Hi folks.
I'm currently building a mid-high gain guitar amp, and I'm struggling with power supply noise. The amp has 4 6V6's and 5 12AX7's, current draw is somewhere around 120mA. I'm using a Hammond 270HX which has a 550Vac centre-tapped HT secondary. I'm using a series rectifier with two 1N4007's on each phase, 47uF as my main reservoir smoothing cap. My HT sits at about 380 on stand-by and has maybe 0.5% ripple. As soon as I switch on I get about 11V p-p of ripple, which is about 3%, and I can live with that, but here's the issue. With series diodes it's a good practice to put capacitors across each to ensure even voltage but when I add those caps, I get audible switching noise from those caps (as in right at the caps!), and also a lot of noise out of the speakers. When I watch my heater voltage (running 6.3Vac heaters with 70Vdc offset) on the scope and switch on the amp I see the sine wave get a nasty sharp edge near the peaks and valleys. Without those capacitors the amp is pretty quiet. I first tried 22nF caps and theorised that they were large enough that they were pumping current back into the secondary and that noise was coupling onto the heater windings, so I tried 10nF and they were also noisy, but somewhat less.
What am I missing here? Is 10nF still too large? I'll go throw some 330p's on there right now and see... Any help would be appreciated!
Thanks
-Dave
I'm currently building a mid-high gain guitar amp, and I'm struggling with power supply noise. The amp has 4 6V6's and 5 12AX7's, current draw is somewhere around 120mA. I'm using a Hammond 270HX which has a 550Vac centre-tapped HT secondary. I'm using a series rectifier with two 1N4007's on each phase, 47uF as my main reservoir smoothing cap. My HT sits at about 380 on stand-by and has maybe 0.5% ripple. As soon as I switch on I get about 11V p-p of ripple, which is about 3%, and I can live with that, but here's the issue. With series diodes it's a good practice to put capacitors across each to ensure even voltage but when I add those caps, I get audible switching noise from those caps (as in right at the caps!), and also a lot of noise out of the speakers. When I watch my heater voltage (running 6.3Vac heaters with 70Vdc offset) on the scope and switch on the amp I see the sine wave get a nasty sharp edge near the peaks and valleys. Without those capacitors the amp is pretty quiet. I first tried 22nF caps and theorised that they were large enough that they were pumping current back into the secondary and that noise was coupling onto the heater windings, so I tried 10nF and they were also noisy, but somewhat less.
What am I missing here? Is 10nF still too large? I'll go throw some 330p's on there right now and see... Any help would be appreciated!
Thanks
-Dave
Yes, I'm using the green winding (with Green/Yellow CT to my 70V divider, which is smooth as glass). Yellow (5V) will be used later for relay control.
QUOTE=DF96;2912320]You could try snubbers instead of capacitors. Make sure the capacitors are linear (so no harmonic generation) and not piezoelectric (no sound). Also, ensure they are rated to withstand constant AC voltage stress.[/QUOTE]
You mean RC parallel snubbers? I'm using standard 500Vdc ceramic caps. http://www.vishay.com/docs/28513/dseries.pdf
-Dave
I just rerouted my main reservoir and rectifier to further reduce my loop area and this seems to have made a marked improvement. The distorted heater supply is still present, but overall amp noise is definitely reduced.
I'm also struggling with control pots picking up noise in my second stage. I'm using a very old chassis (from a Traynor Mark 3 guitar amp, mid-1970's) for my build so the control holes are not large enough for insulators, so I'll be enlarging them all I suppose. I've already enlarged the input and speaker out jacks.
I still suspect there's something else going on causing so much buzz. This old amp (and my Fender Super Reverb) were never terribly buzzy, and I'd like to think that my circuit layout is better than theirs as I've got all my decoupling caps local to each stage right on the point-to-point board rather than in a box on the bottom or way down in one area. I'm going to go through it again looking for ground loops, but I think I've done a fairly good job so far...
Any suggestions are welcome!
Thanks
-Dave
I'm also struggling with control pots picking up noise in my second stage. I'm using a very old chassis (from a Traynor Mark 3 guitar amp, mid-1970's) for my build so the control holes are not large enough for insulators, so I'll be enlarging them all I suppose. I've already enlarged the input and speaker out jacks.
I still suspect there's something else going on causing so much buzz. This old amp (and my Fender Super Reverb) were never terribly buzzy, and I'd like to think that my circuit layout is better than theirs as I've got all my decoupling caps local to each stage right on the point-to-point board rather than in a box on the bottom or way down in one area. I'm going to go through it again looking for ground loops, but I think I've done a fairly good job so far...
Any suggestions are welcome!
Thanks
-Dave
try ac referencing the heaters instead:
remove the circuit that is referencing the heater winding to a b+
the heater circuit should be now just the winding and the heaters.
now install a .1 uf @250V or higher voltage to one side of the heater to ground.
remove caps across B+ diodes.
power on and enjoy.
remove the circuit that is referencing the heater winding to a b+
the heater circuit should be now just the winding and the heaters.
now install a .1 uf @250V or higher voltage to one side of the heater to ground.
remove caps across B+ diodes.
power on and enjoy.
To alleviate the HT supply rectifier noise issue then quite rightly you need to look at 'loop area', and also in that context I suggest bypassing each half HT secondary winding with a small 10nF cap as close as possible to the Tx - this reduces the loop area of the transient current to the winding itself, rather than taking a longer path through the associated diode and supply filter caps. The 10nf cap needs to 630VAC type rating. If possible, twist the 3 secondary HT wires as much as possible from the transformer out to the diodes and 0V (and the bypass 10nF caps if using them).
The elevated DC bias for the heaters could introduce noise from the B+ supply if the bias is not well ac bypassed, or star grounding introduces noise between the bias supply 0V and the input stage 0V reference.
You can dampen the diode noise level by inserting some series resistance with the diodes. Imho I'd remove the capacitor bypass (or RC) across each diode, and focus on the other measures first.
The elevated DC bias for the heaters could introduce noise from the B+ supply if the bias is not well ac bypassed, or star grounding introduces noise between the bias supply 0V and the input stage 0V reference.
You can dampen the diode noise level by inserting some series resistance with the diodes. Imho I'd remove the capacitor bypass (or RC) across each diode, and focus on the other measures first.
Last edited:
Hi,
I think your over loader the 1N4007's diodes. The max. current for the 1N4007 is 1 amp. If you wire all the filaments in parallel your load is about 3 amps. I will use the 1N5408 instead the 1N4007 that is 1000 volts 3 amps. That's maybe what it is causing the noise that your experiencing.
I think your over loader the 1N4007's diodes. The max. current for the 1N4007 is 1 amp. If you wire all the filaments in parallel your load is about 3 amps. I will use the 1N5408 instead the 1N4007 that is 1000 volts 3 amps. That's maybe what it is causing the noise that your experiencing.
yea I forgot to mention diodes. the best overall is to have low impedence. 1n4007 are for solid state circuits. I use 6A10 for my ss rectification. also I see 4 6v6 that is somewhat of a current demand. you need 220uf of capacitance to have a low enough output impedence of the power supply (you'll need 100uf just alone in a pair of 6v6's, not including the drive circuit demands).
Thanks for all the input. A few points:
I need to keep the 70Vdc offset (which is confirmed to be super clean) to protect the cathode followers.
My calculations for 4 6V6's in push-pull-parallel, plus the triodes shows that 47uF is plenty to get me lower than 10% ripple, and I'm currently only seeing 3% ripple. I don't want to go too high with the main reservoir and risk overloading the xfmr.
I'm currently in the middle of improving the layout of stages 1&2. Once I'm finished that, if the problem persists I'll try some of your suggestions.
Thanks everybody!
-Dave
I need to keep the 70Vdc offset (which is confirmed to be super clean) to protect the cathode followers.
My calculations for 4 6V6's in push-pull-parallel, plus the triodes shows that 47uF is plenty to get me lower than 10% ripple, and I'm currently only seeing 3% ripple. I don't want to go too high with the main reservoir and risk overloading the xfmr.
I'm currently in the middle of improving the layout of stages 1&2. Once I'm finished that, if the problem persists I'll try some of your suggestions.
Thanks everybody!
-Dave
Okay, so frustration is reaching its peak here... I've tried everything that has been suggested and still the noise is there. Right now I've got UF4007 diodes in there (which significantly reduced switching transients) with 1nF poly caps in parallel with each of them. I also increased the main reservoir cap to 70uF total, and this made a small difference to the ripple (obviously) but had no effect on the noise.
What's happening is the HT sine wave, just before it goes into the diodes is getting distorted. I think this is because each 1/2 cycle, 1/2 of the xfmr is unloaded, and the collapsing field is being nasty. I suspect this noise is coupling into the heater supply, and the noise on the heaters is coupling into the grids.
Here are some scope captures.
HT top:
HT bottom:
Heater supply:
Noise spikes (120Hz, 20mV/div) at stage 3:
Heaters when on standby:
Any ideas on this? FYI, both phases on the HT show the same noise pattern.
Tomorrow I'll try wiring up a second xfmr to drive the heaters separately to prove the noise coupling into the grids theory, or at least to prove that the noisy heaters are the noise source.
Thanks for any help. I'm at a loss here... considering punching a new hole in the chassis and wiring in a 5U4 but that seems like admitting defeat. Lots of amps use CT xfmrs with series SS rectifiers without trouble...
What's happening is the HT sine wave, just before it goes into the diodes is getting distorted. I think this is because each 1/2 cycle, 1/2 of the xfmr is unloaded, and the collapsing field is being nasty. I suspect this noise is coupling into the heater supply, and the noise on the heaters is coupling into the grids.
Here are some scope captures.
HT top:
An externally hosted image should be here but it was not working when we last tested it.
HT bottom:
An externally hosted image should be here but it was not working when we last tested it.
Heater supply:
An externally hosted image should be here but it was not working when we last tested it.
Noise spikes (120Hz, 20mV/div) at stage 3:
An externally hosted image should be here but it was not working when we last tested it.
Heaters when on standby:
An externally hosted image should be here but it was not working when we last tested it.
Any ideas on this? FYI, both phases on the HT show the same noise pattern.
Tomorrow I'll try wiring up a second xfmr to drive the heaters separately to prove the noise coupling into the grids theory, or at least to prove that the noisy heaters are the noise source.
Thanks for any help. I'm at a loss here... considering punching a new hole in the chassis and wiring in a 5U4 but that seems like admitting defeat. Lots of amps use CT xfmrs with series SS rectifiers without trouble...
The HT diode switching transient has been shown to easily couple in to the heater winding and get amplified by the input stage - so yes that is highly likely to be the mechanism for noise injection.
I suggest you should remove the cap bypasses on the UF4007s, they just aren't needed, and may be exacerbating the problem. Adding a bypass cap across the HT winding itself, and/or adding some series resistance to the HT winding are likely to be your best path - and perhaps adding a bypass cap across the heater winding as close as possible to the transformer. The aim is to minimise dI/dt at diode switching spike, and constrain the energy in the leakage inductances to just the transformer windings.
I suggest you should remove the cap bypasses on the UF4007s, they just aren't needed, and may be exacerbating the problem. Adding a bypass cap across the HT winding itself, and/or adding some series resistance to the HT winding are likely to be your best path - and perhaps adding a bypass cap across the heater winding as close as possible to the transformer. The aim is to minimise dI/dt at diode switching spike, and constrain the energy in the leakage inductances to just the transformer windings.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.