#1  
Old 12-04-2017, 09:55 AM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Cooled Nikon D7000 anybody?

As a Nikon shooter it's hard not to feel a little like a red headed step child when you see just how much info and support there is out there for Canon, Sony and even Pentax cameras when it comes to sensor cooling and all things Astro... But don't let that put you off

So my trusty old D7000 recently started to get a lot more use than my D810 after I learnt about removing the IR filter and nickonhacker's star eater fix, true dark current and a few other things.

Any way long story short, there is absolutely zero info on the entire internet about Peltier cold finger cooling on a Nikon D7000, despite them being quite popular (as far as Nikon goes) for Astro use along with the D5100. I've been trawling through forums and googling the subject pretty hard, and all I could come up with was two examples of a D5100 (which uses the same sensor in a somewhat different layout) with cold finger cooling, but the info was a little sparse, hardly the detailed laboratory testing case studies I've seen for other cameras.

Many said there just wasn't enough room between the motherboard and the CMOS heatsink. And there isn't, but don't let that stop you.

I bought a 150x150mm sheet of 1.2mm copper, which in hindsight is slightly too thick, 0.5-0.75mm would probably have made it a lot easier. Version 1 I trimmed a little metal off the tripod base plate to bring the copper in through the bottom and onto the front side of the sensor as I thought there was no way it could fit behind it on the heatsink side.

I mistakenly thought the sensor was isolated from the aluminium frame by epoxy "rivets" as I could see a very thin air gap between the sensor and the pcb. I had removed both IR/UV filters so there was plenty of room around the front of the sensor and the gap between it and the housing that held the dust piezo looked like it would fit the 1.2mm copper perfectly.

Initial testing got the sensor down to 5c with no load, though it would shoot up pretty quick if I turned on live view. Somewhere along the line I must have cracked the ceramic plate on my eBay sourced tec1-12706 cooler, possibly by over tightening the heatsink which wasn't quite flat across the Peltier.

I then found a picture of a D5100 sensor that had been removed from the aluminium frame and realised the entire back side of it is copper which is obscured by the pcb, and that it is attached with epoxy rivets but the aluminium frame is pressed hard against it, thus making the aluminium heatsink frame the ideal spot to attach the cold finger. A lot more ideal than cooling the sensor through the glass in front of the sensor itself. Derr.

Version 2 I've gone in through the tripod plate again but this time between the sensor heatsink and the motherboard. I very (very very f'ing!) carefully ground down all the raised bumps on the back of the sensor heatsink to make room for the cold finger as well as provide a nice smooth flat surface for the two to aid heat transfer. Talk about hairy, I've got a few more grey hairs after that. It took about 30 minutes of carefully removing a little material at a time and dabbing it with a moist rag, I didn't want to cook the sensor.

To attach the copper to the aluminium frame I drilled and tapped the screw holes for the dust piezo bracket out to M3 and used some small counter sunk screws along with a very careful application of Arctic Silver 5 heatsink compound. Apparently it's non conductive but I'm avoiding contact with the pcb or any pins or components.

There isn't much on the side of the motherboard that faces the sensor, just a few tiny surface mount components and some chips just off to the side to make life hard, and two ribbon cables right where I'm trying to run the copper. Could have done better here, but I think there's a fairly decent cross section of copper coming out of the body.

I put a layer of insulation tape on the side of the motherboard facing the sensor, and cut a piece of 2mm foam sheet to go inbetween the copper and motherboard, and stuffed foam into any air gaps I could see for electrical insulation and to avoid condensation, though I have a bottle of argon gas to be extra sure.

Without the foam it only just fits in, but with it the motherboard has to flex just a tiny bit to be screwed down to its original position, but I'm happy enough with that.

I picked up a 68w Peltier module from Jaycar to replace the broken one. Just gotta mount that, the heatsink and work out where I can fit the temperature sensor from my temp controller before I kick her in the guts, but I'm quietly confident it'll work better than V1.

So far the camera still works as you see it, fingers crossed. More to come.

Full res pics here: https://goo.gl/photos/doQoFbUzSiKHuVS79

PS sorry for the absurdly long post
Attached Thumbnails
Click for full-size image (IMG_20170412_004300-800x600.jpg)
167.0 KB68 views
Click for full-size image (IMG_20170412_000618-01-800x600.jpeg)
154.1 KB60 views
Click for full-size image (IMG_20170412_005731-800x600.jpg)
150.6 KB60 views
Click for full-size image (IMG_20170412_005810-01-800x600.jpeg)
122.1 KB66 views
Reply With Quote
  #2  
Old 12-04-2017, 12:39 PM
sil's Avatar
sil (Steve)
Not even a speck of dust

sil is offline
 
Join Date: Jun 2012
Location: Canberra
Posts: 1,395
interesting, I never looked up cold fingers (not up to dismantling a dslr). But now I have a spare peltier and been eyeing off my old D60, thinking of resurrecting for white light solar. My newly peltiered zwo hopefully gets first light this weekend, clouds permitting. Newly reinstalled laptop, latest drivers and capture software and camera & peltier all tested and ready for action.
Reply With Quote
  #3  
Old 12-04-2017, 01:33 PM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
The cooling bravery award goes to you!

Last edited by rcheshire; 12-04-2017 at 01:48 PM.
Reply With Quote
  #4  
Old 12-04-2017, 03:37 PM
strongmanmike's Avatar
strongmanmike (Michael)
Woohoo it's clear

strongmanmike is offline
 
Join Date: May 2006
Location: Canberra
Posts: 14,714
Quote:
Originally Posted by NinetySix View Post
So far the camera still works as you see it, fingers crossed. More to come.
Full res pics here: https://goo.gl/photos/doQoFbUzSiKHuVS79

PS sorry for the absurdly long post
Not at all, really enjoyed the idea, description and photos, actually sounds like a fun (even if it involved a little trepidation) project, look forward to seeing what you can capture with it

Made me reminisce about a couple of cold camera's my mate and I constructed and used way back in 1983/84 using film....same idea, just a mechanical+chemical version ...to take spectacular (not) photos like these:

NGC 4945

Cen A

....hopefully your modern digital results will be a little better

Mike
Reply With Quote
  #5  
Old 12-04-2017, 09:39 PM
LightningNZ's Avatar
LightningNZ (Cam)
Registered User

LightningNZ is offline
 
Join Date: Oct 2011
Location: Canberra
Posts: 952
As someone with a Nikon D5100 I'm keen to see you post all day long
Cheers,
Cam
Reply With Quote
  #6  
Old 13-04-2017, 07:00 AM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Ok quick update, got it all together last night and gave it a quick bench test. Initially disappointed that it struggled to get much colder than 5c, but I realised the heatsink was getting pretty bloody hot. The fan I started with was quite weak, a little 80mm no name pc case fan that really didn't move much air.

Rustled up a couple year old Intel CPU fan and it's moving a lot more air and the heatsink isn't anywhere near as hot. It got down to 5c again fairly quickly then within about 15 minutes it hit 0c, and very slowly continued to drop from there. I had to pull the pin and go to bed but after about 25-30 minutes it was on -1.8c.

I'm not sure if that's good or not, considering the camera was off. It's probably not going to break any cooling records, but it might well be the coldest d7000 ever described on the internet

I could still improve a few things such as the insulation between the cold and hot side. Also, the heatsink I'm using is far from ideal as it has very thick fins, just something I had laying around. It definitely performs better upside down being that the cold finger comes out the bottom of the camera. Hot up, cold down.

I'll see what more I can squeeze out of it in the coming days, -5c while imaging with the screen off would be nice but we'll just have to see how much low hanging fruit there is with my setup.


Cheers
Attached Thumbnails
Click for full-size image (IMG_20170413_011757-600x800.jpg)
145.2 KB31 views
Reply With Quote
  #7  
Old 13-04-2017, 08:58 AM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Oh and I forgot to mention, ambient room temp was around 22c during testing. When I shut it down to adjust something the temp controller was indicating around 29c once powered back up.

I suppose the delta T isn't too shabby. Earlier I ran the camera for about 10 minutes with live view on and it went from 17c (ambient also 17c) to 27c.

Lastly, I just had one of these donated to my cause:

http://icrontic.com/images/draco/art...u9b_front1.jpg

Noctua CPU cooler. Heat pipes and tiny fins, what's not to like? It takes up about 4 times the volume of the one I've got on there now, yet feels a little lighter
Reply With Quote
  #8  
Old 13-04-2017, 09:23 AM
ZeroID's Avatar
ZeroID (Brent)
Lost in Space ....

ZeroID is offline
 
Join Date: May 2010
Location: Auckland, NZ
Posts: 4,950
Hi Ian,

You will have noticed a few of us were doing peltier cooled sensors with Canons a while back. If you can get down to about 5* above zero you will have low noise and minimal dew\ice problems.

I could get my 450D down to about 15* below but keeping it dew\ice free was a major headache. Sealed Argon Bags and insulation seals were required. But 5* above made enough difference to be useful

Congratulations on joining the DIY nutters group with your modifications though, it only gets worse from here ..
Reply With Quote
  #9  
Old 13-04-2017, 10:16 AM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
That's a good result Ian. Not too shabby at all. You may run into condensation or ice on the camera sensor below 5C. More than likely 0C and below.

The most significant advantage of cooling is temperature consistency, if you do need darks.

Darks are almost redundant below 5C. At 0 and below, they may add noise to your images.

I find (light - bias)/(flat - bias) is ideal for preprocessing cooled DSLR data 0 and below.

EDIT: Looking at the size of your fan, I would look for a more efficient heatsink / fan combo. The heatsink doesn't need to be big, just an effient design. The fan is off an old CPU Phenom cooler.

EDIT: Something like this will keep the sensor face free of H2O down to -10C. 5V. Adds to the heat losd by a few degrees. Most imaging requires only 60 - 80% of total performance for these systems... seldom need the full differential, which takes lots of time and energy.
Attached Thumbnails
Click for full-size image (20150925_203333.jpg)
206.6 KB29 views
Click for full-size image (20150502_152729-1-1.jpg)
195.9 KB31 views

Last edited by rcheshire; 13-04-2017 at 10:45 AM.
Reply With Quote
  #10  
Old 13-04-2017, 04:41 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Ok take 3, massive improvement with a decent heatsink! Ambient temperature 24c, got to -5.1 in about 5 minutes and it was still heading south before I realised I forgot to fill her with argon, hopefully it won't get too wet the heatsink never even felt warm.

With a little luck I might get a few subs in tonight, or at the very least some dark frames to see what difference it makes.

Obviously the heatsink is mounted very temporarily, but I didn't have to drill anything this way. I think the general theory is proven now though, so I think water cooling would put the least weight on the focuser, just a pair of hoses that can be tied to the tube rings.


Cheers
Attached Thumbnails
Click for full-size image (IMG_20170413_172707.jpg)
189.9 KB26 views
Reply With Quote
  #11  
Old 13-04-2017, 04:50 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Quote:
Originally Posted by ZeroID View Post
Hi Ian,

You will have noticed a few of us were doing peltier cooled sensors with Canons a while back. If you can get down to about 5* above zero you will have low noise and minimal dew\ice problems.

I could get my 450D down to about 15* below but keeping it dew\ice free was a major headache. Sealed Argon Bags and insulation seals were required. But 5* above made enough difference to be useful

Congratulations on joining the DIY nutters group with your modifications though, it only gets worse from here ..
Thanks mate

I have a habit of over doing things, but once I'm done patting myself on the back I find a happy middle ground which probably is around 5c.

I've bought the bottle of argon and a reg so I'll be using it, I was thinking of a plastic bag as I've put quite a few holes in the case now
Reply With Quote
  #12  
Old 13-04-2017, 04:57 PM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
No hard and fast rules. Argon is best. I find the electric solution a bit more convenient. Good stuff.
Reply With Quote
  #13  
Old 13-04-2017, 04:59 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Quote:
Originally Posted by rcheshire View Post
That's a good result Ian. Not too shabby at all. You may run into condensation or ice on the camera sensor below 5C. More than likely 0C and below.

The most significant advantage of cooling is temperature consistency, if you do need darks.

Darks are almost redundant below 5C. At 0 and below, they may add noise to your images.

I find (light - bias)/(flat - bias) is ideal for preprocessing cooled DSLR data 0 and below.

EDIT: Looking at the size of your fan, I would look for a more efficient heatsink / fan combo. The heatsink doesn't need to be big, just an effient design. The fan is off an old CPU Phenom cooler.

EDIT: Something like this will keep the sensor face free of H2O down to -10C. 5V. Adds to the heat losd by a few degrees. Most imaging requires only 60 - 80% of total performance for these systems... seldom need the full differential, which takes lots of time and energy.
Cheers mate

Yes I think the test today showed what a good efficient heatsink can do, smaller would definitely be good but so far I've been working almost entirely with salvaged, recycled and donated parts

I also need to sort out better temperature control, I'm using a $2.50 w1209 thermostat and it won't hold to set point within a degree.


I'm hoping to never open the camera again, the Phillips head are looking like hex sockets I had planned for the argon to take care of internal icing, but if that doesn't work out I might put a few wraps of kanthal wire from my ecig around the sensor, that's how I've been making dew heaters.


Cheers
Reply With Quote
  #14  
Old 13-04-2017, 05:22 PM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
Just seal it up in a plastic bag. It doesn't need to look pretty.

Temperature regulation is not overly critical within a degree or so. It took me years to come up with an accurate and repeatable system. Arduino is one route. The other is one of the eBay offerings.

The simplest is linear. A voltage vs temperature differential table is helpful. Then set voltage for a differential based on air temp and required setpoint.
Reply With Quote
  #15  
Old 13-04-2017, 05:38 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Temperature regulation is how I pay the bills, got a few controllers in the cupboard but they run on 24ac, hoping to keep everything 12vdc should I ever take the show on the road. Plus my hardware can only switch 24ac, 0-10V or 4-20mA.

I do like having an all in one standalone box, not sure if such a thing exists, I'll probably leave the thermostat on see if I can't tune it a bit tighter. I've read thermostat switching is no good for Peltier coolers, but I'm happy to learn the hard way with the price of Peltiers on eBay but maybe not Jaycar
Reply With Quote
  #16  
Old 13-04-2017, 08:50 PM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
I started out with a thermostat style system. Unfortunately, the response times are too slow. Temp wanders around too much and the junctions break with continual expansion and contraction.

Driving PWM up and down in response to temperature is just as bad.

A better strategy is to compute the voltage requirement for the setpoint and either use an adjustable linear supply or digital switching (PWM, preferably not) to maintain voltage. As little as 0.1V makes a difference.

The Arduino sketch I use computes a proportional voltage to get near setpoint and then the integral continually recomputes the voltage requirement for setpoint. Takes account of variations in heat load. PID minus the derivative as far as I know.

Once the system finds the correct voltage the temperature readout flicks to and fro, +/- the temp sensor resolution, 0.3C around setpoint. Close enough. Still working on improvements. It competes easily with commercial cameras for temperature control.

Don't chase the temperature, change the voltage incrementally.

Have fun. Always good to get fresh ideas.

Last edited by rcheshire; 14-04-2017 at 11:32 AM.
Reply With Quote
  #17  
Old 14-04-2017, 12:20 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Ok shot some dark frames last night, shot 8 minutes at iso 100 with the thermostat set to -5c on the left and on the right no cooling with the sensor showing around 22.5c with room temp around 19.5c.

I battled the clouds last night and lost, and I also lost against condensation so I like the sensor heating idea
Attached Thumbnails
Click for full-size image (noise comparison-4.jpg)
190.5 KB34 views
Reply With Quote
  #18  
Old 14-04-2017, 03:22 PM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
Quote:
Originally Posted by rcheshire View Post
I started out with a thermostat style system. Unfortunately, the response times are too slow. Temp wanders around too much and the junctions break with continual expansion and contraction.

Driving PWM up and down in response to temperature is just as bad.

A better strategy is to compute the voltage requirement for the setpoint and either use an adjustable linear supply or digital switching (PWM, preferably not) to maintain voltage. As little as 0.1V makes a difference.

The Arduino sketch I use computes a proportional voltage to get near setpoint and then the integral continually recomputes the voltage requirement for setpoint. Takes account of variations in heat load. PID minus the derivative as far as I know.

Once the system finds the correct voltage the temperature readout flicks to and fro, +/- the temp sensor resolution, 0.3C around setpoint. Close enough. Still working on improvements. It competes easily with commercial cameras for temperature control.

Don't chase the temperature, change the voltage incrementally.

Have fun. Always good to get fresh ideas.
Last night I found that the thermostat actually didn't do too bad of a job. Set point plus about 1.5c, temp comes up really quick as soon as the cooling cuts out but it was cycling off for about 3 seconds every 45-60 seconds or so.

The hardware I have access to uses PRAC which does all the PID tuning for you, although you can manually tune it but I've always found that it nails it everytime if your sensor is installed & located correctly and there's nothing funny going on with your control output.

Water cooling for the peltier hot side is definitely looking good now though, the weight saving being a big advantage and also important is reducing the occurrence of me sticking my fingers into a spinning fan in the dark.

With water cooling I reckon I could probably get nice smooth temperature regulation by controlling the flow of water through the radiator, either a variable speed pump or a bypass valve. Sure it won't be the most efficient way, and that might hurt a bit if I'm running off battery power at a dark site but it could be manageable. Maybe that in conjunction with a manual variable voltage supply.


Is that picture above from your setup? Looks like 21 ohms of resistors on the sensor so a bit over a watt with 5v is it? I should be able to swing that using kanthal and some thin foam tape.


The Dark frame comparison shot above I processed the -5 degree photo to look about right then boosted it 5 stops, then applied the exact same settings to the non cooled shot. Very keen to see what difference that will make with the lens cap off, though I'll have to perfect the argon bag technique until I can get a chance to pull the camera to bits yet again.

Good thing it's a long weekend and the forecast is looking good for tonight at least
Reply With Quote
  #19  
Old 14-04-2017, 07:00 PM
rcheshire's Avatar
rcheshire (Rowland)
Registered User

rcheshire is offline
 
Join Date: Apr 2010
Location: Geelong
Posts: 2,610
Looks like it's coming along well. Water cooling Argon and a PID controller. The dark says it all.

Yes... I've pretty much invented every step of the way with an Arduino and the ubiquitous TMP36 digital temp sensor. Trial and try again...

You can wind the voltage back on the sensor defroster depending on the megapixels of your sensor, providing your not cooling too low. -10C is the limit with low relative humidity.

At -5, I've never seen a drop of condensation. There is a but... the heater needs to be on as soon as cooling starts. It's designed to keep the glass clear but not to quickly recover a badly iced sensor glass. That takes ages.

I always get things up and running well before imaging and take a few sample shots. If there's any sign of frosting I up the setpoint by a few degrees. With Argon this is all academic.

EDIT: You can dispense with darks at 0 and below and forget about linear preprocessing methods and do a very simple;

Quote:
Signal = (light - bias) / (flat - bias).
Darks will very likely add noise to the images. It's possible to use linear processing but you need a program that does dark optimisation - personally, I avoid that style of processing with a cooled DSLR.

Have fun...

Last edited by rcheshire; 14-04-2017 at 08:34 PM.
Reply With Quote
  #20  
Old 21-04-2017, 10:30 AM
NinetySix (Ian)
Registered User

NinetySix is offline
 
Join Date: Jan 2017
Location: SE Melbourne
Posts: 21
It's been a while, but it's really coming along now pity Melbourne is going to be cloudy for the long weekend.

So I picked up a water block and radiator and had a bit of a tinker. I'm using a crappy pond pump for the minute and it's just doing the trick despite the longer-than-necessary tubes.

I'm definitely going to yank the thermostat controller as it drops like a stone when it cuts out, but I found that turning the pump off and on allows for quite predictable & smooth control, so I'm looking at picking up an overpriced CPU cooler pump with pwm control, something like an EK D5 to use with an eBay pwm fan temperature controller.

I ran it last night for about half an hour at -10c and the thermostat was switching off every minute or so and the cooler was sub-lukewarm with no fan running, so I reckon I'm on a winner here.

I should add that my temperature sensor isn't actually on the sensors heatsink as I couldn't quite squeeze it in, so it's directly on the copper finger about 2 or 3mm away from the start of the sensor heatsink. I've got some much smaller 10k sensors on the way though so I'll improve that and see what the difference is.

Edit: direct links didn't seem to work, click on the links
https://drive.google.com/uc?export=d...jBVczJPRUlZbVE

https://drive.google.com/uc?export=d...VdvMFhjdTRhZ00

https://drive.google.com/uc?export=d...DVkZUZGUWVGWGs
Reply With Quote
Reply

Bookmarks

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +10. The time is now 05:07 PM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
Advertisement
OzScopes Authorised Dealer
Advertisement
Meade Australia
Advertisement
Bintel
Advertisement
Celestron Australia
Advertisement
NexDome Observatories
Advertisement
SkyWatcher Australia
Advertisement
Lunatico Astronomical
Advertisement
Astronomy and Electronics Centre
Advertisement