You'll be just fine with 5k. If you choose triode, it's easy enough. You could also consider a separate screen grid regulated supply if you wanted to try pentode. Either will work.
I run LT Spice in Wine in my ubuntu machine all the time. Easy to draw with and with some patience, you can also do simulations.
Stay as true to the 5-20 circuit as you can. If you want to get creative later, that's great, but the circuit is proved as is. The big improvements are in the power supply.
Stuart
4.3K loads the tubes a bit more than 5K, usually resulting in more power out and more distortion. Do you any idea what power the transformers are rated for?
If excessive sand in the circuit gives you hives, then TubeMack's soloist is almost there, with a CCS under the LTP. It's using KT88's in a 5-20 variant, although it's UL, not triode.
Also consider Triode Dick's Mono Bill and Mono Bill II, two stage no feedback KT88, or the original Mono Bill, a three stage integrated amp.
I store my digital music on a PC server (the usual, *.FLAC, EAC, etc.) and tired of the inexplicable failures running XP: sudden turn-offs, "dings" in the middle of cuts (when notification signals were supposed to be turned off), interruptions in playback, etc. A friend suggested a Linux variant. The best surprise was that the audio out to my system sounded better than with XP! I use a Hagerman HagUSB, Monarchy DIP and a modified LiteDAC 50.
Changed my main PC to ubuntu also, and I'm not going back!
Re: the amp circuit, you can experiment with different output tubes at your leisure. If you wire it with an eye towards compatibility, you can swap in KT88/6550/EL34/KT66 to your heart's content. But it's really hard to beat EL34s, 6SN7s and EF86s for great sound. Not saying it can't be done, but I don't think you'd be embarrassed by the sound of your amp ever.
Stuart
Yea, I do *like* the idea of KT88, just can't pop for a decent quad right now (no job, kids, school, etc. -- hoping to be employed by fall). I already have a lot of EL34, some of them pretty decent ones.
The transformers were from an amp with PP 7355, which are supposed to dissipate 25W so I figure EL34 in triode mode should be alright(?). Physically they are as large or larger than my set from an amp which is basically a Mullard 5-20 (a Knight KA-95...which ain't working either).
There seem to be WILD descrepencies in the price of 6SN7's on ePay. Are the cheapo ones decent enough to use? otherwise I have piles of 12A*7 tubes, many of which are rather respectable.
I'm mostly lossless codecs too serving a few TB off a couple of machines. I have A LOT of computers in my house. Mostly running Ubuntu. A couple with windows still because of software that doesn't play nice in Wine but mostly because of Windows only driver issues...including my best DAC (Musiland Monitor02). I haven't noticed a difference between Linux and Win with ASIO output but haven't really been able to do an ABX. Windows from DS is awful though.
I am a CS student and know there is a good chance I'll have to develop in and for Windows next year when I (god willing) find a job. I'll manage but I sure do prefer Linux.
The transformers were from an amp with PP 7355, which are supposed to dissipate 25W so I figure EL34 in triode mode should be alright(?). Physically they are as large or larger than my set from an amp which is basically a Mullard 5-20 (a Knight KA-95...which ain't working either).
There seem to be WILD descrepencies in the price of 6SN7's on ePay. Are the cheapo ones decent enough to use? otherwise I have piles of 12A*7 tubes, many of which are rather respectable.
I'm mostly lossless codecs too serving a few TB off a couple of machines. I have A LOT of computers in my house. Mostly running Ubuntu. A couple with windows still because of software that doesn't play nice in Wine but mostly because of Windows only driver issues...including my best DAC (Musiland Monitor02). I haven't noticed a difference between Linux and Win with ASIO output but haven't really been able to do an ABX. Windows from DS is awful though.
I am a CS student and know there is a good chance I'll have to develop in and for Windows next year when I (god willing) find a job. I'll manage but I sure do prefer Linux.
Like you I have lots and lots of 12AU7, 12AT7 and 12AX7s. Please avoid them. The octals are much better. If you must use a 9-pin driver, the 6CG7/6FQ7 is electrically equivalent to the 6SN7. I haven't made an adapter and directly plugged one in for the other, but my sense is that they are equivalent sonically, with allowances for individual types, of course.
Again, within the context of your project, it's a big job, keep it as simple as possible. Buy a couple of inexpensive NOS 6SN7 variants. Even the cheap ones sound good. Check out Jim McShane's website. He's an honest guy. his advice is golden. He suggests new reissue Tun Sols @ $18+ each
Also consider this thorough review of 6SN7s from Brent Jesse. Again, the best information on your choices.
I like to set up the input with stable DC bias and AC couple into it with a 5-10 hz -3db point. I'm not sure for the EF86 if that would be around 250k to ground on the grid and a .1u couple to the input pot. around 6hz -3db point....
You still need the cap on the gate of the LPT lower SN7 biased by the 1M, .05 .1u would work and you have excessive mismatch in the resistors. 33k-27k only need 33k-30k and they can be matched if you use a CCS.
If you use the common 10 ohm bias check resistor, the output tubes will need to matched, if you allow for bias adjustment for each output tube use separate resistors at 10 ohms each. It will allow you to minimize the offset current through the transformer that helps the low end response, preventing early saturation, and you can use tubes that are not "matched".
You still need the cap on the gate of the LPT lower SN7 biased by the 1M, .05 .1u would work and you have excessive mismatch in the resistors. 33k-27k only need 33k-30k and they can be matched if you use a CCS.
If you use the common 10 ohm bias check resistor, the output tubes will need to matched, if you allow for bias adjustment for each output tube use separate resistors at 10 ohms each. It will allow you to minimize the offset current through the transformer that helps the low end response, preventing early saturation, and you can use tubes that are not "matched".
Last edited:
Furthering Roline's comments, consider a 10M45S CCS on the 6SN7 tail (super easy to implement, you need a current set R, and a gate stopper R), and reduce the value of P1 down to 100K-250K or so.
Also, you will need more than -30V to bias the EL34's. Shoot for a range of -30 to -40V or so. Separate bias pots for each tube are handy, and consider designing the bias supply to fail safe if the pot wiper lifts.
Someone with more experience than me may suggest a cap on the input depending on the quality of your transformers. Refer to Sgregory's Opus thread.
FWIW, I've been running my Music machine at 400V/60ma, sounds great.........
Also, you will need more than -30V to bias the EL34's. Shoot for a range of -30 to -40V or so. Separate bias pots for each tube are handy, and consider designing the bias supply to fail safe if the pot wiper lifts.
Someone with more experience than me may suggest a cap on the input depending on the quality of your transformers. Refer to Sgregory's Opus thread.
FWIW, I've been running my Music machine at 400V/60ma, sounds great.........
Last edited:
The schemo you posted is fixed bias..........cathode bias will have much larger cathode R's and no negative rail.
Look at the Opus for the CCS in the LTP tail and the Opus and Music Machine for the bias supply circuit.
Playing around with various sweep tube bias, the 18watt biased at about -25 volts but the larger 35watt tubes biased at closer to -35 volts with one that wanted -55. I believe that you will also find this true going from EL34's to KT88's, so design the bias supply low enough to throttle back the tubes allowing yourself the option of all tube flavors. You may discover that you like the KT88 over the EL34's if you cant afford the holy grail mullard flavor. I just added a couple of the extra 6.3 volt windings to the 55 volt winding used for bias to go low enough. The attached schematic could be modified to just replace the output tubes, it has enough tricks of the trade for a very decent and well behaved amp, local feedback around the output tubes, in the form of plate-to-grid feedback coined Schade.
If you decide to go to cathod bias, the resistors grow in size and you loose voltage swing, output power....
If you decide to go to cathod bias, the resistors grow in size and you loose voltage swing, output power....
Attachments
Last edited:
mfaughn,
1) No need for an input cap, unless you suspect you have DC leaking through some source output cap.
2) DC coupling as many places as possible (1st stage-phase inverter) is a goal to shoot for. Don't add signal circuit caps where not needed.
3) IF you do add a CCS in the tail of the phase inverter (6SN7), THEN the plate loads can and should be equal, as stated by roline. The numbers you have worked for Eico. ;^)
4) Yes, the EL34 will need a lower voltage (greater negative voltage) bias supply. -45V should cover any contingency.
5) As roline stated, better to have individually adjustable bias for each output tube. Reduce those 10ohm cathode resistors to 1 ohm. They are only useful for measuring bias current. Ideally, the output cathodes should have zero ohms between them. 2 ohms is a lot better than 20. If you stick with the larger resistors, consider connecting the cathodes together with a huge film cap, or non-polar electrolytic. Personally, I'd rather use the smaller resistors.
6) At the input, add a 1M resistor from grid to ground. Do not rely on a pot wiper for a grid leak.
7) Do you need the input pot? If you are using a line stage of any sort, best not to add another pot. If you are building an integrated amp, that is, including a volume control and selector switch, I'd use a smaller value pot, especially if you use the EF86 in triode mode. In pentode, Miller capacitance is not an issue, in triode, you have to watch out or your volume pot becomes a tone control.
8) Why not use the EF86 in pentode? It has none of the pentode edge and eliminates the Miller problem. The amp has lots of gain as designed by Eico. Probably it will still be enough with the EF86 in pentode, but you may need an active line stage depending on your source output.
9) For feedback: when you first build it, leave the NFB disconnected. Measure the gain. Then add a pot in the NFB resistor location. Set it for 20dB of feedback to start. Fine tune it by ear, but 20dB will probably be close. When you find the number that suits you best, replace the pot with a fixed resistor. Then, use a 1k or so square wave and a scope, look at the leading edge. Add capacitance, if needed, across the feedback resistor to tame any leading edge ringing. Don't overdo it, one or two waves is okay.
10) If you are considering compatibility with KT88/6550 types, you'll need a -75V bias supply, reduce those output grid leaks to 100k and increase the coupling caps to 0.47uF.
11) That's all the easy stuff, wait until you ask about caps, resistors, wire, chassis materials, etc. LOL!
Stuart
1) No need for an input cap, unless you suspect you have DC leaking through some source output cap.
2) DC coupling as many places as possible (1st stage-phase inverter) is a goal to shoot for. Don't add signal circuit caps where not needed.
3) IF you do add a CCS in the tail of the phase inverter (6SN7), THEN the plate loads can and should be equal, as stated by roline. The numbers you have worked for Eico. ;^)
4) Yes, the EL34 will need a lower voltage (greater negative voltage) bias supply. -45V should cover any contingency.
5) As roline stated, better to have individually adjustable bias for each output tube. Reduce those 10ohm cathode resistors to 1 ohm. They are only useful for measuring bias current. Ideally, the output cathodes should have zero ohms between them. 2 ohms is a lot better than 20. If you stick with the larger resistors, consider connecting the cathodes together with a huge film cap, or non-polar electrolytic. Personally, I'd rather use the smaller resistors.
6) At the input, add a 1M resistor from grid to ground. Do not rely on a pot wiper for a grid leak.
7) Do you need the input pot? If you are using a line stage of any sort, best not to add another pot. If you are building an integrated amp, that is, including a volume control and selector switch, I'd use a smaller value pot, especially if you use the EF86 in triode mode. In pentode, Miller capacitance is not an issue, in triode, you have to watch out or your volume pot becomes a tone control.
8) Why not use the EF86 in pentode? It has none of the pentode edge and eliminates the Miller problem. The amp has lots of gain as designed by Eico. Probably it will still be enough with the EF86 in pentode, but you may need an active line stage depending on your source output.
9) For feedback: when you first build it, leave the NFB disconnected. Measure the gain. Then add a pot in the NFB resistor location. Set it for 20dB of feedback to start. Fine tune it by ear, but 20dB will probably be close. When you find the number that suits you best, replace the pot with a fixed resistor. Then, use a 1k or so square wave and a scope, look at the leading edge. Add capacitance, if needed, across the feedback resistor to tame any leading edge ringing. Don't overdo it, one or two waves is okay.
10) If you are considering compatibility with KT88/6550 types, you'll need a -75V bias supply, reduce those output grid leaks to 100k and increase the coupling caps to 0.47uF.
11) That's all the easy stuff, wait until you ask about caps, resistors, wire, chassis materials, etc. LOL!
Stuart
Last edited:
Oh, one more thing. If you can, build a completely separate power supply for the input stages. No need for any fancy regulation, just simple and straightforward, but:
Tube rectify
Good electrolytics (Nichicon PW series are cheap and excellent) or good film caps like the Mundorf M-Tube Caps
Choke filtering
EI core transformers
For the output (PP) section, a toroid is probably okay, but in general, the low pass filtering given by an EI core transformer is to your benefit.
If you must use a toroid, be sure it has an electrostatic shield and ground it. Antek has those in the product line up.
Damper diodes are best, after that other people can tell you which VT rectifier they like best.
Stuart
Tube rectify
Good electrolytics (Nichicon PW series are cheap and excellent) or good film caps like the Mundorf M-Tube Caps
Choke filtering
EI core transformers
For the output (PP) section, a toroid is probably okay, but in general, the low pass filtering given by an EI core transformer is to your benefit.
If you must use a toroid, be sure it has an electrostatic shield and ground it. Antek has those in the product line up.
Damper diodes are best, after that other people can tell you which VT rectifier they like best.
Stuart
My bad, it must have been the schematic before or after that that had the EF-86 as the amplifier. I single one will certainly give as much gain as both sides of the average 12AX7. I know I saw one in around there somewhere, at least it had EL-34s in it.
Thatch
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.