Hi there !

Please be so kind to help me with a Power Supply design problem for an analog audio mixer.

I have a Behringer QX2442 analog mixer, for which I have made a linear power supply instead of it's inbuild SMPS.

This mixer needs +12V for the effects unit, +15V -15V for the sound mixing section (with OP-AMPS), and +48V for Phantom Power.

I have ordered a custom made toroidal transformer with 230V input (I live In Europe) and secondary side with a 12VAC winding, center tapped 15VAC winding (15VAC-CT-15VAC), and 48VAC winding. There is a screen between primary/secondary, and a screen for the 48V winding in the secondary. This transformer is well build, and can handle the mixer's current requirements with ease on all windings. I have measured it's power consumption for all lines while powering it up through laboratory PSU's, so I am sure there is no insufficient current issue in my problem.

I made a power board pcb, that consists of fuses for each transformer AC output (except center tap), 3 encapsulated bridge rectifiers (8A, I had them at hand), filter caps, and linear voltage regulators ICs.

For the 12V line I've used L78S12CV, ground to the IC provided via mixer chassis - working fine.
For the 48V line I've used TL783C, ground to the IC provided via mixer chassis - working fine.
For the 15V lines I've used L78S15 and L7915CP.

The problem is with this +15V -15V line.

After the bridge rectifier I get 23VDC and -23VDC measured relative to the center tap transformer wire.
This +/-23VDC is filtered with caps: 3 pieces in parallel 35V 4700uF for the positive side, and also 3 pieces in parallel 35V 4700uF for the negative side.
Those caps have the commons connected to the transformer center tap as voltage "zero" reference, as well.
The voltage regulators work fine if I connect their ground pin also to the center tap, outputs are pristine +15V and -15V.

BUT, when I connect this +15V -15V voltages at the mixers power inputs, it all works fine, but with a lot of noise.

Inside the mixers' PCBs, all grounds are connected to chassis, and when I connect the +15 -15V, the chassis also becomes connected to the transformer center tap (because the regulator IC's GND pin is connected to center tap).

I've measured DC voltage between mixer chassis GND and transformer center tap and I get 14V DC. All schematics show that center tap needs to be connected to GND, and yeah, all the voltages are fine then, the 14V are GONE, but the NOISE makes the mixer unusable.

The "workaround" I did:

I left the filter capacitors on the +15 -15V side connected with common points to transformer center tap.
But the +15 -15 regulator IC's ground pin, I connected to mixer chassis.
Sound wise, works perfect - all noise is gone, mixer runs perfectly.
BUT
Due to that 14VDC voltage between center tap and chassis, the negative regulator IC input is at -23V-14V=-37V beyond the absolute maximum input voltage in the datasheet (-35V), but I provided a super huge radiator and it works, it outputs -15V but gets HOT HOT even with this huge thick radiator. On the other side, something happened to the positive IC input voltage, it outputs only 7V, gets only warm-ish (maybe it doesn't regulate...). So the beautiful balance of +15V -15V is ruined, I have +7 -15 and a super HOT negative regulator IC. Sound wise, works like a champ, tested it for lots of hours, no problem, but I kind of hate the idea of having a 1 kilo radiator inside the mixer case for a LM7915 IC. And that 1 kilo radiator being at about 50 deg C. and the OP AMP voltage unbalance.

How can I fix this?
Is there a way I can make a "noiseless" connection from center tap to chassis?

One possible "solution" from my side, maybe a stupid one.
Leave the transformer center tap connected to nothing. Use chassis ground also for the +15 -15 line filter caps, but like this I've measured +37 and -37V V across them, so also the filter caps need to be upgraded to 50V or better yet 63V ones.

Also believed EMI might be the issue, I've moved the toroidal transformer to an outside box, but no, not the problem because when I connect center tap to chassis - again - the same horrible noise.

I was also wondering, if I'm not fooling myself, and maybe my mixer has some bad caps inside, which make it's chassis not really 0V. But I have changed all OP amp power lines caps inside the mixer with brand new ones, before mounting my PSU. On the other hand, if this were the case, why are the +12 V and +48V lines super in spec, because if chassis would not really be zero, I would have problems on those regulator IC's as well.

And yes, the mixer chassis is not connected to the 230V mains EARTH pin, because the transformer is now in a separate plastic box, and there I've connected the transformer safety wire and screens to Earth. Screens/safety wire connected to Earth or not connected to anything - it's not the source of the noise, also tested that.

Please guys, be so kind to help

 

Post a complete schematic of the 15 volt plus and minus supply please.

 

Here is the schematic.

 

What kind of noise? 50 or 100Hz would have a totally different cause than high frequency hiss.

 

What was unacceptable about the original built-in Beringer-Power-Supply ?
I could understand building a custom Power-Supply if You were wanting to run it on 12-Volts DC,
but the Beringer comes with a perfectly serviceable power-Supply, with almost zero noise.

It looks like your problem is that You are using a Positive-Regulator for the Negative-Rail.
Or it may be that it's just wired incorrectly.
The Schematic is not very clear.
.
.
.

 

What kind of noise? 50 or 100Hz would have a totally different cause than high frequency hiss.

It is a mixed thing; hum and hiss. I am very aware of the difference between hum and hiss, I am sure it is both, more on the hum side.

 

What was unacceptable about the original built-in Beringer-Power-Supply ?
I could understand building a custom Power-Supply if You were wanting to run it on 12-Volts DC,
but the Beringer comes with a perfectly serviceable power-Supply, with almost zero noise.

It looks like your problem is that You are using a Positive-Regulator for the Negative-Rail.
Or it may be that it's just wired incorrectly.
The Schematic is not very clear.
.
.
.

Hello,
FIRSTLY,
Original SMPS always developed background noise after a while, the caps had to be changed again and again, although I've recaped with Japaneese Nichicon for high ripple, after a while (3 yeras, or so) the SNR would increase again.
Also, from the start, this factory PSU I think was pushed to the limit, it always made the board hot, as it was very hot at all times.
Also traces were very poorly designed, in the secondary, extremely thin, the pads were metalised and the holes were drilled without any little tolerance, new caps barely entered the holes, old ones were a pain to get out...too much heat could damage the pad/trace, I always monitored the soldering station temperature very close while working on the unit. Also there was poor insulation on the mosfet radiator in the primary which was live...so I hated this PSU - cheapo as can get.
SECONDLY,
No, L78S15 is indeed a Positive regulator and 7915 the negative one, which you are right has a different pinout order, and also you need to isolate the radiator, since radiator tab it is not ground as in the positive regulator, here is INPUT. So I am 100% ++ sure I've connected them correctly, radiator insulation is perfect - I am not a newbie, but I am puzzled by this noise I am having. It's very loud (50Hz hum and sizzzling hiss).

 

You noted on the schematic, that when you break the circuit at the red X, the noise stops.
Is that both the 50Hz and hiss that stops?

Can you remove the connection to the chassis, to see what that does?

 

You noted on the schematic, that when you break the circuit at the red X, the noise stops.
Is that both the 50Hz and hiss that stops?

Can you remove the connection to the chassis, to see what that does?

Hello Sir,
Yes, all noise stops. If center tap of transformer is disconnected from chassis, ALL Noise stops, but the voltages go crazy as I've described in the post, because the regulators use chassis ground, instead of the center tap from the transformer.
Across center tap to chassis, I have a 14V DC voltage drop, which messes everything.
This means that transformer center tap is at a different potential than common mixer ground (chassis), so this voltage drop appears.
If they are connected together, voltages are ok, there is no more a voltage drop, but chassis is then at the potential of the center tap, and so, this horrible noise appears.

 

It looks like your problem is that You are using a Positive-Regulator for the Negative-Rail.
Or it may be that it's just wired incorrectly.

OR, the 7805 came out first. Almost immediately, people started to use it as a negative voltage regulator by putting it in series with the ground, called "regulating the ground". This established the pinout for a 3-terminal negative voltage regulator before there were any actual devices, so the 7905 was designed with the 7805-ish pinout.

ak

 

I am going to guess that without the center tap connected to ground, neither noise nor signal is getting through since there is no common ground.

 

You noted on the schematic, that when you break the circuit at the red X, the noise stops.
Is that both the 50Hz and hiss that stops?

Can you remove the connection to the chassis, to see what that does?

Also, I forgot to mention that connection to the chassis inside the mixer cannot be broken ever, since, even some screws of the mixers PCBs have a ground plated hole, and so, connection to the chassis can never be lifted.
Maybe only if painting those metalised holes with paint, and using plastic screws, but yeah, that's not quite feasible either, because ... considering that the mixer deals with low voltages, low signals, so on - this HUGE ground plane that it gets by connecting to the chassis, is actually very beneficial, since it acts as a good overall shield as well.

 

I am going to guess that without the center tap connected to ground, neither noise nor signal is getting through since there is no common ground.

No, if the center tap is disconnected from Chassis, as you've said, there is no common ground anymore, the noise stops, everything works but the voltages go crazy.
Across center tap to chassis, I have a 14V DC voltage drop, which messes everything.
This means that transformer center tap is at a different potential than common mixer ground (chassis), so this voltage drop appears.
If they are connected together, voltages are ok, there is no more a voltage drop, but chassis is then at the potential of the center tap, and so, this horrible noise appears.

 

Hi,
First of all, what current your mixer do consume? If current is below 5 mA, regulator 7915 and 7815 may generate a "sawtooht" at the output. So, add two 3 kOhm resistors at the outputs, if current is below 5 mA.
Second, restore the broken connection 7815 GND --- 7915 GND--- transformer centertap.
Third, shunt all four rectifier diodes with small capacitors, say 47 nF 100V

 

Try making the connection to the chassis with a small resistor (e.g. 10-100Ω) and see if that makes a difference.

 

Hi,
First of all, what current your mixer do consume? If current is below 5 mA, regulator 7915 and 7815 may generate a "sawtooht" at the output. So, add two 3 kOhm resistors at the outputs, if current is below 5 mA.
Second, restore the broken connection 7815 GND --- 7915 GND--- transformer centertap.
Third, shunt all four rectifier diodes with small capacitors, say 47 nF 100V

Hello,
Mixer consumes about 800mA on the positive 15V rail, and also about 800mA on the negative 15V rail.
Restoring the connection to center tap makes the mixer unusable, the noise is AWFULL.
Yes, in the meantime, I have added some small capacitors on the rectifier bridge, 4 of them, 22nF 100V. They are ok to be there, for stability but have no influence on the noise.
So NO, direct connection center tap to chassis GND is not possible....mixer becomes unusable due to this horrible hum and hiss noise, more on the hum side.
Why does this happen? I am puzzled.....
Can it be because of the transformer itself? When I measure AC voltage on transformer windings they are precisely equal to the center tap, about 18VAC. After rectify, as I've noted on the schematic, it's 23VDC.

 

Try making the connection to the chassis with a small resistor (e.g. 10-100Ω) and see if that makes a difference.

Thanks for this tip. I'll try this tomorrow.

 

I've read on the internet, I've done extensive reasearch on this noise problem, that some analog engineer "gurus", recommend using a thermistor or a bridge rectifier to connect center tap to chassis. What's this all about? you guys have any experirnces with such practices? I've never seen them...

 

One more question- do BOTH regulators- 7815 and 7915 are isolated from the heatsink?

 

One more question- do BOTH regulators- 7815 and 7915 are isolated from the heatsink?

Hello Sir,
Yes they are. Screw isolation and mica foils with thermal paste.
I am very aware that on the negative regulator, the radiator pin is NOT GROUND, but input, so I took extra care with that one especially. Same on the positive - scrww isolation and mica foils ! All checked with multimiter - infinite resistance, working great.