• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

DHT 100hz hum

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi, I´m going to build an all DHT amp and want to avoid DC heaters, so I did a couple on measurements on my DH tubes that some of you might find interesting. I simply fired up the filament without grid or plate voltage, nulled the 60 Hz and measured the residual 100hz hum, it is logical to believe that all DH tubes with same filament voltge will have the same amount of 100 Hz residual hum, but I found out to my amazement that it isn´t nearly so.

I also did a bit of internet research and found that both Dimitry (who´s last name I do not know) and Steve Bench already investigated this effect, but neither measured a significant amount of tubes so I took this task upon myself.
We all agree that although the hum is directly proportional to filament voltage I believe this is not the only factor. These were my results (47 and 46 were triode strapped)

Tube type/ samples tested/ fil. voltage/ hum at plate (mV)/ average for type

47/ 4/ 2,5/ 12, 15, 23, 8/ 14,5
46/ 5/ 2,5/ 30,24,40,28,33/ 31
2A3/ 2/ 2,5/ 20, 29/ 24,5
45/ 4/ 2,5/ 50, 65, 59, 39/ 53,25
50/ 4/ 7,5/ 85, 130 160, 125/ 125

The things to notice are that even though hum for a same type are all within a ballpark there is a high variance.

Filament current does not seem to play a role in this effect, the 45 although having the lowest current has the highest hum of the 2,5V tubes by far.

Mu MIGHT play a role in this, otherwise I don´t know how to explain the 47 being the tube with least hum (Its triode mu is the highest at ~6,7, followed by the 46, 2A3 and finally the 45). All the same, in my web research I found people to report around 8mV of 100Hz hum at the speaker with the type 10/801A/vt25 etc., which in conjuction with a 10k:8ohm tranny implies 280mV at its plate. Even though its mu is double than that of the 50 it has double the hum, so I´m at square one again.

In the next couple of days I´ll measure 4 201A´s which have a 5V filament (low current) and a chinese pair of 2A3´s. I´d like to ask the people who have the means, to measure at least a couple of their tubes so a more serious statistical analysis can be made (If people are interested of course). I´m also going to investigate the viability of relying on hum cancelation in cascaded DH stages, so we can enjoy hum free amps with AC filaments.

As a sidenote, readjusting the hum pot wasn´t at all necessary after changing either tubes or tube types (except for the 50 due to the use of another transformer). I believe this is because the hum pot balances the AC and DC path to ground (DC when there is plate current), so for a same trasformer winding, changing the tube will not affect the balance to a great extent.

Hope someone reads this and finds it useful.
 
Last edited:
Hi!

Thanks for sharing your data.

How exactly did you connect the grid? Leave it floating or connect to ground? If ground how is that related to the filament voltage? center tap of your hum pot to the same ground? I think in order to get meaningful results you need to put the grid on a defined potential.

Then what is happening is that the tube acts as a diode, as the filament swings negative you get a positive voltage from plate to filament and a current flows. I assume that current gives your 100Hz ripple. I would expect that the amount of ripple measured will depend on the emission of the tube. The latter can vary quite a bit between tubes an thus you will get different readings.

I see no reason to avoid DC heaters if you want lowest hum. Properly implemented DC filaments sound better than AC. Use a good current source for the filaments or a good choke filtered supply (LCL), that's what I do.

I just finished a 45 drives 45 amp with LCL DC filament supplies for driver and output. Sounds beautiful.

Best regards

Thomas
 
Thanks for the response. Yes, the grid was grounded to the same ground as the center tap of the hum pot (ground is a defined potential, maybe you meant negative?). I´ll repeat the procedure today first with a negative grid and then with both negative grid and plate voltage at normal operating conditions.

I don´t believe the effect has to much to do with emission, all valves tested have oxide coated filaments and thus work at the same temperature for a directly proportional emmision current per unit area of filament. The 47, 46 and 45 all have roughly the same length of filament (a W with roughly 3cm per line) and therefore roughly the same emission. So if emission played a role then these tubes should share a similar hum level, which they do not.

Furthermore the 2A3 should have a much higher hum level because its emmision is so much greater (I used elecetroharmonix tubes which have a WW filament of roughly 5cm per line) due to the increased filament area, but the measurements showd that both samples had lower hum that both 46´s and 45´s on average.

Thomas I read your blog whenever a new article comes out and i´ve noticed that you prefer amps with the same tube driving itself (45-45, 845-845, etc), I think this is in part due to the very good distortion cancellation produced by the same tube amplyfing the same signal two times, once inverted, as was proven in a Tubecad article, but I suspect that if you were using AC filaments (perhaps you were not) some hum cancellation was also taking place.

As a side-note, when will you do a 50 amp? :D

My wanting to avoid DC is a pet peeve of mine since I don´t want anything to do with silicon, not even in the filaments (yes, I know it is pretty stupid).

PD: I apologise for my logorrheic posts
 
Hi!

Ok, that's what I meant so grid was at a defined potential. I think it would be a good idea to measure emission of the tubes. I would suspect that there is a correlation when you measure the emission and hum within the same type of tubes.

Still I expect variation of hum which I guess is due to mechanical differences of the filaments and also spacing.


Thomas I read your blog whenever a new article comes out and i´ve noticed that you prefer amps with the same tube driving itself (45-45, 845-845, etc),

Not necessarily. Depends. For example I like the 801A as driver a lot for pretty much any DHT (45, 2A3, 300B, 211, 845 and of course to drive 801A)

I think this is in part due to the very good distortion cancellation produced by the same tube amplyfing the same signal two times, once inverted, as was proven in a Tubecad article, but I suspect that if you were using AC filaments (perhaps you were not) some hum cancellation was also taking place.

This cancellation is a myth IMHO. The output tube amplifies a much larger signal so that cancellation effects are not really big. Cancellation would require proper tuning which I want to avoid. Once built an amp shall play for years without touching it. I use power tubes as driver for the headroom. In some cases it happens to be the same as the output tube. For DHT drivers I always use DC filaments

As a side-note, when will you do a 50 amp? :D

When someone orders one. ;)

My wanting to avoid DC is a pet peeve of mine since I don´t want anything to do with silicon, not even in the filaments (yes, I know it is pretty stupid).

I used to be the same. But when you want to use DHTs as driver hum becomes very difficult to control with AC and impossible in a line stage with a 801A.
Just 4 Schottky diodes is not that bad. I use LCL filters afterwards.

For a 45 in the output position AC filaments are ok hum wise. With a 300B it becomes difficult and depends on the tube samples. Anyways I found DC filaments to also sound better. Of course the filament supply matters.

Best regards

Thomas
 
As a sidenote, readjusting the hum pot wasn´t at all necessary after changing either tubes or tube types (except for the 50 due to the use of another transformer). I believe this is because the hum pot balances the AC and DC path to ground (DC when there is plate current), so for a same trasformer winding, changing the tube will not affect the balance to a great extent.

Yes, the hum pot adjusts the balance of 50/60Hz currents of opposite phases - which is added to the anode-current, and returned to ground. When the pot is balanced, equal and opposite 50Hz currents flow in each half of the pot, and the resultant 50Hz current-component is near zero.

But the 100/120Hz components are not nulled. Dmitry Nizhgorodov believes that their source is a distortion product of the triode itself.

The intermodulation spectrum generated by ac-heating is 100/120Hz sum and difference products attached to every component of the music spectrum. These IM components have no natural relationship with the music - which is why they sound very obviously out-of-place. A power valve with ac-heating can be easily improved with a high quality dc filament supply - but when the driver is also a DHT, ac-heating produces very bad results:

When the power valve receives a music signal which already contains intermodulation from the previous stage, the situation is worsened by sum-and-difference spectrum on frequency components that are already unnatural. A multiplication of unwanted frequency components!
Yes, with ac-heated DHTs you get distortion-multiplying effect, rather than cancelling.

Try some quality dc heat - The LCL filters that Thomas uses are easy to implement, or use purpose-designed current-Regulators like my DIY kits - which are designed expressly to avoid influencing the music-signal at all.

.
 
Rod,

In a DHT, as the two filament ends are at different potentials, there is an audio modulated current that flows between these two points. With a CCS DC supply like your module, this current cannot go through the heater supply and so it modulates the cathode/filament, adding something to the music signal.
What do you think, is this really happening or I am missing something?
 
Rod,

In a DHT, as the two filament ends are at different potentials, there is an audio modulated current that flows between these two points. With a CCS DC supply like your module, this current cannot go through the heater supply and so it modulates the cathode/filament, adding something to the music signal.
What do you think, is this really happening or I am missing something?

hi Vincent, It's a good question.

In a DHT operating with dc-heat, the (geometrical) side of the filament connected to the +ve of the filament supply will be biased hotter than the -ve side, because the effective Vgk is smaller (effective Vak is also smaller, but has less influence). And naturally, the effective bias increases as we move along the length of the filament, from +ve to -ve.

So the whole DHT is like a set of parallel-connected triodes, all biased slightly differently.

If you connect the anode-current return (ground) to the filament +ve (for example), then the (music) current from all the parallel triodes adds up, inside the metallic wire of the filament, and we get the full music signal conducted from the B+ through the transformer, to the anode, filament and down to ground.

If you heat the filament with a Filament Supply that possesses high dynamic impedance, the music signal will not be altered at all - it will still be the arithmetic sum of all the small current generated by the (set of ) gm inside the triode. This is what you get with a Current-control supply.

If you use a voltage regulator, or any kind of Filament Supply that places a capacitor across the filament - then you are creating a short-circuit to signals around the filament. If you do this, the differential-signal voltage at the ends of the filament is shorted out, and the current is forced to flow in the filament itself. Since the filament is resistive (a few ohms) the power generated by this differential music signal is dissipated in the resistance of the filament.

So there is the difference: With current-control filament supply, you neither add nor subtract anything. With a Voltage regulator (or just rectified dc) you are taking something away from the triode, and wasting it. This is not an advantage!

Something similar happens with ac-heat. The trafo winding presents a partial short across the filament - and it is reactive, too - so you get a load that is not constant with frequency (unwanted equalisation!!). This is probably why toroidal transformers sound different for ac-heat, compared to EI types. This is the observation that got me looking at filament supplies in the first place.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.