120Hz spikes at output of my amp.

I don't have a preamp, just the power amp, there is no hiss, it's really quiet other than the buzz.

At this point, seems like changing the bridge rectifier is the only way to go. I don't think any RC is going to do it after all the experiment I tried.
 
It sure looks like transformer resonance to me. And 10R is such a low resistance that it might not materially reduce resonant Q. There's another possibility here as well: This noise could be coming from the outside world, through the AC line, and it could be common-mode, differential mode, or both.
 
I am going to stop saying anything as I don't know $hit on this. I have been searching and reading. I just finished reading this article that I think is very good:https://www.microsemi.com/document-portal/doc_download/14617-rectifier-reverse-switching-performance

I am reading this one and I am not going to say anything more until I go through this:http://www.hagtech.com/pdf/snubber.pdf

Anyone interested in this subject should at least read the first article, it's good. I reserve my comment on the second article as I am going through the math step by step. I am only on page 5.
 
Last edited:
That is good reading to start. I have both of those stored away on my computer.

Just as an aside, you will find that some of the science applies to snubbers used to protect switches.
 
I spent some time on the Hagtech article, http://www.hagtech.com/pdf/snubber.pdf

I cannot follow after page 5. I have no problem with the equation up to page 5. BUT it did talk about instead of trying to find the equivalent L and C from the open and shorted primary of the transformer measurement, just look at the oscillating frequency ( frequency of the ringing) which in my case is 71KHz.

But if you look at page 6, Rs is defined by either L or C of the transformer. And in Fig7, the author introduced Lt and Ct terms without defining it.

Bottom line, the paper did not connect how to find Rs from the ringing frequency. It is incomplete. Sounds like you have to measure the L and C of the transformer.

Question is how to determine Rs from the oscillating frequency.
 
He is not overly clear, he does explain it all but he jumps around. It may help to read to the end and then go back.

Yes, the inductance of the transformer is measured and you know the frequency of oscillation, so there you go. Now you have the L (Lt) and the C (Ct) of the transformer (calculated). Of course Rt is the resistance of the transformer winding.

Then it just all plugs in. If you use 0.5 for the damping coefficient and 2pi to lower the corner frequency by about 6.3, all of the higher math can go away and you can just use the reduced forms.

Again, he could have been clearer.
 
He is not overly clear, he does explain it all but he jumps around. It may help to read to the end and then go back.

Yes, the inductance of the transformer is measured and you know the frequency of oscillation, so there you go. Now you have the L (Lt) and the C (Ct) of the transformer (calculated). Of course Rt is the resistance of the transformer winding.

Then it just all plugs in. If you use 0.5 for the damping coefficient and 2pi to lower the corner frequency by about 6.3, all of the higher math can go away and you can just use the reduced forms.

Again, he could have been clearer.
I was hoping I don't have to measure the L and C of the transformer. He said you can use the frequency of oscillation (71KHz), but all his formulas involve either L or C.

I understand all the formulas assume the damping factor or 0.5 and the Rs Cs snubber frequency fo is 1/(2pi) the fn. Just he jumped and the later formulas have L and C where he said you don't need to do any measurement and just use the fn measured by the scope.

Do you have any formula or any article that is better?

Thanks
 
Another thing that complicates the situation. I read article that you find the value of snubber cap by putting different caps across the secondary and find a cap that lower the oscillation frequency by half. I went from 0.022uF all the way to 0.47uF and the frequency did not change far as I can see on the scope.
 
To be clear, you see this ringing right at your filter capacitors?

The method that he explains, although I did not get it from him, has worked well for me.
 
Does it matter? Well, if the buzzing doesn't bother you, then no. It's not going to hurt anything. If you wanted to fix it, start with a cap across ALL the diodes in the bridges. Bypassing just one, as you showed, isn't likely to eliminate the problem. Faster rectifiers, or better wiring routing might help too, but the solution usually isn't as simple as we'd all hope...

In my Nikko Alpha, for example, they had to use .01uF caps across all the diodes in both bridges. Hard to say if that'd help in your case, but it's probably the easiest experiment to try.
Capture.PNG
 
Last edited:
To be clear, you see this ringing right at your filter capacitors?

The method that he explains, although I did not get it from him, has worked well for me.
I do not have ring at the big filter caps. I only look at the output of the amp. The scope picture is at the output of the amp with no input signal.

I played with the wiring, IT'S THE +/-V and GND wires for powering the input section that pick up the noise. I can move the wires and make it totally disappears.

I was hoping to fix it at the source, but looks like I have to experiment moving the wires.

Problem is the amp is very low THD, moving these wires affect THD. Don't ask me why, it's black magic to me at this point. Now I have to experiment looking at the noise and THD and find a location that optimize both!!!!
 
That opens up more possibilities.

If you can not see it on the filter capacitors, do you believe there is enough parasitic reactance to allow the diodes/power supply transformer to ring and the ringing not make it to the power supply filter capacitors? This is not totally uncommon in the world of RF.

You need to look at all of the power supply. Do you see the ringing on the transformer secondary or possibly even the primary?

If you do not see it at the power supply, it is possible, even though it is synced to the AC mains frequency that is is not being generated in the power supply.

This might help explain the change in distortion when moving some of the internal wiring. I had to build a full Faraday screen room for my lab when I needed repeatability when measuring some very small stuff.

Then it would go to the EMC facility for validation.

So many cans, so many worms.:eek::D

Let us know what you find.
 
Update

Moving the wires work ONLY when I shorted out the input of the amp, when I hooked up the Quant Asylum but not running the FFT, the spike still there and moving the wire will not help. So far, only the schottky diode help. So back to drawing board again.
 
I am planning to try putting a 0.0uF ceramic cap across each of the 4 diodes in the bridge rectifier and see what happen. Any comment?

Also, I am going to try tucking the wires along the heatsink to get the lowest spike when I shorted the input. then put the top cover of the amp on and try hooking up to the Quant Asylum and see whether that help. The chassis including the top cover are made of thick aluminum, over 1/8", that should block the 71KHz ringing from getting out of the chassis.
 
Last edited:
Sometimes it helps to ponder the fundamental source of the noise. If the source is rectifier reverse recovery current, ponder the path that current 'wants' to flow around; then minimize its loop area and keep it away from sensitive inputs. Adding the cap across the diode is a simple means to give this fast-risetime pulse of current a small, low-loop-area path.

BTW, are you hooking the amp outputs to your QA401 differentially? just curious...
 
Sometimes it helps to ponder the fundamental source of the noise. If the source is rectifier reverse recovery current, ponder the path that current 'wants' to flow around; then minimize its loop area and keep it away from sensitive inputs. Adding the cap across the diode is a simple means to give this fast-risetime pulse of current a small, low-loop-area path.

BTW, are you hooking the amp outputs to your QA401 differentially? just curious...
I am pretty sure it's the diode turn off from the waveform of the sine wave.

No, I did not hook up to QA401 differentially, just single end and terminate the other side with a 50ohm termination plug.

Thanks for reminding me on the AC side. I kept looking at the output side. Yes the two AC line has a loop!!! The way the two leads coming out of the transformer is far apart and I wired to make the leads shorter and form a loop. Did not expect 71KHz ringing. I have to do some surgery on the wires to close the loop, I'll be back with the result later.
 
Last edited:
Thanks guys for all the help. I reduced the loop area of the secondary ( a lot of work!!!) I twisted the two wires as much as possible. I sure shows improvement. I cannot trigger the scope when I adjust the threshold beyond 7mV. That is down from over 10mV.

I am still going to try the 4 0.01uF cap. I think the reduced the loop area on both the secondary and the grounding as much as I can already. Now I have to tame the rectifier more.
 
Last edited:
Nice work! I do like the idea of the bypass caps at the source. Seems better than allowing that fast di/dt pulse to push its way into your power transformer. I'd use ceramic caps - they seem to have decently low ESR...

You'll want 8 caps as it looks like you have two bridges (one for each channel?). And those caps obviously do the most good when mounted with short leads soldered direct to the bridges.
 
Remember that the combination of a capacitor and resistor can have more impact, although just using capacitors in your case may solve your problem.

It was not uncommon for loop area to be an issue (usually one of several) for us in terms of EMC compliance.

It is interesting that you said that you do not see the oscillation on the power supply filter capacitors. Did you see it anywhere down stream on the power supply rails?
 
Back
Top Bottom