Simple circuit, say 9V AC to a regulated 6.3 V at at least 2 amps?

[Paul Price]

Rugus says,"Two 7805s can have a much as 400mV difference in output and typically only 10mV of load regulation at 1.5A. "

It is extremely unlikely that you will ever encounter a 7805 within the same lot of purchased regulators to differ as much as that.
Quality control from any reputable manufacturer ensures they beat this spec.

I say it is not even necessary, but is there effort at all to very quickly, maybe spend 5 seconds to quickly press a 7805 into a breadboard to check it's output voltage before adding useless components or seeking a more complex or more costly alternative?

[Paul Price]

ChrisWilson,

I hope I might be speaking for the all of us, that we have scared the OP away from trying to use any regulating circuit, use any of our circuit ideas, just by our comments.


No one has said here that the regulators could cause damage, we are just ranting over current sharing.

Must we always say, without a reasonable doubt, that for the safety of all,  and without extensive  proof and laboratory testing,  that there will be fire and smoke and burnt filaments, that if you tried this idea if would ruin your equipment, that it is in fact so dangerous, so experimental, and that it doesn't really work well because of some assumed "sharing" problem?

[Paul Price]

Mikeselectronicstuff,

Don't forget what the OP asks for,
----------------------------------
Re: Simple circuit, say 9V AC to a regulated 6.3 V at at least 2 amps?
----------------------------------
Why try to find the right size and wattage resistor when you can use regulators?

First, it might be harder to find the right size and wattage resistor than to find 7805 regulators.

Secondly, and more likely, he might be trying to reduce AC hum in a sensitive audio amplifier or other equipment by removing the AC filament contribution to this problem in low level amplifiers.

[Paul Price]

Ok, if there was an attempt to parallel two or more 7805's to function as a regulator in "Mission Critical" piece of equipment such as what might be used in a hospital or other medical use, or to make a 747 fly, then the most critical evaluation and test of any circuit is required.

But what if you were in Zimbabwe or some other third world land and your old tube-based design of an EKG needed 6.3V at 2amps desperately to save some lives...? Would you say, "Oh No,  it can 't be done! The spec sheet doesn't recommend this! "

I haven't seen a spec sheet that said this should not be done.

[Rufus]

For a fixed load like a filament, you could just use a resistor - it will get hot, but so will a linear regulator

If it is a single filament where you don't care what voltage is applied if it has failed.

You could also provide some of the current with a resistor and let a smaller regulator control voltage and provide the balance.

[richard.cs]

cs.richard says, "Is there a particular reason for using a 9V transformer and d.c. heating? Would you consider running the heaters in series?"

Yes, this tampers with an antique circuit and creates more work soldering than necessary with possible damage to brittle old tube socket pins and wires with rubber or fragile cloth insulation that is ready to drop off at the first toggle or breeze.

I've clearly missed the bit where he states this is a modification to existing equiment, rather than a new design - in existing equpment I'd expect a 6.3V transformer winding specifically for the heaters. Having said that valve heaters are useually specced for +/- 10% anyway and running them 9V ac through a suitably chosen resistor should be fine, the mains should stay within 10% in any developed country and certainly so in the UK where Chris W is based. Why complicate things with a regulated supply?

[Chris Wilson]

The amp I am making the power supplies and antenna switching for is a Russian linear with a driver stage using 2 tetrodes in parallel. A French ham has a lot of experience with these and always uses a DC filament supply to the driver stage to reduce hum. I am striving for the cleanest signal possible, and if a DC filament supply helps, I don't mind the slight extra complication and expense. I am not experienced enough to say whether it's OTT twaddle or a viable means to gain a cleaner signal, so I am erring on the side of caution. Amp is part of the R-140 station. Just the amp is here: http://www.gatesgarth.com/r140.jpg


Thanks.

[richard.cs]

Thanks for the info Chris. It's not exisiting equipment then and you can build it how you like, but there seems to be reasonable justification for d.c. heaters.

I don't think you've mentioned part numbers for the valves or posted any schematics, are they directly heated?

[Chris Wilson]

Input valves a 2 off 6E5P in parallel, Class A.

[richard.cs]

With indirectly heaed cathodes I'm suprised there's significant hum from a.c. heating, but ok.

I would lean towards running the heaters in series then as you can keep the voltage across the regulator the same (at just a little higher than the dropout voltage) but halve the current and therefore halve regulator dissipation. It could be as simple as a single 7812 (or there's a 1.5A variant that's otherwise the same and a little less marginal) with a diode in its ground connection to bump it up to 12.6 Volts. This would need a d.c. input of about 15V which could happily come from a 12V transformer winding.

[edavid]

Why doesn't anyone try paralleling of 7805's to test their current sharing operation before commenting?

Because I don't have enough 7805s on hand to pick out a worst case mismatched pair, and I don't know how to determine the worst case conditions for the test, so it wouldn't prove anything.  I know it will work if the 7805s are well matched and the load is not too extreme, but I don't know what the limits are, so I would never recommend it, especially in this case, when there's no benefit to doing it that way.

Anyway, I can't figure out what the OP is building... is it part of a radio transmitter or receiver?  Either way, he doesn't need DC heater power in the first place.

[Paul Price]

Why try to find mismatch to create trouble for yourself if you desired to find a simple way to create a regulated power supply quickly?
 
edavid, you say,"so I would never recommend it, especially in this case, when there's no benefit to doing it that way."

The benefit is simple, a very simple circuit, low parts count, low cost, 7805 parts likely on hand, it works!

KISS

[Chris Wilson]

I breadboarded a circuit today using a single LM317 regulator and fed it 10V from my variable DC supply. It would regulate OK at 6.3V until it got hot, then needed a bigger differential voltage to stabilise. Perhaps another 1V.  I attach photos of the heat sink used. I also measured current draw. It was 1.2 Amps. Bigger heat sink and all OK? Need to rethink or try two LM317 in parallel? Thanks.

[SeanB]

Much bigger heatsink for sure, about 5-10 times bigger would do. If the LM317 is so hot that you cannot touch it then it is too hot.

[Chris Wilson]

I can just about keep a finger on it, but then again I am working on red hot engines a lot of the time, and I admit I was once called asbestos fingers What do I know, but to me it felt like it was too hot for long term sensible operation? The heat sinks came gratis with the regulators, so no big deal. I will find a big block of aluminium in the workshop and try it on that temporarily! Thanks Sean

[Rufus]

I also measured current draw. It was 1.2 Amps. Bigger heat sink and all OK? Need to rethink or try two LM317 in parallel? Thanks.

The valve datasheet you linked specified 600 +/-40mA so driving 2 in parallel is 1.2A - figures.

Can't you split them and have a regulator each?

[Chris Wilson]

Rufus: I probably could, would need advice on the necessity of changing the capacitor C3 in schematic attached.

But.... I have it on an ugly triangle of .75 inch thick aluminium and it's run driving the filaments for about an hour and its temperature is just warm. Will monitor for a few hours tonight and see what happens. I am monitoring current and voltage and using my "finger probe" to monitor temps, although I could get scientific and use a K type thermocouple surface probe and calibrated meter, which I use in my day job. But for sure it's not too hot at the moment, and I can't see this amp running all day, for example. I am learning quite a bit about what sort of heat sink size things need per watts dissipation and especially if they are run near the edge of their specs. I can input 9 volts now and it regulates fine, which should reduce dissipation? Thanks for all the great advice. Will see how it goes longer term now and report back! Cheers.

[Rufus]

Rufus: I probably could, would need advice on the necessity of changing the capacitor C3 in schematic attached.

Add another one it is just some rf decoupling.

Testing with a dc power supply you don't have any ripple on the input to the regulator. With a transformer/rectifier there will be ripple and the average input will have to be at least half the ripple voltage higher. The LM317 has a data sheet minimum input to output voltage of 3v. It needs 9.3v in for 6.3 out.

LDO (low drop out) regulators are available with much lower minimum input to output requirements. They would allow lower input voltage which reduces dissipation in the regulator.

Thermal considerations are not too difficult to work out. 9v in 6.3v out at 1.2A is 3.24W. Heatsink data will give you a degree C/W rating so for example a 13C/W heat sink would have to run 13*3.2 C above ambient. Looking at heat sink data will give you a feel for it.

Semiconductors can be run hot. The LM317 recommended maximum is 125C junction temperature. The 'junction' will be about 3C/W higher than the metal tab. It is possible to run them on a heat sink hot enough to boil water.

[Chris Wilson]

The supply has been left running the filaments overnight, and the regulator is a sensible temperature, I can easily leave a finger on it long term. The heat sink is just slightly warm. Voltage and current stable. 

Thanks for all the great info.

The next supply for this thing is 12.6 V DC at 7 amps. I am tempted to use a switch mode. The Chinese 12 V ones have a trimmer that gives + - 10%. The guru for the amp in question has started using them, as they are small, cheap and light, and have proven 100% reliable if you pick one a bit bigger than you should really need.....

Then on to the screen and grid supplies. Luckily the HV anode supply is something I have already built. Thanks again everyone!

[Chris Wilson]

Oh dear, sorry I am back again with another darned question!

Scoping the output across the 10,000 uF electrolytic, before the regulator shows what I think is a lot of ripple. After the reg it looks OK. Measurements taken with circuit under load. I am pretty new to all this, I just normally buy a PS So comments please. Thanks.

As an aside I have tried mounting the regulator direct to the heatsink, with no insulator washer and thermo pad, or whatever the thermal transfer, electrically insulating thin rectangular pads are called. It runs barely warm to the touch now, so the pads are presumably not terribly efficient at heat transfer? I realize the tab on the reg is also the output side, and now the sink is linked to the circuit.

[richard.cs]

800 mV pk-pk ripple before the regulator is fine, I would have expected a bit better than the 3 mV pk-pk on the output side but it's plenty good enough for any valve heater. How much capacitance do you have after the regulator?

Yes the thermal pads aren't great. The silicon ones are pretty crummy and the mica washers (with thermal grease) are considerably better. But if you don't need the isolation from the heatsink just bolt it straight on as you have with a little thermal grease. Sometimes the circuit allows it eg: grounded tabs on a 78 series regulator or sometimes you just float the heatsink from the chassis.

[SeanB]

800mV is not terribly bad, you would need to add more capacitance ( or a better lower ESR one than the one you have) to drop it much further.

As to the insulating kit, you need to add the thermal compound there between the device and the washer and heatsink. It is not just for show, but does do a very important function to reduce thermal resistance. It is rather like a bearing will work without grease in it, but adding the grease will make it last longer and run quieter ( as you say you are accustomed to cars so this is a good analogy) and this is very necessary. If you do not have the thermal compound ( not silicone sealer or gasket maker) use a thin smear of regular GP grease instead as a temporary substitute, and go to Maplin and buy some heatsink compound ( about the only thing they stock, just buy the cheapest per volume one, you do not need fancy here) to use there.

[Chris Wilson]

OK, that's a relief, thanks! I have used thermal grease between the tab and the h/sink. I used the thermal pad dry, should it have a thin film of the grease each side? The heat sink is now at output positive, and the heaters are run positive grounded, so I shouldn't have to worry about the sink being bolted to the chassis of the PSU. How much capacitance would be needed to get things smoother? Is 10,000 uF not quite a lot already? For future reference... I have a scrap Tek analogue scope here, is that likely to have some suitable caps in the PSU? I would imagine a scope needs a pretty smooth supply?

Also for future reference, is 3 Amps as much as these variable regulators will manage? Or can you get heftier ones at a price?  Any good sources of simple circuits to introduce a transistor to use for high current applications? Are they called "pass transistors"? Thanks.

[SeanB]

Thin film on each side, does the trick. Buy new caps at Craplins while you are there, get 4 2200uF 35V 105C electrolytics there ( they will have some, any value like 2200uF 3300uF or 4700uF will do as long as the voltage is over 25V) and use them in parallel in your circuit. Will drop it a few mV more, and the big chunk of metal will keep it cool.

As to higher current, look at the LM317 datasheet and you will see how to increase the output current using an external power transistor like a 2N2955 on a big heatsink.

[richard.cs]

10,000uF is a lot, more than is needed. To get really low ripple it's usual to use a fairly crude pre-regulator to a voltage a little above what you want then a second regulator down to the desired voltage. No-one does it unless they really need it because of the inefficiency and additional complexity. There will also be regulators that perform better than the LM317, better in the sense that for a given value of input ripple they give less output ripple. Also you didn't mention how much capacitance you have to ground on the output side of the regulator?

You've reduced the a.c. on the heater from  18 V peak to peak (a.c. powered heater) to 3 mV peak to peak. Any hum you might have got from the heater will have been reduced by 6000 times already - you really don't need it any smoother than this. Remember the heater is not electrically part of the valve circuit, it just has a very small capacitance to the cathode. At 50 Hz this capacitance looks like a very high impedance and can usually be ignored, especially in the circuit you posted where the cathode connects straight to ground anyway. Usually heater hum is only a problem when the cathode is many kilohms from ground, typical with certain audio pre-amp circuits.

3A is the biggest of this type so far as I'm aware although I'm sure someone must make adjustable linear regulators up to 5-10A. But really by 3 amps you're in the region where it makes sense to build a switching supply. There are ways to add an external transistor to a standard regulator and some regulators don't include the pass transistor at all and you have to use an external one.