LM317 With 46v supply - is this okay?

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
Working on an old Harman Kardon 75+ and the 30V supply is flaking out.

This "regulator board" is fed from a rectified power source of aprox 46 volts. It's sole responsibility is:

1) Provide regulated 12vdc
2) Provide regulated 30vdc
3) Control the "stereo tuned" indicator bulb (12v) via a simple transistor (base controlled from the tuner).

I am basically deleting the board and replacing it with my own. I have no need to keep the functionality of #3 because I have replaced the old indicator bulb with an LED. Ahh technology.

Now... for #2... I need 30 volts DC. I understand (correct me if I am wrong) that there is (theoretically) no voltage limit of an LM317.. just a differential limit. So, am I correct to say that using an LM317 in this application is safe?

Regarding #1, I plan to feed a SECOND lm317 from the 30vdc out of the first in order to keep a safer margin of error. Again, am I on the right track here?

The total current requirements of the 30vdc is less than .05ma. The 12 volt is even less and only used for the AM tuner. (it was originally ALSO for the stereo indicator, but adios to that!).

This board originally used 3 bjt's, a couple zeners and a handful of other passive components. I mainly want to completely refresh this using parts I already have on-hand and simplify the design in the process.

- Dean
 

crutschow

Joined Mar 14, 2008
25,446
Yes, it's the differential voltage that's important to the LM317.
The problem is that during startup or a momentary output short, the LM317 differential can be exceeded and zap the regulator.
Use an LM317HV, at least for the first regulator, which is good up to a 60V differential.
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
I have used there with great success.
https://www.recom-power.com/pdf/Innoline/R-78HBxx-0.5_W.pdf
Make sure yo get the "H" variety as the have a 72V max i/p voltage.
The advantage is efficiency, and no heatsink required.
Yes, normally I would jump all over something like that vs a linear regulator.. BUT...

This is a vintage 70's receiver and I am concerned about switching noise. Kinda wanna stick to something linear.
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
Okay.. further looking into this, can someone take a look at the attached schematic?

Notice capacitor C032 and C033? Now take a look at how the rectified output is sent to them both. The +46V hits C032 first. From there, it is run through R03 (10 ohm) before it can charge C033. Now, the actual regulator board was connected to C033 (after the resistor).

So what I have done (so far) is connected an old analog multimeter to the 46V feed (that goes into the regulator board) and powered on the receiver. Because of the capacitor/resistor/capacitor network, there is no actual spike. The voltage takes about 1 second to climb to full 46v.

Given this setup, I have been going over-and-over how the voltage comes into the LM317 and I would imagine that 1 second slow-climb should give it plenty of time for the circuit to equalize out and prevent any spilking. No?

In case anyone DOES happen to look at the schematic, I will give an outline of what this regulator board is doing.. When powered on - it does not go straight to the required 30 volts. It starts at 15, perhaps 20.. and slowly (over a minute or more) climbs until almost 36-37(!!) volts. At times, it drops back out (mainly when cold) or dips down to 18, 20, etc. Since this powers ONLY the pre-amp section it causes a serious issue for my speakers. Because as it is dropping or climbing, it is causing a huge DC offset between the pre-amp and the amplifier stage... pulling in or pushing out the speakers causing a huge DC spike to the speakers. Since the only protection this amp has against such situation is fuses (no protection relay, clamping circuit, etc), this can be bad on the speakers and I have already smoked 2 bench speakers (cheapos just for testing stuff like this).

Any help would be greatly appreciated.

When I got this amp, almost all the transistors on the pre-amp board were damaged (working, but the audio was really choppy and only came through at high volume levels). As though the transistors were not properly operating in their "active region" - they were kinda acting like "completely on or completely off". Now, that I replaced all pre-amp transistors and caps, the amp (and preamp) sound amazing. Just have to wait for the regulator board to settle down (usually evens out and stops fluctuating after 2-3 minutes).

My experience is more in the MCU realm, analog circuits in general are my weaker point. And in neither situation am I an electrical engineer. I understand a lot of theory, match transistors, figure out substitutions and check many components, but this aging board kinda had me saying "to hell with this old garbage". Especially since figuring out substitutions in wake of the zener/trans (current buffer?) is certainly over my head.

View attachment 75.pdf

MOD:
Clipped the PSU section, easier to follow. E
AA1 28-Sep-18 07.47.gif
 
Last edited by a moderator:

Dodgydave

Joined Jun 22, 2012
9,468
If you want to go with Linear regulators, then use a Lm317HV for the 30V, and a LM7812 for the 12V, bear on mind that for every 1A drawn the 317 will be dissipating 16W wasted heat , and the 12V will be dissipating 18W!

Personally i would use buck down regulators, like LM2596.
 

AnalogKid

Joined Aug 1, 2013
8,608
Running the two 317 regulators in series (as in the original schematic) will work well. The only issue might be that, compared to running both regulators directly off of C033, the 30 V regulator will run warmer because the 12 V output current is going through it, and the 12 V regulator will run cooler because of the lower Vdiff.

Either way, both outputs will have less hum than the original circuit.

ak
 

bertus

Joined Apr 5, 2008
20,773
Hello,

What is the current of the stereo indicator lamp?
That is also connected to the 12 Volts.
When the stereo indicator is activated, its current will add to the 12 Volts AND the 30 Volts.

Bertus
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
Hello,

What is the current of the stereo indicator lamp?
That is also connected to the 12 Volts.
When the stereo indicator is activated, its current will add to the 12 Volts AND the 30 Volts.

Bertus
The stereo indicator is a non-issue. Originally it was powered from the 12vdc. It was connected through a small npn transistor and the base of that transistor went to the FM board. The FM board outputs 4-ish vdc and can easily source the 20ma needed to directly drive a small LED.

So literally, the only thing the regulated 12vdc does [now] is power the AM/FM tuner board.

--UPDATE--
So... I whipped up a simple dual-lm317 board with what I had. I added a small indicator LED to each (12 and 30 volt) side of each regulator so to ensure there is always *some* minimum current draw on the circuit (which I think is important to ensure correct regulation?). Added a 1000uf capacitor to each regulated end as well.

Works great except... At startup, I get a STRONG negative DC offset (2 seconds) followed by a strong positive DC offset (about the same).. Each offset basically "rolls away" (speaker cone slowly comes back to center).

Given the completely "dual mono" construction of this amp, I cannot see this being an issue of the amp boards or pre-amp (directly). Rather, I have a feeling it is the way my new regulator board "comes up"... I have a feeling that - like the capacitor/resistor/capacitor network just before it, that part of that regulator board circuit is an additional "slowly come up" delay that gently brings the 30vdc up. Originally, these receivers usually take 5-8 seconds before any sound is heard.. And given there is no "soft start" or other protection elsewhere, I would think this responsibility (and original feature) landed on the regulator board somehow...

Thoughts?
 

AnalogKid

Joined Aug 1, 2013
8,608
At startup, I get a STRONG negative DC offset (2 seconds) followed by a strong positive DC offset (about the same).. Each offset basically "rolls away" (speaker cone slowly comes back to center).

Given the completely "dual mono" construction of this amp, I cannot see this being an issue of the amp boards or pre-amp (directly). Rather, I have a feeling it is the way my new regulator board "comes up"... I have a feeling that - like the capacitor/resistor/capacitor network just before it, that part of that regulator board circuit is an additional "slowly come up" delay that gently brings the 30vdc up. Originally, these receivers usually take 5-8 seconds before any sound is heard.. And given there is no "soft start" or other protection elsewhere, I would think this responsibility (and original feature) landed on the regulator board somehow...
You are correct. R902. C901, C902 form a 20 second time constant exponential ramp that is clipped at around 32 V by CR901. A typical LM317 circuit comes up much faster, before all of the amplifier capacitors have had a chance to charge up and stabilize. The charging time to a full 30 V output is a little more than one R-C time constant - one TC is 63.2%, and 32/46 = 69.5%.

ak
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
You are correct. R902. C901, C902 form a 20 second time constant exponential ramp that is clipped at around 32 V by CR901. A typical LM317 circuit comes up much faster, before all of the amplifier capacitors have had a chance to charge up and stabilize. The charging time to a full 30 V output is a little more than one R-C time constant - one TC is 63.2%, and 32/46 = 69.5%.

ak
Thanks AnalogKid!

So, sorry to bug everyone, but I want to be sure I understand this - because zener diodes and how they work in regulated power supplies is certainly not my area of expertise.

I am attempting to understand how the original circuit worked...

The R902. C901, C902 network (originally) would slowly come up until they hit the zener voltage (CR901/RD35A/36V), and at the specified 36 volts, it would start shunting the dual-transistor-cascade to ground - stopping current until voltage fell again and then opening up once more?

Next, it seems R905 (100 ohm/2W) was used to further take some of the grunt of droppig from the 36v -> 12v before hitting the next transistor. Kinda to help dissipate some of the heat?

From there, R904/C904/CR902 allow Q904 to be wide open until the zener voltage of 13v is reached - at which point, again - the transistor is shunted to stop current flow until voltage falls below the threshold..

I also suspect that that first R03 is vital in the power regulation as well? As it serves part of the current limiting network for the zener/transistor/resistor network (for the 36 volt) on the regulator board?

Now that I have looked up each part.. it seems
-> B5 should be 36 volts (not 30)
-> B6 should be 13 volts (not the 12.1 I currently have it running at)..
I suspect that this may be why the tuner is now not pulling in any stations (just static).. That extra volt must be important.

For the soft-start, I may stick a small Arduino pro mini in here and a couple relays. This way, I can also monitor for sustained DC voltage (after startup period) and cut power to the speakers as a safeguard. I buy them things 20-a-time via China for less than $2/each.
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
Okay.. so soft-start aside..

-> Changed the "12v" regulator resistors: 330/3.22k == 13.44 volts
-> The "30 volt" regulator is as-was: 330/7.5k == 29.65
-> Radio works now
-> Confirmed B5 is 29.6 volts
-> Confirmed B6 is 13.4 volts
-> New regulator board is receiving exactly 40.5 volts when all is powered on
-> The above 40.5 volts drops to about 38.4 volts when FM is on

Have carefully measured the current BOTH circuits (while radio is on)
-> B5 circuit (29.6 volt) is drawing exactly .143 amps
-> B6 circuit (13.4 volt) is drawing .08

But these buggers are getting hot. Almost too hot to touch (the LM317). Mainly only when the FM radio is on.

So please tell me if my math is correct:

-> B5 is at 29.6 volts, from ~40 volts, it drops ~10 volts
-> 10 volts at the max measured current of .143... lets call it at .2 is (10*.2 is 2 watts)
-> B6 is at 13.4 volts, from 29.65 volts, it drops ~16.25 volts
-> 16.25 volts at the max measured current of .08... let's call it at .1 is (16.25*.1 is 1.6 watts)
-> A total of 3.6-4 watts will be dissipated by this circuit - correct?

Attached is a photo of the new setup. It is - literally - attached to the same spot the previous board was and the two LM317's are attached to the same two holes in the chassis (18 guage steel) that the original two transistors were mounted to (for heat dissipation). When mounted, thermal paste was used, and all the tab insulation hardware was included so there is no short to the chassis.

Should I re-locate these to a real heatsink?

- DeanIMAG0117.jpg
 

AnalogKid

Joined Aug 1, 2013
8,608
At those power dissipation levels, a TO-220 package will melt fingerprints. Depending on the internal construction, 1.5-2.0 W is the absolute max dissipation it can stand in +25C free air with no heatsink (1970's thermo class). The die is in the 125-150C range, and the case is over 100C.

For Q902 and Q904, what are the package types and how are they mounted?

ak
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
The original Q902 and Q904 were also TO-220 packages. Hence, I literally mounted the 2 LM317 chips exactly where the old transistors sat.

As an update... I robbed the 100-ohm (2-watt) resistor that was originally on the previous board and added it BETWEEN the 30v and 12v regulators. This way, - just as in the original - the 12v regulation gets about a 10 volt drop (via the resistor). This knocks out an entire watt of heat dissipation.

Overall, the original design seemed kinda poor and not very reliable. Then again, this WAS the 70's. I mean.. my implementation -vs- the original.. The same amount of heat has to be dissipated, regardless of how it is done. Linear is still linear, right? So I cannot imagine the heat generated WITH the stereo light before.

I still need to figure out a way to mute the speakers for 10-15 seconds, the more I think about it, the more I may say "to hell with it" and find a good PWM-operated regulator. Because, I am likely going to control the soft-start with a small cheapo atmel MCU.
 

Thread Starter

Dean Rantala

Joined Sep 27, 2018
27
Okay gents.. so, just wanted to update on my progress, and check a few safety margins.

So, after getting my "ramp up" circuit working (thanks to everyone on the other thread who helped me!), I am about to wrap this project up and - before I do that - I would like to get some professional advice.

1) In theory, this circuit should not dissipate any more heat than the previous circuit.. Right? I added [the original] 100 ohm (2 watt) resistor between the 30 volt output and the input of the 12 volt regulator. Now, the LM317's are not heating like they were, but that resistor is really how. Like.. around 70-80C hot. Looking at the previous circuit, this seems to be essentially the same heat it originally would have put off. What do you all think? Okay at this temp?

2) The two LM317 regulators at full load (radio on, pre-amp up and music playing after 10 minutes) get to around 60-65C. Are these running in a safe margin? I looked over the datasheet and - again - in theory, it should be fine. But rather check.

- Dean
 

dendad

Joined Feb 20, 2016
3,602
The circuit will dissipate just the same heat as before. All you have done is to spread the heat over a couple of components now. The only way to run cooler, with the same voltages and currents, is to go switchmode.
 
Top