L7809C Pedal Power Supply Noise

[camsy83]

A number of years ago I made a 10-way pedal power supply to cut out ground loops and reduce noise.

The problem is, my power supply introduced a whole lot of noise.  So a less than successful project, shelved... until now!

The regulator I've used is an L7809C.
The introduced noise is a static hiss.
The datasheet(PDF P18) states "Output noise voltage; B = 10 Hz to 100 kHz, TJ = 25 °C; 70 µV/VO" which adds up to 630µV at 9V VO

Here's the circuit, ten of these in the enclosure:


Some oscilloscope plots to quantify the noise -
Plugging the supply into a pedal, connecting to the guitar amp, and scoping across the amp speaker output:
Pedal off, baseline (spectrum, averaged, 0-100kHz):


Pedal on (spectrum, averaged, 0-100kHz):


Superimposed for easier comparison (spectrum (not averaged), 0-100kHz, blue - pedal off, red - pedal on)


The noise I'm experiencing may be an inescapable fact of the L7809C that younger-me decided to use without taking the noise spec into account...
I realize the input to the regulator is very high - younger-me neglected to account for the fact that the 12v from the transformer secondary was RMS.

Does anyone have any ideas for curing this power supply induced hiss?

My plan-B is to use an LM350 instead of the L7809, with noise stated as 0.001% VO, totalling 90µV, 1/7th the noise (with the bonus of possible variable voltage). But if a few extra passives can cure it, I'd prefer that!

[Edited to correct my images]

[ElectricDruid]

630µV is only 2/3rds of a millivolt, and there's no way that's "a whole lot of noise". So I think there's something else going on. If the regulator is operating according to the spec, the noise shouldn't be objectionable.

But you said you're measuring the amp output across the speaker? What makes you think the noise you're seeing is *anything* to do with the pedal power supply? It could be the pedal itself, pre-amp, the power amp, the amp's power supply, the substation in the neighbours garden, etc etc. There's a lot else going on beyond the pedal power supply.

I'd put your scope trace on the 9V output from the power supply. See how much noise you've got there (My guess - less than 1mV, per the spec). If that's fine, your noise is coming from somewhere else.

HTH,
Tom

[camsy83]

I think the amp being the source of the noise can be ruled out, as turning the pedal off cuts the noise completely.

That leaves the pedal or the power supply, right?

I get the same result with a few different pedals which leads me to suspect the supply. From memory, the cheapo wall wart I used to use didn't have the hiss problem, but I can't confirm right now because I can't find it [as an aside, I stupidly fried an op amp with a wallwart I did find, forgetting they're supposed to be negative-tip... whoops!]

Here are a couple of timebases of the supply output:
(please note this is the AC component of the DC output only, due to the resolution limitations of the scope)




And the spectrum (linear, peak, to 12kHz):


What do you reckon, is this pretty clean supply? I don't have a benchmark to compare against.

[camsy83]

Also, here's the input to the regulator (again, the AC component only):

[ElectricDruid]

What do you reckon, is this pretty clean supply? I don't have a benchmark to compare against.

I think that's a pretty clean supply, yes. From  those graphs, you've got about 10mV ripple on the output and some level of noise much less than that - perhaps 1 or 2mV. That doesn't sound serious to me.

The only interesting thing is the spike around 9KHz on the spectrum. It's not clear where that's come from - although a linear regulator seems an unlikely culprit. It's also close-but-not-the-same-as the 7KHz spike you had on the graphs from the amp.

HTH,
Tom

[camsy83]

I feel exonerated in my component choices now, after the bitter disappointment when I first plugged it in after building it a few years ago! Really need to find that old wall wart for a proper comparison

The multiple different timebases and frequency ranges of my plots have confused the issue, but the spikes on the power supply are around 9kHz, whereas the band on the amp output was around 70kHz.

[davent]

Have you tried powering the pedal(s) with a 9v battery? A battery snap wired to a 2.1mm power plug or whatever you're using if the pedals won't take a battery internally.

dave

[Rob Strand]

FYI, you need to measure the ripple with full (or max intended) load.  When you increase the load a number things happen, the average voltage drops, ripple increases, the regulator drop-out increases.  That can push the minima of the input waveform below the dropout point of the regulator.

I've seen cases where the device works fine with a few low power effects but you plug in a delay effects pedal and you get hum due to the increase load.

[ElectricDruid]

+1 agree. Rob's got a good point. L7809 is 100mA max, and I wouldn't want run it at that. Now, 100mA is quite a lot for most pedals, but there are some that would get close.

[Rob Strand]

+1 agree. Rob's got a good point. L7809 is 100mA max, and I wouldn't want run it at that. Now, 100mA is quite a lot for most pedals, but there are some that would get close.

I was actually talking about dropout due the dips in the ripple.  The L7809 is something like 1A or 3A but the 78L09 is 100mA.  Confusing as hell.  The only reason I know the L7809 is big is I've pulled some out of printers.

[camsy83]

You may consider me confused and embarassed... Powering with batteries results in the same hiss. I don't have any 9v batteries, but have tried with 2x Li-Ions and 3x Li-Ions in series, so that's tests at approx. 8v and 12v, both give the hiss!

My memory is obviously faulty, I was positive that both battery power and wall wart power was better than my DIY supply. I really need to locate that wall wart, and get some 9v batteries to properly convince myself.

The chips are definitely L7809, not the 78L09, rated for 1.5A. (Though I wouldn't want to go anywhere near that when I'm dropping so many volts across the regulators, I think they'd get pretty hot)

The plots provided for the supply output were while powering a cloned compressor pedal. Not sure of it's power requirements, but I'd guess it's up there (and I do mean guess), with 7x 2N5088 and a CA3080.
The input ripple plot was with no load, though I wouldn't expect a difference.

Interestingly, a delay pedal (PT2399 based) didn't sound so bad.

Any testing is on hold for now, as my amp's power transformer has packed in!

[ElectricDruid]

+1 agree. Rob's got a good point. L7809 is 100mA max, and I wouldn't want run it at that. Now, 100mA is quite a lot for most pedals, but there are some that would get close.

I was actually talking about dropout due the dips in the ripple.  The L7809 is something like 1A or 3A but the 78L09 is 100mA.  Confusing as hell.  The only reason I know the L7809 is big is I've pulled some out of printers.

Oh, sorry, my mistake. As you say, confusing as hell, and it got me confused.

[Rob Strand]

Interestingly, a delay pedal (PT2399 based) didn't sound so bad.

Sometimes you need two pedals.  If you have circuit which is sensitive to the supply, something with transistors, like say a fuzz face, you can hear the hum more easily.   The delay loads the power supply and causes the regulator to drop out and the fuzz-face lets you hear that has happened.  Sometimes you don't need a fuzz face and you can hear it with only the delay.  It comes down to the specifics of the pedals.  The old boss pedal can hide the hum of the power supply whereas the new ones and the ibanez ones are less forgiving.

[camsy83]

My years-ago tests were with a bunch of pedals daisy-chained on the wall-wart, so maybe my DIY supply is much worse when pushed a little with multiple pedals.

Actually, that doesn't make a lot of sense, the whole point of it is that I have ten separate ground-isolated power supplies, so each will only be supplying one pedal...

I hope to be back with more once I have my amp back on it's feet, and once I find the missing wall wart.
OT: I've identified a rectifier diode that is reading 1.5Kohms in both directions  , so it shouldn't take too long, provided I haven't cooked the power transformer...

Edit: Typo

[camsy83]

So, problem solved! Details follow for anyone interested

First, apologies to everyone trying to help - the hiss I was asking about in the OP is in fact very quiet relative to signal, and not a problem.
The problem I actually had years ago (it's all coming back to me now...  ) was mains hum. This hum was at 'normal' levels when my pedals were powered by wall-wart or by battery. But the hum was atrocious when powered with my diy supply. I couldn't figure it out, so I gave up and moved onto other projects...

Now, in my new probing and testing, I found that if I disconnected the power supply chassis earth, the hum disappeared. This is not a solution for obvious reasons, especially considering the power supply case is made from folded stainless steel sheet! But it did lead me to the solution.

Voltmeter between the chassis and the mains earth - 12v!
12v DC, AND nearly 12v AC?? Also the power supply circuit boards are isolated from the chassis with mounts made from rubber hose. Very confusing.

Separate one of the boards from the chassis and test. No hum!

After much head scratching, ohmmeter test the rubber hose. Probing the OUTSIDE of the hose, as is accessable with everything built up reads open circuit, >20Mohms
However, probing through the cross-section... 500 ohms!!

Well, there's your problem.

Now probing between chassis and the transformer secondaries, around 500 ohms for each.

Just so happens that with my board layout each piece of hose was in electrical contact with one side of each pair of transformer secondaries, connecting them all to the chassis, resulting in the terrible hum.

Cut one veroboard trace for each of the ten supplies. Problem solved.

[Rob Strand]

So, problem solved! Details follow for anyone interested

Thanks for posting.

FYI, If you need to ground there's the cap + resistor + diode network:

http://sound.whsites.net/earthing.htm

[ElectricDruid]

Yeah, absolutely agree - thanks for the follow-up. It's the kind of difficult-to-track-down problem that seems to turn up far too frequently, so having one that actually got fixed is (a) encouraging and (b) gives other people an idea what to be looking for.

Nice work!

T.

[snarblinge]

Was just confirming my setup for a similar supply, and a word of warning, punching l7809 into safari produced an inappropriate result!!