Troubleshooting my step down opamp power supply (TCA0372)

[brndvnrdn]

Hi everyone,

Hope someone can impart some wisdom since I seem to be at a dead end.

I'm using a TCA0372 power opamp to step down voltages from +- 16V to 15V and 8V for a DC-DC converter and LDO going to a tube filament/heater.

I don't have the DC-DC converter load connected yet since I am running into some trouble with the heater supply.

Whenever I connect the load going to the tube heater, it's only drawing 150mA but the TCA0372 starts overheating very quickly. This is strange, since it should be able to supply up to 1A, yes ? I don't see how it could be non functioning like it is. It's not oscillating I think since I grounded any unused inputs when troubleshooting.

 

[JohnRoberts]

Heat is a function of voltage times current so 150mA times 8V (16-8) = 1.2W.

Check the device spec sheet for power dissipation capability... >1W can get hot without a heatsink.

JR

 

[ccaudle]

150mA times 8V (16-8) = 1.2W

Datasheet says 80 degC/W, so that would make the part over 80 degrees above ambient. Unless you keep it in your refrigerator that part will be over 100 deg. C.

I'm surprised those parts don't have a thermal pad on the bottom of the case, keeping good contact between a big heatsink and an SOIC package seems a little difficult.

 

[brndvnrdn]

Thanks for the suggestions! Will look at some heatsink possibilities and see if that alleviates the problem.

 

[JohnRoberts]

You can touch a drop of spit on top of the IC package, if it boils away the package is 100'C or higher...

JR

 

[brndvnrdn]

Will try to spit on it a bit haha. Good thing my HT supply is not yet connected.

 

[Bo Deadly]

This is a bad combination of parts.

By dropping +16 to 8V at 150mA you are effectively wasting 1.2W of power. Presumably you did this because your DC-DC converter cannot accept 16V. In this case you could have just used a zener or, if the load is constant, a sufficiently large resistor to get you under the DC-DC converter maximum input voltage. Or, if you used an AC-DC converter, you would eliminate a lot of unnecessary parts and not see excess heat at all.

Also, consider that you are probably operating the op amp at the edge or over the output swing range.

Perhaps you should explain what problem you are trying to solve rather asking about problems with a bad solution.

 

[Bo Deadly]

You can touch a drop of spit on top of the IC package, if it boils away the package is 100'C or higher...

Does it have to be spit? Why not just clean water?

 

[JohnRoberts]

Does it have to be spit? Why not just clean water?

of course it doesn't need to be spit... spit is just something that most people have handy. This is just an old technician trick. 100'C (boiling temp of water) is generally considered too hot for reliable operation. Less precise if you can't hold your finger on it, it's too hot. Back in the day I burned little round circles into my finger tips from touching old metal can transistors.

JR

PS: If the junction to case relationship is 80'C/W and it is dissipating 1.2W that means the junction will be at least 100'C +96'C if the spittle (or water) boils away. Don't pour water on an energized circuit you can dip your finger in water and then touch that wet finger to the package. Maybe be careful probing with a wet finger around high voltage circuitry.

 

[Khron]

I have a little spray-bottle of 99.8% IPA that i use for general cleaning and whatnot. Spraying some on energized circuits seems to have been pretty safe so far, and one can gauge roughly how hot a component is getting at least by how much faster it evaporates compared to stuff around it.

 

[Bo Deadly]

I have a little spray-bottle of 99.8% IPA that i use for general cleaning and whatnot. Spraying some on energized circuits seems to have been pretty safe so far, and one can gauge roughly how hot a component is getting at least by how much faster it evaporates compared to stuff around it.

I wouldn't do that on higher voltage circuitry and especially not tube stuff. There's all sorts of things that could happen like the shock of the cold IPA causes a hot part to blow up and ignites the IPA.

Similarly, I agree that no one should stick wet fingers into energized high voltage circuits.

Also, no matter how compelling the thought might be, do not stick a fork into a mains socket.

 

[brndvnrdn]

This is a bad combination of parts.

By dropping +16 to 8V at 150mA you are effectively wasting 1.2W of power. Presumably you did this because your DC-DC converter cannot accept 16V. In this case you could have just used a zener or, if the load is constant, a sufficiently large resistor to get you under the DC-DC converter maximum input voltage. Or, if you used an AC-DC converter, you would eliminate a lot of unnecessary parts and not see excess heat at all.

Also, consider that you are probably operating the op amp at the edge or over the output swing range.

Perhaps you should explain what problem you are trying to solve rather asking about problems with a bad solution.

Am trying to make a 6.3VDC filament/heater supply.

Doing this for 500 series so trying to keep current spec low.

Also trying to feed an appropriate voltage to the L7806 regulator so it doesn't get very hot.

What would you suggest ? Buck DC-DC converter ?

 

[radardoug]

You can use an LM2576-5 simple switcher and put some resistance in the control voltage lead to make it produce 6.3volts. Low component count and switcher, so low dissipation.

 

[scott2000]

Fischer Electronik-ICK heatsinks come in 6mm and various lengths and properties. I know Jim Williams recommended in the past using some Loctite 498 for attaching heatsinks directly to some hot running smd op amps.?.... idk... 498 working sweet for a rear view mirror at least.....

 

[Bo Deadly]

Am trying to make a 6.3VDC filament/heater supply.

Can it be 12.6V instead?

Doing this for 500 series so trying to keep current spec low.

Also trying to feed an appropriate voltage to the L7806 regulator so it doesn't get very hot.

What would you suggest ? Buck DC-DC converter ?

Just thinking about efficiency, the ideal part would be a DC-DC converter that accepts 30V and outputs 6.3V + enough for an RC or capacitance multiplier to filter / step-down the output (so let's say 7.5V output from the converter if a CM was used as the filter / step-down but could be higher if you used a mosfet CM).

Note that 30V is used instead of 15V because you don't want a large load imbalance on the 500 series supply. DC-DC converters are equally efficient regardless of input voltage range spec.

I would not try to create your own SMPS if you don't have to. Off-the-shelf SMPS usually come in quantized output voltages and 7.5V is not one. However, MeanWell SLW05B-09 accepts 18-36VDC and outputs 9VDC. This is about the right current for one 12AX7.

What are your precise heater current requirements? What type tubes and how many? Note that one 12AX7 draws 300mA when heaters are wired in parallel (using 6.3V) and 150mA when in series (using 12.6V).

Stepping down 9V to 6.3V burns 0.81W which is marginally better than the 1.2W.

But, if you step down from 15V to 12.6V instead, that's 0.36W which is quite a bit better.

So I would look at SLW05B-15 and then follow that with either a really simple RC filter of say 10 ohms and 100uF or whatever you need to drop the voltage to 12.6V. Use surface mount parts with a lot of copper top and bottom stitched together with vias to make a make-shift heat-sink.

The only caveat to all of this is that SMPS can emit high frequency noise so having the DCDC converter inside the 500 series enclosure might be an issue. Especially for high gain, high impedance circuits like tubes.

 

[brndvnrdn]

Can it be 12.6V instead?

Unfortunately not. Using a 6AS6 pentode and an EF86 pentode. Both only support 6.3VDC.

Going to try out the LM2576-5 with a feedback resistor as suggested by Radardoug and see how it goes.

 

[Bo Deadly]

Creating your own switcher is not trivial. It is very likely that you will end up with noise somewhere. Most of the $3 step down switching regulators on Ebay use LM2576. Those are actually pretty good [1] but probably not for stuffing inside a 500 series chassis. If you follow the Layout Guidelines section in the datasheet like gospel, it might work. IMO you would be better off using an off-the-shelf switcher that was carefully designed by professional engineers and tested extensively because they sell thousands of parts.

[1] I've used one of those Ebay switches to power my Lexicon MX200 for years and it's still going. The second channel is a little flaky but I somewhat doubt the supply is responsible. Although the first one I tried didn't work at all and the second died quickly. Fortuntely, I bought 10 so I was able to find one that worked.

 

[brndvnrdn]

Creating your own switcher is not trivial. It is very likely that you will end up with noise somewhere. Most of the $3 step down switching regulators on Ebay use LM2576. Those are actually pretty good [1] but probably not for stuffing inside a 500 series chassis. If you follow the Layout Guidelines section in the datasheet like gospel, it might work. IMO you would be better off using an off-the-shelf switcher that was carefully designed by professional engineers and tested extensively because they sell thousands of parts.

[1] I've used one of those Ebay switches to power my Lexicon MX200 for years and it's still going. The second channel is a little flaky but I somewhat doubt the supply is responsible. Although the first one I tried didn't work at all and the second died quickly. Fortuntely, I bought 10 so I was able to find one that worked.

You are right, that is a much better idea.

I'm now using this switching regulator which is working beautifully: TSR 1-2465 | Traco Power

 

[Bo Deadly]

I'm now using this switching regulator which is working beautifully: TSR 1-2465 | Traco Power

Bear in mind, that is not an isolated converter. But if it works and you're not getting any noise, then great.

But running at 500kHz, you should carefully check for switching noise using a scope. If you find that there is some significant 500kHz, you could try an isolated SMPS and float the header supply relative to 0V with two resistors or elevate it if desired / necessary. Can't do that with the one you're using.