LM317 on steroids

[Terry_VK5TM]

Came across this adjustable voltage regulator - LR8N3 - which is a high voltage regulator in a TO92 package.

Quick specs:

Up to 450v in

output range 12v below input

max 20mA without pass transistor

Voltage is set just the same as with the LM317 - two resistors.

Datasheet http://ww1.microchip.com/downloads/e.../20005399B.pdf

There is a HV power supply project in Issue 5/2019 Nut & Volts mag using this device.

A bog standard 317 will take about 45V across it. Input voltage can be anything, it doesn't appear across any of the terminals.

[MrBungle]

If you know what the current and voltage is up front you can just shove a few zeners in front of the bog standard LM317 to do the same. In fact that’s fairly normal.

This Ic appears to be designed for hot side SMPS controller preregulators.

[chriswood1900]

Terry
A useful device, I used one as the screen supply in my valve tester project in conjunction with a pass transistor. Most of the Nuts and Volts article appears here:-
http://nutsvolts.texterity.com/nutsv...&folio=48#pg48
There is enough info if you care to have a go at an HV PSU.

[mhennessy]

The LM317 is only good for 37V dropout voltage. The absolute maximum is 40V, but you'd be mad to try it there. If there is a version that can withstand 45V - presumably from one manufacturer specifically - then that would be useful information, as 37V is often not quite enough.

The LR8 is a pretty amazing device though. Could be challenging designing the PCB if you want to use the SOT89 version for the voltages at the higher end of its range though

output range 12v below input

I based my comment on that. Reading the datasheet, what a handy device!

[G6Tanuki]

That's a chip I'm definitely going to make a mental-bookmark of; it could come in useful for a 90V/120V power-supply I need to build.

I always like to keep an eye on new developments, whether it's this sort of chip, low-dropout low-voltage regulators designed for automotive applications, high-side-driver switching power-FETs or low-cost surface-mount microwave devices with noise-figures we would have killed for in the 70s.

Technology moves on. Embrace it! I'm sure there were some people around who shook their heads in disgust when high-voltage medium-current silicon diodes appeared on the market and their old mercury-vapour rectifiers were rendered obsolete, or lost sleep about what would become of their beloved 807s when in the 60s Motorola introduced the first bipolar transistors that could generate 150W of RF at 30MHz...

[As far as the LM317 goes, I still make use of the 3-amp-rated TO3-cased LM117/317 "STEEL" range, even though they're getting harder to source these days, There was also a similar TO3-profile regulator from National Semiconductor - LM140T-somethingorother? - that had four 'legs' and was rated to 5A - back in the 80s I used loads of them in situations where I needed a +/- 24V supply. Those were really hard to kill! ]

[MrBungle]

The current “best” device in the LM317K replacement class is the LT108X series: https://www.analog.com/media/en/tech...s/108345fh.pdf

They go up to 7.5A and have much lower dropout.

[mhennessy]

Originally Posted by merlinmaxwell

I based my comment on that. Reading the datasheet, what a handy device!

Yes, much confusion here

12V is the minimum dropout voltage of the LR8. But you and I were talking about the maximum dropout voltage of the LM317.

So I'm guessing you added 12V to 37V to get 45V. Yes, if you're making 12V with an LM317, then you can go up to 45V at the input, as the dropout voltage is 37V. But if you draw enough current (or get it hot enough) to cause the output to fall below 12V, then that increases the dropout voltage beyond 37V and puts the device at risk of failure from that. Too risky, IMHO. But people do it...

The minimum dropout voltage of an LM317 is around 3V, which is fairly typical for a "standard" voltage regulator. So on the face of it, 12V for the LR8 is unusually high. But that's because the pass device is a MOS-FET - which, apart from other things, helps to keep the quiescent current to an absolute minimum. Given the intended application (stepping down lots of volts at hardly any current), it's fair enough

[Terry_VK5TM]

Originally Posted by chriswood1900

Most of the Nuts and Volts article appears here:-
http://nutsvolts.texterity.com/nutsv...&folio=48#pg48
There is enough info if you care to have a go at an HV PSU.

I get the "You must be a subscriber" message to access that page.

[chriswood1900]

Terry
That is strange it works fine for me and I'm not a subscriber, it does not give the whole article just the first few pages.
I'm using Chrome with Windows 10 and it also worked with Safari on IOS13.

[joebog1]

Works for me Terry!.
Its a very nice little device indeed!!. It is also the most stupid package device indeed.
Could be one of the reasons I hate SMD stuff with a passion.

Joe

[Diabolical Artificer]

Works for me Terry, see att. The LT1083 looks very good as Mr B says, far more current and low dropout V.

Andy.

[Terry_VK5TM]

OK, much weirdness happening .

In Chrome I get as mentioned above



and in Firefox I see the same as you guys.

Anyway, at the going price of $1 ea (RS), I think I just might stick a few in the junk box.

[buggies]

The chip is used for a variable supply in the Sussex valve tester (long thread in homebrew section of the forum) e.g a couple of years ago so maybe some modifications since this post:-
https://www.vintage-radio.net/forum/...postcount=1456

[Philco-Pete]

Actually there IS such a device; the imaginatively named LM317HV.
Features a 60V in/out differential at up to 1.5A. I have used them quite a few times having first discovered their existence after one of the young lads at work shot me down in flames for quoting the standard “37V is the most you can get across a 317” mantra.
Never too old to learn...

[mhennessy]

Err, it's a fact, not a "mantra"

Obviously the LM317HV is a different part. Very similar, but different. Quite a bit more expensive too. You can see why Marketing thought it was a good idea to maintain the "LM317" part of the name, but you have to wonder how many standard 317s got fitted in place of the HV part in busy repair workshops...

60V is the absolute maximum rating for dropout voltage. I would be careful about operating too near to that.

Another one is the TL783.

That's good for a bit less than 125V (125V is the absolute maximum, but the datasheet does say that it's OK to use it there - but better safe than sorry!). Cheaper than the LM317HV too. Handles less current (700mA), but at these voltages the thermals will be more of a limit. It has a higher dropout voltage as like the LR8 it's based on a MOS-FET, but again, that's fine at these sorts of voltages.

[joebog1]

I have ordered 10 of them. @$1.01 each from Element14 I couldnt refuse. When they arrive I will be requiring help to up the output from the nutsnvolts article to fetch something more than 50 mA from the output. Nutsnvolts article is in fact quite elegant!!
( yes from me )It uses the KISS principle. Its cheap. As I read it the short circuit protection bit of the chip will protect the rest of the circuit. ( I would love to have the whole article) Its lots cheaper than my original design for my valve tester that uses 3 807,s for the plate and screen supplies ( looks VERY much more sexy than some black blobs on a circuit board though).

I think this thread may last a while.

Joe

[mhennessy]

The TIP50 transistor is used as an emitter follower. As such, the current limit is determined by whatever the LR8 limits to, multiplied by the hFE of the transistor.

Obviously, hFE varies wildly, and the current limit of the LR8 is also pretty loose (10 to 30mA, but read the whole datasheet).

So I would advise caution with a circuit like this. It'll be fine for many applications, but it might play up under certain conditions.

A better way to beef up the output of a regulator like this is to put the transistor across the regulator, and have the regulator supply base current according to what the regulator is doing. See figure 23 of this datasheet: http://www.ti.com/lit/ds/slvs044x/slvs044x.pdf

You've probably seen this sort of thing before - there are many variations. You can use a single PNP power transistor, a PNP Darlington pair, or a CPF as shown in this example. The 22 ohm resistor in this example sets a current threshold of around 27mA; anything above that is supplied by the pass transistor. For the LR8, you'd obviously use a higher resistor as the LR8 probably ought not supply more than a milliamp or thereabouts (perhaps something in the 560-680 ohms area would be a good starting point?).

This circuit gives much better voltage regulation than the EF version, though that is probably not a major concern in a circuit like this. But so far, the definition of the current limit is no better than the EF version. So to fix that, we need to add another resistor and transistor. I did a quick web search, but didn't find a nice diagram of that, but could easily produce one. In essence, the resistor goes in series with the input to the pass transistor, and the extra transistor senses the voltage across it. When that voltage reaches 0.6V, the transistor turns on to shunt away the drive to the pass transistor. This extra transistor must be able to carry the short-circuit current of the regulator IC, which isn't much for an LR8, of course. The overall current limit isn't as precisely defined as it would be in a "proper" bench power supply that uses an op-amp for the task, but it's a lot better than what we started with

Also, with the addition of a couple of resistors, it could be changed to a "foldback" action, which is well worth considering for a high voltage PSU.

[joebog1]

As I mentioned, I am seriously considering using this for my valve tester. I should explain.

I am not so much interested in a valve tester per se, but more a gadget to match valves.

I quite often make fairly large class A (yes I'm an audiophool) valve amps that use multiple output bottles. Making everything balance nicely before one starts, selecting 10 or 15 bottles with essentially the same characteristics is more than halfway there.

I am still planning on building myself a "Laboratory Golden Ear " ( design by Joseph Marshall ) and that needs 8 valves. Four per channel. He used KT66's, but as I haven't won the lotto yet I will be using 5B/254's made by ITT in Britain. I have about 50 in stock, and access to many hundreds more.

Hence my interest.

Joe