Direct Bias Amp Off Experiments for SC1 and SC3 modified astrophotography webcams

Introduction to long exposure webcams and telescopes


This page is a summary of some recent experiments in overcoming the "amp glow" that can be found in our astrophotography astronomy images taken with modified webcams.

I primarily use SC1 and SC3 modified Toucam webcams to do my astrophotography astronomy imaging. This ideas was following the original concept by Steve Chambers. The CCD sensors in these modified webcams have certain on-board components that tend to glow very faintly, and show up as a brightening, usually in the top left hand corner of the image. This often spoils our astrophotographs. The main culprit is a small amplifier (hence AmpGlow), but there are others.

There are a number of different amp off modifications for SC1 and SC3 modified webcams. These are well documented to allow this glow to be reduced on modified webcams. These normally involve reducing the supply voltage of the CCD sensor during the exposure. The reduced volts means a reduced glow. When the picture is read into the computer (by k3ccdtools or your chosen astronomy imaging program) after the long exposure the full supply voltage is turned back on briefly so that all the electronics in the CCD chip can do their job.
Useful links on this topic include:

Steve Chambers
Greg Beeke
Martin Burri
William Behrens
Julian Palmer

I've used the "Zener Amp Off" method myself for some time. This does a good job of reducing the amp glow, but its still there. I was beginning to get frustrated by this residual glow - especially on longer exposures. This image of M1 taken with a narrowband filter with 30s exposures shows the problem. The effect can be removed from our images with darkframes (nb I've never understood why people try to remove amp glow with artifical flats.. never worked for me). The processed image shows this - but this is a poor work around. Detail and dynamic range in the effected area is badly compromised, and one has to be careful to avoid this part of the chip for very long exposures using narrowband filters and gratings (for spectrums).

The other downside is the effect the reduced supply voltage causes in the chip. Without going into technical details, reducing the supply voltage to the CCD can reduce the dynamic range of the CCD by effecting the "substrate bias" on the ccd. Images of bright targets saturate the pixels before you've done an exposure long enough to collect faint detail. eg The core of a galaxy is often much brighter than the arms. A reduced supply voltage makes it harder to preserve core detail whilst going deep enough into the spiral arms. Our webcams only have 8 bits, compared with "astronomical" ccd cameras that usually have 16 bits - reducing ones bits to save some ampglow is clearly a bad compromise to make. In the ICX424 and ICX098 chips, the substrate bias is produced inside the chip using the supply voltage. Drop the supply voltage, drop the amp glow, drop the bias, drop the dynamic range.

In the extreme case, one could reduce the supply voltage to the point where the CCD turns off! In which case you get a white screen.

Steve's website has a good general guide on the inner workings of CCDs that I recommend reading at some point. Our substrate bias is the voltage holding the electrons in the well at the bottom of figure 2. Thats my understanding anyhow.

An Idea

In April 2005 Steve Chambers made some posts to the QCUIAG group regarding support for the ICX424 chip in the Artemis CCD camera. He made available a darkframe from this camera. This frame showed impressively low noise, and no visible amp glow, even after histogram stretching. Obviously Steve was using something other than the webcam ampoff ideas to control his amp glow!. Steve explained what was going on, and I decided to have a go at making this work on my SC3 modified webcam.

The principle was to lower the supply voltage to the CCD during the exposure but inject your own bias voltage into the CCD chip. This way the amp glow is reduced, but dynamic range is preserved.

My knowledge of electronics is "reasonable" at the best of times, and the inner workings of CCDs I put firmly to the left of witchcraft. Consequently I had to spend a few evenings playing around with my precious cameras, guided by Steve, experimenting with these ideas.

The Details

First off, some handy datasheets... these are on the sony site, these links may not last forever.

ICX098 Datasheet
ICX424 Datasheet
Peter K3 CCD Comparison

After much playing I got a system that appeared to be working. What follows are some simple steps to replicate my setup.


1. Change your ampoff circuit. You need an amp off circuit that interrupts the supply voltage pin of the CCD. Some ampoff methods use the ground pin of the CCD - this won't work here. Supply voltage style ampoff circuits either use a variable resistor or zener diode to set the amp off supply voltage. This needs to be adjusted to give about 6 to 6.5v during amp off. My original ampoff mode gave about 8.8v supply during amp off, so I changed my zener diode for one rated at 9.1v which gave 5.9v supply during amp off. Also, your amp off modification needs to switch the volts to both pins 9 and 10 of the ICX424.

2. Measure the existing substrate bias. The important pin here is the substrate clock. On the ICX424 this is pin 12. On the ICX098 its pin 10. With the webcam running normally, ie amp on, use a multimeter to measure the voltage at this pin. My ICX424 was 9.77v

3. Now set the camera running long exposures with amp off enabled and watch the voltage on this pin. During amp off you'll see this voltage drop. If it doesn't, you probably haven't interrupted the voltage to pin 10 (SUBCIR).

4. Next setup a variable resistor. I used a 5k pot. Connect the resistor between the pad for pin 9 (+15v) and ground. Use it as a potential divider to inject volts via a diode into pin 12. Putting a fixed resistor between the variable resistor and ground is probably good for safety. Set the VR wiper to its midpoint before switching on the camera. Below is a rough diagram of the setup. Note that pin 12 is not lifted. The zener and the "switch" at the top belong in the existing amp off circuit.



5. Now fire up the camera and set it doing long exposures with the amp-off. The key is to adjust the variable resistor until the voltage at pin 12 is to within 100mV (but just under) the value measured in (2). I think it is important to adjust this value when the camera is at operating temperature - ie peltier on, if you have one. I've found the well depth seems to change as the camera cools.

6. And thats it. Bet you were expecting something with more components in!

Some Results

I took a series of darkframes to keep track of my progress. Each is a small stack, and each has been histogram stretched by the same amount to illustrate the effects.
First of all, here is my original darkframe with my standard "drop the supply to 8v" amp off modification.
Amp Off. Supply voltage 8.8v. No Bias injection. 60s frames. Max gain.

And then the darkframe when the amp off was changed to 5.9v.
Amp Off. Supply voltage 5.9v. No Bias injection. 60s frames. Max gain.

Final darkframe with the bias applied to pin 12.
Amp Off. Supply voltage 5.9v. Bias injection 9.74v. 60s frames. Max gain.

As you can see from the final darkframe, the amp glow has gone! Next I tested on a target using an SLR lens. As the laws of astronomy dictate, it was rather overcast outside, so I darkened the room and used an SLR lens pointing at a chart with some flags on - above (well, below, but we're upside down) this I tacked a printed out gradient of black to white.

After much phaffing about with the aperture and the ambient lighting (had to point the led on the bottom of my mouse at the wall to illuminate things, there aren't many streetlamps in these parts) I was able to get some frames....

First of all, with amp on.
Amp On. Supply voltage 15v. Bias injection 9.74v. 30s frames. Max gain.

And with amp off.
Amp Off. Supply voltage 5.9v. Bias injection 9.74v. 30s frames. Max gain.

For those of you that like fits, this zip file contains both FITS files.

Conclusions

It seems to work. :-)

There is no visible amp glow - I think there might be a slight brightening down the left hand side, buts its hardly noticeable.

In these test images there is also no visible problem with dynamic range. Only an astronomical test will clarify this.

The amp off images are potentially a little more noisy, but I am not certain of this.

The amp off image seems a little dimmer overall - this may be because the bias voltage is slightly lower.

Further work...

There is a fair bit of work left to do.
1. Experiment to see what effect slight variances in the supply voltage during amp off has. Steve tells me the ideal is 6.4v, but this needs playing with.
2. Experiment with changing the bias voltage
3. Change the layout/improve the circuit to try and reduce noise.
4. Dynamic range tests (any ideas??).
5. Noise tests and comparisons.
6. Proper astronomical tests and comparisons.

Some more tests.

SC1 Cameras

I've managed to make this work with an SC1 camera using a colour 098 CCD chip. I did this first as a test before I trashed my 424 camera. I don't have any results from the 098 camera, and have not made a careful analysis of the outcome. However, I don't see why it shouldn't work on an SC1.

and Finally...

This has been an interesting test - but it needs some more fiddling and experiment. Off you go!

Please feel free to email me any questions etc on tomh (at) tgh dot demon dot co dot yoo kay (also in image on the front of this website)

As usual, large amounts of thanks to Steve and the Artemis gang for giving me the idea and helping me get a result. And thanks to all the QCUIAGers, without their endless mutual support, help and encouragement, I'd still be looking down an eyepiece.

Finally, be careful. Don't short out anything important with your multimeter probes. Don't blame me if you break anything. :-P

Update Aug 2005

Not much else to add at this stage - I am still using this technique with great sucess and I hope other amateur webcam astronomers will do the same!

Nov 2005

A thought occours to me - it is important to adjust the bias to get the best dynamic range. Maybe the terminator of the Moon would be good for this.

Tom How, Curdridge Observatory, April 2005.