I made some simulations with the latest circuit you have posted.
At present the OPT makes a bit too low load impedance. Some 3k...3k5 would give almost same output power but with much lower THD.
The expected output power is just below 4 W. THD is about 3.7 %.
With 3k OPT the output power would be 3.5 W at THD = 2.7 %
The overall linearity could be improved by biasing the 12AT7 to better distortion canceling.
At present the OPT makes a bit too low load impedance. Some 3k...3k5 would give almost same output power but with much lower THD.
The expected output power is just below 4 W. THD is about 3.7 %.
With 3k OPT the output power would be 3.5 W at THD = 2.7 %
The overall linearity could be improved by biasing the 12AT7 to better distortion canceling.
Last edited:
For my learning, Can I ask what are the negative side effects if grid stopper is 10K+ for a triode connected EL34?
The pentode wired EL34 has negligible stray capacitances, so using a large value grid stopper doesn't present high frequency response problems. When you triode wire the EL34, the Miller capacitance becomes a bit of an issue!
Volume pot is wired backward; when full down it shorts the source, half-up it adds a lot of resistance to the grid. Assuming 100pFd at triode grid, -3dB at 33KHz, -1dB @ 16KHz. Right-way, worst-case 25K, -3dB@66KHz. (And 12AT7 Cin is probably less than 100pFd.)
The triode's "Rout" does not "load the output valve"-- the EL34 grid loads the triode.
EL34 in pentode has <2pFd grid to plate. Taking grid to plate gain as around 10, this is an added 20pFd load on the triode driver, which is probably insignificant. (10K rp and say 30pFd total at EL34 grid is 500KHz.) I don't see EL34 triode Cgp data. Assuming not too unlike 300B, 15pFd Cgp, times EL34(tri) gain of 7, is 105 say 120pFd at EL34 grid. 135KHz. Since we don't hear 135KHz well, and there is no overall NFB to wonk-out, I'd say this is unimportant. (I'd tend to much lower current in the driver.)
The 33K has 0.6 Watts in it. Always double then round-up. 2 Watt part. EL34 cathode resistors, 1W actual, use 2W, but I'd use 5W for looks. OTOH the 470r's actual dissipation is so low you can't find a resistor so small it will be in trouble. I'd use 1/2W because it is traditional, and because 1/4W part leads look lame on 9- and 8-pin tube sockets.
Do you really want to spend for a 5-Watt trimmer? Frankly, EL34 at rational voltage normally bias-up just fine withOUT trimming. Trimming adds trouble, also cost.
And if you feel you must trim, you don't need a huge pot. Lever the grid voltage, not the whole cathode current.
2K load, 4K load.... yes, we can find "more optimum" for whatever we want optimumed. 2.5K and less may make more raw power with heavy tube flavor. 3.5K and up makes less distortion and better speaker damping. But triodes are tolerant.
The triode's "Rout" does not "load the output valve"-- the EL34 grid loads the triode.
EL34 in pentode has <2pFd grid to plate. Taking grid to plate gain as around 10, this is an added 20pFd load on the triode driver, which is probably insignificant. (10K rp and say 30pFd total at EL34 grid is 500KHz.) I don't see EL34 triode Cgp data. Assuming not too unlike 300B, 15pFd Cgp, times EL34(tri) gain of 7, is 105 say 120pFd at EL34 grid. 135KHz. Since we don't hear 135KHz well, and there is no overall NFB to wonk-out, I'd say this is unimportant. (I'd tend to much lower current in the driver.)
The 33K has 0.6 Watts in it. Always double then round-up. 2 Watt part. EL34 cathode resistors, 1W actual, use 2W, but I'd use 5W for looks. OTOH the 470r's actual dissipation is so low you can't find a resistor so small it will be in trouble. I'd use 1/2W because it is traditional, and because 1/4W part leads look lame on 9- and 8-pin tube sockets.
Do you really want to spend for a 5-Watt trimmer? Frankly, EL34 at rational voltage normally bias-up just fine withOUT trimming. Trimming adds trouble, also cost.
And if you feel you must trim, you don't need a huge pot. Lever the grid voltage, not the whole cathode current.
2K load, 4K load.... yes, we can find "more optimum" for whatever we want optimumed. 2.5K and less may make more raw power with heavy tube flavor. 3.5K and up makes less distortion and better speaker damping. But triodes are tolerant.
Attachments
Thanks for the response.
Am I right in thinking you're suggesting I redraw my loadlines for a 3.5K load, rather than simply substituting such a load into the current circuit?
Matt.
An old pair of Mission 752 Freedoms. With a sensitivity of 90dB 1W@1M, they're probably borderline for compatibility with a SET. However, feeling your pain...
Buying new speakers is not an option at present so they'll have to do until my winning ticket comes in.
I asked around in the speaker forum before I started the project - the consensus was that I'd just be ok, but the frequency response might need some tweaking.
I hope your project works out well for you. Not a million miles away from mine, on paper at least. Looks like you're pushing for more power though!
Matt.
Yes, frequency response, in general you are using a speaker designed for the 40watt transistors era, the impedance/inductance of most 80s speakers is terrible as far as my experience goes for SE amps. Screaming midrange sound. The exception is paper alnico Tannoys. I saw tannoys -30 years for 300$.
Further down the road you have the possibility to bi-amp. Plan everything to the last detail before starting the project otherwise your basement will fill with spare parts.
Thanks for the reply. A lot of information to digest there, I think I've got my head around most of it
Took me a while to see the difference, but it's patently obvious now I have
.
Interesting. It was my - presumably erroneous - understanding that the output of one stage loaded the input tof the next. Have I got that backwards, or am I fundamentally failing to grasp the concept?
Aside from that, I've taken your advice and ditched the trimmer. The mathematical value for Rk on the EL34 is 212R, what's 8 ohms between friends.
I've also taken a look at the loadlines for the ECC81, and redrawn them to give a slightly higher output, with considerably less current, which surely can't be a bad thing.
However, that's increased r out to 16K, proof, should we need it, that there's no such thing as a free lunch.
Am I better going for the higher current, lower r out, or the inverse? Or perhaps trying trying again with an anode load of 50K?
Thanks for the help,
Matt.
Took me a while to see the difference, but it's patently obvious now I have
. Interesting. It was my - presumably erroneous - understanding that the output of one stage loaded the input tof the next. Have I got that backwards, or am I fundamentally failing to grasp the concept?
Aside from that, I've taken your advice and ditched the trimmer. The mathematical value for Rk on the EL34 is 212R, what's 8 ohms between friends
.I've also taken a look at the loadlines for the ECC81, and redrawn them to give a slightly higher output, with considerably less current, which surely can't be a bad thing.
However, that's increased r out to 16K, proof, should we need it, that there's no such thing as a free lunch.
Am I better going for the higher current, lower r out, or the inverse? Or perhaps trying trying again with an anode load of 50K?
Thanks for the help,
Matt.
Yes, that's backwards.
Without using computer simulation, you'll have to go by the spacing on your load lines. If all else fails, look at the operating points provided in the datasheets.
Thanks.
Better is probably the wrong term, but if you were designing a circuit with an ECC81 driving an EL34, would you opt for a driver stage with a higher Ia and lower r out, or vice versa?
Rl = 33K, Ia = 9.4mA and r out = 10K.
Rl = 75K, Ia = 4.1mA and r out =16K.
Is there a right or wrong choice between the two?
I haven't drawn the loadline, but presumably an anode load of 50K would result in Ia of >4.1mA but <9.4mA, and an r out of >10K but <16K. Would this be a better match for the EL34?
Matt.
No.
"Wrong" would be when the driver runs so starved it can't slam the EL34 properly. EL34 comes from a line of tubes that are "easy" to drive. You could run much less current. Note that assuming 20V into a 200K grid, it is only 0.1mA actual load. The driver current must be more, but needn't be 100X more.
The other "wrong" is when you run so much driver current that it jacks-up the retail price and/or reduces your net Profit. Essentially all audio companies eventually go broke; this hastens the bankruptcy. (DIY economics is very different.)
Further to the above, is it acceptable practice - assuming there's enough current available - to power both channels from a single transformer?
Seems that a lot of amps have a seperate mains transformer for each channel. I'm assuming there must be a good reason to add extra cost/weight?
Matt.
Seems that a lot of amps have a seperate mains transformer for each channel. I'm assuming there must be a good reason to add extra cost/weight?
Matt.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.