Gina's DIY All Sky Cam - Complete Redesign

[Gina]

No joy yet   Beefed up the 0v conncetions and added a 47nF capacitor across the 5v to Gnd on the Nano very close to the device.

I've checked that everything is wired correctly.  What I don't understand is that the cooling circuit was working fine before but now it isn't yet I haven't touched that part of the circuit.

[Gina]

I'm wondering if it would be helpful to move the power MOSFETs away from the Arduino (or not).  Maybe the return current shouldn't go through the same wires as other (Arduino) functions.

[Horwig]

nothing beats a good star earth Gina.

[pete_l]

Try a 100R resistor between Arduino D12 and the MOSFET gate.

Here's why: http://irf.custhelp.com/app/answers/detail/a_id/215

[Gina]

OK - This is literally how the power MOSFETs and their loads are wired up now - ready to test.  The gate resistors are 120R as I didn't have any 100s left.

[Gina]

It's just the same - MOSFET showing 2.7v on my DMM and drawing about half the designed current

Note the Arduino is running off USB power - there is no connection to the 5v power input and just the one wire to the power 0v.

I have used these IRFZ44N N chanel power MOSFETs dozens of time in the past connected directly to the Arduino both for on/off and PWM applications and never had a problem with them before.  In Peltier TEC cooling and dew heating.  Am I finally losing it???

Now where did I put my knitting needles and wool?

[pete_l]

Hmmm, that's a stumper!

What I'd suggest (just to prove you aren't imagining it all) is to pull the gate connection off the Arduino and physically connect it to 0V and +5V alternately and then check the drain voltage.

After that, I'm out of ideas.

[Gina]

I can try that

[Gina]

Right...  Here are the results of that experiment :- 

1. 0v - OFF
2. +5v from PSU - ON fully dew heater drawing 0.7A as designed.
3. 5v (nominal) from Arduino, actually 4.13v, half on - 0.33A.
4. 4.1v from PSU applied to both gate and dew heater - 0.35A.

MOSFET must be oscillating since it doesn't warm up.  If it were half on it would dissipate 2v at 0.35A which is 0.7W and I would have thought that would be enough to warm up the MOSFET (without heatsink) but it stays cold to the touch.

[Gina]

A couple of points from the above. 

1. 4v should be well enough to switch MOSFET on as, according to the spec, over 2v at the gate is an ON state. 
2. The MOSFET isn't drawing any DC current as the state remains when the gate is disconnected due to the intrinsic capacitance.
3. The Arduino Nano is not producing the specified 5v at it's 5v output pin.  I have added the DHT22 - I wonder if that has just tipped the balance.

OK that's three but the second is an expected result.

I could run the Nano off the external 12v as I gather that with the CH340 USB to serial chip the USB port can still be used.  With the other sort it can't.  Unfortunately 5v isn't enough - though I guess I could test wit 6 or 7v (input voltage limits are 6-20v : 7-12v recommended).

[pete_l]

Arduino D12 is also multiplexed with MISO. There's not something else initialising that pin, is there?

[Gina]

I don't think so.  But I can easily use another pin.  OTOH D6 behaves the same vis-a-vis the MOSFET.

[Gina]

I've now connected the Vin on the Nano to 12v and the 5v pin is now reading 5.4v.  The feed to the gates is reading 5.3v.  I have the cooling on pin D6 and the dew heater on pin D11.  Both drains are reading just over a volt when ON.  The TEC is drawing 3.3A and the dew heater 0.6A which is not much different from what it should be but over a volt in the cooling MOSFET with 3.3A means some 3.5W if the MOSFET was working steady-state.  Now a TO220 package uncooled would certainly get hot if it were dissipating 3.5W!  So I conclude the MOSFET must be switching.

Must see if I can find my hand-held oscilloscope

[Gina]

Well...  Done some more experimenting trying to stop any oscillation.  No joy yet

When I've used PWM in the past I've used an LC circuit to cut RFI - the L being a few turns of insulated wire threaded through a ferrite ring and a ceramic capacitor (currently 47nF, value not critical).  No change whatsoever.

Then I tried a ceramic capacitor from gate to ground at the MOSFET - also no change.  Still dropping about a volt.

Had a thought... mostly in the past I have used 12v or so for the power supply rather than 5v.  But I don't see how this could affect the issue as the current drain to source voltage is about a volt.  I have used similar currents in the past so it's not that.

I'm stumped

[Gina]

Thought I'd check out the datasheet for the IRFZ44N again.  It does seem to be the best sort of MOSFET in that it's a HEXFET.  Gate threshold voltage spec is 2-4v which is quite a range but well within the 5v outpout of the Arduino.  Also in Googling I found hundreds of examples of using an Arduino to drive the IRFZ44N for controlling power loads.  With the Arduino running off USB power and around 4v output I can see that some samples of IRFZ44N could be near the limit.  However, if the drive were insufficient the MOSFET wouldn't switch and dropping a volt would cause very noticeable heating, which is simply NOT happening.  Both samples stay cold.

The GS voltage rating is +- 20v so I could use a signal MOSFET or NPN transistor and pull-up resistor to switch the gate at 12v ie. 0v to +12v logic levels.  Now I've moved the IRFZ44Ns off the circuit board there's plenty of room .  Further studying of the datasheet shows that the RDS(on) is specified at 10v GS and drain current of 25A.  Looking at the graphs the drain current limit v Vgs curves start at 4.5v and around 10A so don't go down to the lower currents that I'm using.  Maybe this power MOSFET is better for really high power switching and 12v logic.  But that doesn't explain why so many people are using the IRFZ44N with Arduinos.  It's a puzzle.

[Gina]

More Googling has come up with a lower gate threshold version of the IRFZ44N with slightly higher RDS(on) but still low enough for my application.  It's the IRLZ44N so I've ordered some from ebay to try.

[Gina]

Meanwhile...  I have connected in a pair of 2N7000 signal MOSFETS plus pull-up drain resistors to feed the two gates of the IRFZ44Ns so the latter are running on 12v gate logic.  Drain voltages have dropped to around 0.6v better (but no cigar as they say) - should be millivolts.  Anyway, they are staying cold and I'm running the Arduino on USB power only again so the voltage regulator on that isn't getting hot.  Of course the control logic in the sketch is now inverted but that's no problem except that if the Arduino is disconnected from USB or the laptop shut down while the 12v and 5v PSU supplies are on, both heating and cooling are ON.  I would prefer OFF as the default state.

The low voltage MOSFETS should arrive Thursday or Friday so the inverted logic can be put back to normal then   In normal night-time use the cooling will be on and I would have thermostatic control of the dew heater, but the full on heater power is not excessive and I don't think anything would suffer if left full on.  It's unlikely anyway as I switch the 12/5 PSU off when not imaging. Or testing.

[Gina]

Put it all back together and taken it out to the obsy.  Now back on roof and all connected up and mostly working (except for parts of the sketch I haven't done yet).  The connection from the hotside temperature sensor seems to have come off   Not really surprising with all the wiggling about the circuit baord has had to endure lately   Once I've got everything working properly I'll tidy up the wiring and stick it in place.

[Gina]

A nice AVI file from overnight   Clear sky with lots of stars.  No dome dew and dew heater working well   Cloudy now.

[Gina]

Been out on a trip for most of the day but this evening I brought the unit indoors and reconnected the hotside temperature sensor.  As before cooling is working well and so is the dew heating.  The dew heater when left on heats the dome to a good 10C above ambient.  This should be quite sufficient but I shall be interested in seeing what happens in heavy dew conditions and also rain.  The 12v logic feed to the IRFZ44N power MOSFETs is producing adequate power switching

[Gina]

Overnight AVI was good - no dew   A light cloud passed over for a couple of minutes and a number of aricraft trails.  Here is a screenshot of a single frame from the AVI an hour or two before dawn showing the Milky Way.  Captured with Sharpcap with exposure of 60s gamma set to 2 and contrast to 1.5. 

Two versions - first just clipped and flipped right way up (north at top) in Photoshop, the second enhanced slightly with curves.  Both saved in JPG.

[Stu]

Looking very good now Gina

[Gina]

Thank you

[Gina]

James - I have now used the command line to copy oacapture v 0.5.0 to the appropriate directory using "sudo"   I have it running and I'll try it out shortly on the QHY5.

[Gina]

I now have oacapture v 0.5.0 imaging with the QHY5.