Yes, and they can be calculated using a Taylor Series, but the math gets nasty after about the third harmonic.
My point is that one can easily measure the harmonics and show that second harmonic is not necessarily greater than third in Triodes, and harmonics out beyond the 24th can easily be measured, including odd harmonics.
This is contrary to what many DHT and Triode only advocates claim.
They need to learn to look at graph and estimate sounds in their mind. It helps. It is trained skill.
The SET amp sounds similar with headphones outputting 0.03W as it does with speakers near max output at ~2W, even though with headphones the distortion is much lower. In theory, with 0.1% distortion, it should sound colder, but it still sounds warm and harmonically rich. The same headphones through an SS amp I have sounds bright and metallic, much more analytical.
Most of us beleive in the theory that harmonic distortion is what is responsible for that lucious SET sound, but how sure are we?
What "everybody says" is not necessarily true. Yes, pentodes can make more H3 and higher order garbage, but so can triodes. Whether they do or not is dependent on loadlines and signal swing. Given that VTs are low gain devices, how the circuit operates is also dependent on the harmonic "personality" of the type selected. I've seen that personally: 807s advertise a THD= 1.8% (PP - Class AB1) yet the residual from the Twin T test shows a nearly perfect sawtooth: lots of H3 and higher order distrotion. PP Class AB1 6BQ6s show more distortion, but it looks like an almost perfect sine: mostly H3 with very little H5+. Open loop, 6BQ6s sound better than 807s. This determines how to treat the problem: give 807s local NFB (as advocated by the developer of the type: O. H. Schade) and that'll clean 'em up. 6BQ6s just need some gNFB to take the "edge" off.
That's the halibut right there: your low gain triodes may make less distortion all by themselves, but what about the driver that needs to swing far wider than that of a pentode while dealing with a likely higher C load? What you gain by using a triode just might be offset by higher distortion in the front end. There are always design trade-offs regardless of what you're designing, or active devices, both hollow state and solid state.
0.1% distortion, if low order, is likely to be inaudible. 'Warm and harmonically rich' could indicate distortion or frequency imbalance, possibly caused by the headphones themselves. At 30mW a SET amp will have quite low distortion, but the headphones might not.cotdt said:
'Bright and metallic' sounds like something is wrong. 'Analytical' sounds like everything is right and truthful.
Not the most. I would say, most of people who spread this belief. Actually warmness and richness does not mean presence of some additions. But absence of some artificial, dead additions. Even SS amps can be done properly to sound rich and warm, but with tubes it is easier. You can never add "richness and warmth" to the sound that had been damaged by "death" and "stirillness".
Edit- here is good analogy. Suppose you have some vessel that has clay particles on it's bottom. You fill it with clean water from the creek and drink it.
Then you wash the vessel with gasoline. It looks clean, no clay particles on bottom. But the water does not taste good anymore.
Similarly to "adding tube distortions" adding some clay particles to the water in the vessel cleaned by gasoline does not return that freshness and taste of the water.
Last edited:
Interesting point and very true.
I just changed my SSE EL34 SET amp from a 12BH7 driver with a CCS plate load to a 6SN7 Aikido and there was a very audible decrease in distortion, along with an increase in detail, faster transient response and an overall "cleaner" sound with less noise.
My SET ranges from 2-5% thd with the 12BH7 driver, the Aikido is about .05%, but I have not yet measured the thd of the Aikido/EL34 combo.
The 2HD relative to 3HD level is dependent on the operating point and how hard the amplifier is being driven. It looks like you are driving it quite hard or using a non-optimal operating point because there are lots of harmonics visible above 9th. A nicely cruising valve amplifier will show a steady decay of 2HD > 3HD > 4HD > 5HD. Harmonics beyond that are usually buried in the noise floor.
There must be something else going on here. We are not seeing just the distortion of the output tube. The transfer function shape of ALL triodes is similar and for normal excursions within this function 2H dominates with other harmonics at least 10dB below. The only time this does not occur is when you explore the extremities of the characteristic when all the harmonics pop up in greater proportions. I find it very hard to believe that the harmonic sequence illustrated can occur where the second harmonic is itself 50dB below the fundamental. The only simple thing I can think of is the 2H from the 6SN7 driver is cancelling some of the 2H from the 300B.
Cheers
Ian
glennb,
The amp is not my design, just a document I found which I thought was interesting. interesting .
Look at the rest of the plots. They are comparing some changes in the amp and changes in tubes (chinese vs GEC 6SN7s).
Possible partial harmonic cancellation, and possible operating point issues.
The amp is not my design, just a document I found which I thought was interesting. interesting .
Look at the rest of the plots. They are comparing some changes in the amp and changes in tubes (chinese vs GEC 6SN7s).
Possible partial harmonic cancellation, and possible operating point issues.
Last edited:
The circuit on the associated web page is rather strange. The first half of the 6SN7 is run at a fairly standard 4mA plate current and is directly coupled to the second triode which drives the 300B. The driver stage appears to be run at about 4.5mA which I would suggest is perhaps a little low for this application.
Working backwards from the 8 ohm speaker the primary impedance is 3K so the transformer turns ratio is about 19:1. For 1W into 8 ohms there needs to be 2.828V rms on the secondary or about 55Vrms at the plate. If we are generous and allow the 300B stage a gain of 3 times this means there need to be about 18V rms at the 6SN7 plate. I know from numerous experiments that at this signal level, the intrinsic 2H distortion of a 6SN7 CC stage is somewhere between 0.2 and 0.4% or around -50dB or so. That's probably just about the right level to cancel much of the 300B stage 2H as indicated by the measurements.
Cheers
Ian
I have another question about SE distortion. I know that IM distortion increases as power output increases. But does harmonic distortion increase in the same way? I'm designing an SE amp using 6688 preamp and a 6550 output tube. The amp should put out about 20 watts, hopefully. TDSL lists harmonic distortion for 6550 at 2% or less for PP designs but doesn't give a spec for SE. Anyone no what it is? And will it become unnacceptable at 20 watts?
Since long time ago I not were watching but in this link http://frank.pocnet.net/sheets/135/6/6550A.pdf say that with 20 watts you get more than 13%. Maybe with paralleling you going to avoid in froze them.
Yes THD increases with output power, @20W, THD=13.5%, is that too much
An externally hosted image should be here but it was not working when we last tested it.
IM and harmonic distortion come from the same source: device nonlinearity. Except where there is some cancellation you can expect them both to vary in similar ways with signal level. When there is cancellation then there will be higher order terms arising from that cancellation, in exactly the same way as those which arise from feedback - except that cancellation distortion only goes up to 2N in order, when each stage goes up to N in order.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.