If you choose to have a dedicated coupling cap for the feedback signal (in my opinion the better option) then you would need a resistor to ground to get the proper bias.
I'm going to post a separate thread on the anode local feedback design (will welcome your input) as I still struggle to work out how best to use this setup. As a Christmas present to myself I've just bought some breadboarding sockets to make it easier to experiment, but may need to build a new amp (oh dear...) so I don't do any further damage to my existing one!
I'm going to post a separate thread on the anode local feedback design (will welcome your input) as I still struggle to work out how best to use this setup. As a Christmas present to myself I've just bought some breadboarding sockets to make it easier to experiment, but may need to build a new amp (oh dear...) so I don't do any further damage to my existing one!
Hi Guys
It is good to see some creativity in preamp design. But...
The original post stated that a modern metal sound is desired and this is most easily attained with four gain stages that each runs free. Free-running stages clip soft and allow a gradual buildup of harmonics. It must be remembered that where feedback-controlled stages are common in hifi and broadcast work, the goal there is to REDUCE distortion not to add it. So, I would suggest you try a more traditional preamp.
The original circuit shows a bootstrapped cathode follower driving the EQ. This is good if you want extremely high clean signals fed to the EQ, but it is contrary to the purpose of the stage in most builds. As TUT6 details, the CF-driven EQ lacks in tonal dynamics and is a flaw in any amp that incorporates it; EQ is much more dynamic when plate-driven, so wire all four stages for gain.
If you want to retain the CF, place it after the EQ or better yet after the MV so that long cables to a PA can easily be driven.
In an economical high-gain amp with SE output, it is tempting to keep the tube count low. However, this easily cripples tone yet can be done. In this case, a 3+1 architecture is useful with the following layout:
A1 > gain control > A2 > interstage attenuator > A3 > EQ > second-gain pot > A4 > MV > PA
Having the fourth stage to drive the output tube grid assures enough signal is available at that point to do so easily. I've tried many topologies and have generally been disappointed with too few gain stages. In the above layout, the first three stages may need cathode bypass caps to exhibit the gain at high frequencies required for a good metal tone. The fourth stage may or may not need Ck. The interstage attenuator should be adjusted so that no grid-rectification occurs, as usual (see TUT5 for a method for doing this).
As always, every grid should have a grid-stop. It is best not to rely on these for actual frequency shaping and thus their values will be kept low enough to not contribute too much noise.
Variable gain for the first stage is more usual, but as above, the free running stages are preferred especially in a low-tube-count design.
Have fun
It is good to see some creativity in preamp design. But...
The original post stated that a modern metal sound is desired and this is most easily attained with four gain stages that each runs free. Free-running stages clip soft and allow a gradual buildup of harmonics. It must be remembered that where feedback-controlled stages are common in hifi and broadcast work, the goal there is to REDUCE distortion not to add it. So, I would suggest you try a more traditional preamp.
The original circuit shows a bootstrapped cathode follower driving the EQ. This is good if you want extremely high clean signals fed to the EQ, but it is contrary to the purpose of the stage in most builds. As TUT6 details, the CF-driven EQ lacks in tonal dynamics and is a flaw in any amp that incorporates it; EQ is much more dynamic when plate-driven, so wire all four stages for gain.
If you want to retain the CF, place it after the EQ or better yet after the MV so that long cables to a PA can easily be driven.
In an economical high-gain amp with SE output, it is tempting to keep the tube count low. However, this easily cripples tone yet can be done. In this case, a 3+1 architecture is useful with the following layout:
A1 > gain control > A2 > interstage attenuator > A3 > EQ > second-gain pot > A4 > MV > PA
Having the fourth stage to drive the output tube grid assures enough signal is available at that point to do so easily. I've tried many topologies and have generally been disappointed with too few gain stages. In the above layout, the first three stages may need cathode bypass caps to exhibit the gain at high frequencies required for a good metal tone. The fourth stage may or may not need Ck. The interstage attenuator should be adjusted so that no grid-rectification occurs, as usual (see TUT5 for a method for doing this).
As always, every grid should have a grid-stop. It is best not to rely on these for actual frequency shaping and thus their values will be kept low enough to not contribute too much noise.
Variable gain for the first stage is more usual, but as above, the free running stages are preferred especially in a low-tube-count design.
Have fun
Wow that's a load of information thanks, maybe I got carried away with al the various cool looking schematics of other amps and what not.
And I think I'm gonna build,test and rebuild a few times with your tips included.
When(if) I've a got a working amp, I'll post the results and we'll see how it goes from there.
Cheers
I've been designing and building guitar amps for over a decade, and I must say it's most intriguing after designing hi-fi equipment for 40 years before that. Very different rules, very subjective, etc. I don't recommend using negative feedback unless you have a full understanding of phase margin. I have the full understanding, and still prefer to not use any, in my guitar amps. I'm not into heavy metal though. I'm more into the edge of distortion thing that a blues guitarist might prefer. I try to play and sound like Clapton, although he didn't always have the best sound either. Learning about all the different distortion mechanisms, and which ones are good vs bad, is what guitar amps are all about. It's a complex subject, and very subjective.
One quick contribution is to put a passive Rf filter right at the input. Maybe a 10k R in series with a roughly 1nF cap to gnd. Set it to be 3 dB down at about 25kHZ, so the guitar acting as an antenna won't cause as much slewing and demodulation of IM products.
One quick contribution is to put a passive Rf filter right at the input. Maybe a 10k R in series with a roughly 1nF cap to gnd. Set it to be 3 dB down at about 25kHZ, so the guitar acting as an antenna won't cause as much slewing and demodulation of IM products.
Thanks for the input, really interesting. And making me reconsider my recent build...
My goal when I set about building a higher gain amp was to have a 'lean to mean' element - increasing the distortion without a large increase in perceived volume, ideally using the guitar volume pot. The dream living room / partner friendly amp.
I understood that local NFB reduced distortion until a certain point (input signal voltage), when it rapidly flipped to heavy clipping.
So my impression was that the AX84 L2L (& P2) design enabled this by allowing you to 'live on the edge' of the transition point - adjust G1 pot to get the level of feedback (therefore clipping & lean to mean-ness), adjust G2 to get max breakup. Then roll back on the guitar to the 'other side' of the transition point, lowering clipping on the 2nd stage.
The 1st two stages seem to be of limited (any?) use in tone shaping - likely a sticking point WRT getting a good tone. My LTSPICE sims (happy to share if anyone can help) seem to indicate this. I struggled to find much literature on this and I'm no expert!
I've just got myself a cheap USB scope / signal generator to help with this voicing - automated bode plot generation to compare to LTSPICE outputs.
I was hoping to start a dedicated thread on this process and what I found - my current build sounds good to me, but could be better and I hope the voicing (changing coupling caps, high bypass caps etc) will help.
Tristan
Bob, this is an interesting aspect that I've yet to consider - does this apply to local NFB around a single stage? What is the likely result of the phase margin difference if introduced into the 2nd stage of a 4 stage build? Any pointers?
Thanks,
Tristan
Ok so at some point I guess I made a mistake with ordering the right transformer , it only has 2x12 V outputs, I expected it to also have a 230V output , so my question is : Can I rectify the current straight from the wall socket I.e. 230 v wall socket-> fuse -> switch -> rectifier? Without using the transformer, it's a stupid question probably , but I'm just checking
No negative feedback wrapped around one or more stages.
NO! This is a good way to get yourself DEAD. It creates a direct connection between the wall socket and something you will wrap your hand around in such a manner that prevents you from letting go during an electrical shock. Yes, people have died.
Look at it this way. Would you plug your guitar cord into a wall socket, wrap one hand around it (the strings are connected to the ground side of the cord) and then put your other hand or mouth on a grounded metal object, like say a microphone.
Last edited:
Wow, i wouldn't have guessed that, good thing I'm checking , one could assume that 230Vac = 230 Vac , but apparently not !
Hi Guys
Susu: As voltwide says, you can't do that. However...
All is not lost. If this is a preamp, then you can buy a second 12+12V PT and wire it backwards to the first one, i.e., 12V to 12V. The primary of the second PT is now its secondary, which will produce the 230Vac, which bridge rectified and filtered will be about 320Vdc.
Note the first PT tied to the mains has to supply all the power to the heaters AND for the plate. Assuming the plate power is only a fraction of the total, the second PT can be a lower VA rating.
Note also that when using the tiny toroids and potted toriods in the blue cases, you MUST use a unit of twice the rating compared to your load if you want to have good reliability and low EMI. The designed temperature rise at full load is 60C - way too high for reliable operation and the EMI output is comparable to an EI transformer.
Have fun
Susu: As voltwide says, you can't do that. However...
All is not lost. If this is a preamp, then you can buy a second 12+12V PT and wire it backwards to the first one, i.e., 12V to 12V. The primary of the second PT is now its secondary, which will produce the 230Vac, which bridge rectified and filtered will be about 320Vdc.
Note the first PT tied to the mains has to supply all the power to the heaters AND for the plate. Assuming the plate power is only a fraction of the total, the second PT can be a lower VA rating.
Note also that when using the tiny toroids and potted toriods in the blue cases, you MUST use a unit of twice the rating compared to your load if you want to have good reliability and low EMI. The designed temperature rise at full load is 60C - way too high for reliable operation and the EMI output is comparable to an EI transformer.
Have fun
Ok thanks, I'm using a 50 W rated 2x12v toroid, maybe i'll just buy. An extra 230->230v toroid if it's better available , the one I wanted in the first place was out of stock and I got a bit hasty with changing the order, I feel stupid now ...
The 230 volts from your local utility has one side, or the center tap of the power companies transformer grounded, depending on what country you live in. This must be the case to avoid real ugly things if one of their transformers short out, power lines get tangled in a storm, or lightning hits the lines.
This means that touching one of the lines in your wall socket, and a grounded object, will complete the circuit, connecting you across the mains!
A power transformer in your amp removes this path, because there is no direct connection between the primary and secondary. All the energy is transferred magnetically. Your guitar will be connected into the electronics that are attached to the secondary of this transformer, but there is no path from the electronics back to the mains through the transformer. This is why a good quality transformer must be between any electronics that can be touched by a user, and the mains.
Note, there WERE guitar amps manufactured back in the 1950's and 1960's that did not use a transformer, and DID kill people. An isolation transformer MUST be added to these old amps before using them. They attempted to isolate the user from the mains with a capacitor, which could get leaky or short out over time. This cap is known as the death cap!
Hi Guys
As I said above, you can wire two PTs back-to-back. Since this is a preamp, use the 50VA unit you already have as the input device. The plate power for a preamp is quite low, typically just a couple of watts, so you can get a 10VA PT rated at 12+12V for this.
The 50VA part will tie to the mains as usual, producing 12V output with the windings in parallel. You can send this directly to the tube heaters. 12V wiring is preferred anyway as it places all of the heater wiring along one side of the sockets. The wiring to the triodes does not have to cross this, so less danger of noise ingress.
The small PT is wired with its 12V windings in parallel and tied to the 12Vac from the large PT. The output of the small PT is the 230Vac, which using a full bridge and filter cap produces about 320Vdc. Use a CRC filter before going to the plate of the final preamp stage. Across the second filter make a voltage divider with 330k to the positive end and 47k to ground. Add a 10uF cap acorss the 47k. This divider provides a DC standoff voltage for the heaters. Wire a pair of 330R resistors in series across the 12Vac and then tie the junction of these Rs to the DC standoff. As TUT3 demonstrates, this connection provides 30dB of hum rejection for heater-related noise - the same as if you used DC heaters but without the hassle.
Note that were you to plan this a bit more precisely to begin with, you would start with the number of tubes and their heater requirements. Each 12A_7 requires about 2W of heater power - so 2VA per tube (12V x 150mA). If the preamp uses three tubes, then there is about 6VA of heater power. The plates require only about 1mA or two per plate, multiplied by the supply voltage. For three tubes there are six plates, and even at 2mA each at 320Vdc the total power is just 4W. Add this to the heater power to designate the VA rating for the input PT, then 10VA for PT1 and 4VA for PT2.
Have fun
As I said above, you can wire two PTs back-to-back. Since this is a preamp, use the 50VA unit you already have as the input device. The plate power for a preamp is quite low, typically just a couple of watts, so you can get a 10VA PT rated at 12+12V for this.
The 50VA part will tie to the mains as usual, producing 12V output with the windings in parallel. You can send this directly to the tube heaters. 12V wiring is preferred anyway as it places all of the heater wiring along one side of the sockets. The wiring to the triodes does not have to cross this, so less danger of noise ingress.
The small PT is wired with its 12V windings in parallel and tied to the 12Vac from the large PT. The output of the small PT is the 230Vac, which using a full bridge and filter cap produces about 320Vdc. Use a CRC filter before going to the plate of the final preamp stage. Across the second filter make a voltage divider with 330k to the positive end and 47k to ground. Add a 10uF cap acorss the 47k. This divider provides a DC standoff voltage for the heaters. Wire a pair of 330R resistors in series across the 12Vac and then tie the junction of these Rs to the DC standoff. As TUT3 demonstrates, this connection provides 30dB of hum rejection for heater-related noise - the same as if you used DC heaters but without the hassle.
Note that were you to plan this a bit more precisely to begin with, you would start with the number of tubes and their heater requirements. Each 12A_7 requires about 2W of heater power - so 2VA per tube (12V x 150mA). If the preamp uses three tubes, then there is about 6VA of heater power. The plates require only about 1mA or two per plate, multiplied by the supply voltage. For three tubes there are six plates, and even at 2mA each at 320Vdc the total power is just 4W. Add this to the heater power to designate the VA rating for the input PT, then 10VA for PT1 and 4VA for PT2.
Have fun
I recommend getting the right power tranny, rather than turning one around and using it backwards. The backwards thing can work in theory but a 12 volt winding may not be designed to deal with 240VAC and could arc or burn out somehow. I would want to be more of an expert on transformers before doing that myself.
Phase margin is where the high frequency rolloffs of all the major devices (tubes, output transformer, routing of wiring, and any high end rolloff capacitors) cause time delay, more often called phase shift (same thing in this context). When the time delay adds up to the point of causing the negative feedback to become positive feedback at any frequency, the amp oscillates, and probably blows up your speakers and often itself.
The way to minimize the likelyhood of this happening is to pick the device that rolls off the soonest in frequency (usually the output tranny in a tube amp), and you design the circuit so all other roll-offs (which also cause time delay) are as far out in frequency as is possible. The goal here is to have the open loop gain roll off to where the closed loop gain is, before there's enough delay (or phase shift) to cause the circuit to oscillate. Each rolling off device will usually roll off at a 1 pole rate (not necessarily, but lets assume here). If a device rolls off with a 1 pole rate, it will introduce up to 90 degrees of phase shift (45 degrees at its -3dB frequency). If one rolling off device (the output tranny for ex.) can roll off the open loop gain to the point that the difference between open loop gain and closed loop gain is less than one, the amp will never oscillate. If a bunch of devices (tubes and trannys) roll off at nearly the same frequency, the delays will add up fast and there is a high probability that there will be enough phase shift, before the differential loop gain goes to less than one, to cause the amp to oscillate. I hope this makes sense.
When you take neg feedback from the output all the way back to the input, you have more devices that can introduce phase shift in that region of frequency where the open loop circuit is rolling off. When the neg FB is only going back one or two stages, there are less devices that introduce this time delay or phase shift. Feedback that is only in one stage is the best scenerio. the R off the cathode in a cathode follower stage is said to be a form of local negative feedback. There's also the case where the output tranny is wired "ultra-linear" giving negative feedback to the screen grids of the output tubes. That's also "local neg. feedback". Some rare amps have a second secondary winding that is put in the cathode circuit of the output tube(s), and that is considered local negative feedback. Hope this helps.
Phase margin is where the high frequency rolloffs of all the major devices (tubes, output transformer, routing of wiring, and any high end rolloff capacitors) cause time delay, more often called phase shift (same thing in this context). When the time delay adds up to the point of causing the negative feedback to become positive feedback at any frequency, the amp oscillates, and probably blows up your speakers and often itself.
The way to minimize the likelyhood of this happening is to pick the device that rolls off the soonest in frequency (usually the output tranny in a tube amp), and you design the circuit so all other roll-offs (which also cause time delay) are as far out in frequency as is possible. The goal here is to have the open loop gain roll off to where the closed loop gain is, before there's enough delay (or phase shift) to cause the circuit to oscillate. Each rolling off device will usually roll off at a 1 pole rate (not necessarily, but lets assume here). If a device rolls off with a 1 pole rate, it will introduce up to 90 degrees of phase shift (45 degrees at its -3dB frequency). If one rolling off device (the output tranny for ex.) can roll off the open loop gain to the point that the difference between open loop gain and closed loop gain is less than one, the amp will never oscillate. If a bunch of devices (tubes and trannys) roll off at nearly the same frequency, the delays will add up fast and there is a high probability that there will be enough phase shift, before the differential loop gain goes to less than one, to cause the amp to oscillate. I hope this makes sense.
When you take neg feedback from the output all the way back to the input, you have more devices that can introduce phase shift in that region of frequency where the open loop circuit is rolling off. When the neg FB is only going back one or two stages, there are less devices that introduce this time delay or phase shift. Feedback that is only in one stage is the best scenerio. the R off the cathode in a cathode follower stage is said to be a form of local negative feedback. There's also the case where the output tranny is wired "ultra-linear" giving negative feedback to the screen grids of the output tubes. That's also "local neg. feedback". Some rare amps have a second secondary winding that is put in the cathode circuit of the output tube(s), and that is considered local negative feedback. Hope this helps.
Hi Guys
Bob said: "The backwards thing can work in theory but a 12 volt winding may not be designed to deal with 240VAC and could arc or burn out somehow. I would want to be more of an expert on transformers before doing that myself. "
You are correct, Bob, you are not an expert and should read more carefully.
Back-to-back transformers work fine. There is only 12V on the 12V windings and 230V on the 230V winding. No arcing. No problems. It's a way the OP can make his ordering mistake turn into something useful. There is a nice side benefit that mains noise is fully isolated from the plate supply even with standard PTs that don't have electrostatic screens.
This is an old technique used since transformer were invented by Mr. Faraday. RCA illustrates it in their tube app notes from the 1950s and people like myself have used the method since the 1970s.
Also, this is a preamp not an integrated amp, so free-running is as it applies to voltage gain stages. The OP's first schematic had a feedback-controlled second stage. Not of much use in a high-gain preamp for guitar. A free-running stage is exactly that: a tube running "wild" with just its grid reference, cathode biasing (if not fixed biased - yes, you can fix bias a voltage gain stage), and the plate load. We will ignore the fact that EVERY amplifying device has intrinsic feedback and just refer to what is VISIBLE and ASSEMBLED by the builder.
Have fun
Bob said: "The backwards thing can work in theory but a 12 volt winding may not be designed to deal with 240VAC and could arc or burn out somehow. I would want to be more of an expert on transformers before doing that myself. "
You are correct, Bob, you are not an expert and should read more carefully.
Back-to-back transformers work fine. There is only 12V on the 12V windings and 230V on the 230V winding. No arcing. No problems. It's a way the OP can make his ordering mistake turn into something useful. There is a nice side benefit that mains noise is fully isolated from the plate supply even with standard PTs that don't have electrostatic screens.
This is an old technique used since transformer were invented by Mr. Faraday. RCA illustrates it in their tube app notes from the 1950s and people like myself have used the method since the 1970s.
Also, this is a preamp not an integrated amp, so free-running is as it applies to voltage gain stages. The OP's first schematic had a feedback-controlled second stage. Not of much use in a high-gain preamp for guitar. A free-running stage is exactly that: a tube running "wild" with just its grid reference, cathode biasing (if not fixed biased - yes, you can fix bias a voltage gain stage), and the plate load. We will ignore the fact that EVERY amplifying device has intrinsic feedback and just refer to what is VISIBLE and ASSEMBLED by the builder.
Have fun
Last edited:
I have another question,
I was going to elevate one of the two 12 V winding to B+ via a divider, say +/- 40 Vdc. But now that same elevated ac signal will be used in a backwards 12-230 transformer. Is this going to be a problem ???
And Thanks for al the help so far .
EDIT .. sorry missed that bit
where does one find these TUT (orials i guess)?
Last edited:
No, elevating the windings to +40 volts will not cause a problem as long as DC does not flow through the transformer. It won't as long as there is no connections other than the heaters, the two transformers, the two resistors, and a cap to ground.
TUT = The Ultimate Tone. A series of books available via the link in Struth's posts. Not in my price range, so I haven't seen them.
- Status
- Not open for further replies.