I want to split a guitar input signal early in the circuit, and im trying to figure out if i'm going to run into signal attentuation problems, noise problems or phase difference problems.
Imgur: The most awesome images on the Internet
If a picture is worth a thousand words - is there a superior method to follow?
An externally hosted image should be here but it was not working when we last tested it.
Imgur: The most awesome images on the Internet
If a picture is worth a thousand words - is there a superior method to follow?
I would recommend the first option, as a typical triode preamo stage has a high input impedance relative to a guitar pickup or FX pedal output. The 3rd one also looks promising, but would cause interaction between the tone stacks.
Sent from my LG-D801 using Tapatalk
Sent from my LG-D801 using Tapatalk
Thanks!
Is there anything special I could do for that, or do you think it would be good to do the normal 100k over 1.5k fender type biasing?
To reduce interaction between the two tone stacks in your figure 3, I assume?
One way would be to drive two separate cathode followers from the output (anode) of the triode. Each cathode follower would then drive one tone stack. There won't be any interaction to speak off between the two tone stacks.
Of course two cathode followers is an entire 12AX7. If all you want is a completely transparent buffer (no audible distortion or "tone" of its own), you could use a pair of high voltage MOSFET source-followers rather than cathode followers. A pair of the little LND-150 Supertex MOSFETs is probably all you need, as long as you keep within the limited power handling capability of the tiny little package. Run them at 1 mA or so, with around 150 V or less across source and drain, and they'll do the job perfectly, and stay cool and happy.
You didn't mention what will happen to the two signals after they're split off - will they go through different processing, and then be heard simultaneously (mixed back together, or sent to separate amps)? If so, you could potentially run across phasing issues when mixing the two signals back together, and it would be wise to include some way to reverse the phase of one of the two signals at will.
By the way - there's also the possibility of making a little custom buffer/splitter in a guitar effects pedal, external to your guitar amp. You can do the job very nicely with one dual opamp like a TL072, which can provide both the splitting function, and a switchable phase reverse for one signal.
-Gnobuddy
I forgot something. I'm sorry. I completely forgot. This amp I'm putting together is supposed to be a low watt all-in-one - uh... it's sorta like a wet/dry rig. I want to distort one circuit and keep the other one clean. I don't mind if there's a VERY SLIGHT chorus effect sometimes but I don't want anything too weird.
http://chasingtone.com/dvnator/wp-content/uploads/sites/6/2016/01/JCMSL3_schematic.jpg
That's my inspiration for the "B" circuit - which is not going to be straight into the tone stack. I'm planning to pretty much copy that circuit.
The "A" circuit is to be a lot like a Fender Princeton, but with an ECC99 push-pull power amp.
I am interested in saving filament current so maybe I could try a bit of solid state as you suggested.
http://chasingtone.com/dvnator/wp-content/uploads/sites/6/2016/01/JCMSL3_schematic.jpg
That's my inspiration for the "B" circuit - which is not going to be straight into the tone stack. I'm planning to pretty much copy that circuit.
The "A" circuit is to be a lot like a Fender Princeton, but with an ECC99 push-pull power amp.
I am interested in saving filament current so maybe I could try a bit of solid state as you suggested.
Last edited:
That combination will tend to give you fairly neutrally-biased stage with a good bit of clean headroom, so ideal for your clean channel.
AbsolutelyQUOTE=Gnobuddy;4969286] in which case, you might as well, make the dirty channel more like a regular JCM800 preamp.
Yes, depends how transparent you want. Cathode followers have a pleasant asymmetry helping to add some even harmonic warmth, which their MOSFET equivalent doesn't. However, the latter wins in terms of sheet simplicity.
IMO a FET capable of handling at least 250 to 300V would be a safer bet, with more safety margin.
To keep it simple, just keep in mind that each gain stage inverts the signal phase, while a cathode follower (or source follower) maintains the same phase. Then count b the stages and make sure both channels end up being in the same phase when you combine them back together again. Hope that made sense.
Sent from my LG-D801 using Tapatalk
Last edited:
A quick couple of comments about that schematic. The 1.5k/22uF combination on the 3rd stage cathode will tend to give you muddy/woofy bass. I would recommend taking the cap down to around 1uF to tighten it up. With the lower cathode resistor you could go up to a 2uF cathode cap, and you'll really have to try both cap values to see which sounds best to your ears.
Also if you're not going to use a cathode follower, then I recommend reducing the third stage's cathode resistor to around 600 to 620 ohms, to provide a similar asymmetric soft limiting effect, as you would get with a CF (note, FET source followers also don't have this asymetric characteristic of a tube CF).
I would also recommend using "Jalapeno" tone stack values for a a Marshall style stack that works well anode-driven. It's quite similar to what you already have there, but the slope resistor is 56k instead of 47k, the values of the Bass and Mid caps are swapped, and the 10k fixed resistor is replaced by a 50k pot with the now 0.01uF Mid cap going to its wiper. If you want to stay with a 2-knob control, you can substitute the 50k pot for a pair of 27k resistors in series, with the 0.01uF cap connected to the centre connection of the two resistors. Hope that made sense.
Last edited:
The LND-150 is actually capable of withstanding 500V (Vds max), so no worries on that score.
It is, however, a little TO-92 package that can only dissipate somewhere around 350 mW maximum. (Thermal resistance from junction to ambient is 170 deg C/W!)
That's why I recommend operating these (quiescent) at around 150 volts between source and drain, and 1 mA drain current. That way, the steady-state dissipation is only around 150 mW, and the device runs cool and stays happy.
This shouldn't be a problem if direct-coupled to the typical Fender (RCA, really!) 12AX7/100k/1.5k gain stage. There will usually be roughly 100 volts across the 100k anode resistor, and that's the voltage that will appear across a direct-coupled source-follower MOSFET. The rest of the B+ will appear across the source resistor, so just choose that to keep drain current to 1 mA or so, and everything should be fine.
Note that this might involve using quite a large source resistor - if the anode is sitting at, say, 220 V, you'll want to use a 220k source resistor for the LND 150 to keep its source current to 1 mA.
Even with a 220 k source resistor, output resistance will still be very low compared to the unbuffered triode.
-Gnobuddy
I'm more worried about the bass and midrange cancelling out, if the phasing is wrong and you mix the signals.
It sounds like everything is under your control, though! (No FX loop with potentially unknown phase shift to worry about.)
Sounds like a fun project, I hope you share a few more details once you get it built and working!
-Gnobuddy
I will definitely be back around. The lead time on my Edcor PTs is 6 weeks and I'm not a fast worker, but I'll be back no matter how long it takes.
I don't intend to re-combine the electrical signals at all. Both circuits will be tucked inside an oversized wooden enclosure. I did already mount two small 3.5" speakers in the front of the enclosure but they're not intended to be paired with either circuit. The back of the enclosure will have separate 1/4" speaker outs for each circuit, plus a 1/4" plug that goes to the internal speakers. It will be possible to connect either circuit to the internal speakers for quiet practicing, but the internal speakers are meant to be used for listening to music.
About fx loops. One disadvantage of my concept is that I can't separate the effects. Any effect would have to be in front of the amp and would hit both circuits. Unless I come up with some kind of an fx-loop solution I'm afraid that certain effects (like reverb and delay) may not work very well.
I do want to have an old-fashioned 12ax7 cathode follower driving the tone stack on the "dirty" circuit. That's actually what I meant when I said that the "B" circuit isn't exactly like the schematic that I shared. I'm planning to add a cathode follower to it. I'm studying up about that.
For an executive-level overview of the project - I want to distort my tone on channel B and then use channel A to "fill out" the sound. For the B channel I just want something like a JCM800 circuit. The A channel just needs to have some low end and some pretty highs. I want it to be kinda like Silverface/Blackface Fender clean.
Of course I'm making alterations. I decided against octal tubes already (for various reasons) and I'm not expecting to outdo any of the classic circuits. I just want something unique that works for me.
I don't intend to re-combine the electrical signals at all. Both circuits will be tucked inside an oversized wooden enclosure. I did already mount two small 3.5" speakers in the front of the enclosure but they're not intended to be paired with either circuit. The back of the enclosure will have separate 1/4" speaker outs for each circuit, plus a 1/4" plug that goes to the internal speakers. It will be possible to connect either circuit to the internal speakers for quiet practicing, but the internal speakers are meant to be used for listening to music.
About fx loops. One disadvantage of my concept is that I can't separate the effects. Any effect would have to be in front of the amp and would hit both circuits. Unless I come up with some kind of an fx-loop solution I'm afraid that certain effects (like reverb and delay) may not work very well.
I do want to have an old-fashioned 12ax7 cathode follower driving the tone stack on the "dirty" circuit. That's actually what I meant when I said that the "B" circuit isn't exactly like the schematic that I shared. I'm planning to add a cathode follower to it. I'm studying up about that.
For an executive-level overview of the project - I want to distort my tone on channel B and then use channel A to "fill out" the sound. For the B channel I just want something like a JCM800 circuit. The A channel just needs to have some low end and some pretty highs. I want it to be kinda like Silverface/Blackface Fender clean.
Of course I'm making alterations. I decided against octal tubes already (for various reasons) and I'm not expecting to outdo any of the classic circuits. I just want something unique that works for me.
So you're going to have at least one more bottle than the schematic you linked to earlier - and, if it's a 12AX7, that means you'll have two more triodes available.
One of those could be the cathode follower in the "dirt" channel.
The other one could drive the "clean" channel tone controls.
My thought is to split the signal at the anode of the input triode (V1a in the schematic you linked). Have V1a feed two pots, one the "gain" pot for the dirt channel, the other the "volume" pot for the clean channel.
The "dirt" channel topology then becomes something like: input triode, "dirt" gain control, triode gain stage, Twin-Tee mid scoop/ boost filter, cathode follower, tone controls, master volume (dirt), and on to the power amp.
The clean channel topology would be: input triode (shared with dirt channel), clean volume control, triode gain stage, tone controls, master volume (clean), and on to the power amp.
The only real change in the "dirt" channel topology is that the twin-tee notch/ boost filter would go between a gain stage and a cathode follower, rather than between one gain stage and a second gain stage. You can still direct-couple the cathode follower (through the two 220k resistors in the Twin-Tee filter).
This scheme shouldn't have any phasing problem, either: the extra stage in the "dirt" channel is a cathode follower, which doesn't invert phase. So both "clean" and "dirt" channel outputs should be nominally in-phase with each other (subject to minor differences due to tone control settings, etc.)
That's usually just about the best reason to DIY something!
-Gnobuddy
This is definitely a helpful discussion. Thank you.
So far I'm down to just 5! total tubes for two complete amps and if I follow the self-split PI method used in the firefly amp I could even free up another triode.
Now I'm not sure if I misunderstand something or maybe you overlooked it. The "twin-tee notch/boost" segment wouldn't work to boost a cathode follower stage, right? I don't see where the "boost" could connect. I also have understood the "deleted" stage (with the 15k cathode resistor) as a component that contributes to the way those Marshall amps sound. Is this a "pick-your-poison" decision? Either accept the phase difference or lose that gain stage?
As for the new input idea. Wouldnt the volume/gain controls after the first triode stage still interact? If I turn down the clean volume and it makes the dirty channel dirtier, that wouldn't be ideal.
So far I'm down to just 5! total tubes for two complete amps and if I follow the self-split PI method used in the firefly amp I could even free up another triode.
Now I'm not sure if I misunderstand something or maybe you overlooked it. The "twin-tee notch/boost" segment wouldn't work to boost a cathode follower stage, right? I don't see where the "boost" could connect. I also have understood the "deleted" stage (with the 15k cathode resistor) as a component that contributes to the way those Marshall amps sound. Is this a "pick-your-poison" decision? Either accept the phase difference or lose that gain stage?
As for the new input idea. Wouldnt the volume/gain controls after the first triode stage still interact? If I turn down the clean volume and it makes the dirty channel dirtier, that wouldn't be ideal.
You're welcome. 🙂
It may be a challenge to keep everything stable, with two active devices crammed into each tiny glass bottle.
I've read about a number of similar builds that ended up with incurable (or very difficult to cure) oscillation problems, simply because all those electrodes are too close to each other, and stray capacitance between them causes unwanted (positive) feedback at high frequencies.
One day I do want to try something similar with a pair of 6DX8 valves, though, just because I have a few in the junk-box.
"Self-split" is a can of worms. It only works when both output tubes are biased very hot (all the way into class A), and works quite badly even then (very poor symmetry between the two phases, inadequate drive for full output power, etc.)
I used a little LND 150 MOSFET "sourceodyne" in my amp, doing the same job as a conventional cathodyne phase splitter.
A cathodyne has 100% negative feedback, which makes it so clean it has no "tone" of its own. Unless you overdrive it, and then it tends to make quite nasty noises (Merlin Blencowe has a section about this in his preamp book).
A MOSFET "sourceodyne" also has 100% negative feedback, it is also so clean it has no "tone" of its own. It also needs no heater power or valve socket, takes up very little room, is very cheap (65 cents at Mouser.com if you buy 10 or more), and will last essentially forever as long as you keep power dissipation to 150 mW or less, and use a pair of back-to-back zener diodes between gate and source to prevent excessive voltage there.
It would connect back to the triode stage with the 10k cathode resistance, just as in the schematic you linked.
Take another peek at the way that switch works - it either grounds the "tail" of the Twin-Tee filter, or it grounds the 15k resistor, putting it in parallel with the 10k cathode resistor. It doesn't matter if that 10k/15k stage is before, or after, the Twin-Tee filter itself.
You can also put the Twin-Tee between V1b, and V2a, if you prefer. Same deal with the "boost" position, it will still operate on V1b.
There's no deleted stage - you'd have the same three stages in the same sequence in the dirt channel. You're just adding on a fourth stage, your cathode follower, between V2a and the tone stack.
I was wrong about the phase. 😱 Three inverting stages in the "dirt" channel plus a cathode follower, net result is one inversion. Two inverting stages in the "clean" channel, net result, no inversion. 😱
Silly me, I screwed up. Sorry!
Now, you can still fix this in the (separate) power amps, or even by reversing the leads to one speaker! But that last trick will only work for the "combo" version with built-in speakers; you can't rely on external speaker cabs to have the phasing you want.
Looks like you do need one more inverting stage in the clean channel, so best to go back to the idea you started with - two parallel and quite independent amps.
They wouldn't interact to any meaningful degree. The first stage has an output impedance around 43 k (rp of 12AX7 in parallel with 150k anode resistor). Use this to feed two 500k or 1M pots in parallel, and there is essentially no loading. Those volume/gain knobs will be quite independent, and won't interact.
-Gnobuddy
The size of my enclosure is relatively large. It's around 12"x9", and about 10" in height as well. I'm not going to have to pack things TOO terribly tight.
The PTs are two EDCOR xpwr024 and they each have up to 2.0A available filament current. The ECC99 will take 0.8 and the 2 ECL84s each have 0.72. I could add as many as five, maybe even six 12a_7 type tubes to that if I was willing to test the tolerance of the PTs, but instead I would prefer to do the opposite and stay very conservative.
After your explanation about the first stage volume controls I'm feeling that's the basic direction that I should go, but I also got distracted by a new thought. I'm trying to decide if a stacked "blend" knob that controls them both simultaneously would be nice to have. Or if it's even possible to find one that would suit the purpose. It has the feel of an idea that seems good at first, but might just create new problems that I don't need.
The PTs are two EDCOR xpwr024 and they each have up to 2.0A available filament current. The ECC99 will take 0.8 and the 2 ECL84s each have 0.72. I could add as many as five, maybe even six 12a_7 type tubes to that if I was willing to test the tolerance of the PTs, but instead I would prefer to do the opposite and stay very conservative.
After your explanation about the first stage volume controls I'm feeling that's the basic direction that I should go, but I also got distracted by a new thought. I'm trying to decide if a stacked "blend" knob that controls them both simultaneously would be nice to have. Or if it's even possible to find one that would suit the purpose. It has the feel of an idea that seems good at first, but might just create new problems that I don't need.
Sorry I wasn't clear - that JCM-SL3 schematic shows that the output valves combine both a triode and a pentode in one bottle. This is the "tightly packed electrodes" part that I'm concerned about.
I know that there were a lot of triode-pentode valves, usually designed for use in TV sets, where the two valves would perform very different functions that didn't interact. The JCM-SL3 is different - for instance, V2a is two valve stages "ahead" of the output pentodes, yet is crammed into the same little glass bottle as output pentode V2b.
If even a little bit of the output signal (from V2b anode) leaks back into the grid of V2a (via stray capacitance due to V2a and V2b being a fraction of an inch apart), you can end up with an unstable amp that oscillates like crazy.
I've never built an amp like this, but have read about a few people who did have this sort of problem. For instance, there are reports online from several people who tried and failed to build a three-stage guitar amp into one 6AF11 compactron, because of instability.
There is also an Australian guitar amp called a Valvetone Impact, created by a very experienced builder (Darryl Hoy) with decades of commercial and hobby experience building guitar amps. The Impact uses 6DX8 triode-pentode output valves, and even with all his experience, Darryl mentioned online that he had encountered some stability issues with those output valves, and eventually modified the circuit to only use one of the two "free" triodes.
On the plus side, I'm using small-signal triode-pentode valves from the valve-TV era, in my own design guitar preamp, with no problems. But I use them a lot like the two triodes in a 12AX7 - usually we use them in adjacent parts of the signal chain, and don't encounter any problems as a result.
But if we used one half of a 12AX7 as the input stage, and the other half as, say, the fourth gain stage in a high-gain preamp, we'd be asking for trouble!
Excellent, you have a Plan B in case you encounter incurable stability problems from triode-pentode output valves!
A very frequent problem around here. Oh look, a squirrel!
😀
-Gnobuddy
Hey I found his amp once. I'm familiar with the name and what his amp does. The 6DX8 is basically the exact same valve as the ECL84/6F4P that I'm trying to use.
I made a slightly fancier block-diagram for myself and I figure I might as well share it.
I spent yesterday exploring gain staging. I have a small stash of NOS Sylvania 12au7a that I think I ought to find a use for and I was wondering if it would be an appropriate fit for V1.
My elementary understanding is that a 12ax7 can handle up to around 5vp-p on the grid before it starts to distort (with lots of caveats, depending on how it's biased) and it seems to me that a 12au7 might be sufficient for what I'd like to do. On the both amps it should just barely be enough (at the max gain setting) to push the second stage toward compression, but probably not enough to create an appreciable amount of dirt.
I like that on the clean side because I want this to be a pretty clean amp. I don't want "so clean it sounds sterile" but I'm thinking of a mini Fender Twin sort of sound where it mostly only fattens up and sounds thick. Definitely not looking for the Champ sort of breakup. So it would be 12au7 > Gain control > 12au7 > Tone stack and Volume control > 12ax7 gain recovery stage > PI > PA
I think it might also be healthy on the dirty side. I should have plenty of gain to work with. I'm not looking for hard-core heavy metal sounds, and high gain builds are supposed to be more challenging, right? In this case the V4a (12ax7 this time) would be able to amplify the signal all the way up to the limits of my rail voltage. Then through the Bridge-T to the cold-clipping stage and the cathode follower. I should have enough signal to get distortion at every stage.
Now if I were going to add an fx-loop - I know it's unusual but do you think there'd be any particular problem if I make it a series loop and put it in the clean amp in between the clean gain knob and V1b? I figure it would give me at least a level control going out, and since it's early in the circuit then it shouldn't be too much of a problem if I put the return signal straight into the grid of V1b. Does that sound reasonable?
Wow, this design is getting so complicated my head is starting to hurt! Here are some of my thoughts FWIW.
- Delay and modulation effects work absolutely fine straight into the input jack of an amplifier (I do that all the time). You don't actually need an FX loop. If you're really dead set on having one, you can create a simple series loop connection between the preamp and PI, which works really well.
- A typical PP ECL86 power stage will give you around 10W to 13W. To avoid instability problems mentioned earlier, I recommend using the triode sections as an LTP PI, which will normally give a little more imbalance and warmth to the amp's tone. I've built an ECL86 amp with this PI/PA configuration and it's actually very loud. 🙂
- Instead of two separate amps inside the box, whi not just have two preamps and a blend pot between the preamps and PI? Just a suggestion.
- If you use a 12AU7 for V1, it will then limit the gain and distortion of the dirty amp side.
- Delay and modulation effects work absolutely fine straight into the input jack of an amplifier (I do that all the time). You don't actually need an FX loop. If you're really dead set on having one, you can create a simple series loop connection between the preamp and PI, which works really well.
- A typical PP ECL86 power stage will give you around 10W to 13W. To avoid instability problems mentioned earlier, I recommend using the triode sections as an LTP PI, which will normally give a little more imbalance and warmth to the amp's tone. I've built an ECL86 amp with this PI/PA configuration and it's actually very loud. 🙂
- Instead of two separate amps inside the box, whi not just have two preamps and a blend pot between the preamps and PI? Just a suggestion.
- If you use a 12AU7 for V1, it will then limit the gain and distortion of the dirty amp side.
6F4P/6F3P Amps | DvNator's Amp Projects
Dvnator goes back and forth between PI designs. There's a lot to read so I'll sum it up: he and a few other builders seem to have come to a rough agreement that the ECL84s sound best at around 208v B+ and they also agree that you can't hit the grid of the ECL84 pentode with a really strong signal.
His most recent design used the cathodyne PI. (I suspect he wanted to limit the amp to just 3 pieces of glass.) The prior 2 designs did as you suggest - using the enclosed triode to make a LTP PI. I'm not as experienced so I was thinking about using a pragmatic approach here - redundancy - I will leave room to insert another 12ax7 just in case I run into the feared oscillation problems. Worst case is that I just won't use the triode sections.
Aside from that - I won't know until I try a few things.
Dvnator goes back and forth between PI designs. There's a lot to read so I'll sum it up: he and a few other builders seem to have come to a rough agreement that the ECL84s sound best at around 208v B+ and they also agree that you can't hit the grid of the ECL84 pentode with a really strong signal.
His most recent design used the cathodyne PI. (I suspect he wanted to limit the amp to just 3 pieces of glass.) The prior 2 designs did as you suggest - using the enclosed triode to make a LTP PI. I'm not as experienced so I was thinking about using a pragmatic approach here - redundancy - I will leave room to insert another 12ax7 just in case I run into the feared oscillation problems. Worst case is that I just won't use the triode sections.
Aside from that - I won't know until I try a few things.
Last edited:
Sorry my bad, I thought you were talking about ECL86s. I've not used ECL84s, just ECL86s, which are pretty much half of a 12AX7 sharing a bottle with an EL84. I think Dvnator likes to experiment with different PIs, but I would still suggest that the cathodyne (concertia) is better suited to HiFi, and you would probably do better with an LTP or paraphase PI - at least n theory.
You have convinced me to plan for an extra tube and to use the LTP version of the PI.
As for the overall circuit - it's not that bad if you just remember to think of it as (almost) two discrete amplifiers. I came up with the idea because I wanted to build both amps anyways, and I said "why not just build them both then?" and then the shared input idea came next - and from there it's just a matter of making certain that they don't interact with each other in unpleasant ways.
There is extra planning in the here-and-now but there'll be less cabinet construction and just a single device to put on my shelf when it's done. And I didn't say it but I'm going to do this in stages. I'm doing the dirty amp first. That way if there's any bad effects from adding the clean amp then it'll be really obvious that it's from the new stuff rather than just a poor design.
As for the overall circuit - it's not that bad if you just remember to think of it as (almost) two discrete amplifiers. I came up with the idea because I wanted to build both amps anyways, and I said "why not just build them both then?" and then the shared input idea came next - and from there it's just a matter of making certain that they don't interact with each other in unpleasant ways.
There is extra planning in the here-and-now but there'll be less cabinet construction and just a single device to put on my shelf when it's done. And I didn't say it but I'm going to do this in stages. I'm doing the dirty amp first. That way if there's any bad effects from adding the clean amp then it'll be really obvious that it's from the new stuff rather than just a poor design.
- Status
- Not open for further replies.