Reasons LT regulator fails under load

[CSlee]

This is the first LT regulator I have built so I was following the schematic and building instructions to the letter, I guess it's a case of MJ says jump...

The supply chain is transformer - bridge - reservoir - cm choke - regulator.

It must work though, so it will be interesting to make it stable with the chokes and find out where the construction mistake are.

[DF96]

Did he give a PCB design too? Exact placement of components can affect stray capacitance and hence stability.

Personally, I can't quite see the point of a CM choke there unless your mains is particularly dirty. Better to put a filter on the primary side.

[CSlee]

No it was just a schematic with some notes on construction.

I just moved house and the mains has been an issue so built a little CLC mains filter which helped a lot. So maybe this is overkill...

[abraxalito]

Originally Posted by CSlee

I will keep trying to get back to the original schematic as there must be a reason for the instability.

Can anyone offer suggestions on this?

My guess is the regs don't like seeing inductive reactance between the input and the reservoir cap(s). Normally datasheets for 3 pin regs say things like 'input cap needed if reg is more than 6 inches from res cap' implying input reactance is an issue for stability. By putting a CM choke in there you'll get a fistful of leakage inductance (tens to hundreds of uH typically) - potentially equivalent to putting the res cap in a different room from the reg. Like DF96, I can't see any point in having a CM choke there.

If you try AndrewT's suggestion of turning the regs into CCS, I've found 317s can go unstable in those configurations too.

[jitter]

The application hints in the datasheet say that when caps are used on adjust and/or output that bypassing at the input is needed to keep the regulator stable. Abraxalito's explanation seems plausible since you have none between choke and input and you do have 10uF on the adjust.

Personally, I'd just leave out the CM choke.

If you don't like the idea of current peaks powering up cold heaters and the CCS configuration might be prone to instability, then perhaps the slow turn on config I linked to in my previous post is the best solution. Here's a video showing the slowly rising output voltage of a slow turn on regulator. From the comment below the video: "Slow turn on - adjustable voltage regulator - 1.5v to 35v - 1.5A ideal for tube filament supply 6.3v", I' wouldn't be surprised if it's also based on an LM317.

[DF96]

I have found that 'over-improving' a circuit can make it behave worse, or stop it operating at all. Despite knowing this, I still sometimes do it!

[AndrewT]

a CCS 317 is just two components: 317 + resistor.
There is no connection to ground. Therefore no need for decoupling of either the input or the output. There is no complex circuit.
It just works.

The Tube/valve Members have used them and there are some good reports for this style of heater drive.

BUT !!!!
each CCS can only drive heaters with the same current requirement. The rules on heater to cathode voltage differential still apply.

[Preamp]

Originally Posted by DF96

Did you follow any advice in the datasheet about local decoupling etc.?

That's my pick, too. Those regulators are happy with something like 100nF from input and output to ground, physically close to the IC itself. Did you do this?

[CSlee]

There are different heater current requirements here so I guess a CCS is not an option this time.

I have a 1uF tant cap on the output as recommended in the datasheet, but actually I have only a 10nF and not 100nF on the input, so maybe this is the reason it is misbehaving.

Regulator circuits of this type are quite new to me, so learning a lot. Thanks for the help.

Edit: ditched the chokes, if it works without them...

[cotdt]

Did you get your LM317's from China?