Voltage divider resistor choice for DC elevated heaters

[Electrobumps]

A few pertinent points could be:

If the amp has simple heater grounding, for example if the heater mid-point was grounded through a heater winding CT, or externally with a humdinger, then for starters retrofit a humdinger pot and see how significant that pot can lower any observable hum. That addresses one of the main hum ingress paths - stray capacitance coupling to input stage grids.

Heater elevation addresses a different hum ingress path - that goes via the heater stray resistance to cathode. That path is very dependent on the tube being used - so imho it is worthwhile firstly swapping in/out different tubes to observe any noticeable change. The 'elevation' aims to use a difference in voltage between the heater and cathode to increase the stray resistance value, and hence suppress that hum leakage path. As the decoupling capacitor is the part passing the hum current, the capacitor should connect to the quiet 0V point of the input stage, rather than say a noisier chassis or power supply 0V point.

More detail on this in the linked article: https://dalmura.com.au/static/Hum article.pdf

The Power TX is centre tapped. A while ago I tried a humdinger and removed the centre tap for an artificial centre tap/Humdinger and the hum was really loud, maybe worth investigating again.

As for the tubes I have noticed some pre amp tubes are good and others are moderate and a couple are terrible. While some hum is being injected at some point in the preamp stages I also have hum that is post pre amp.

 

[Electrobumps]

If I remember correctly, some revisions of B15N have elevated heaters and some don’t. All perform acceptably but the ones with heater elevation can be *dead* quiet

They also all(?) have hum balance pots (aka “humdinger”) on the filaments, so someone really cared about avoiding heater noise

I rebuilt a totally-wrecked B15 last year and was very careful about heater lead dress, with very tight twists and careful routing near sockets. I employed the elevation scheme from a later revision and of course the hum balance control

I should also concede that I did the grounds in a non-stock fashion… all jacks and pots were isolated from chassis and the entire amp grounds to chassis at only a single point (added .01uF capacitors shunt RF to ground at the input jacks)

In its quiescent state it is so impressively quiet that the first time I powered it up I didn’t think it was working—it sounded like it was in bypass.

There’s really so, so much performance you can get with just careful implementation

I have most of the parts including an orignal chassis to build a B15 heritage. I just need to get the board made. With my Twin clone I tried grounding many ways and the quietest seemed to be stock.

The cleaning of the brass plate was also mentioned in this thread. I run a bus wire across the back of my pots and this seemed to be the quietest method. Although I have seen on YouTube a guy who adds the brass plates into clones.

 

[Electrobumps]

Finally found some time for some elevation on my AB763 Twin clone. These are my findings measuring at centre of 1 speaker against grill cloth

A Channel has a lead mod and also feeds trem & reverb. No longer out of phase with B channel (which can cancel some noise).
Also some additional filtering in the preamp and bias circuits. To note a considerable amount of noise is coming from the hiss of the carbon comp resistors.

Measurements Pre and Post Elevation

Amp idle (all Knobs at 0) 52.6dB - Post elevation not measured
Amp idle pre amp metal valve shields added 51.8dB - Post elevation 48.5dB = -3.3dB
Copper shield added over open chassis 51.8dB, stopped a ticking from my mobile several feet away. Post elevation not measured
Both channels Pre amp Bass/Mid/Treble set to 5 53.2dB - Post elevation 49.2dB = -4dB

 

[abbey road d enfer]

. To note a considerable amount of noise is coming from the hiss of the carbon comp resistors.

I don't understand your measurements. dB referenced to what? Measured how?

 

[Electrobumps]

I don't understand your measurements. dB referenced to what? Measured how?

dB SPL in REW with a umik-1 calibrated mic. I’m measuring the noise floor at the speaker. The reference point is the amp in an idle situation pre elevation.

I’m interested in any reduction in noise floor. when recording loud amps the lower the noise floor the longer the fade out I have to play with.

while this test may not be worthy of a white paper or amp technical specs it was helpful to me to have evidence of any improvements of the noise level. Rather than just listening and presuming it is or isn’t better.

 

[trobbins]

Well done Electrobumps for using what you have available for test instrumentation. Perhaps also worth taking an 'ambient' dB measurement during such a test 'campaign', to get a better appreciation of what your test environment noise floor is. For example if your ambient level was say 47.5dB, and so only 1dB lower than your lowest reported level, then that could make interpretation of results a bit more difficult. The other aspect of reporting would be to include a spectrum plot and identify the frequency peak (or frequencies) that is effectively being reported, and confirm that the dB of the frequency peak aligns with the general SPL being calculated by REW, as you may need to bandwidth limit how REW is calculating its general SPL (if that is what you were reporting) otherwise it can be prone to 'out-of-range' noise levels that aren't obvious.

 

[Electrobumps]

Well done Electrobumps for using what you have available for test instrumentation. Perhaps also worth taking an 'ambient' dB measurement during such a test 'campaign', to get a better appreciation of what your test environment noise floor is. For example if your ambient level was say 47.5dB, and so only 1dB lower than your lowest reported level, then that could make interpretation of results a bit more difficult. The other aspect of reporting would be to include a spectrum plot and identify the frequency peak (or frequencies) that is effectively being reported, and confirm that the dB of the frequency peak aligns with the general SPL being calculated by REW, as you may need to bandwidth limit how REW is calculating its general SPL (if that is what you were reporting) otherwise it can be prone to 'out-of-range' noise levels that aren't obvious.

Thanks for the feedback. I was using the live SPL meter set at dBa which I believe is filtered to 20hz to 20khz. The ambient noise in my control room is 29dB. I left the amp under test for several minutes and checked in several times to check the readings where consistent.

 

[Philbo King]

dBa is good to use around machinery to see if you'll go deaf.

Better to use dBz or none.

 

[Electrobumps]

dBa is good to use around machinery to see if you'll go deaf.

Better to use dBz or none.

I’ve always used dBa to level match my monitors.

Please can you you explain why this wouldn’t be an accurate reading.

 

[Philbo King]

Red - A weighted
Blue - C weighted


Additional info:
https://www.noisemeters.com/help/faq/frequency-weighting/

That said, all of them give an equal reading with a 1 KHz sine wave, so if that's what you did for level matching it'd be fine. But if there was bass in the test signal the results could be skewed.

 

[Electrobumps]

That said, all of them give an equal reading with a 1 KHz sine wave, so if that's what you did for level matching it'd be fine. But if there was bass in the test signal the results could be skewed.

As mentioned in first post amp is in idle conditions. The main aim was to see what effect elevated heaters would have on hum. My measurements were the simplest way for me to compare the level of amp noise pre and post procedure.

I've read a fair amount of information on elevating heaters in guitar amps and I've only found subjective accounts of improvement.

 

[Philbo King]

Well... if the hum was the same 50 or 60 Hz for both measurements, I suppose the -30dB cut at those freqs would cancel out since it was present for both measurements. My bad...

 

[Matador]

When I last tested this, I used a different setup: I hooked up the amp into my Torpedo Reload reamp box so I wouldn't go deaf, and then used an FFT plugin to zero in on the relative level at 60Hz, with no guitar plugged in (aka. shorted input). This gave the cleanest display with the strongest (relative) 60 Hz signature.

Stock configuration was heater secondary center tap at 0V, and I measured -22dB at 60Hz. After elevating the center tap to around 20V, the relative level dropped by about 7dB (a littlle less than half). Going past 20V had little effect that I could see (I typically use 40V in practice).

In addition, lifting the center tap, and deriving a virtual center tap via a 250ohm pot (elevated voltage to center of pot), I could then dial out another 2dB (average). Measuring the pot afterwards I saw roughly 100 ohms on one side and 150 ohms on the other.

So after all changes roughly 9dB reduction, which on a high gain amp is huge.