Grid 2 voltage affects
I am revamping/modding a push/pull(El34 output, 1963 vintage) monoblock set. The original power supplies were quite limited IMO. I have added regulation for input/driver stage B+ and grid 2(el34s), change to soft recovery diodes and increased most of the capacitor sizes. My question is what sound changes are to be expected with grid 2 voltage changes. Presently B+ for the El34's is 645v and the grid 2 voltage is 380v. Original supplies for the grids ran through a 1.2 henry/45 ohm reactor with a 390v supply, but I remove this reactor after regulating this supply. I have seen modern applications of very similar amps with a regulated 300v supply for the grids. Is there any guidelines other that the 425v max for el34s?
Excellent explanation & site, great answers for questions like these!
Essentially the loadline affects the screen grid dissipation, since there is a pronounced knee in the pentode's characteristics...as the plate voltage approaches 0, the screen voltage takes over to attract electrons toward the plate, increasing its dissipation. You will note that screen current curves and distortion curves tend to look very similar in the Phillips datasheet for class AB. Lower screen voltages are easier on the tube, but allow for less power, while higher screen voltages demand a lower impedance load (to get over the knee) and increase screen dissipation because the plate goes lower than the screen sooner. Learning to plot the loadline will *really* help make it clear what screen voltage you may want for your application.
In 1963, the Mullard EL34 often used at that time was rated for 500V screengrids, 800V anode.
Applying those values to modern EL34s usually results in short lifetime of the power valve. I think 580V anode, 390V VG2 is about the limit if you want a reliable amp with today's components.
If the amp was designed to run 380-390V VG2, then reducing to 350V will have no perceptible effect on the power output, provided the original designer set the output transformer to give an anode load appropriate for these operating conditions.
As m6tt says, you can work through a load line to see whether the design is appropriate. Or you could evaluate the turns ratio of the output trafo, and work out the impedance presented to the anode using the speaker you have in mind.
The Mullard 1965 data sheet gives operating conditions somewhat near to yours:
795V anode, 400V VG2 (both idle values), 11K a-a load, VG1=-39V for 2x25mA anode current 100W at 5% distortion.
So I would expect your trafo to present 8 to 10K a-a. Regulating the screen supply should give improved sound in my experience, providing the regulator has been designed to give graceful dynamic behaviour (no nasty overshoots or ringing in response to a fast transient). Increasing cap values is usually helpful here too.
Thanks for the reply
Thanks for the input. I will look into this a bit more. I am not completely familiar with this subject. I did set the regulation on one of the monos at 330v and the sound seemed to soften a bit, maybe it was the power decreasing(?). Playing with the output transformer is out as that would be well beyond my abilities. Would applying the reactor to the grid circuit give the system something, as I thought that the original idea was to smooth/regulate the circuit in the first place. Also, this reactor was in series with 4 paralleled tungsten bulb(?) before being applied to the El34s. Some kind of power limiter?
IIRC, Telefunken mentioned in one of their "Laborbuch" a supposed life time of less than 1000h for a 100W Po PP class B operation point (mind, for and from a single pair of EL34) with Ea=800V and Eg2=400V (separate feed, preferably stabilized).
Tubes just are expendables - the more you push them, the faster they expend.
The cool thing about output transformers is that there's not much to do with them other than select amongst taps available. They transform, mostly! So a transformer set up for 8 ohms secondary with 8k a-a (primary) becomes 16k a-a if you put a 16-ohm speaker on the 8-ohm tap. This can and will change the frequency response, sometimes the global feedback network will "fix" this until clipping. This adjusts the loadline (since a speaker isn't a resistor, tubes swing between multiple loadlines when amplifying depending on frequency and the speaker characteristics, don't worry about that!).
Essentially for a given speaker tap, a lower than indicated load will present a lower load on the output tubes, and a higher than indicated load will provide a large load to the power tubes. It is dangerous to do this without making sure the new load line doesn't bust the tubes specs too much :)!
The tungsten lamps are a good idea inasmuch as they don't get in the way of normal screen current, but quickly show high impedance to high currents, and protect your vacuum.
dropping the screen voltage to 330V may be OK, but did you readjust the bias? At 390V the designer may have intended 2 x 25mA to flow. At 330V you may need a slightly smaller cathode resistor for the same current. Or to drop the fixed bias voltage from -39V to say -36V or even lower.
learning all the time
Concerning the tube life, the output power is handled by 8 El34s and the same Svetlana flying cs have been in the amps for 6 winters(I only have them running in the winter half of the year). Still seem to sound fine and the bias points have been the same since new.
About the load variation, the amps are driving impedance compensated Dynaudios, so the load is about as constant as possible. I have a 3.75 ohm tap.
I did have to adjust the bias up(-36 to around -32) when lowering the regulated voltage, back to 30-35 millamps per tube, where the bias had been the previous years. It does take just about the -36vdc to control the bias at the 390v point. I did put the voltage back to the 390v point, and yes forgot to rebias but quickly relies the mistake without mishap.
These amps have some very curious circuit implementations, and not being an expert in the tube design area, it has been a real struggle along the mod/improvement path. I appreciate the information and was hoping that someone might be interested in explaining some of the details of the circuit. I do have the schematic. Output transformer primary side tertiary winding for global feedback, feed forward compensation etc. Any takers....
Other than biasing, which you have already have under control, the remaining general effects of screen voltage are:
- higher screen voltage yields a higher peak current (at Eg1=0). But in an audio power amp, this does not increase power output once a critical level of Eg2 has been reached, because of the way the impedance between the anodes overlaid on the anode curves of the EL34 develops power. The link given in post #2 explains this effect thoroughly.
Notably the GEC data sheet for the TT21 (a KT88 with a topcap) informs us that "there is no advantage in setting the TT21/KT88 screen voltage above about 300V" for fixed bias push-pull duty.
For the EL34, exceeding the 360V Eg2 at which the pentode is largely characterised, is similarly futile.
I do not believe that there is any substantial change in the sound (except near & above overload) with differing Eg2 levels, in the normal range: 300..400V.
- High screen voltages risk overheating of the screen during big peaks, and risk destroying the valve.
- 630V MKP capacitors are probably best for the G2 supply; but if you use electrolytics, I find the performance of 400V rated types is better than for higher voltage ranges. This is another incentive to run EL34 screens at 360V.
I routinely install a 900V power-FET gyrator to EL34 guitar amps I have in for rebuilding. I adjust it to give 360-380V on the screens.
I'm not a fan of complex feedback schemes, but would still be interested in seeing your schematic, if you're going anywhere near the scanner...
|All times are GMT. The time now is 03:08 PM.|
Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio