Strange Forest of Noise with Linear PSU

Status
Not open for further replies.
xrk971 said:
And since only a single rectifier is used, I suspect that the secondaries on the transformer are 34VAC, which when rectified makes 48v, and if center is referenced to 0v, then we get +/-24v from a single secondary.
No. If you want a centre-tapped DC supply then you either need a centre-tapped secondary plus one bridge or two secondaries and two bridges. That amp appears to have gone for the latter option, as you have.
 
xrk971,
I can see schematics in #52, #89, #105.

What would be good when presenting frequency spectra, or any noise measurement, is a sketch of the total measurement system. That would include all equipments used in the measurement, and their connections to mains AC, and a sketch of the grounding within the amp.

Although such a sketch is not really required for simple indication of results (especially results showing improved performance), it does provide confidence for others that the measurement result is what it appears to be.

Within the amp, with ref to #89, it helps if separate connections to chassis ground are shown separately - as that schematic implies that circuit ground connects to chassis ground half way along the wire between the AC mains plug and its connection to chassis. And #45 indicates your transformer may have a static shield (that will also connect somewhere).

Also within the amp, if the sketch can indicate the location of the local star point taken to chassis (eg. is it effectively at the power supply common power distribution take-off to the amp modules, or at the amp module signal inputs, or ...).

The measurement system devices need to identify if they are mains or battery isolated powered, and if you interfaces or signal isolators before a soundcard for example.
 
xrk971,
I can see schematics in #52, #89, #105.

What would be good when presenting frequency spectra, or any noise measurement, is a sketch of the total measurement system. That would include all equipments used in the measurement, and their connections to mains AC, and a sketch of the grounding within the amp.

Although such a sketch is not really required for simple indication of results (especially results showing improved performance), it does provide confidence for others that the measurement result is what it appears to be.

Within the amp, with ref to #89, it helps if separate connections to chassis ground are shown separately - as that schematic implies that circuit ground connects to chassis ground half way along the wire between the AC mains plug and its connection to chassis. And #45 indicates your transformer may have a static shield (that will also connect somewhere).

Also within the amp, if the sketch can indicate the location of the local star point taken to chassis (eg. is it effectively at the power supply common power distribution take-off to the amp modules, or at the amp module signal inputs, or ...).

The measurement system devices need to identify if they are mains or battery isolated powered, and if you interfaces or signal isolators before a soundcard for example.

Ok I will try to get a sketch. Basically the USB powered sound interface is connected only to a battery powered laptop not connected to mains. Signal output from sound interface had ground connected to signal ground of preamp which is connected to mains. Previous measuremt of preamp shows it is clean. Signal input of sound interface is connected speaker output ground of amp.
 
Ok I will try to get a sketch. Basically the USB powered sound interface is connected only to a battery powered laptop not connected to mains. Signal output from sound interface had ground connected to signal ground of preamp which is connected to mains. Previous measuremt of preamp shows it is clean. Signal input of sound interface is connected speaker output ground of amp.

One hassle with a single soundcard is that the line in/MIC and the line out are typically connected to the same 0V in the interface. As such, inadvertent noise/hum could be injected in to the measurement plots if you make soundcard in and out connections to different ground locations in the preamp/amp.

It's also good to do a loopback noise floor measurement with everything as similar to the real measurement. It's easy for soundcards to pick up radiated hum and noise from nearby equipment (I've got a particularly noisy LED magnifier lamp on my work bench, and usually have to turn that off as well as a 12V halogen lamp with nearby transformer to remove minor high frequency spikes and some mains fundamental from my spectrum plots).
 
Noise that pops up on the FFT

Hello All.

This is an interesting thread about noise that pops up on the FFT.

While you all have been chatting, I have been poking about on the bench attempting to optimize the noise and distortion of a 6C4 tube. (one half of a 12AU7)

First off I thought that 60Hz and the series of harmonics were originating in the bench power supplies. I turned off all the power supplies and came back in the morning and ran the FFT again with all the power supplies off and the 6C4 cold, all of the 60 Hz and series of distortion harmonics remained in the FFT.

It is my impression that power supply ripple is in the air everywhere. I also found that my LED lamp over the bench also generates a lot of junk in the FFT output.

See attached FFT’s

Analyzer loop back with generator turned off. (noise base line)

Analyzer loop back with 0.1 volt 1000 Hz signal applied. (analyzer baseline noise and distortion)

LED lamp turned on over bench. Analyzer input leads attached to 6C4 breadboard.

Leads attached to 6C4 breadboard with power supplies turned off, tube cold and no signal applied. (note ambient 60Hz and distortion series)

6C4 breadboard energized by bench power supplies, analyzer on.

I am thinking of a metal box or noise cage to isolate the power line hum and harmonics. I am also thinking that placing the transformer and diode based power supply inside the amplifier enclosure nearly guarantees powerline 50/60HZ and distortion series will be present to some degree in the amplifier output.

Comments?

DT
 

Attachments

DT - good stuff.

I guess one main aspect is to know/appreciate what is parasitic noise/hum, and not part of the actual measurement information being investigated. That often comes from many cross checks, and a good dose of scepticism when looking at plots.

I use an aluminium foil tray to sit my $1 USB simple soundcard, USB isolator, batteries and old preamp interface in - that surprisingly shields a lot of benchtop related hum/noise. Whereas a quality interface (EMU0404) has excellent shielding and power supply and almost no need for extra shielding. But shielded cable for interconnects are almost mandatory, and of course appreciating when the soundcard/interface in and/or out levels are clipping and causing extra signals.

It's certainly easier when the only mains connection is via the DUT, and when the DUT has a single internal 0V point for input and output signal connections.

It's not easy to setup and work in a Faraday cage!
 
hi,
iam planning on using single 600 va center tapped , dual bridge rectifier 35 A ,and two power supply pcb for stereo amp class ab(250 watts @ 4 ohm) as exactly in rod elliots diagram attached below with ground loop breaker , i just added couple of more caps for each amp , going by the discussion here is it not advisable

http://sound.whsites.net/p04_fig1.gif

another option i have is with ,single 50 amps diode bridge and connect pcb one after another totalling 4 caps each rail ( single psu), avoiding loop as in post 23 by mark
which wound be better
thank you

Hi,
No body has answered mine , is it a stupid one or it has been answered
 
the ESP sch is not wrong.
But it does connect the two supplies via the common Zero Volts wire.
The two supplies are not isolated.
That could lead to a loop if you "think" that you have two separate supplies.

There is no advantage to splitting the rectifiers and capacitance.
You will get the same performance with a single rectifier and single bank of capacitors.
 
When I first saw the ESP schematic I wondered why he had wired it like that. Using the same components he may as well have rectified the secondaries separately and joined the two supplies at the output to give the 0V rail. Some consider this to give a slight advantage
 
The ESP schematic doesn't have parallel return paths through ground/0V like in post#23.
Current will flow BR1-C1-GND or BR1-C1-C2-BR1. Use a GLB like ESP has done and add the option of an HBR on the amp boards. You can always replace the HBR resistor with a wire if you don't need it.
 
I really like the ESP PSU. It manages to get as close to two monoblock PSU's from a CT transformer as possible. Uses only one bridge per side and applies the double voltage to the single bridge and references the CT to 0v. This is sort of what I was thinking was being done in the Pass F7 amp. I am going to take my P2P CRC's and switch to this topology and see how to works.

An externally hosted image should be here but it was not working when we last tested it.
 
Status
Not open for further replies.