It's music peak power, probably you don't hear any distortion, just missing some details or what I am no expert, but that how good amp is rated, isn't?
Most amplifiers out there use RC coupling to the finals with the associated risk of blocking distortion at high listening levels. Ballpencil's design is not especially bad in this respect. However to avoid this in a conventional design requires an extra stage (cathode or source follower) whereas in this design it can be done for "free". Ballpencil already stated that the capacitors were there due to an issue around the biasing method he was planning. It is up to him to decide whether the benefits of eliminating the RC drive outweigh the possibly added complexity in the power supply.
How about we simply reduce the opamp power supply so that it clips first just before Vgk reaches zero? Then the peak power is determined by the opamp just like in the ubiquitous gainclones.
Even better if we can make it +/- 12v since now we can power the heaters from the same supply as the opamp.
Even better if we can make it +/- 12v since now we can power the heaters from the same supply as the opamp.
It doesn't work that way. You will end up setting the control to what you think is a reasonable setting, then in certain musical passages, you'll get bad distortion.
You will end up with a volume so small as to be completely unsatisfying, yet occaisonally certain musical passages will still give bad distortion.
It isn't like setting the volume on a lo-fi AM transistor radio, where there is a definite point where turning the volume further up brings immediate obvious distortion. In quality audio systems with bad clipping behaviour fed from CD's or good vinyl, theree is no definite point where it goes from good to bad. You can get several minutes of good and then a second of so of bad. At various volume control settings, you can get some tracks playing ok and other tracks not ok.
You can easily try this experiment if you have an oscilloscope: Monitor the amplifier output on a 20W or so amp while playing music at a volume whwere conversation remains comfortable and there' no desire to shout. It comes as a surprise to many how often the CRO shows clipping. Certain recordings show this more than others.
Over the years I have built up a collection of recordings that show up various sorts of amplifier impairments. Certain tracks in The Buddy Holly Story (London Cast) really put an amplifer's clipping behaviour to the test, at volume settings nobody would regard as loud.
It is just as important to engineer an amplifier to clip cleanly and show no recovery artifacts as it is to design for low harmonic distortion on steady tones. Especially if the amp is not high power (say 50W or more with efficient speakers).
Last edited:
Certainly why not according to my sim data reveals just that. But even gird current can be eliminated in output tube by properly adjust loading resistor in IT sec. Maybe to reduce transformer ringing while clipping. Transformer has to be correctly terminated. Again please see my earlier post today, that is my view based on sim.
Last edited:
Most amplifiers out there are Class A or Class AB1. So no grid current and little or no problem.
BallPencil's design is Class B which is prone to bad blocking distortion. Class B outputs are normally designed like Koonw's sims, with the bias in series with the driver transformer secondaries and no grid capacitors.
That will work. However, you would be significantly reducing power output (about a 56% reduction with 12V supplies) in an amplifier whose power output is barely adequate.
You can instead increase the bias to 70V and greatly improve the situation without crippling power output. And if you also increase the HT to +,- 180V and get rid of those 220 nF capacitors, the problem is pretty much gone.
Last edited:
I doubt the amp will be stable (ie, global Fdbk) with that interstage xfmr with the Op. Amp loop gain through it. Ultimately, this will likely transition back to the Futterman like designs to get rid of the xfmr.
The 10 year old "SE sound from P-P" scheme was originally intended as an add-on for a low distortion (like class A) and stable amp. design. It would just replace the Fdbk attenuator with the 1/Mu triode attenuator for some coloration. Using it with overloaded class B outputs and an interstage xfmr is far from the intended design scene. One may have to consider a Hawksford style Error Correction loop (EC) to fix the class B crossover distortion sufficiently to get this to work cleanly for an OTL setup.
In fact, I consider OTL designs to be an insult to tube amplifiers. They need EC just to get them respectable.
The 10 year old "SE sound from P-P" scheme was originally intended as an add-on for a low distortion (like class A) and stable amp. design. It would just replace the Fdbk attenuator with the 1/Mu triode attenuator for some coloration. Using it with overloaded class B outputs and an interstage xfmr is far from the intended design scene. One may have to consider a Hawksford style Error Correction loop (EC) to fix the class B crossover distortion sufficiently to get this to work cleanly for an OTL setup.
In fact, I consider OTL designs to be an insult to tube amplifiers. They need EC just to get them respectable.
Last edited:
My thoughts almost exactly.
Oh, they're ok if you are willing to use 400 ohm or 800 ohm speakers.
I've been commissioned to produce a 40 watt tube OTL design (my design is Class A and does use a form of feedforward EC) and need 400 Ohm speakers to audition it. I had no trouble buying on eBay 40 year old Philips 800 ohm woofers in perfect condition, but have been totally unsuccesful getting 400 or 800 ohm midrange/tweeters. I've bought half a dozen but they all turned out to be useless due to voice coil faults.
I think the jury is still out on this question. There is nothing logically wrong with the possibility that your SPICE simulation showing Vgk always negative is perfectly correct. On the other hand, it is also logically possible that Koonw's SPICE simulation that shows Vgk becoming somewhat positive is correct. It all comes down to the question of exactly what the characteristics of the 6C33C tube are in the "uncharted territory" where the anode current is of order a couple of amps or so.
One person's SPICE model says that Vgk is still negative, while another person's SPICE model says that Vgk is positive. As long as the debate is restricted to SPICE simulations, then it is not really possible to decide who's model is nearer to the truth. As it happens, the SPICE model of the 6C33C that I am using supports your conclusion, that Vgk will always stay negative in the set-up you are considering. This also accords with my impression of how things work in practice, in the OTLs that I have built. But, they are currently inaccessible to me, being across the other side of the ocean from where I am right now. If I had access to them, it would be about five minute's work to make the necessary measurements that would settle the issue, at least for the specific examples of the 6C33C tubes in my OTL. (There may well be significant variation in the anode current that results from Vgk=0, as between one 6C33C tube and another.)
My money would be on your SPICE simulation being reasonably near to the truth on this particular question. But I don't think I would want to bet too much, until I had made the crucial simple measurements on the real system!
Chris
Hi cnpope,
I think the problem stems from the differing approach on what we consider as the full output power of the amp.
If we consider Vgk = 0 being the limit of the output power, then by that definition alone, Vgk remains negative for the full cycle.
But if we consider the limit is determined by the op-amp clipping first, then the amp is able to go to positive Vgk before the op-amp clips.
I think the problem stems from the differing approach on what we consider as the full output power of the amp.
If we consider Vgk = 0 being the limit of the output power, then by that definition alone, Vgk remains negative for the full cycle.
But if we consider the limit is determined by the op-amp clipping first, then the amp is able to go to positive Vgk before the op-amp clips.
True. But one can make a reasonable guess that outside ratings the tube is likely going to need a more postive grid on signal peaks than needing less.
There is also the matter of interpretaion on SPICE results. In a capacitively coupled circuit, grid current will charge the capacitor and reduce the amount by which the grid goes positive. And that's bad, because it leads to blocking. One can also do hand calculations. They aren't difficult. Hand calculation indicates grid current.
True.
When you operate a tube outside its ratings, you expect more tube-to-tube variation.
Last edited:
But nobody with a lick of sense would rate an amplifier that way. It simply isn't the limit.
It's like saying a mad dog isn't vicious because you keep him isolated in a room and don't ever allow people or animals to come near.
But I'm glad you now recognise your design does result in grid current, and that that is a flaw.
Last edited:
But isn't designing for the driver to be able to to reach Vgk=0 on the output tube, the usual approach?
Mullard 5-10, Mullard 5-20 just to name two. I think both amps are able to reach Vgk=0 on the output tubes and yet they are AC coupled.
i think this is what Tikiroo is saying on post #142.
Keit said Vgk = 0 is not the limit, meaning Vgk can be up to +10 or +20 or more. That is A2 B2 class, like 833 SET operates with zero volt grid bias.
Last edited:
Ok my post was poorly worded. Here it is rewritten:
But isn't designing for the driver to be able to clip the output tube with positive Vgk, the usual approach?
Mullard 5-10, Mullard 5-20 just to name two. I think both amps are able to clip the output tubes to positive Vgk and yet they are AC coupled.
i think this is what Tikiroo is saying on post #142.
But isn't designing for the driver to be able to clip the output tube with positive Vgk, the usual approach?
Mullard 5-10, Mullard 5-20 just to name two. I think both amps are able to clip the output tubes to positive Vgk and yet they are AC coupled.
i think this is what Tikiroo is saying on post #142.
Yes, it might operate but do not mean it does not draw grid current, and much effort to maintain the linearity above zero. The question here is about this particular 6c33c whether it actually can be made to work the same or draw grid current since it is not exactly reveal or hidden from us.
BTW I have to add that in sim, I can show you amp can deliver more than 5A or about 50W output, unbelievable, is'it.
Last edited:
What i am saying is that, i think designer of that Mullard amps doesn't care what happens when Vgk goes positive. In other words they are happy the determine the max output power to be equal as when Vgk = 0, not positive. This is why the AC coupling.
It hasn't been mentioned on this thread, but a quick search on 6C33C and grid current turned up a few comments along the line of "don't go pass Vgk=-0.5V or the grid current gets ugly...", which was based largely on the datasheets, I have not come across any actual measurement data for near Vgk=0V or +Vgk operation though...
- Status
- Not open for further replies.