I have same idea for all my products, AC heater for power tube from 45, 2A3, VT52... to 300B and DC heater for driver or pre directed heater tube.
71A has low mu, low heater voltage so we can use AC for them with least hum noise.
Other tubes with more mu and/or more heater voltage , e.g. SV811, 2E22, 50, VT25, 801A/VT62... need DC heater.
Last edited:
yes, you tame the noise in the first gain (front-end)stages and you will have a quiet amp...
Is the Pope Catholic?Yes, current regulated DC is SUPERIOR to a simple CRC filter.
Start out by rectifying the AC with Schottky diodes. Zero switching noise, from the outset, is a good foundation on which the filament supply can be constructed.
i am sure it is
but i am adamant to put another ss amp on my tube amp..again this is only me...YMMV....
I was considering making an elaborate measurement of intermodulation distortion of DHTs running on ac heating - but then found that Steve Bench has already written up a useful account of it:
Effects of AC Heating Power Applied to Directly Heated Triodes
There are many interesting points to his work:
- For 801, a single IMD spur at a frequency twice the mains frequency down from the test signal 880Hz for 1kHz input yields as much as 0.35% IMD.
- when we consider that there are many other cross-products generated by this same mechanism even on a single tone in the music (ie signal frequency + n x 100/120Hz and - n x 100/120Hz). Now imagine what happens when an orchestra strikes up: all the tones being played, plus all the rich harmonics naturally present in real musical instruments are each plagued by these unwanted sidebands. The sidebands will appear all across the spectrum to produce a blanket of distortion running into fractions of a percentage point.
- None of these IMD components has a natural harmonic relation to the music, so they will sound .... unnatural.
- What's worse, the IMD levels are not a constant fraction of the music. Instead, the IMD increases non-linearly with level. The end result is an unnatural distortion that bounces up & down with the programme ( & volume).
- I have already described what this sounds like, but in case anyone thinks my opinion tendentious, it appears that George (Tubelab) has very similar things to say about ac heat, including applications of the 2A3:
Read the paragraph "Output Stage: "
TubelabSE
All all this talk would be for nothing, if the IMD were too difficult to tackle. But this is a rare case when the problem can be reduced to effectively zero.
The precaution is only that we must not make the problem worse, by changing to dc-heating, and this is where DIYers run into trouble. The sound can be severely degraded by producing a filament supply using any of the usual methods for a dc supply.
Failings include: allowing rectifier forward and reverse current-pulses into the filament (raw rectified dc's failing) or allow noisy bandgap references to corrupt the result (LM317, LT1084 and most other IC-based regulator's failing). Remember that bandgap noise will generate IMD in the same mechanism as 100/120Hz, but this time you have broadband noise, hence massive numbers of sum & difference products, not just one generator tone.
This is one of the reasons that the 2A3-based TRAM-2 DIYers preamp found that the Coleman Regulators make more difference to the overall sound than just lowered noise. (They are comparing it to the LT1084-based DIYhfs CCS-Regulator)
http://www.diyaudio.com/forums/diy-hifi-supply/156854-dht-otl-linestage-tram-2-a-9.html#post3198498
Of course, this is DIY, and DIYers should heat filaments however they feel is best; but let everyone know the outcome of using ac really is, even on low voltage filaments like 2A3. It's far more than a question of hiding the hum.
.
Effects of AC Heating Power Applied to Directly Heated Triodes
There are many interesting points to his work:
- For 801, a single IMD spur at a frequency twice the mains frequency down from the test signal 880Hz for 1kHz input yields as much as 0.35% IMD.
- when we consider that there are many other cross-products generated by this same mechanism even on a single tone in the music (ie signal frequency + n x 100/120Hz and - n x 100/120Hz). Now imagine what happens when an orchestra strikes up: all the tones being played, plus all the rich harmonics naturally present in real musical instruments are each plagued by these unwanted sidebands. The sidebands will appear all across the spectrum to produce a blanket of distortion running into fractions of a percentage point.
- None of these IMD components has a natural harmonic relation to the music, so they will sound .... unnatural.
- What's worse, the IMD levels are not a constant fraction of the music. Instead, the IMD increases non-linearly with level. The end result is an unnatural distortion that bounces up & down with the programme ( & volume).
- I have already described what this sounds like, but in case anyone thinks my opinion tendentious, it appears that George (Tubelab) has very similar things to say about ac heat, including applications of the 2A3:
Read the paragraph "Output Stage: "
TubelabSE
All all this talk would be for nothing, if the IMD were too difficult to tackle. But this is a rare case when the problem can be reduced to effectively zero.
The precaution is only that we must not make the problem worse, by changing to dc-heating, and this is where DIYers run into trouble. The sound can be severely degraded by producing a filament supply using any of the usual methods for a dc supply.
Failings include: allowing rectifier forward and reverse current-pulses into the filament (raw rectified dc's failing) or allow noisy bandgap references to corrupt the result (LM317, LT1084 and most other IC-based regulator's failing). Remember that bandgap noise will generate IMD in the same mechanism as 100/120Hz, but this time you have broadband noise, hence massive numbers of sum & difference products, not just one generator tone.
This is one of the reasons that the 2A3-based TRAM-2 DIYers preamp found that the Coleman Regulators make more difference to the overall sound than just lowered noise. (They are comparing it to the LT1084-based DIYhfs CCS-Regulator)
http://www.diyaudio.com/forums/diy-hifi-supply/156854-dht-otl-linestage-tram-2-a-9.html#post3198498
Of course, this is DIY, and DIYers should heat filaments however they feel is best; but let everyone know the outcome of using ac really is, even on low voltage filaments like 2A3. It's far more than a question of hiding the hum.
.
Last edited:
Intermodulation distortion due to AC-heated filaments is not an issue. IMD will occur every time whenever signals of different frequencies are present together. For example, in amplified symphony sound, violin, flute, oboe, etc. will be intermodulated by drum, clarinet, harp, trombone, etc, in all imaginable combinations, and we are talking not about measly 0.1% filament hum contribution, but contribution coming from multiple signals of equal strength. So, please apply some common sense and stop fearmongering.
IMD exacerbates other nonlinearities (as nicely demonstrated by Steve Bench), so when THD is low, IMD is not a problem.
AC filaments is a simple and viable solution in many, if not most cases. Solid state DC regulators have their problems and should be used judiciously.
IMD exacerbates other nonlinearities (as nicely demonstrated by Steve Bench), so when THD is low, IMD is not a problem.
AC filaments is a simple and viable solution in many, if not most cases. Solid state DC regulators have their problems and should be used judiciously.
IMD produced by cross-products of different music signals is natural and benign.
But it is completely different to IMD from ac-heated filaments: this produces 50/60, 100/120, 150/180Hz sidebands around any tone in the music signal. It is also highly non-linear with signal level, since the IMD levels are very low near the idle (operating point) level, where the hum has been nulled-out, and becoming increasingly pronounced as the DHT operates further toward the edges of the triode-curves.
No need to restrict the discussion to such hand-waving, either (although it makes it easier to understand). It is very easy to correlate the measurement of ac-heat IMD with the fuzzy sound it causes: just try George (Tubelab's) test. Connect the amp as normal to ts speaker, and also to a PC sound-card. Start up a spectrum analyser app., and run. Instead of a sine source though, plug in a guitar or other stringed instrument to the input.
When the strings are played gently, the IMD is small or nonexistent, but as you play harder, the IMD sidebands appear prominently, and the sound is easily heard to be muddled and fuzzy.
Repeat with dc-heated filaments, and witness the effect entirely absent.
But it is completely different to IMD from ac-heated filaments: this produces 50/60, 100/120, 150/180Hz sidebands around any tone in the music signal. It is also highly non-linear with signal level, since the IMD levels are very low near the idle (operating point) level, where the hum has been nulled-out, and becoming increasingly pronounced as the DHT operates further toward the edges of the triode-curves.
No need to restrict the discussion to such hand-waving, either (although it makes it easier to understand). It is very easy to correlate the measurement of ac-heat IMD with the fuzzy sound it causes: just try George (Tubelab's) test. Connect the amp as normal to ts speaker, and also to a PC sound-card. Start up a spectrum analyser app., and run. Instead of a sine source though, plug in a guitar or other stringed instrument to the input.
When the strings are played gently, the IMD is small or nonexistent, but as you play harder, the IMD sidebands appear prominently, and the sound is easily heard to be muddled and fuzzy.
Repeat with dc-heated filaments, and witness the effect entirely absent.
No, it isn't. Physics is the same in both cases. There is no noticeable IMD in live orchestral music because there is no distortion. Similarly, an ideal amplifier with zero distortion would also have no noticeable IMD. IMD is exacerbated by THD.
From Steve's paper, IMD from 120 Hz thermal modulation in AC-powered filaments becomes considerable only when power peaks and THD levels surge. Under a scenario of low THD, e.g. push-pull Class A triode at half power, this kind of IMD is very benign and hardly requires any correction. Heater-borne IMD becomes a problem only when amplifier is driven to a point where it starts seriously distorting.
By no means I am saying that DC filament supply is a bad idea. DC is good for high voltage filaments and directly heated tubes in pre-amplifiers.
Thermal modulation is gives tiny IMD, but that is not the problem. This is apparent as soon as you take real measurements, because 50/60Hz sidebands dominate, not 100/120Hz.
The substantial levels of IMD found in real measurements arises from failure of the nulling process established at idle. A hum-pot is adjusted to minimise hum at idle, but this cancellation is not effective when large grid swings move the operation into areas of the triode-curves when the symmetry (which allows the nulling) is invalid.
This all comes out in real measurements. Bela (euro21 here) has posted some carefully-taken FFTs on this forum, that illustrate this perfectly. If you have an ac-heated DHT amplifier now, please try George's test, as described above - it is very instructive.
The point about dc heating is that this non-trivial distortion goes to zero when comparing ac to dc heating. Since the only reason to use DHTs is to benefit from their lower overall distortion (compared to indirect heated valves), then it seems pointless to undermine their advantages with a wholly-avoidable flaw.
If you read Steve's paper carefully, the fundamental 50/60 Hz heater hum is easily nulled out and is not a problem. By contrast, the thermal modulation, which has 100/120 Hz frequency, is very difficult to cancel, and that was the focus of Steve's investigation.
I looked at George's test. It is basically spectral analysis of IMD and deliberate injection of distortion. An artificial test that has nothing to do with the way we listen to the music using an amplifier. As Daniel von Recklinghausen said, "If it sounds good but measures bad, you are measuring the wrong thing".
I did a more relevant test myself. The A4 piano sound (440 Hz fundamental) was picked up by a high quality condenser microphone and amplified by a PP tube amplifier whose driver (26) and output (2A3) tubes were AC-heated. The speaker and the piano were in different rooms. I asked my daughter (a trained musician and piano tech) lo listen to the amplified sound for the presence of IMD tones with ±60 and 120 Hz, which would be approximately C5, B4, G4, and E4. She couldn't detect any.
Last edited:
You didn't read carefully. There is no deliberate injection of distortion but just two pure tones at 1000 and 950 Hz. The amp is working near its max power and there are few tweaks to the amp to minimize IMD.
AC filaments cause more IMD. No question about that.
It is surely for me. I can clearly hear the difference.
As always if one cannot hear it doesn't automatically mean that it doesn't matter. Likely there are serious limitations somewhere else in the system, including the amplifier itself. But at the end of the day the amplifier is the easier part of the system to sort out. So why don't make sure that it doesn;t happen at all or it is at least minimized?
If I know that something is potentially harmful, I have the possibility to get rid of it and I don;t do then there is no point is DIY. I just buy a commercial product that does it better and is cheaper and save my time for other things. That's my opinion.
Wrong again.
A zero feedback triode single ended amplifier hasn't got 0.000001% THD at half power. At half power THD is not a tiny fraction at all.
I'm fully aware of Stephie's work - it was me that linked it, above. But that was the year 1999, and that work has been extended, and - more importantly - corrected subsequently. Dmitry Nizhgorodov performed some useful measurements that showed that the hum products at idle were distortion harmonics of the electrical ac-heat waveform. Dmitry's investigation showed that the hum-harmonics of ac-heat did not diminish sufficiently with frequency of the ac-heating current, to allow a thermal explanation - at higher frequency, the heating-and-cooling cycle becomes shorter, and for a constant thermal mass, the hum-amplitude would decline as frequency rises; but measurements showed that even for 600Hz heating, no decline in hum output was seen.
Dmitry's measurements are here:
On Correlation Between Residual DHT Filament Hum and AC Frequency
This work was performed the collaboration and encouragement of Stephie, and should be taken as superceding the 1999 Bench experiments. Neglecting very lighweight filaments (eg battery triodes), the thermal-variation theory of hum has been completely discredited.
We have to picture the DHT as a Grounded-grid amplifier, to see how this hum-harmonics are generated - with the ac-heat waveform as the (low-impedance) input. The mains fundamental is nulled at idle by the hum-pot, as well as it can be. The nulling is achieved by equalizing the effective ac-amplitude around the centre-tap, where the anode-current is returned to ground. But this null is only valid in the region of the triode curves, near to the operating point; at larger swings, we move into a different curves-region, and the null is lost. In these regions the IMD rises sharply, and shows up easily in a spectrum analyser.
Arguments about ac-heating are perfectly easily settled by simple measurements. Any kind of computer and soundcard is enough to resolve the large amounts of IMD generated by ac-heating, so there is no excuse not to measure.
There are already a number of good assessments posted on our forum, but this one from Bela (euro21 here) shows a direct comparison of ac- and dc- heating of 300Bs (5V, as per the type 71A in the OP's question).
http://www.diyaudio.com/forums/tubes-valves/258870-sound-300b-set-satoru-kobayashi.html#post4030013
The power output in these plots is only 1W, but the ac-heated PP sample (the RIGHT-hand picture) is already generating large numbers of sidebands - the mains-fundamental sidebands (±50Hz in this case) are only 50dB down. Keep in mind that these sidebands will arise around every tone in the music, creating a widespread fog across the spectrum (a fog is rather like what it sounds like, too). This is certainly not my idea of Hi-fi.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.