7815/7915 dual supply hum problem
A friend’s Cayin ss amp suffers from audible hum. The hum is caused by a narrow pulse of about 1 mV amplitude and 10 ms intervals. The pulse generates a rich spectrum making the hum objectionable.
Output stage and front end have separate power supplies. Output stage supply is quiet, but the +- 65 volt front end supplies carry the narrow 100 Hz pulses. Voltage is stabilized by 7815/7915 regulators on top of 48 volt zeners. Zeners are decoupled by film caps and outputs by 470 µF electrolytics. Input DC is 70 volts.
My scope shows a 100 mV (pk-pk) triangular input ripple and a 1 mV narrow pulse at the output. The pulse is also present at the adj pin. The pulse occurs exactly when input ripple voltage has its peak value.
I fail to find a probable explanation to this phenomenon. I have tried a few decoupling schemes like small electrolytics over the zeners, small ceramics at input and output pins, to earth or adj pins - all with no effect. Since decoupling does not seem to alter anything, I’m beginning to suspect the behavior is a property of the chip or the circuit design. Symptoms are the same on both rails; except the pulse on the positive rail is somewhat lower in amplitude than on the negative.
I fail to see why a pulse should occur at regulator outputs when ripple voltage has its peak value (on both positive and negative rails)? What am I missing here? Any idea or suggestion is gratefully appreciated…
Best wishes,
Lennart
A friend’s Cayin ss amp suffers from audible hum. The hum is caused by a narrow pulse of about 1 mV amplitude and 10 ms intervals. The pulse generates a rich spectrum making the hum objectionable.
Output stage and front end have separate power supplies. Output stage supply is quiet, but the +- 65 volt front end supplies carry the narrow 100 Hz pulses. Voltage is stabilized by 7815/7915 regulators on top of 48 volt zeners. Zeners are decoupled by film caps and outputs by 470 µF electrolytics. Input DC is 70 volts.
My scope shows a 100 mV (pk-pk) triangular input ripple and a 1 mV narrow pulse at the output. The pulse is also present at the adj pin. The pulse occurs exactly when input ripple voltage has its peak value.
I fail to find a probable explanation to this phenomenon. I have tried a few decoupling schemes like small electrolytics over the zeners, small ceramics at input and output pins, to earth or adj pins - all with no effect. Since decoupling does not seem to alter anything, I’m beginning to suspect the behavior is a property of the chip or the circuit design. Symptoms are the same on both rails; except the pulse on the positive rail is somewhat lower in amplitude than on the negative.
I fail to see why a pulse should occur at regulator outputs when ripple voltage has its peak value (on both positive and negative rails)? What am I missing here? Any idea or suggestion is gratefully appreciated…
Best wishes,
Lennart
Hi jackinnj, thanks for your response.
I’m sorry for being unclear; this is a commercial amplifier and I’m not trying to anything other than finding the cause of the hum (which I btw suspect to be a design flaw).
I don’t know about current consumption, but the regulator is mounted in free air with no heat sink, it is not warm to the touch, it provides voltage to the front end only, and it can supply 1.5 amps – so my guess would be that the regulator is not current limiting.
The zeners are biased by a resistor from the output, and yes, the transformer has its center tap grounded. Actually, I can detect nothing unusual on the PCB. Everything looks “normal” to the eye, except I would have used another circuit for front end voltage supply.
I’m sorry for being unclear; this is a commercial amplifier and I’m not trying to anything other than finding the cause of the hum (which I btw suspect to be a design flaw).
I don’t know about current consumption, but the regulator is mounted in free air with no heat sink, it is not warm to the touch, it provides voltage to the front end only, and it can supply 1.5 amps – so my guess would be that the regulator is not current limiting.
The zeners are biased by a resistor from the output, and yes, the transformer has its center tap grounded. Actually, I can detect nothing unusual on the PCB. Everything looks “normal” to the eye, except I would have used another circuit for front end voltage supply.
Sounds like you may have a problem on the return side of the supply. 1mV of ripple should be insignificant at that supply voltage.
OR, the problem is at one of the I/O jacks. Sometimes there is a low value ceramic between shield and ground on one of the jacks. If this part is broken, the unit will hum.
OR, the problem is at one of the I/O jacks. Sometimes there is a low value ceramic between shield and ground on one of the jacks. If this part is broken, the unit will hum.
Hej, Lennart! I think your problem is related to the ground and the return path of ground currents. The ripple rejection of a 7815 is at least 54 dB and you have 40 dB right now. How about investigate where the ground connection of the smoothing caps?
When you measure it's also important where you put the ground cable of your oscilloscope probe.
BTW: Why don't you have your homepage with your beautiful tube amps published here?
When you measure it's also important where you put the ground cable of your oscilloscope probe.
BTW: Why don't you have your homepage with your beautiful tube amps published here?
Possible explanations:
1 - grounding. The ripple peak also corresponds to peak charging current into the reservoir cap, so it does not take much ground impedance to generate 1mV.
2 - induction. The charging pulse loop is inducing a spike into the regulator output loop.
3 - servo bandwidth. Any regulator has a finite bandwidth, signals above this frequency will be passed through with less attenuation.
My guess is 1 or 2.
1 - grounding. The ripple peak also corresponds to peak charging current into the reservoir cap, so it does not take much ground impedance to generate 1mV.
2 - induction. The charging pulse loop is inducing a spike into the regulator output loop.
3 - servo bandwidth. Any regulator has a finite bandwidth, signals above this frequency will be passed through with less attenuation.
My guess is 1 or 2.
Hi all,
I'm truly amazed. This was five years ago, almost to the day
Thanks for your response anyway! And yes, this was a grounding problem. Too many commercial amps based on large PCBs fail to correctly manage ground systems. My own designs always feature three ground systems; signal ground, power supply ground and chassis ground. Treated separately, and correctly joined when the amp is finished, they can almost guarantee quiet ground systems.
peranders: Good question. My explanation is expressed in terms of available time. I have spent way too much time on local forums, but this is subject to change...
I'm truly amazed. This was five years ago, almost to the day
Thanks for your response anyway! And yes, this was a grounding problem. Too many commercial amps based on large PCBs fail to correctly manage ground systems. My own designs always feature three ground systems; signal ground, power supply ground and chassis ground. Treated separately, and correctly joined when the amp is finished, they can almost guarantee quiet ground systems.
peranders: Good question. My explanation is expressed in terms of available time. I have spent way too much time on local forums, but this is subject to change...
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.