CCD PSU - Low Ripple DC-DC Power Conditioner

[kbrown]

Hi,

Thought I'd share some first results of a project I've been working on for some time. The background to this is that I have an Atik 383L+ CCD that does not like low supply voltage. There's been numerous posts online about how its noise levels go up, when the supply voltage goes down. Some say this starts happening already below 12.5V!

My original remedy to this was to get a cheap XL6009 chip based DC-DC boost converter to keep the voltage at steady 13.5V and to be honest things worked okay. But at some point I had a look at what the voltage actually looks like under load with an oscilloscope. As you can guess it wasn't pretty. The more current you draw the higher the ripple voltage.

So after studying the subject a bit and testing things on a breadboard I came up with this device:

There's basically three stages in this:

1: The XL6009 DC-DC Boost Converter pumping up the input voltage up to 19V.

2: A Capacitance Multiplier circuit to significantly reduce the ripple produced by the boost converter. Similar to the top half of Figure 4 on this page.

3: A hight current LM317 based linear voltage regulator similar to Figure 23 on the LM317 datasheet page 17 except that mine isn't adjustable. It's set to about 13.6V.

 

Here's a test of the Capacitance Multiplier stage. The red channel shows the ripple on the output of the DC-DC booster which is 620mV Peak-to-Peak while drawing about 1A into a dummy load resistor. The yellow channel shows the ripple after the Capacitance Multiplier stage at the same scale. A massive difference!

 

Now that I've got everything put together, I have been testing it with the Atik 383L+ set to cool -30C below ambient for about an hour or so. The cooler is at about 67% power and drawing occasionally up to about 1.5A but it varies a lot. Average is probably about 1.1-1.2A. And since it's not purely a resistive load, the ripple is also very different:

Above you can see the ripple after the DC-DC converter (yellow) is a whopping 2.5V Peak-to-Peak! Yet the ripple going into the camera is only about 140mV.

There are a couple of drawbacks: The system isn't very efficient. If the camera is drawing 1.5A, the PSU could be drawing something like 2.5-3A occasionally! This means that things do get a bit hot. Overall I'm quite happy with the result anyway. I'll be taking new dark frames soon. Be interesting to see whether I see any difference to my old ones

Edited July 23, 2022 by kbrown

[Dr_Ju_ju]

If you truly want a clean output voltage, instead of using a "switched" supply, with its' inherent switching noise, I'd suggest a linear supply, based around a toroidal transformer, bridge rectifier etc.....

[kbrown]

43 minutes ago, Dr_Ju_ju said:

If you truly want a clean output voltage, instead of using a "switched" supply, with its' inherent switching noise, I'd suggest a linear supply, based around a toroidal transformer, bridge rectifier etc.....

That is true. But I'm running things off a car battery hence the need to boost the voltage for the Atik. Wish I had a RoR obsy...

[skybadger]

I use an automotive 8-40 to 13.5v converter on my solar/battery system in the obs. 

It solved a lot of 12v reliability problems putting the converter in. And it will supply 10A. But I guess I will have to look at the ripple now. Having said that I power the entire pair of scopes, cameras, mount, heaters etc all off the same DC sources supply and don't have any issues I can see. 

https://www.amazon.co.uk/your-orders/pop/ref=ppx_yo_mob_b_pop?_encoding=UTF8&gen=canonical&lineItemId=kjsptuirninwqoy&orderId=202-9967652-6373158&packageId=1&returnSummaryId=&returnUnitIndices=&returnUnitMappingId=&shipmentId=DVXPFlZDp I use an asi178mm cool and a qhy9 with it. 

 

 

[kbrown]

Been testing this a while now and while it works; It's not without quirks...

Originally I was going to power just the CCD with the power conditioner. But I also wanted to get rid of the cable clutter running to the scope so I divided the 12V input between the power conditioner and other (direct) outputs which I have connected to various devices on the system. By doing this I encountered an unexpected side effect: When the CCD cooler is working hard it occasionally draws a lot of current momentarily. While this is not a problem at the output of the power conditioner, it does show up as a voltage drop at the input (i.e. the other direct outputs as well). The drops in my case last only about 100ms but it does go down to something like 9.5V. This manifested as flickering on my flat panel for example...

First I tried adding more capacitance at the input and output but this didn't really fix the problem. In the end I ended up connecting all the devices to the power conditioned output. Everything works fine and is stable but I'm now drawing more current out of the device than I really wanted to. Below is my current circuit. While it works my concern is the Q4 getting a bit too hot. It's a TO-247 package with roughly a same size and 6mm thick heat sink on top of it. But it still gets too hot to touch under load.

 

Are there any electronics gurus out there who could give me tips on how to get this circuit to run a bit cooler? Perhaps my Q3/Q4 transistor choices are not the best here?

 

 

[Len1257]

Not sure that you need all of that circuit complication but.............

Linear regulators always generate heat, nature of the beast. Get a fan in there at least. The old 'standard' heavy current supplies we used in our 1970's mini computers at work had multiple 2N3055's in parallel with suitable sharing resistors. But I guess your heat issue is the heatsinks in your photo look positively puny with a constrained air flow. Better to get a heatsink that can be bolted to a wall of your case with fins to the outside and mount your Q1 and Q4 with insulated mica's to that. Something like a Fischer Elektronic SK48 65 x 20 x 50mm at 2.8C/W or equivalent. Some on Ebay £6.39. I would have looked for one that maybe was wider with room for two devices and spanned across the side of your case. Lots of choice out there. Rapid Online Heatsinks have lots of variety. Else get a PC fan!

[kbrown]

1 hour ago, Len1257 said:

Not sure that you need all of that circuit complication but.............

Linear regulators always generate heat, nature of the beast. Get a fan in there at least. The old 'standard' heavy current supplies we used in our 1970's mini computers at work had multiple 2N3055's in parallel with suitable sharing resistors. But I guess your heat issue is the heatsinks in your photo look positively puny with a constrained air flow. Better to get a heatsink that can be bolted to a wall of your case with fins to the outside and mount your Q1 and Q4 with insulated mica's to that. Something like a Fischer Elektronic SK48 65 x 20 x 50mm at 2.8C/W or equivalent. Some on Ebay £6.39. I would have looked for one that maybe was wider with room for two devices and spanned across the side of your case. Lots of choice out there. Rapid Online Heatsinks have lots of variety. Else get a PC fan!

Not sure if I can get any bigger heatsinks in the enclosure which is pretty much at it's maximum size already. Maybe a thin laptop style cooling fan...  Think I'll build everything again from C3 towards right on the schematic on a breadboard again and see whether I can find a solution that would naturally run a bit cooler under load.

[Mandy D]

9 hours ago, kbrown said:

Been testing this a while now and while it works; It's not without quirks...

Originally I was going to power just the CCD with the power conditioner. But I also wanted to get rid of the cable clutter running to the scope so I divided the 12V input between the power conditioner and other (direct) outputs which I have connected to various devices on the system. By doing this I encountered an unexpected side effect: When the CCD cooler is working hard it occasionally draws a lot of current momentarily. While this is not a problem at the output of the power conditioner, it does show up as a voltage drop at the input (i.e. the other direct outputs as well). The drops in my case last only about 100ms but it does go down to something like 9.5V. This manifested as flickering on my flat panel for example...

First I tried adding more capacitance at the input and output but this didn't really fix the problem. In the end I ended up connecting all the devices to the power conditioned output. Everything works fine and is stable but I'm now drawing more current out of the device than I really wanted to. Below is my current circuit. While it works my concern is the Q4 getting a bit too hot. It's a TO-247 package with roughly a same size and 6mm thick heat sink on top of it. But it still gets too hot to touch under load.

 

Are there any electronics gurus out there who could give me tips on how to get this circuit to run a bit cooler? Perhaps my Q3/Q4 transistor choices are not the best here?

 

 

Obviously, as others have said, more cooling is required for your pass transistor (TIP35). You have 5.4 volts across that transistor and up to 1.5 amps collector current. By my reckoning, that is a dissipation of 8.1 watts. For a comfortable rise of 50°C, that requires a heatsink rating of 6K/W.

You note that adding more capacitance to the input and/or output does not improve things. If you want to continue with this design of converter, I would suggest proper input and output LC filters, rather than just capacitance. So, a choke (inductor) in between pin 1 of J1 and the filter capacitors and similar on the output side will improve your situation.

[kbrown]

10 minutes ago, Mandy D said:

Obviously, as others have said, more cooling is required for your pass transistor (TIP35). You have 5.4 volts across that transistor and up to 1.5 amps collector current. By my reckoning, that is a dissipation of 8.1 watts. For a comfortable rise of 50°C, that requires a heatsink rating of 6K/W.

You note that adding more capacitance to the input and/or output does not improve things. If you want to continue with this design of converter, I would suggest proper input and output LC filters, rather than just capacitance. So, a choke (inductor) in between pin 1 of J1 and the filter capacitors and similar on the output side will improve your situation.

I don't think it's quite 5.4V across the Q4 as the capacitance multiplier circuitry before the regulator stage lowers the 19V down a bit but I get the point. Have to probe what's actually going into the regulator stage at C3. I'll play around with some chokes tonight as well. Thanks for the tips!

Edited August 5, 2022 by kbrown

[Len1257]

23 hours ago, kbrown said:

Not sure if I can get any bigger heatsinks in the enclosure which is pretty much at it's maximum size already. 

That's your problem then. Constraining the size is a restriction you are placing upon yourself. You could easily re-hash the physical layout to make it more thermally efficient.  Your flat minimal height layout isn't helping at all. Then making it out of plastic is another shot in the foot.

This is what I use.  Nevada Power Supply Note how the case is the heatsink. You can buy thermally efficient finned aluminium cases in all sorts of sizes.

Maybe your underlying problem is actually your battery. If it drops below your quoted 12.5V and it's a lead-acid one then maybe look for one that holds up better under load like a suitably rated LiFePo4 one? When I'm away from home I use a 36Ah one that easily lasts a full night supplying my rig.  Definitely a more expensive solution and the battery managment system will chop the supply off if you should exceed it's minimum voltage. Can't remember offhand how far my volts drop though, I'de have to dig my notes out.

[kbrown]

8 hours ago, Len1257 said:

That's your problem then. Constraining the size is a restriction you are placing upon yourself. You could easily re-hash the physical layout to make it more thermally efficient.  Your flat minimal height layout isn't helping at all. Then making it out of plastic is another shot in the foot.

This is what I use.  Nevada Power Supply Note how the case is the heatsink. You can buy thermally efficient finned aluminium cases in all sorts of sizes.

Maybe your underlying problem is actually your battery. If it drops below your quoted 12.5V and it's a lead-acid one then maybe look for one that holds up better under load like a suitably rated LiFePo4 one? When I'm away from home I use a 36Ah one that easily lasts a full night supplying my rig.  Definitely a more expensive solution and the battery managment system will chop the supply off if you should exceed it's minimum voltage. Can't remember offhand how far my volts drop though, I'de have to dig my notes out.

Yes I know the enclosure is a bit constraining. But height wise there isn't more room available where I want to have this sitting. I'll see if I can find a suitable aluminium case. At very least I could easily replace the existing lid with an aluminium one and perhaps have the Q4 mounted on it.

Nonetheless I managed to get it run a bit cooler by lowering the +19V at the booster output down to +17.2V. Still seems to perform well under load. I also added a 15uH 10A choke before the input caps and that seems to help with the interference to other devices.