• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

LM317 for power amp regulation?

Hi all,

I have been looking for regulator to power a pair of 2a3 amp. Most of the designs look complicated and the most simple zener/voltage divider does not appeal that much to me.
Then i found several designs over the net using LM317 for regulation (for preamp).

Consulted the datasheet, it says nothing about maximum absolute voltage. It only specifies Vout-Vin<40v, something that is manageable in my case. Furthermore, the current rating of 1.5A is more than enough for any power amp application, am I right?

Has anyone ever tried this and what is the outcome of it? If this chip is applicable, it would make a really simple and yet moderately good performing regulator, I think.



2004-11-09 11:09 am

To do this, you need to include a pass transistor and a zener to limit the voltage across the 317 - 40V is not a lot to play with when dealing with ~300V. The zener/transistor combo limits the voltage across the 317 to ~ 8V. It is connected such that the incoming (unregulated) voltage goes to the collector (NPN transistor), emitter goes to input of the 317, a ~8V zener is connected from the 317's output to the base of the transistor - this limits the voltage across the regulator to the zener voltage - ~0.65V (be junction of transistor). A resistor from collector to base of the transistor is also required to turn the transistor on.

With this high voltage, you need high voltage transistors - my preamp has this configuration, I originally had a MJE340, but I am now upgrading and I need more current (~200mA). However, as the voltage across the 317 is held constant at ~8V, most of the volts are dropped across the MJE340 - which isn't able to handle too much power. Another problem is that the beta (current gain) of the MJE340 can only be guarnteed to be 30 - this means that for ~200mA output, you need ~6mA at least through the resistor for a voltage input differential of about 10-20V (when the mains voltage drops). This means that the resistor will be ~15/6 ~ 3k, and when the mains voltage is normal, you might get ~40 to 50V across this resistor, which will dissipate ~ 50^2/3000 ~ 1W, so at least a 2W component is required.

To solve this problem, I insted used a darlington connected pair of transistors, with the MJE340 driving a BUT11A (TO220 device, so it can handle the power dissipation). This gives a much higher current gain, which allows the use of a smaller valued c-b resistor.

So, yes I think you can quite easily make a regulator for your amplifier(s), but I reccomend using a darlington pair for the series pass transistor.

If this is not clear, I'll draw you a schematic. Also, a read of Morgan Jones' Valve Amplifiers would certainly not go astray - that is where I learnt this from (the basic circuit - I modded it to suit my needs).

Have a look around the Cursio Audio Engineering Website.

You'll turn up several examples like:

Also somewhere here there was a thread (do a search) which talked about LM317 based high voltage regulators and SY posted a link to the original National Semiconductors Application Note upon which these high voltage regs are based.

These things are usually called "Maida" regulators after Michael Maida the aithor of the NS Aplication Note. So perhaps search on "Maida".

I've only used the design to regulate the front end supply (approx 10mA current draw at 285V) BUT it worked very well for that.

how about if I just use a zener diode across the input and output pins to prevent >40v? Furthermore, I am using a dht rectifier, so voltage rise will not be that fast.
One reason I am looking at chip regulator is because I cannot afford to have too much voltage drop on my supply, maybe 20v maximum. Discrete regulators are complex and requires too much drop and current drain.

Ian, Andy,
thanks for the link. will take a look on that.

I am using 317 to regulate my 2a3. But look for the t-suffix, because the rest does not have enough current output to support the 2a3 tube. Also, use a reasonably large heatsink because the regulator will be running at maximum capability.
At least 10,000 uF is required before feeding into the regulator as well.

Thanks for the help guys!
Hi Skyraider,

If you only plan to have ~20V across it, then a 33V or so zener across the regulator may work (with a series resistor before the regulator of course!). However, something to consider is that with such low difference (20V) between input and output can be asking for trouble - at such high voltages, 20V is only 20/300 ~ 6% of your B+ rail. Mains is allowed to vary by +10/-6%, so you would be right on the limit (over, in fact if you include the ripple voltage).

Another problem is if the voltage rises a bit too much - your zener may be conducting most of the current to the output - at ~200mA *33V this is ~ 5.5W, quite a lot! Also, would the series resistor actually be ok? Just thinking of it, the resistor would normally be passing all of the current, when there is too much voltage, the regulator still maintains the same output voltage, hence the current draw is the same. Thus the voltage drop across the resistor is the same - i.e. the input voltage rises and so does the voltage at the 317's input. This configuration won't work - there is nowhere for the zener to conduct extra current to bring the voltage at the 317's input down. Since the zener is 33 (or so) V and the input to the 317 cannot be pulled down, the output voltage will be this input minus 33V, i.e. the 317 is no longer regulating!

Just going over your post, you say you are using a valve rectifier. Why not use diodes (since you are using a silicon regulator), which will give you more volts to play with, and use a series pass transistor (or two as you will most likely need) - it is only 2 transistors, a resistor and a zener extra to connect! Whilst this will mean you lose any warm up delay inherent with the valve rectifier, you can just build a simple timer to switch on the B+ ~20 seconds after the heaters are turned on.



2003-01-01 7:21 pm
I am using 317 to regulate my 2a3. But look for the t-suffix, because the rest does not have enough current output to support the 2a3 tube. Also, use a reasonably large heatsink because the regulator will be running at maximum capability.
At least 10,000 uF is required before feeding into the regulator as well.

Thanks for the help guys! [/B][/QUOTE]

Actually I was asking something different, the question was if the 317 could be used to regulate ac-ac, in stead of ac-dc for the filaments. The datasheet has an example of the 317 used as an, ac regulator. I was wondering if this could be done with dht's in order to reduce the noise you can typically get with ac heating........

SY said:
Yes, and it's a valid approach. But active regulation, trading off headroom, gives a rail which stays at a constant voltage. Additionally, the lower output impedance attainable at low frequencies makes the regulated stage more effective at decoupling. Motorboats need not apply.

You really notice the benefit from regulation -- at least for those of us who are more interested in measuring things than listening :) -- in the low power consuming state -- this is where the absence of ripple is made apparent.

Sy: I had Levente Szasz for physics, I wonder if they were related?
So people like Jones and Curcio are just complete idiots to use those in their designs?

That's a matter of your interpretation of 'idiot' as related to amp design. I built 'Daniel' the same year as the article in AA came out and was monumentally disappointed. The fact that he had emptied an AR SP6 case to house the design was impressive but the sound was not. We obviously do not share the same sonic taste.