Odd behavior or normal voltages in heater supply?

[trobbins]

It is worth appreciating how peaky the current pulse is going through the two diodes, and how short a duration it is - especially when a large filter capacitance is being charged.  The current peak can be easily 5-10x the 'AC' peak current for a similar output load power, and so has a significant impact on the transformer secondary winding voltage drop at that time - which would be 'flat-topping' the ac waveform fed to the AC heaters.

The distortion in the AC voltage may also cause some error in your ac voltage measurement, depending on how good a true rms reading your meter can make.

With respect to the DC heater voltage and hum, those particular heaters will not now see 6.3Vac pushing leakage current across the heater-cathode interface, but rather the low AC ripple voltage level.  To my mind, there is no net benefit in trying to reduce the ripple from 100mV down to 1mV, as both of those levels are hugely below the incumbent 6.3Vac.  And there is an adverse affect in adding more and more filter capacitance - the diode current pulse peak value gets higher and the duration of the pulse gets shorter - that means the stray harmonics are pushed higher up in frequency and so are more likely to become audible elsewhere.

The AC powered heaters are 'grounded' in that schematic, as the voltage of each AC winding terminal is bounded as it floats from near 0V ground to near ~7-8V as it is pushed from one level to the other as the diodes conduct in one direction and then the other.  The valves using AC power may then exhibit higher hum levels than if they were grounded by a humdinger, as the voltage swing (relative to ground) is about twice.

You can use schottky diodes to lower the diode voltage drop during conduction, and hence raise the DC voltage available for those heaters.  And some designers purposefully lower the heater voltage on input valves to reduce other forms of noise - if that is at all measureable.

 

[abbey road d enfer]

Yes I know, but since I described the total current draw and the voltages I thought that you could discuss the problem anyway? I guess I was wrong?

Best regards

/John

Th ecapacitor value, the way the RC filters are arranged, the place of the inductor and its value may all matter. Indeed, the main suspect is still the transformer. You know, in an inquiry, it is equally important to exonerate than proving guilty.

 

[johnheath]

Th ecapacitor value, the way the RC filters are arranged, the place of the inductor and its value may all matter. Indeed, the main suspect is still the transformer. You know, in an inquiry, it is equally important to exonerate than proving guilty.

Yes, as an investigating police officer I am very well familiar with the context of finding evidence in all directions

I will be more detailed in the future when posting schematics

Best regards

/John

 

[johnheath]

It is worth appreciating how peaky the current pulse is going through the two diodes, and how short a duration it is - especially when a large filter capacitance is being charged.  The current peak can be easily 5-10x the 'AC' peak current for a similar output load power, and so has a significant impact on the transformer secondary winding voltage drop at that time - which would be 'flat-topping' the ac waveform fed to the AC heaters.

The distortion in the AC voltage may also cause some error in your ac voltage measurement, depending on how good a true rms reading your meter can make.

With respect to the DC heater voltage and hum, those particular heaters will not now see 6.3Vac pushing leakage current across the heater-cathode interface, but rather the low AC ripple voltage level.  To my mind, there is no net benefit in trying to reduce the ripple from 100mV down to 1mV, as both of those levels are hugely below the incumbent 6.3Vac.  And there is an adverse affect in adding more and more filter capacitance - the diode current pulse peak value gets higher and the duration of the pulse gets shorter - that means the stray harmonics are pushed higher up in frequency and so are more likely to become audible elsewhere.

The AC powered heaters are 'grounded' in that schematic, as the voltage of each AC winding terminal is bounded as it floats from near 0V ground to near ~7-8V as it is pushed from one level to the other as the diodes conduct in one direction and then the other.  The valves using AC power may then exhibit higher hum levels than if they were grounded by a humdinger, as the voltage swing (relative to ground) is about twice.

You can use schottky diodes to lower the diode voltage drop during conduction, and hence raise the DC voltage available for those heaters.  And some designers purposefully lower the heater voltage on input valves to reduce other forms of noise - if that is at all measureable.

Thank you sir

If I understand you correctly (poor english in my case) you say that it might not be beneficial to use DC for the heaters in this case?

best regards

/John

 

[trobbins]

Every circuit application is different. Every valve is different. Every construction and layout is different.
Everyone has a different level of measurement amd listening capability, and interest level.  Enjoy the journey.

 

[abbey road d enfer]

With respect to the DC heater voltage and hum, those particular heaters will not now see 6.3Vac pushing leakage current across the heater-cathode interface, but rather the low AC ripple voltage level.  To my mind, there is no net benefit in trying to reduce the ripple from 100mV down to 1mV, as both of those levels are hugely below the incumbent 6.3Vac.

Comparing 6.3V 50/60Hz with rectified voltage is inadequate. The problem is that the rectified voltage has higher harmonics; since most of the coupling between heaters and sensitive nodes is capacitive, they may be more audible, even if the amplitude is less.

  And there is an adverse affect in adding more and more filter capacitance - the diode current pulse peak value gets higher and the duration of the pulse gets shorter - that means the stray harmonics are pushed higher up in frequency and so are more likely to become audible elsewhere.

At one point increasing capacitance has no effect on harmonics because the transformer's inductance and resistance become dominant.

The AC powered heaters are 'grounded' in that schematic, as the voltage of each AC winding terminal is bounded as it floats from near 0V ground to near ~7-8V as it is pushed from one level to the other as the diodes conduct in one direction and then the other.  The valves using AC power may then exhibit higher hum levels than if they were grounded by a humdinger, as the voltage swing (relative to ground) is about twice.

Actually, the voltage swing is exactly the same, but theamplitude is now about 0-10V instead of +/-5. Does it matter? No. Buzz can happen when the cathode is at a positive voltage vs. heaters, because the cath acts as an auxiliary anode for the heaters (which in turn behave as a cathode). That's why some designers elevate the AC heater voltage with DC; that is exactly what happens in the case in hand. If the positive leg of the heater voltage was grounded, that would be different, but I doubt it would really matter, axcept if some cathodes are at a particularly high positive voltage, such as in a cath-follower.

 

[trobbins]

I agree that a rectified heater changes the heater waveform from what the mains provides to DC plus higher order harmonics.  The constructor has to use appropriate layout and first filter capacitor to suppress the harmonics.  Nowadays, the mains AC waveform can have significant levels of harmonics and other signals in the audio range, and some of that can come from the B+ rectifier operation, and those don't get any attenuation for standard AC heater use.  Nowadays, a good radial electrolytic of not too large a size has a pretty low and consistent ESR- impedance, and if the impedance reaches the ESR in a few hundred hertz then there is little else to be gained.

The heater to cathode interface typically shows a symmetric resistance characteristic around 0V difference - there is typically no diode conduction related characteristic for that interface resistance behaviour (even though some have  described the behaviour as semiconductor like).  As the interface resistance characteristic is typically symmetric, elevating the heater positively or negatively achieves the same benefit in that the interface resistance increases significantly.  In practise it is generally easier to apply a positive heater elevation.  Valve to valve variation of interface resistance has been shown to be huge,  so any noticeable poor performance is usually remedied by tube swapping.  Interface capacitance however is a constant and is unlikely to vary much between valves. 

Of course knowing for certain if hum is caused by one or another particular path is very hard to predict, and often tube swapping and circuit isolation faultfinding techniques, and substitution techniques like using a battery for heater powering, is often the long road taken if a hum has always been or becomes annoying.

 

[abbey road d enfer]

As the interface resistance characteristic is typically symmetric, elevating the heater positively or negatively achieves the same benefit in that the interface resistance increases significantly.

Can you elaborate on that? Where is that "interface resistance" located? How come it increases with the absolute value of the DC voltage applied to it?

 

[trobbins]

There is a summary of Rhk and some recent measurements in:
https://dalmura.com.au/static/Hum%20article.pdf

I came across a Chk datasheet value for the 12AX7 the other day so need to add that in.

 

[abbey road d enfer]

There is a summary of Rhk and some recent measurements in:
https://dalmura.com.au/static/Hum%20article.pdf

Very interesting article; it's the first time I read about these non-linearities in the oxide coating.

 

[johnheath]

The transformer only gives 6.1VAC and rectified I get some 6.05VDC @ 0.84A... It is within the 5% margin for the heaters so I won't fight about it with the transformer dealer... for now anyway

Thanks for the inputs


Best regards

/John

 

[radardoug]

Are you going straight from the rectifier into a resistor before the filter cap? If so, you limit the charge current for the filter cap, and never achieve peak d.c.

 

[abbey road d enfer]

I won't fight about it with the transformer dealer...

And I would think you'd have no cause to do so. Usually, xfmr heater winding specs are for a resistive load; I would think if you loaded the xfmr with a 1.8 ohm resistor, the voltage would be correct. Rectification puts an additional set of constraints on the xfmr, that are certainly not taken into account by the manufacturer. If he had, he would have advertized it, because taking into account these constraints costs money, and he would have to justify the higher price.