So tonight I decided to start wiring up the power supply for my new PP 429A amp. The schematic shows 1N4007 diodes but they're UF4007 in real life. The output seems to have some random LF noise and I'm not sure about how to clear it up. Here is the schematic
http://i88.photobucket.com/albums/k162/astouffer/psu.jpg
Now here are the screencaps from my scope
http://i88.photobucket.com/albums/k162/astouffer/5s.jpg
http://i88.photobucket.com/albums/k162/astouffer/10s.jpg
Adding a small film cap like .047uF or .1uF across the first filter cap had no result. Maybe snubbers for the diodes?
http://i88.photobucket.com/albums/k162/astouffer/psu.jpg
Now here are the screencaps from my scope
http://i88.photobucket.com/albums/k162/astouffer/5s.jpg
http://i88.photobucket.com/albums/k162/astouffer/10s.jpg
Adding a small film cap like .047uF or .1uF across the first filter cap had no result. Maybe snubbers for the diodes?
When you look at the dc power supply noise you will get random fluctuations , sometimes quite large , due to fluctuations in the mains supply. Anything switched in or out in the house or in the vicinity is supposed to show up on your rectified line. LF garbage being quite hard to filer out fully.
Someone suggested that if you find this on the dc supply and you want to look closely at the supply noise , build a sharp filter at about 30 Hz to block LF garbage.
If I use an ac meter to look at a tube dc supply ( in the 10 mV range ) , the noise fluctuates at random with ocassional large peaks. Haven't yet tried a filter to cut out LF garbage.
Someone suggested that if you find this on the dc supply and you want to look closely at the supply noise , build a sharp filter at about 30 Hz to block LF garbage.
If I use an ac meter to look at a tube dc supply ( in the 10 mV range ) , the noise fluctuates at random with ocassional large peaks. Haven't yet tried a filter to cut out LF garbage.
Boris_The_Blade: The load is a large 3k resistor drawing 90mA.
SY: No I have not looked at the power line yet. The transformer does go through a normal IEC filter. What is the best way to look at the incoming AC? Float the ground of the scope or just measure the unloaded secondary of a lower voltage winding?
infinia: The scope is an old HP 54200A. Can you elaborate on what you mean by trigg or aliasing error?
Tomorrow I'll take a closer look at the AC line and post my findings. Thanks everyone.
SY: No I have not looked at the power line yet. The transformer does go through a normal IEC filter. What is the best way to look at the incoming AC? Float the ground of the scope or just measure the unloaded secondary of a lower voltage winding?
infinia: The scope is an old HP 54200A. Can you elaborate on what you mean by trigg or aliasing error?
Tomorrow I'll take a closer look at the AC line and post my findings. Thanks everyone.
Hi
Your description sounds like a scope that is not triggered properly esp on measurements related to the power line frequency. The trigger usually can be switched to line and then re-adjusted?
Aliasing errors due to the variable sampling on a wide band non-filtered input can give amplitude and time based errrors. The sampling rate has to match the expected timebase of the TBD signal to display properly. http://en.wikipedia.org/wiki/Aliasing
Your description sounds like a scope that is not triggered properly esp on measurements related to the power line frequency. The trigger usually can be switched to line and then re-adjusted?
Aliasing errors due to the variable sampling on a wide band non-filtered input can give amplitude and time based errrors. The sampling rate has to match the expected timebase of the TBD signal to display properly. http://en.wikipedia.org/wiki/Aliasing
That's nothing a few hundred Henries won't fix 
.
It makes sense that the only things coming out of your low pass filter are very low frequencies. I don't think there are any scope errors here.
I get a small amount of random very low frequency noise out of my choke too. It bothered me enough that I am building a Maida regulator.
.It makes sense that the only things coming out of your low pass filter are very low frequencies. I don't think there are any scope errors here.
I get a small amount of random very low frequency noise out of my choke too. It bothered me enough that I am building a Maida regulator.
My reasons for thinking this is not a scope error are the following:
1. All digital scopes that I have used don't trigger at very slow time scales. They run like a strip chart. Triggering is not necessary when you can see everything as fast as the scope puts it out. The term for this mode is 'untriggered roll'. I don't know if this HP scope does this or not, but the screen captures are valid whether the scope triggered or not. It saw those voltages on its input during that time period.
2. The screen captures are quite continuous for this to be a high-frequency (>50MHz) signal aliased into this display. Does this scope have any kind of anti-aliasing filter built in? If not, how can you trust any measurement that it makes?
3. The measurements make intuitive sense. If you feed a 60Hz sine wave (+random noise) into a low-pass filter with a cutoff frequency less than 60Hz, you should get out some residual 60Hz and very low frequency noise.
Truly, you can never completely trust any measurement that you make, I just see no reason to doubt this one. As I stated, I have seen comparable levels of very low frequency noise (not oscillating at any regular frequency) on the output of my choke as well. I assume that it is just fluctuations in line voltage with time. I guess my power isn't perfect.
If one is bothered by it, there are much more complicated circuits that can get rid of it.
1. All digital scopes that I have used don't trigger at very slow time scales. They run like a strip chart. Triggering is not necessary when you can see everything as fast as the scope puts it out. The term for this mode is 'untriggered roll'. I don't know if this HP scope does this or not, but the screen captures are valid whether the scope triggered or not. It saw those voltages on its input during that time period.
2. The screen captures are quite continuous for this to be a high-frequency (>50MHz) signal aliased into this display. Does this scope have any kind of anti-aliasing filter built in? If not, how can you trust any measurement that it makes?
3. The measurements make intuitive sense. If you feed a 60Hz sine wave (+random noise) into a low-pass filter with a cutoff frequency less than 60Hz, you should get out some residual 60Hz and very low frequency noise.
Truly, you can never completely trust any measurement that you make, I just see no reason to doubt this one. As I stated, I have seen comparable levels of very low frequency noise (not oscillating at any regular frequency) on the output of my choke as well. I assume that it is just fluctuations in line voltage with time. I guess my power isn't perfect.
If one is bothered by it, there are much more complicated circuits that can get rid of it.
I see this sort of stuff all the time on the inputs to my supply regulators and it is definitely from small voltage variations in the AC line. It is very difficult to observe these very low frequency line voltage variations when you are looking at a 60Hz 170Vpk sinewave on your scope - you really won't see it clearly, although you will see plenty of other even more horrifying things like a clipped sine waveform, (lots of distortion) whole dropped cycles, sags and surges of a couple of cycles or more, and huge transients. (Use a very high speed scope and you will even see multi-kV transients that last up to a couple of uS - often lightning strikes many miles away..)
A few hundred mV to several volts of LF voltage variation will be typical and should not be a cause of concern unless your audio circuitry has relatively low psrr (which most do...) - still the only place this should be a concern would be in a high gain situation such as one a phono or tape stage.
In short I would not worry or obsess over it, I don't even bother to try and "fix" it any more. You can reduce it a lot, but it is almost impossible to get rid of short of active regulators.
Long ago I had an HP 54200A and it certainly has enough sampling bandwidth that aliasing would never be an issue at 60Hz and below. Somewhere beyond 10 - 20MHz (I don't remember the one shot sampling rate any more, but it was not very high) with non repetitive waveforms aliasing becomes a real concern. I will also say I found it less than ideal for low level analog work due to quantization noise, and that is why I got rid of mine.. YMMV
A few hundred mV to several volts of LF voltage variation will be typical and should not be a cause of concern unless your audio circuitry has relatively low psrr (which most do...) - still the only place this should be a concern would be in a high gain situation such as one a phono or tape stage.
In short I would not worry or obsess over it, I don't even bother to try and "fix" it any more. You can reduce it a lot, but it is almost impossible to get rid of short of active regulators.
Long ago I had an HP 54200A and it certainly has enough sampling bandwidth that aliasing would never be an issue at 60Hz and below. Somewhere beyond 10 - 20MHz (I don't remember the one shot sampling rate any more, but it was not very high) with non repetitive waveforms aliasing becomes a real concern. I will also say I found it less than ideal for low level analog work due to quantization noise, and that is why I got rid of mine.. YMMV
Astouffer,
By now you would have realised that your scope patterns are very normal! It would be better after R1, where I presume you take off the more sensitive stages of your amp (although your R1=50 ohm is quite small. Can that be increased?) One usually take off power stage anodes only, right after a choke input choke (I did not look up your circuit).
Circuits like the Quad II took off the anodes right after the rectifier (C-input filter), where matters looked even more ugly, without audible effect. Hook up and listen.
By now you would have realised that your scope patterns are very normal! It would be better after R1, where I presume you take off the more sensitive stages of your amp (although your R1=50 ohm is quite small. Can that be increased?) One usually take off power stage anodes only, right after a choke input choke (I did not look up your circuit).
Circuits like the Quad II took off the anodes right after the rectifier (C-input filter), where matters looked even more ugly, without audible effect. Hook up and listen.
First if you don't have a good trigger using any scope esp a digital sampling one, the measurement is garbage (big period).
http://www2.tek.com/cmsreplive/tire..._3GW_22049_0_2008.11.18.10.09.30_14525_EN.pdf
Using Your Oscilloscope to Measure a Noisy Signal
Stable Triggers Required
Before you analyze your signal, you need a stable display
which can be a problem if your signal is noisy, making it
difficult to create a stable trigger. Most oscilloscopes come
with several features that assist you with this problem.
I encourage you to visit Tek.com for their vast amount or technical app notes
As others have pointed out a wideband sampling scope is not the best tool in the shed for these type of measurements.
Using a digitizing sampling oscope to measure low freq small signals you just have to be a little smarter than the avg bear.
1) You are measuring a small 120 Hz signal (not 60Hz). limit your sampling rate.
2) ADC input BW to the lowest possible (diode switch noise can be an aliasing influence here) An active LPF is a good idea here.
3) Get a stable power line related trigger (build an external trigger if you have to)
4) LF random noise is not a problem here.
http://www2.tek.com/cmsreplive/tire..._3GW_22049_0_2008.11.18.10.09.30_14525_EN.pdf
Using Your Oscilloscope to Measure a Noisy Signal
Stable Triggers Required
Before you analyze your signal, you need a stable display
which can be a problem if your signal is noisy, making it
difficult to create a stable trigger. Most oscilloscopes come
with several features that assist you with this problem.
I encourage you to visit Tek.com for their vast amount or technical app notes
As others have pointed out a wideband sampling scope is not the best tool in the shed for these type of measurements.
Using a digitizing sampling oscope to measure low freq small signals you just have to be a little smarter than the avg bear.
1) You are measuring a small 120 Hz signal (not 60Hz). limit your sampling rate.
2) ADC input BW to the lowest possible (diode switch noise can be an aliasing influence here) An active LPF is a good idea here.
3) Get a stable power line related trigger (build an external trigger if you have to)
4) LF random noise is not a problem here.
Hi there,
I can confirm the existence of the described LF variations in the DC output of choke-input supplies.
The setup is: 480-0-480 transformer > 5AR4 rectifier > 5H, 87R choke > 120µF cap > 5H, 87R choke, 120µF cap
The 100Hz ripple on the DC is about 10mV, just as Spice simulations predicted. Superimposed on the ripple are chaotic LF variations of about 200mV with frequencies of sub-Hz to a few Hz. Although the posted oscilloscope images of the OP are no longer available, this sounds very much like the behaviour the OP described.
If I do a frequency response analysis of the filter chain, there is almost no filter effect up to about 10 Hz (-0 up to -5dB max), so it sounds sensible that LF variations make it through the filter chain.
It is definitely no scope (sampling) error, I used an analog oscilloscope.
Question to the tube experts here: Are you sure that this effect is due to line voltage variations and does not indicate a problem with the choke-input supply? If yes, I will leave everything as it is and live with it. Remains to be tested if this LF garbage makes it to the output of the amp - which is not the case I hope.
Question two: Is this a special behaviour of choke-input supplies? I fired up the negative rail of my amp (180-0-180 transformer > EZ81 > 20µF > 220R > 220µF > 110R > 220µF > 110R > 220µF) yesterday and there were absolutely no LF variations visible.
But current draw is much lower in this case, and filtering capacitance much higher.
Frequency response analysis showed about -20dB in the LF area for the second filter chain.
Greetings,
Andreas
I can confirm the existence of the described LF variations in the DC output of choke-input supplies.
The setup is: 480-0-480 transformer > 5AR4 rectifier > 5H, 87R choke > 120µF cap > 5H, 87R choke, 120µF cap
The 100Hz ripple on the DC is about 10mV, just as Spice simulations predicted. Superimposed on the ripple are chaotic LF variations of about 200mV with frequencies of sub-Hz to a few Hz. Although the posted oscilloscope images of the OP are no longer available, this sounds very much like the behaviour the OP described.
If I do a frequency response analysis of the filter chain, there is almost no filter effect up to about 10 Hz (-0 up to -5dB max), so it sounds sensible that LF variations make it through the filter chain.
It is definitely no scope (sampling) error, I used an analog oscilloscope.
Question to the tube experts here: Are you sure that this effect is due to line voltage variations and does not indicate a problem with the choke-input supply? If yes, I will leave everything as it is and live with it. Remains to be tested if this LF garbage makes it to the output of the amp - which is not the case I hope.
Question two: Is this a special behaviour of choke-input supplies? I fired up the negative rail of my amp (180-0-180 transformer > EZ81 > 20µF > 220R > 220µF > 110R > 220µF > 110R > 220µF) yesterday and there were absolutely no LF variations visible.
But current draw is much lower in this case, and filtering capacitance much higher.
Frequency response analysis showed about -20dB in the LF area for the second filter chain.
Greetings,
Andreas
Both a choke input supply and a CLC supply are LC resonators, typically at a few Hz. Under constant load and constant mains input they will behave similarly, because the rectifier conduction will impose a short circuit at one end (almost all the time (L in) or twice per cycle (C in)) and so in effect you have an LC in either case. This may ring if the damping provided by the load is inadequate - often the case nowadays when people seem to think that bigger caps are necessarily better.
When the mains input varies they will behave differently. The choke in will simply pass through any variations below its rolloff, so you see the mains variations appearing at the output. The cap in will be temporarily disconnected from the mains during any sharp reductions in mains voltage, until the caps have discharged enough to get the rectifier conducting again or the mains has risen again. You will see mains variations looking more like a sawtooth i.e. somewhat smoothed out. This assumes that the choke in has sufficient load that the rectifier conducts for almost all of the cycle.
When the mains input varies they will behave differently. The choke in will simply pass through any variations below its rolloff, so you see the mains variations appearing at the output. The cap in will be temporarily disconnected from the mains during any sharp reductions in mains voltage, until the caps have discharged enough to get the rectifier conducting again or the mains has risen again. You will see mains variations looking more like a sawtooth i.e. somewhat smoothed out. This assumes that the choke in has sufficient load that the rectifier conducts for almost all of the cycle.
Depends on how good is "good", and how it is done. Subsonic noise will not be directly audible (that is what subsonic means) but it could modulate the signal and so generate IM sidebands. This would be most noticeable, other things being equal, in circuits which use cancellation to get good PSRR rather than filtering etc. For example, a push-pull output will have good PSRR because of the balance but it will be just as prone to IM from the supply as a similar single-ended stage.
How would it sound? In extreme, as unwanted random tremulo!
How would it sound? In extreme, as unwanted random tremulo!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.