-board with wires running around? Full commercially made ECB? DIY etched? Other?

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know it is a hassle to tweak good code that has been working for years, but:

If you have a "spare" LED on the PCB, and a spare I/O pin into which you can add a PB switch, then it is easy to insert a few lines into the project's start up code.

In the start up code, after your usual initializations, but before starting your Main Loop, you read the PB switch.

If it is pressed, then turn off the LED.

If the PB switch was pressed during power up of the device then turn off the LED.

Then run your real project.

 

When you turn the device on you press the PB switch, and the LED won't light up, and the device will run as per usual.

 

If the device is resetting itself, (power source, power supply, errant code, etc.), then it will turn on the LED, (because you were not pressing the PB switch during power up).

 

That gives you a simple, built-in, monitor for startup/reset of the micro, and it can monitor the PCB for days on end.

 

As mentioned above, if the system was working for years, and recently has some odd behavior, EMI would be high on my list, as would a power supply issue, (poor noise regulation, i.e. partial failure...).

 

If your project has several PCBs with interconnecting cables, then remove and re-seat them 50 times to rub off any oxide that has built up on the connectors.

 

Are there any cables on the moving parts?

If so, that that is a big source of failure, and the mechanical stress eventually takes its toll on the wires and their end-point connections.

 

With the moving cover there will be vibration in the box, so again check for any loose / worn connectors.

 

Any back-emf diodes on motors or solenoids or servos?

They are another common failure item, and then allow intrinsically generated EMF back into the system, (i.e. not the Ham next door's fault!).

 

When you built it, did you have several PCB's made?

Perhaps its time to build a new board with new parts?

 

JC

 

Edit: Typo

 

Last Edited: Sat. Jul 6, 2019 - 06:05 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
Make sure you have the leads of that bypass cap as short as possible and connected as directly to ground as possible. What kind of enclosure is it in (metal vs plastic)?

I put the bypass cap on the perfboard containing the RFID circuit where two long wires from the button connect to a 2x5 10 pin header.  From there goes a 10 wire ribbon cable about 4" long to the microcontroller board and plugs into Port A, so I am not as close to the microcontroller as possible.  I know it is not ideal, but it is the best I could do.  The microcontroller board is something I got from Futurlec a long time ago and they still sell:   http://www.futurlec.com/ATMEGA_Controller.shtml    The enclosure is styrene.

 

david.prentice wrote:
Alternatively measure the voltage with a DMM across the button when kit is powered.

The button has 5V across it.  The 5V comes from the power pin of Port A on the Futurlec board, through 4" of ribbon cable to the perfboard RFID board, and then 8 or 10 inches of a pair of 24g wires to the button.  Does that make sense?  I should probably twist the pair of wires going to the button.

 

MrKendo wrote:

A few random thoughts.

1) How are you detecting button press in software eg. are you doing something like read button every 10ms, need 5 consecutive low readings to consider the button pressed, or just polling for the button pin to go low.

2) Maybe problem isn't with the button input. What about the circuit the microcontroller is driving to open/close door. Is that circuitry held in a defined state when microcontroller is not driving it eg. if micrcontroller was resetting for some reason (what's the power supply?)

I am just continuously polling the Port A pin to go low.  Checking for 5 consecutive lows is a good idea and something I should always do.  Thanks MrKendo  The microcontroller is continuously polling the RFID circuit for data to appear.  When it sees something on the RFID data pin it starts looking for the "101010101010101010" header transmitted by the fob on the cat's collar and then proceeds to decode the Manchester code for the fob number, and if it is a good number opens the door.  The door is held open for 5 seconds while continuing to detect the fob.  If there is no fob detected for 5 seconds the door closes.  This leads me to believe if it is something wonky in the RFID circuit tripping the door to open, it wouldn't open and close continuously like it does because of the 5 second wait time.  Which brings me back to the button circuit.  The power supply is a wall wart.  I thought about that.  The power to the button is coming from the microcontroller board so if power is interrupted to bring the button low it also takes power away from the microcontroller too.  If the wall wart starts oscillating it might do what it is doing.  The stepper motor is also powered by the same 12V wall wart and driven from Port C.

 

kk6gm wrote:
Is the board in a place where it could have slowly collected moisture along with grime, creating a leakage path in the switch circuit?

It is in the kitchen, so yes.

 

ka7ehk wrote:
How is the "board" constructed? Perf-board with wires running around? Full commercially made ECB? DIY etched? Other?

The microcontroller board is commercial, and the RFID board is perfboard.  I ran the wires from the button to the RFID board 2x5 header so I could use the same ribbon cable from the RFID board to the Futurlec microcontroller board for the button detect.

 

I see I just got a post from Doc.  I will respond to that after I feed the cats.

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What happens when you plug in/unplug power (or reset) ? Lets say you have it set to open, then close on power up (to get to a known state, whatever). If the micro is resetting due to brown out, power supply flaky, voltage regulator bad, etc., then you may be seeing resets. If you do nothing with the door on reset, then disregard.

 

Or maybe the Y2K bug is about 20 years late on this one.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarkThomas wrote:

kk6gm wrote:

Is the board in a place where it could have slowly collected moisture along with grime, creating a leakage path in the switch circuit?

 

It is in the kitchen, so yes.

Maybe just give the board(s) a good scrub with alcohol and blow-dry it/them.

 

It sounds like you have two triggers, a button and an RFID.  Having 2 LEDs (or a bicolor LED) to tell you which trigger happened last, or which happened in the past 24 hours, might be useful.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:
I know it is a hassle to tweak good code that has been working for years, but:

The real problem was finding the code.  It was in the OneDrive folder, which was called SkyDrive when I was using it when it was new.  It turned out to be more trouble than it was worth, and now I archive/back up on Dropbox.

 

DocJC wrote:
As mentioned above, if the system was working for years, and recently has some odd behavior, EMI would be high on my list, as would a power supply issue, (poor noise regulation, i.e. partial failure...).

Yes, that is what I am thinking too.  That the button pin is held high by the microcontroller Port A input pin internal pull up resistor enabled, and then pulled to ground with a button push it is hard for me to imagine that a microcontroller reset would cause the door to open and close continuously for as long as I let it until I unplug the wall wart.

 

DocJC wrote:
If your project has several PCBs with interconnecting cables, then remove and re-seat them 50 times to rub off any oxide that has built up on the connectors.

Maybe that is why it worked for several months after the last time this happened and I replaced the button.  But I am having a hard time imagining how the Port A button pin which is held High by the internal pull up resistor could suddenly go low because of oxidation.  I would think if the pin is oxidized, pushing the button would have no effect and not pull the pin to ground.

 

 

 

DocJC wrote:
I know it is a hassle to tweak good code that has been working for years, but:

The real problem was finding the code.  It was in the OneDrive folder, which was called SkyDrive when I was using it when it was new.  It turned out to be more trouble than it was worth, and now I archive/back up on Dropbox.

 

DocJC wrote:
As mentioned above, if the system was working for years, and recently has some odd behavior, EMI would be high on my list, as would a power supply issue, (poor noise regulation, i.e. partial failure...).

Yes, that is what I am thinking too.  That the button pin is held high by the microcontroller Port A input pin internal pull up resistor enabled (not as I said above from the Port A power pin.  The RFID circuit is powered by the Port A power pin, and the push button is powered by a Port A pin with internal pull up resistor enabled) , and then pulled to ground with a button push it is hard for me to imagine that a microcontroller reset would cause the door to open and close continuously for as long as I let it until I unplug the wall wart.

 

DocJC wrote:
Are there any cables on the moving parts?

Nope.

 

DocJC wrote:
With the moving cover there will be vibration in the box, so again check for any loose / worn connectors.

When I looked at the wiring on the home made RFID board I sort of shuddered.  It may be that one of the RFID rat's nest is shorting out all power to Port A and causing the pin to go low.  But it starts up when nobody is near it, and suddenly the door starts opening and closing on an endless cycle.

 

DocJC wrote:

Any back-emf diodes on motors or solenoids or servos?

They are another common failure item, and then allow intrinsically generated EMF back into the system, (i.e. not the Ham next door's fault!).

 

When you built it, did you have several PCB's made?

Perhaps its time to build a new board with new parts?

I do have back emf diodes on the stepper motor.  The microcontroller board is commercial, but I bought 5 at the time and have 3 left and they are all 5 years old.  3 of them have been powered up.  When I was doing RFID, I built 5 RFID circuits on perfboard using an EM4095 chip and a bunch of silver mica caps.  So I could replace the RFID board.  I think I have 3 left, and some of them even have the pin number reversed on the header with the red stripe on the ribbon on the left side instead of the right.  Remember, this project was made early on in my career in this hobby, so I am finding flakey stuff.  But the thing is, it starts up doing this with the nearest animal at least 3 meters away.

 

curtvm wrote:
If you do nothing with the door on reset, then disregard.

No, I do nothing on reset.  The code just starts up at the beginning.  Inits everything, starts the infinite while() loop, checks for button push and then starts checking for RFID coming in on the Port A data pin.

 

kk6gm wrote:

Maybe just give the board(s) a good scrub with alcohol and blow-dry it/them.

 

It sounds like you have two triggers, a button and an RFID.  Having 2 LEDs (or a bicolor LED) to tell you which trigger happened last, or which happened in the past 24 hours, might be useful.

It got a good cleaning the first time i replaced the button.  There was lots of interesting stuff in there, besides cat food and cat hair.  I cleaned it all out.  It lasted about 6 months.  It is almost always the case the last trigger was from RFID while Simone ate.  I only use the button at the next meal to open the feeder and get the dirty bowl out and put in one with food in it.

 

Thanks all for ideas and suggestions.  I appreciate it.    Now I will wait for it to happen again.  Being intermittent makes it hard to diagnose.

 

david.prentice wrote:
Or your cat knows how to operate the push button.

I worry about that.  Otis tries putting his head in the same place Simone does and moves it around like Simone does to get it to open, much to his demise.  One of these days he is going to figure out the button on the side.  Once opened by the button, it wont close unless with another button push, or an RFID read, which closes the door after a 5 second delay from the last read.

 

Thanks all.  I think it is a pretty interesting problem, and revisiting my 5 year old code and perfboard design shows my code has not degraded too much, but my perfboard building has definitely gotten better.

 

Now I have to go help socialize a 3 month old feral kitten, which is sort of old for it, and I expect to come home with scratches and bites, hopefully none deep enough to send me to the ER for antibiotics.  I'm a little weird because I like it when they scratch me and blood come out.  For some reason I like seeing my blood leaking out.

 

more later,

mark 

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarkThomas wrote:
It is very strange.  I am wondering if I have a ghost in the house.  It just started doing it all on its own when nobody was near it, except the ghost.
or the "Ghost in the Machine"

The interplanetary magnetic field (IPMF) event of 29-Jun was amazing though similar effect has occurred before (but IIRC, there was no CME impact 29-Jun'19); 29-Jun'19 was felt with reverb though IIRC 3-Jul'19.

Might be as simple as Earth flew through Jupiter's IPMF (each planet has an EM circuit with the Sun) then Earth's magnetic field bounced around for a while.

How geo-effective such had on San Francisco and San Francisco's current loops with its AC generators will vary.

There's a research project that has a prototype EM 3D (latitude, longitude, time) graph of CONUS that's amazing to watch (I don't have the URL)

Some geo-effective EM events will damage sub-stations.

As Curt mentioned, check the mega128's power supply; re-review the power supply's design (bulk cap hold-up long enough, bulk cap ageing, measure the power supply's gain and phase margins and noise, etc)

There are "large" capacitors for circuits and for whole house (mounted onto the circuit breaker box) to hold-over brief glitches.

 

https://iswa.ccmc.gsfc.nasa.gov/IswaSystemWebApp/StreamByDataIdServlet?allDataId=1649536385

USA NOAA might also have some data on magnetic.

Delta Lightning Arrestors | Surge Capacitor Information

 

P.S.

relative to electrical circuits, "The Mangler"

relative to electromagnetic, "Maximum Overdrive (1986)"

Top 10 Horror Movies With Killer Machines… | AwesomeBMovies.com

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

gchapman wrote:
As Curt mentioned, check the mega128's power supply; re-review the power supply's design (bulk cap hold-up long enough, bulk cap ageing, measure the power supply's gain and phase margins and noise, etc)

I think I will take a look at the power supply wall wart, although I dont understand exactly how that would have the effect I am seeing since the power for the button is coming from the microcontroller.  I have some more 12 V power supplies.  Maybe I will be proactive and change them out.  That is probably the most likely.  I think all this started happening after we put in a surge protector for the refrigerator some months ago and the feeder wall wart is plugged into it. 

 

It keeps life interesting.  Feeding the cats without that feeder will be a major hassle keeping them from eating each other's food, so I really want to keep it working.  It freaks my wife out when it starts it's open-close-open-open...routine for no apparent reason and with nobody around.  My next door neighbor works for Cisco, and he is running some pretty high power enterprise WiFi equipment that sometimes radically interferes with ours.  Maybe it is that interacting with the solar wind 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MarkThomas wrote:
I think I will take a look at the power supply wall wart, although I dont understand exactly how that would have the effect I am seeing since the power for the button is coming from the microcontroller.
Wall warts are low risk though can over-voltage; XMEGA challenge wall warts.

MarkThomas wrote:
I think all this started happening after we put in a surge protector for the refrigerator some months ago and the feeder wall wart is plugged into it.
Notebook PC would run but its battery charger was non-functional; cause - very leaky MOV in the surge suppressior in the outlet strip.

There are SOV for a circuit or a package of SOV for whole house.

MarkThomas wrote:
My next door neighbor works for Cisco, and he is running some pretty high power enterprise WiFi equipment that sometimes radically interferes with ours.
Wi-Fi is approx 100mW ERP though antenna gain is a factor.

More likely is cellular as that's an order of magnitude greater ERP.

MarkThomas wrote:
Maybe it is that interacting with the solar wind 
laugh

 


https://www.avrfreaks.net/forum/xmega-sram-slow-turnaround-solved-glitchy-power-supply#comment-1028546

Delta Lightning Arrestors | Modern Surge Protectors

Technical Tidbit - August 2007

Mobile Phone Induced EMI

by Douglas C. Smith

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

gchapman wrote:

Wall warts are low risk though can over-voltage

 

I'll say!  I have a fairly large collection of wall warts

saved from equipment which is no longer in use.

Among these were three identical supplies labeled

9V DC.  However, when I measured with a meter

they were all putting out more than 20 volts. So I

did the only sensible thing and saved their cords and

chucked the rest.

 

--Mike

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I got a 7.5V wall wart that came with a BK Precision signal generator to run it off battery that was putting out something like 12 or 15 V.  I would have expected better from BK.  I put it in the box of 12V wall warts.  I think some of them change voltage under load, but my experience has been not by much.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

PolyZen are one way to protect loads from such wall wart defects.

9V wall wart holds are 1.3A and 2.3A.

PolyZen Devices for Overvoltage-Overcurrent Protection - Littelfuse

 

TND6093 - Low VCE(sat) BJT's in Automotive Applications

(page 5)

Over Voltage Protection

avr-mike wrote:

So I

did the only sensible thing and saved their cords and

chucked the rest.

There are relatively generic series of SMPS transformers though "might" consider salvaging those if can get through the defective wall wart's case/encapsulation.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Possibly more likely is that the nylon leisure suits you were wearing accumulated a 20KV charge & gave the AVR quite a shock each time you hit the button---over time  causing internal damage.  Is your switch pin protected with a transorb or similar scheme?

 

Could also be line transients, if plugged in....neighbor had a lot of electronics in strange-permanent zombie mode after nearby lightning.

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Go on.   Wall-warts of the heavy transformer variety are just transformer + bridge rectifier + reservoir capacitor.

 

I am fairly certain that the "over-voltage" comments come from readers of mature years.    I bet that you have all dabbled in electronics since teenage.

 

Yes.    If you have a 12V 2A wall-wart and you only draw 20mA it will have a high voltage.

If you draw 2000mA I bet that you will get a typical 12V.

 

If you can find a 12V 20mA wall-wart you will probably get 12V out of it.

 

SMPS wall-warts are a different matter.    They will probably adjust to the load current. 

 

I still suspect dirt and grime on the switch.    But a 10k external pullup would be an improvement on the internal (30k) pullups.

As others have suggested.     Make a record of any RESET activity.   e.g. Brown-Out,  Watchdog, Power-Up, ...

This could record in EEPROM or just light an LED.

 

My friends have cats.    Cat feeders that open at set times mean that you can go away for a day or two.

Obviously they will never be as sophisticated as Cow feeders.    But they are only cats after all.

 

David.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As David point out, there are (generally) two kinds of wall warts. Transformer kind and switch-mode kind.

 

Transformer kind will give you very close to the rated voltage AT the RATED LOAD. It will be higher at lighter loads. Think of a series resistor (though it is really transformer resistance). To get rated voltage at rated load, the open circuit (e.g. unloaded) voltage has to be higher. Transformer kind will have ripple at 2x or 1x the line frequency but  has galvanic isolation between line and load.

 

Switch-mode kind has internal feedback regulation and usually stays pretty close to the rated voltage over a pretty wide range of load currents. Tradeoff: high frequency noise but low line ripple, and sometimes mediocre isolation. 

 

There ARE specialized wall warts that don't quite fit either of the "rules" listed above (high voltage chargers for "quartz" photographic flashes, for example). Phone chargers with USB output tend to be more than a bit noisy and often very poor isolation (because it is expected that there are no user accessible conducting surfaces connected to the internal circuitry). In the latter case, you often get what you pay for.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In my defense....  ;-)

 

Initially I trusted the label and used one of the "9V"

regulators to power a circuit which had a 7805 to

generate the required 5V.  This regulator got really

hot!  Nearly burned my fingers.

 

--Mike

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just do the sums. I would expect a 9V wall wart to have about 14V at "low load".
A 7805 is plain stupid. ( 14V - 5V ) × 100mA = 900mW
Modern LDO regulators will do the same job but with much less dissipation.
Of course motors, relays, ... will take much more than 100mA.
.
David.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It was even dumber at 23V !!

 

--Mike

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

david.prentice wrote:
I bet that you have all dabbled in electronics since teenage.

 

Oh, LONG before teenage!

 

These defective walwarts are probably fine. Put a small load on it and you'll see its voltage fall within range. They're not regulated, the device they operated had any regulation it needed inside it.

If you don't know my whole story, keep your mouth shut.

If you know my whole story, you're an accomplice. Keep your mouth shut. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LDOs DO NOT have lower dissipation when operated under the same conditions as a standard linear. 100ma load x 9V (e.g. Vin - Vout) is still 900mW, whether standard or LDO. Where the LDO wins is the ABILITY to operate with MUCH lower input-output voltage drop (sometimes less than 100mV). LDO will regulate 5V down to 3.3V easily; standard will not. The dissipated power is still the load current (plus quiescent) times input-output voltage drop, no matter what kind of LINEAR regulator it is.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The transformers use the cheapest and fewest windings possible---leading to high ohms & wide voltage swings under loading.  A beefy 3 pound transformer would be much more steady, but much more $$$.

They make 'em cheap & in bulk...i saw some surplus ad 20 years ago for 5 tractor trailer loads of scrap units piled high...est quantity 2 million ..must take all.

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!