If you have a PP EL84 amp and you want smooth output without global negative feedback, then you strap the EL84s in triode - yes?
Well, having read somewhere about operating in ultralinear with a series resistor in the ultralinear line, I tried this out. It does seem to take off a bit of the "edge". The original suggestion was 300 ohms - I had some 220 handy so I tried those. Without NFB it does seem to smooth the sound somewhat.
So can somebody explain what's happening here, and how do I size the resistor for best results? With 220 the volume is largely the same, so I guess the resistor could go up a bit and still get more output than in triode?
Well, having read somewhere about operating in ultralinear with a series resistor in the ultralinear line, I tried this out. It does seem to take off a bit of the "edge". The original suggestion was 300 ohms - I had some 220 handy so I tried those. Without NFB it does seem to smooth the sound somewhat.
So can somebody explain what's happening here, and how do I size the resistor for best results? With 220 the volume is largely the same, so I guess the resistor could go up a bit and still get more output than in triode?
I can't explain what you experienced, unless you compared pentode mode with UL mode. UL mode gives close to triode operation without sacrificing as much power as triode mode.
I know of a couple of reasons for connecting a resistor between grid 2 and the UL tap on the OPT. The main reason is to prevent parasitic oscillations, for which purpose the resistor must be connected very close to the g2 pin with a short lead. The second reason is to provide a voltage drop, so that the screen voltage won't rise above the plate voltage.
Values of around 100-200 ohm seem to be ususal but I've also seen values as high as 1k. You could try different values for yourself and see if it makes any audible difference.
I know of a couple of reasons for connecting a resistor between grid 2 and the UL tap on the OPT. The main reason is to prevent parasitic oscillations, for which purpose the resistor must be connected very close to the g2 pin with a short lead. The second reason is to provide a voltage drop, so that the screen voltage won't rise above the plate voltage.
Values of around 100-200 ohm seem to be ususal but I've also seen values as high as 1k. You could try different values for yourself and see if it makes any audible difference.
Oscillations? I've had the same effect you describe using small values in series with grid and/or plate depending on tube and layout. In simulations and measurement large values of resistance in the screen circuit - starting around 10k typically - results in a nasty string of low-level distortion harmonics which continue past the ~20 kHz limit of my measurement gear. The same behavour has never appeared, or been below the simulation/measurment floor with resitances the size you used, but given the large value behaviour I'm skeptical even small values like 220 ohm were making the circuit more linear in the audio band.
I can't explain what you experienced, unless you compared pentode mode with UL mode. UL mode gives close to triode operation without sacrificing as much power as triode mode>>
I have a Leak stereo 20, typical EL84 PP amp, where I put a diff pair of ECC40 with CCS as input and no global feedback. I tried it with UL connection straight to OPT and then with 220 ohms in series. Seemed to sound less edgy with the resistor in circuit. Interesting that larger values increase distortion - I never knew that. So there may be some optimum point. I'm not sure how the actual physics of the electrons varies with these changes in G2.
I have a Leak stereo 20, typical EL84 PP amp, where I put a diff pair of ECC40 with CCS as input and no global feedback. I tried it with UL connection straight to OPT and then with 220 ohms in series. Seemed to sound less edgy with the resistor in circuit. Interesting that larger values increase distortion - I never knew that. So there may be some optimum point. I'm not sure how the actual physics of the electrons varies with these changes in G2.
You can see an example using a KT100 here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=77388&highlight=
Last week I confirmed identical behaviour using a fixed-bias, triode-connected EL84 with cathode feedback and 10K screen resistors. The EL84 has much lower high order distortion components than the KT100 so the effect was far more pronounced. 220 ohm screen resistors showed no measurable change. I can't recall the exact value but I think something in the 5K range poked an extra harmonic above the noise floor. I hesitate to extrapolate this behaviour to all scenarios (can't see why it would happen in pentode mode for example) and values. Like you suggest there may indeed be an 'optimum' value that minimizes high order harmonics that's below the resolution limit of my sound card, and any consumer audio format for that matter.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=77388&highlight=
Last week I confirmed identical behaviour using a fixed-bias, triode-connected EL84 with cathode feedback and 10K screen resistors. The EL84 has much lower high order distortion components than the KT100 so the effect was far more pronounced. 220 ohm screen resistors showed no measurable change. I can't recall the exact value but I think something in the 5K range poked an extra harmonic above the noise floor. I hesitate to extrapolate this behaviour to all scenarios (can't see why it would happen in pentode mode for example) and values. Like you suggest there may indeed be an 'optimum' value that minimizes high order harmonics that's below the resolution limit of my sound card, and any consumer audio format for that matter.
Have a look at this site under DIY kits:
http://www.mennovanderveen.nl/eng/index.html
Menno is really useful here - his circuit doesn't require extra windings on the OPT. This is of interest to EL84 or EL34 amp owners because it can be done with a handful of resistors and capacitors. I've transcribed all his mods from his site and books, and I'm ready to have a go.
I started with 1K resistor between screen grid and ultralinear tap on the OPT - that sounds nice and clean so that stays. 1K is his value for this so who am I to argue. He's obviously been here before and some.
After that is the output tube mod - 100k pot instead of 10k grid stopper (with optional 100pf silver mica/polystyrene wrapped around it), and 2.2Meg from EL84/34 control grid via series 10nF cap 500v to the anode of EL84/34. That's the mod. This is interesting stuff! Andy Evans
http://www.mennovanderveen.nl/eng/index.html
Menno is really useful here - his circuit doesn't require extra windings on the OPT. This is of interest to EL84 or EL34 amp owners because it can be done with a handful of resistors and capacitors. I've transcribed all his mods from his site and books, and I'm ready to have a go.
I started with 1K resistor between screen grid and ultralinear tap on the OPT - that sounds nice and clean so that stays. 1K is his value for this so who am I to argue. He's obviously been here before and some.
After that is the output tube mod - 100k pot instead of 10k grid stopper (with optional 100pf silver mica/polystyrene wrapped around it), and 2.2Meg from EL84/34 control grid via series 10nF cap 500v to the anode of EL84/34. That's the mod. This is interesting stuff! Andy Evans
We had some of this before, so apology for some repitition.
This is not as simple as changing only one thing, e.g. adding a serie resistor to g2. For optimum conditions the bias and anode load impedance are also factors. There is the classic example of the EL34 which definitely performs better with serie g2 resistors of as high as 1K - manufacturer's curves confirm this. Other tubes do not want this, e.g. where the g2 current of a beam tube varies substantially during operation. One must remember that g2 is not just another (signal input) grid; it draws current and therefore acts as a 2nd anode, and thus has an internal impedance similar to plate resistance (rp) for the anode. In this respect a serie resistance may or may not cause increased non-lineariy. A tap on the output transformer "drives" the g2 with a low impedance, making its own "internal impedance" relatively unimportant with low serie resistors.
Those with distortion analysers are at an advantage here; simulation can give only vague results because of the difficulty of modelling a UL output transformer especially for h.f. And all this discussion has not taken into account the effect of the G2 resistor also acting as an oscillation stopper. But in cases where h.f. oscillation is a factor, I am in favour of a serie R.C network between anode and G2 tap firstly - let us deal with one thing at a time!
Regards.
This is not as simple as changing only one thing, e.g. adding a serie resistor to g2. For optimum conditions the bias and anode load impedance are also factors. There is the classic example of the EL34 which definitely performs better with serie g2 resistors of as high as 1K - manufacturer's curves confirm this. Other tubes do not want this, e.g. where the g2 current of a beam tube varies substantially during operation. One must remember that g2 is not just another (signal input) grid; it draws current and therefore acts as a 2nd anode, and thus has an internal impedance similar to plate resistance (rp) for the anode. In this respect a serie resistance may or may not cause increased non-lineariy. A tap on the output transformer "drives" the g2 with a low impedance, making its own "internal impedance" relatively unimportant with low serie resistors.
Those with distortion analysers are at an advantage here; simulation can give only vague results because of the difficulty of modelling a UL output transformer especially for h.f. And all this discussion has not taken into account the effect of the G2 resistor also acting as an oscillation stopper. But in cases where h.f. oscillation is a factor, I am in favour of a serie R.C network between anode and G2 tap firstly - let us deal with one thing at a time!
Regards.
Tubes4e4,
Yipes, I remembered your request but forgot on which thread!
I am hunting for this; have to confess that I often speak from memory of stuff I had when still employed, but which did not come with me or got lost when I moved. Not a good excuse, but at least there to ask for some patience. I will come back with what I can find (much of it still in boxes!) I have EL34 pentode graphs, but am now looking for the UL ones.
Regards.
Yipes, I remembered your request but forgot on which thread!
I am hunting for this; have to confess that I often speak from memory of stuff I had when still employed, but which did not come with me or got lost when I moved. Not a good excuse, but at least there to ask for some patience. I will come back with what I can find (much of it still in boxes!) I have EL34 pentode graphs, but am now looking for the UL ones.
Regards.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.