100Hz Hum in Lehmann Linear clone

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have a 100Hz hum in my Lehmann linear clone.
It is constant, the volume knob does not affect it. I am certain it is NOT a 50Hz hum.
The hum stops when i pull the plug (the music goes on for a few seconds because of the large supply caps). So i am certain it is rectifier-related.

It is barely audible with my most efficient headphones. But it should not be audible at all.

Does somebody have any experience with this problem? Especially Lehmann Linear clone owners (Lovely Cube etc)?



The board is a cheap board from aliexpress, but I have replaced all the semiconductors in case they were counterfeit. I used output transistors with about the same hFe (BD-139-16 and BD140-16, the 16 indicates the hFe group). The driver transistors were upgraded to their low-noise variants.
I also upgraded all but the largest electrolytics to Panasonic FC.
Input caps were changed too. All other components were measured and tested OK.

The diodes (1N4007) were replaced with something I had laying around: BYV-27. Those are ultra-fast recovery diodes (for SMPS), not really meant for audio applications. Can those be a problem?

I measured about -20mV DC offset on both outputs. Isn't that a little too much for a headphone amplifier? Inbalance in the output stage might explain the hum. How can I correct for this?

I cannot measure 100Hz hum on the output with my scope, it is not sensitive enough. The only signal I can measure is the DC offset, and a 10mV RMS 100Mhz signal (Radio interference?).

Any ideas? I included the schematic.
 

Attachments

  • lovelycubecircuit02a.jpg
    lovelycubecircuit02a.jpg
    120 KB · Views: 1,259
  • 838x500px-323219fc_LehmannAmpVolts.jpg
    838x500px-323219fc_LehmannAmpVolts.jpg
    34.3 KB · Views: 1,257
What do you mean by that... two connections to the chassis? Then you have a ground loop. All ground connections eventually go to one point... safety earth.

No no, only one connection to the chassis, but it's not a direct connection, there is a resistor in between (power ground =/= signal ground).
It's not a ground loop (that would be 50Hz i think).

I found out something else: Changing the gain setting on the preamp makes the hum louder or softer. So the opamp in the preamp reacts to this.

I tried replacing the opamp (an OPA2134) with a spare NE5532 i had lying around, but unfortunately, on of the pins was bent and I didn't see it. All I heard was a loud 'pop' and now my headphones are dead, so probably big DC on the output.
I can't test it anymore, because I don't have a spare pair of crappy headphones lying around...

By the way, maybe the preamp reacts to ripple on the power supply or large current spikes from the big power caps charging. This might be because I replaced the preamp's 470uF capacitors by 1500uF Panasonic FC (I only had these lying around). They also have very small ESR so big charging currents...

EDIT: i found another pair of crappy headphones. Unfortunately, these cans are much less efficient so the noise is almost impossible to hear (if you didn't know it was there, you wouldn't hear it). But it is still there, with a NE5532 or a OPA2134.

(by the way, the amp itself sounds brilliant and volume-dependent input noise is almost non-existent.)
 
Last edited:
100Hz or double the mains frequency points to issues with your power supply at the secondary, mainly your caps but can be a bad regulator too.

Double the frequency due to AC to DC retification.

Or if it's 100Hz fundamental but with higher order components (e.g. a chirp that happens 100 times a second) I would suspect rectifier. A scope with FFT comes in useful, or a sound card and you can download either ARTA or Audacity to do the FFT.

Does your rectifier have all sorts of fancy snubber caps attached and are you using some fancy fast or low dropout diodes. How is your ground routing? And how symmetric are your + and - rails and what is the impedance of +/- and ground? All to consider when troubleshooting.
 
Last edited:
100Hz or double the mains frequency points to issues with your power supply at the secondary, mainly your caps but can be a bad regulator too.
The main supply caps are cheap and chinese 'Nover' caps. I have tested them for capacitance (they test ok) but I can't test them for ESR.

Or if it's 100Hz fundamental but with higher order components (e.g. a chirp that happens 100 times a second) I would suspect rectifier. A scope with FFT comes in useful, or a sound card and you can download either ARTA or Audacity to do the FFT.
I have a scope, but it does not have FFT capabilities. The signal is so weak I cannot see any form of 100Hz signal in it on my scope. I can see mainly the DC offset, some 100Mhz interference and some other noise.

Does your rectifier have all sorts of fancy snubber caps attached and are you using some fancy fast or low dropout diodes.
There are just some small caps over the diodes. The diodes are indeed fancy fast diodes, I can swap them for 1N4007's and see the problem is still there.

How is your ground routing?

And how symmetric are your + and - rails and what is the impedance of +/- and ground? All to consider when troubleshooting.

I'm gonna check on that.
 
FFT is useful in that you're unlikely to visually see a component of a signal that is 100x smaller than the main signal (or noise), but with FFT it will show up as a -40dB which is noticeably high.

If you suspect high ESR, spam more CRC filtering. Especially the R part, since it sounds like you have lots of spare capacitance.
 
which version is your clone? I've read lots of problems like this in china's forum,
there's a solution if you really don't know what to do. make sure the version is the same though. i've never tried clone one.

Step1.
connect the 3 red lines with cable, the red square in the middle is actually same point.
this should reduce the hum to barely audible. u can stop here if you want.
An externally hosted image should be here but it was not working when we last tested it.


step2.
cut through the black line to separate ground.
An externally hosted image should be here but it was not working when we last tested it.


step3. connect the yellow line.

That's it. just translated for you. if you really don't know what to do next.
 
No no, only one connection to the chassis, but it's not a direct connection, there is a resistor in between (power ground =/= signal ground).
It's not a ground loop (that would be 50Hz i think).

That depends on the source that drives the groundloop.

I found out something else: Changing the gain setting on the preamp makes the hum louder or softer. So the opamp in the preamp reacts to this.

Which would have me suspect that the noise is coupled or conducted into the inputs of the opamps. Probably differential mode as this opamp has good common mode and power supply rejection.

I tried replacing the opamp (an OPA2134) with a spare NE5532 i had lying around, but unfortunately, on of the pins was bent and I didn't see it. All I heard was a loud 'pop' and now my headphones are dead, so probably big DC on the output.
I can't test it anymore, because I don't have a spare pair of crappy headphones lying around...

That's unfortunate.

By the way, maybe the preamp reacts to ripple on the power supply or large current spikes from the big power caps charging. This might be because I replaced the preamp's 470uF capacitors by 1500uF Panasonic FC (I only had these lying around). They also have very small ESR so big charging currents...

One thing to try is to put the original caps back in and listen.

It's not the lower ESR that results in higher current peaks but the lower impedance of the bigger caps.
You may know the formula for capacitance reactance: Xc = 1/ωC. ω is the angular frequency that we will ignore here as it doesn't change in this case.
By tripling C you effectively bring the impedance down to 1/3 of the original caps which is what causes the increase in current spikes (Ohm's Law).

Have you actually tried what happens if you put some more distance between the transformer and the pcb? You may assume it's not the tranny as the hum seems to be 100 Hz, but over the years I've learned to never assume anything and test it anyway.

Since you have a scope, did you measure the cleanliness of the power supply?
If not, do this as close as possible to the power supply and do not use the clip in earth lead of the probe. At low levels this acts as an antenna. If possible, bare the probe's tip (the shield part) and slide on the small spring clip that should have come with the probe (along with some other small accessories). Also, enable the bandwidth limit function, maybe that attentuatues the noise from the full bandwidth (you wrote you were able to measure 100 MHz) enough to see the 100 Hz hum.
This way I recently discovered a fault in the PSU of my own headphone amp. This was not audible, but I corrected the routing of the ground traces and lowered their resistance anyway.
 
Last edited:
Hi, thanks all!
I measured again with the bandwidth limit on, that is indeed much better, thanks!
I measured the power supply ripple:
Main capacitors: 140mV ptp . That is almost 10% ripple and to me, that's a lot.
After regulators: 2.5mV ptp . Almost below measuring threshold, so very good.

Everything else seemed to be in order, so I tried something else: grounding the inputs to different ground points, and listening.
The "regular" input ground has a bit of hum, this was my starting point.
I then grounded the inputs to the main supply caps ground: the same hum, but much louder. I could definitely hear the charging currents in the ground path.
Then, I tried grounding the input to the - vcc opamp decoupling cap : the same hum, but almost completely quiet. Then, the opamp +vcc capacitor: completely quiet.

It definitely seems like a ground routing problem.
Flying fish, thank you very much! This is indeed my board, and it seems to be a solution to a grounding problem too... Maybe I can find another solution that does not require to cut traces.
 
Hi, thanks all!
I measured again with the bandwidth limit on, that is indeed much better, thanks!
I measured the power supply ripple:
Main capacitors: 140mV ptp . That is almost 10% ripple and to me, that's a lot.
After regulators: 2.5mV ptp . Almost below measuring threshold, so very good.

Everything else seemed to be in order, so I tried something else: grounding the inputs to different ground points, and listening.
The "regular" input ground has a bit of hum, this was my starting point.
I then grounded the inputs to the main supply caps ground: the same hum, but much louder. I could definitely hear the charging currents in the ground path.
Then, I tried grounding the input to the - vcc opamp decoupling cap : the same hum, but almost completely quiet. Then, the opamp +vcc capacitor: completely quiet.

It definitely seems like a ground routing problem.
Flying fish, thank you very much! This is indeed my board, and it seems to be a solution to a grounding problem too... Maybe I can find another solution that does not require to cut traces.

I'm no designer, so please correct me if I'm wrong, but looking at the modifications flyingfishtw posted, I'd say the designer of that pcb put in one big groundplane and thought he/she could use it as some sort of star ground. The proposed cutting across the black line and connecting the red and yellow lines with wire brings a real star into the design.

Your experiments seem to show that that big groundplane indeed doesn't act as a real star ground. Cutting that trace is IMHO something to consider after all.

Don't worry too much about the amount of ripple on the unregulated side. As long as the regulators can stay in regulation (i.e. the ripple dips do not drop below the minimum required Vin-Vout difference for the regulator to work correctly), there's no problem. Putting in bigger caps arbitrarily in the hope to "cure" the ripple may actually lead to other undesired phenomena, as already mentioned.
That 2.5 mVpp ripple sounds very good to me (esp. if you did use the clip on earth lead of the probe). It might in reality be even less.
Even industry quality laboratory power supplies starting at € 500 will hardly do better.
 
I convinced myself to cut the ground plane and connect the wires as shown on the picture that flyingfish posted.
Result: no more hum. It doesn't look very good, but the hum is gone and that's the most important thing. Thanks again! :cheers:

I am now running some endurance tests before I hook up my expensive Ultrasones. It seems like the output transistors are running really hot. That's not surprising, since it is biased very hot. But still, I'm a bit worried. How hot is 'too hot'?

Thanks again,
Olivier
 

Attachments

  • DSC_0190.jpg
    DSC_0190.jpg
    906.4 KB · Views: 1,025
  • DSC_0189.jpg
    DSC_0189.jpg
    971.7 KB · Views: 759
Since the re"designer" seems to have messed around with the layout, I wonder if he also messed around with the bias with the idea that "more and hotter is better". Now I shudder too...

In electronics, you don't want components to run hotter than needed, even if they can take more. Heat kills or at least shortens useful life, especially if your amp is going back into a cramped or not too well ventilated casing....

Without thermometer, if you get an "ouch" reaction when you touch the heatsinks, then rule of thumb says it's too hot.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.