I'd be very grateful if, and you may find it interesting to, have a look at an article I've put up on my website...about my first attempt at building a valve amp!
http://www.brahms.demon.co.uk/software/electronics/ecl82_1/index.html
However I'm pretty sure a lot of my original assumptions were wrong and that this amp shouldn't work!
My first need in tracking this down is to know how to obtain the characteristic curves for ECL82 valves in UL mode. Am I right in thinking that when I plot these I should:
Pick a value for B+ = Vb
Use the formula Vs = Vb - (Vb - Va) * 43% to obtain the voltage on the grids.
On the amp I built I also have 4k7 resitors in the screen lines so would need
Vs = (Vb - Va) * 43/100 - Is * 4700 ? To get the relevant screen voltages.
I've done this and get something that looks roughly right but I really am not sure this is the correct way of doing it? The lines on the graph with the 4k7 resistors look to be lower and more evenly spaced!?!
I have looked all over and can find no information on how to do this.
Cheers
Dom
http://www.brahms.demon.co.uk/software/electronics/ecl82_1/index.html
However I'm pretty sure a lot of my original assumptions were wrong and that this amp shouldn't work!
My first need in tracking this down is to know how to obtain the characteristic curves for ECL82 valves in UL mode. Am I right in thinking that when I plot these I should:
Pick a value for B+ = Vb
Use the formula Vs = Vb - (Vb - Va) * 43% to obtain the voltage on the grids.
On the amp I built I also have 4k7 resitors in the screen lines so would need
Vs = (Vb - Va) * 43/100 - Is * 4700 ? To get the relevant screen voltages.
I've done this and get something that looks roughly right but I really am not sure this is the correct way of doing it? The lines on the graph with the 4k7 resistors look to be lower and more evenly spaced!?!
I have looked all over and can find no information on how to do this.
Cheers
Dom
Hi Dominic,
I am going to give a quick and incomplete reply here, hopefully of use to you. Firstly, although your arithmetic seems in order, the curves that you show do not represent UL - I do not know whether you are able to plot the dynamic curves with your programme.
The characteristics of a power tube in UL look more like triode than pentode, with the knee only going towards pentode style. Somewhere on the web (sorry that I do not give the reference for your convenience) you can look up characteristics of e.g. KT88 and 6550, for pentode, UL and triode operation. Should you want a comparison of output power, Rp and distortion vs. G2 tap position going from pentode to triode, you can go to my post #46 in the "UL to triode" thread here under "Tubes". From the points given you can draw graphs. These are for a KT88 which was what I could get hold of, but I also state there that they generally represent all UL operation (obviously with changed values). You are so right; UL characteristics for tubes are hard to find.
I am not able to open your complete circuit for some reason or other (right hand side will not come on), but I am worried by the G2 resistor of 4.7K. This is very high and would compromise UL operation - they form separate loads for the G2 instead of direct connection to the transformer taps. Do I understand this correctly? Why do you use these resistors? EL34 circuits seem to work better with a 1K series resistor (manufacturer data), but in most other applications there are no series screen resistors (or maybe just 100 ohms to swamp h.f. oscillation tendencies).
For the rest you should be OK if you follow basic principles, which at a first glance you seem to be doing. Just a warning if necessary: Any red glow on the anode is out; if this happens you are exceeding the max. plate dissipation quite substantially and should switch off and investigate.
Perhaps more later, have fun.
I am going to give a quick and incomplete reply here, hopefully of use to you. Firstly, although your arithmetic seems in order, the curves that you show do not represent UL - I do not know whether you are able to plot the dynamic curves with your programme.
The characteristics of a power tube in UL look more like triode than pentode, with the knee only going towards pentode style. Somewhere on the web (sorry that I do not give the reference for your convenience) you can look up characteristics of e.g. KT88 and 6550, for pentode, UL and triode operation. Should you want a comparison of output power, Rp and distortion vs. G2 tap position going from pentode to triode, you can go to my post #46 in the "UL to triode" thread here under "Tubes". From the points given you can draw graphs. These are for a KT88 which was what I could get hold of, but I also state there that they generally represent all UL operation (obviously with changed values). You are so right; UL characteristics for tubes are hard to find.
I am not able to open your complete circuit for some reason or other (right hand side will not come on), but I am worried by the G2 resistor of 4.7K. This is very high and would compromise UL operation - they form separate loads for the G2 instead of direct connection to the transformer taps. Do I understand this correctly? Why do you use these resistors? EL34 circuits seem to work better with a 1K series resistor (manufacturer data), but in most other applications there are no series screen resistors (or maybe just 100 ohms to swamp h.f. oscillation tendencies).
For the rest you should be OK if you follow basic principles, which at a first glance you seem to be doing. Just a warning if necessary: Any red glow on the anode is out; if this happens you are exceeding the max. plate dissipation quite substantially and should switch off and investigate.
Perhaps more later, have fun.
Dom,
I had a quick look at what you've done so far. Johan has it right - drop those screen resistors on the Ultralinear taps to 470R or even 220R.
Use separate cathode resistors of 560R on each output pentode cathode rather than a common 270R (this will give you much better balance and it will run the tubes at slightly lower anode current). Use bypass cap across each one. I think these two simple things will improve the sound significantly.
Cheers,
Ian
I had a quick look at what you've done so far. Johan has it right - drop those screen resistors on the Ultralinear taps to 470R or even 220R.
Use separate cathode resistors of 560R on each output pentode cathode rather than a common 270R (this will give you much better balance and it will run the tubes at slightly lower anode current). Use bypass cap across each one. I think these two simple things will improve the sound significantly.
Cheers,
Ian
"Click for bigger" just opens the same page.
I understand why narrow column widths are more readable, but this is too narrow for even my narrow brain (see photo).
> I also have 4k7 resitors in the screen lines
Why? That is a substantial drop in G2 voltage, peak current output, and a substantial reduction in UL feedback. A small, ~hundreds of ohms, resistor is wise to keep screen current from soaring in overload,essential in guitar-duty; but in normal use it should be less than maybe 10% voltage drop and probably much-much less than the ~10K and very-variable impedance of the screens. 100-1K, not 5K.
ECL82 is a "cheap" tube for "cheap" TV sets. Probably an excellent tube, but nobody was interested in its ultralinear uses. You are a pioneer.
Here is a set of Triode, Pentode, and UL-mode curves for a larger more famous tube. You can see the trend. There are others in the list, but no small power pentodes.
Frankly, the ECL82 pentode smells enough like a half-size 6BQ7/EL84 that you could probably use a 6BQ5 plan but cut the supply voltage down to 75% or 70% to stay inside the plate dissipation rating. The world is filthy with 6BQ5 P-P UL plans, and even data.
ECL82 power-side is not a bad power triode. Rp is like 1K3 so it should make decent power with low distortion at 5K SE, maybe 10K P-P. We could wish for lower Rp, but we do get easy drive. While some data suggests UL has better overall performance balance than pentode or triode, I always think of UL as a half-way approach, and tend to go to either extreme: pentode for POWER, triode for sweet and simple.
I disagree about separate bias resistors, but I'm an old grump and we never used two parts when one would do. The few exceptions, like the Heathkit Williamson, didn't run any better and sure were more work to assemble. But fashions change, and you may like Ian's way better.
I understand why narrow column widths are more readable, but this is too narrow for even my narrow brain (see photo).
> I also have 4k7 resitors in the screen lines
Why? That is a substantial drop in G2 voltage, peak current output, and a substantial reduction in UL feedback. A small, ~hundreds of ohms, resistor is wise to keep screen current from soaring in overload,essential in guitar-duty; but in normal use it should be less than maybe 10% voltage drop and probably much-much less than the ~10K and very-variable impedance of the screens. 100-1K, not 5K.
ECL82 is a "cheap" tube for "cheap" TV sets. Probably an excellent tube, but nobody was interested in its ultralinear uses. You are a pioneer.
Here is a set of Triode, Pentode, and UL-mode curves for a larger more famous tube. You can see the trend. There are others in the list, but no small power pentodes.
Frankly, the ECL82 pentode smells enough like a half-size 6BQ7/EL84 that you could probably use a 6BQ5 plan but cut the supply voltage down to 75% or 70% to stay inside the plate dissipation rating. The world is filthy with 6BQ5 P-P UL plans, and even data.
ECL82 power-side is not a bad power triode. Rp is like 1K3 so it should make decent power with low distortion at 5K SE, maybe 10K P-P. We could wish for lower Rp, but we do get easy drive. While some data suggests UL has better overall performance balance than pentode or triode, I always think of UL as a half-way approach, and tend to go to either extreme: pentode for POWER, triode for sweet and simple.
I disagree about separate bias resistors, but I'm an old grump and we never used two parts when one would do. The few exceptions, like the Heathkit Williamson, didn't run any better and sure were more work to assemble. But fashions change, and you may like Ian's way better.
Johan Potgieter said:
I did them by hand by actually measuring with the valve in a tester. And yes
I plotted pentode curves (with the screen at 225V) I realise now thats wrong!
I've bookmarked that thread...I'll read it tomorrow. I've read your bit, but still not sure how I plot a real UL curve. I was hoping to do it with my valve tester and a real valve. I think using a formula of "Vs = Vb - 43% (Vb - Va)" where Vb is my chosen working point should work? Only trouble is I can only set Vs to a granularity of about 25V so will have to interpolate the results. I've done one graph and it "looks" right. With all the lines going to zero and a mixture somewhere between a pentode and a triode with straight evenly spaced lines but a more closely packed round 0. I'll post a copy of the graphs up tomorrow to check if you like?
Will do both. The large screen resistors were definitely a dead end botch to try and "alter" the pentode curves - but as I'm running in UL I realise this was a mistake!
Not that I understand why having two cathode bypasses will have better balance though, but others elsewhere have suggested it!
Cheers
Dom
Not that I understand why having two cathode bypasses will have better balance though, but others elsewhere have suggested it!
Cheers
Dom
Thanks PRR,
I've changed my website and expanded my horizons!
The UL curves on that site seem to confirm that I'm on the right track so will do some detailed measurements today, or may even try and write some kind of computer model!
I'll put up a second follow-up page with some curves later this week
Cheers
Dom
I've changed my website and expanded my horizons!
The UL curves on that site seem to confirm that I'm on the right track so will do some detailed measurements today, or may even try and write some kind of computer model!
I'll put up a second follow-up page with some curves later this week
Cheers
Dom
Hi PRR,
Nah, there are enough PP/UL circuits out there for ECL82/6BM8 and their 300mA heater brothers like PCL82/16A8. Even commercial ones from Supertone, Pamphonic, and last but definitely not least, and maybe most famous, from Mullard themselves.
Edit: A quick ´net search reveals that there is a PP/UL kit in current production, have a look at Stoetkit junior. And WAD (World Audio Design, UK) had their KECL82 PP/UL kit for sale until recently.
BTW, if someone is interested in the plate curves of the power pentode section being triode strapped, the plot, among others, can be found at my site here.
Tom
Nah, there are enough PP/UL circuits out there for ECL82/6BM8 and their 300mA heater brothers like PCL82/16A8. Even commercial ones from Supertone, Pamphonic, and last but definitely not least, and maybe most famous, from Mullard themselves.
Edit: A quick ´net search reveals that there is a PP/UL kit in current production, have a look at Stoetkit junior. And WAD (World Audio Design, UK) had their KECL82 PP/UL kit for sale until recently.
BTW, if someone is interested in the plate curves of the power pentode section being triode strapped, the plot, among others, can be found at my site here.
Tom
Thanks for the triode strapped graphs, they helped me discover that the model I've just been trying to use seems to be wrong - will double check it - I get
Ia=20ma, Va=210, Ig=-20, but you get Ia=~40ma on the graph so I'm going wrong somewhere again!
To everyone else:
I tried lowering the B+, altering the quiescent point to something below the graph, plotted a full push-pull load line, took out the screen stoppers and replaced as advised, just tried it: it sounds bloody awfull!
Think I've been looking at this too long...will have to leave it a few days and try again...I'll be posting back again in the near future.
Maybe I should just go with the original (good sounding design) and get better PSU trannys!
I do want to try and get the theory right though!!!
Cheers
Dom
Ia=20ma, Va=210, Ig=-20, but you get Ia=~40ma on the graph so I'm going wrong somewhere again!
To everyone else:
I tried lowering the B+, altering the quiescent point to something below the graph, plotted a full push-pull load line, took out the screen stoppers and replaced as advised, just tried it: it sounds bloody awfull!
Think I've been looking at this too long...will have to leave it a few days and try again...I'll be posting back again in the near future.
Maybe I should just go with the original (good sounding design) and get better PSU trannys!
I do want to try and get the theory right though!!!
Cheers
Dom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.