gofar99 wrote:
Hi I believe you are on the right track. Any amplifying device can pick up and amplify noise. I have seen a number of cases similar to yours. The ground side of the AC mains is full of junk. Since an amplifying device senses the difference between ground and signal imput it has no clue if the stuff is signal or not and it gets amplified. The AC mains noise is being mixed with the input signal. What I have found needed in high noise conditions is to first have three wire AC mains power going to everything it can with noise filters on each input. I like the one piece IEC power entry style with combined fuse and line filter built in. Second (seems redundant) I run a separate chassis ground from each piece of equipment. The only exceptions I have in my own system are an OPPO Blue Ray player that only has a two wire mains cord and the final mono-block power amps that do have three wire power but I find they do not need the extra ground. You are correct though in that there are many sources of noise that have crept into our homes. I went through 4 external A to D external boxes to get one that would not containimate the signal going into the PC (most were powered by the PC, and its power supply is really full of crud). In my shop ...regardless of what I do, I have extensive grounding and use APC and Triplet power filters and still can not get the noise as low as I would like. The best I can get is about -100dbv. This is not good enough if you want to measure things that are quieter. There is a way to make the amp more immune to above band noise. Increase the 1K resistor in the grid to a higher value. This will form a RC filter when combined with the tube's internal capacitance. You could start with a value of about 10K and see what happens. If you increase it too much the actual response of the amp at high frequencies will be reduced. I would expect values above about 25 K or so to slightly reduce the gain as well.
Good listening
Bruce
Hello Bruce again !
I have re-boxed my amp, and used a 10K resistor in the grid, and the spikes are all gone. I haven't yet installed a line-filter, as I have trouble to get one that has the house-ground filtered, that's not allowed by my local regulation it seems. But as the spikes are gone, I might skip this step.
I tried to trace the frequency response with my "poor" tools. High-freq seems unaffected by using 10K on the grid. Low freq is about -3 dB at 50 Hz and -7 dB at 20 Hz. Using my Sennheiser HD449, I don't feel I miss basses, but it might be nice to compare my results with someone else's.
Attachment:
freq_response.jpg
I have two remaining problems (noob in asymptotic search for perfection) :
1. Hum.The level is very low, higher on the left than on the right. Evaluated at 2mV on the scope, hard to be precise, cheap DSO. It doesn't really disturb when music is playing, but with high-sensitivity headphone you can hear it when no signal at input.... and I want silence, no 50Hz hum.
Below the pic of my arrangement:
Attachment:
boxing_metal_1.jpg
There is that 1mm-thick metal sheet with mu-metal foil sticked on both sides. I thought it would be enough to avoid 50Hz going from the PSU transformer to the output transformers. Apparently not. The distance between the transformers is about 4cm (1.6 inch). As the noise is stronger on the left channel and the L-opt is closer to the PSU than the R-opt, I concluded the 50Hz hum comes over from magnetic induction. When I carefully shift another metal sheet in between, the noise drops a bit. So here, my plan is to modify the amp side (low part on pic) to have the OPT's as far away as possible from the PSU and use 3mm steel in between.
Edit : the hum level doesn't change when I change the volume setting.2. High-frequency noiseThe other issue is noise above about 200Khz on the output. Below two graphs, input (yellow) at 400 mV 1KHz sinus, output (blue) loaded with 150 ohms resistor, output amplitude about 60 mV.
The first graph shows the output without any filter on the DSO :
Attachment:
1khz_without_filter.jpg
The second graph shows the output with a 100KHz filter on the DSO :
Attachment:
1khz_with_filter.jpg
Now I guess this is not oscillation of the amp itself, I would guess instead power-line pollution, when I zoom it I see low-level sinus on top of my signal, this is what makes the trace "thick" when unfiltered on the scope. So I will probably still need a filter. Not sure if it is a real issue and if I really need to do something about it. Another option would be another condensator in parallel on B+, something like 100nF.... Not sure it will bring something.
Kind regards,
Charles