question about new high-gain project

[stormbringer]

Hi all,

I'm a novice DIYer working on my first non-kit amp - a high-gain design with 4 preamp stages (following the preamp illustrated in TUT Fig. 5-53). I"m sure I'll have many questions about design as I go along, and I'll start with these two questions:

1. lead dress. In his preamp book, Merlin recommended twisting the leads from each preamp stage, whereas others have recommended keeping the wires far apart (or intersecting at right angles). I'd like to try the twisted method. Question: for the input stage, where a shielded cable comes directly from the input jack, shall I twist the two remaining wires (to HT and cathode) together ?

2. cathode bypass. Is there any downside to using a 400V metalized poly cap instead of a 50-100V electrolytic cap for the cathode bypass? I understand that it won't form to handle the full 400V (if that makes sense), but is that a bad thing?

Cheers!
John

[Alan0354]

You don't want to twist any of the signal wire or anything together. That's the last thing you should do. Wire twisted together not only form capacitor coupling. The current in one wire will generate a circular magnetic field around the wire, if you have another wire running along, you will induce current into the other wire. Of cause it is perfect to have everything perpendicular to each other, but that is not realistic. So keep them far away as possible. Having two wire perpendicular is the best, the magnetic field from one wire will not circle the other wire and no current can be induced from one wire to the other.

You twist the two heater wires together and tight to get more coupling between the two. This will minimize the 60Hz hum from radiating out towards the sensitive circuits. Tight coupling will reduce radiation. But you better know what you are doing before twisting anything together. Unless you know for sure the the other wire contain the return current of the first wire, twisting together any other wires is always bad news.

For cathode bypass cap, you need large value cap like 10uF or up, your choice is very limited other than electrolytic ones. I always want to challenge the convention believe, I even put tantalum cap for cathode bypass on mine and I like sound of my amp. Don't buy too much into those stuff, it need to be proven by a skeptical eye.

[stormbringer]

Thanks for the input. I was surprised to read Merlin B.'s recommendation (on the last page of his book). It is obvious he knows what he is doing, so I assume there is some merit to this. To be clear: he does NOT advise twisting together the wires from both stages but rather the individual stages. Before I start twisting, I am hoping to get input from several people who know what they are doing.

My schematic calls for a low-value bypass cap, so there are lots of choices. Of course, I might change it to a larger value later on, but for now I will stick to the on-paper value...

[JoeM]

.....
2. cathode bypass. Is there any downside to using a 400V metalized poly cap instead of a 50-100V electrolytic cap for the cathode bypass?....
John

What capacitance value? You can use 50v film caps for cathode bypass, but they will (most likely) be physically larger than an electrolytic. 400v is OK too.

[stormbringer]

What capacitance value? You can use 50v film caps for cathode bypass, but they will (most likely) be physically larger than an electrolytic. 400v is OK too.

ah, there must have been a reason I left out that detail...

This design calls for .1uF bypass in the first stage. The 4-stage preamp illustrated in TUT Fig 5-53 has no bypass cap, so I might remove it in the end (or increase the value to taste).

[Alan0354]

Thanks for the input. I was surprised to read Merlin B.'s recommendation (on the last page of his book). It is obvious he knows what he is doing, so I assume there is some merit to this. To be clear: he does NOT advise twisting together the wires from both stages but rather the individual stages. Before I start twisting, I am hoping to get input from several people who know what they are doing.

My schematic calls for a low-value bypass cap, so there are lots of choices. Of course, I might change it to a larger value later on, but for now I will stick to the on-paper value...

You just think about whether you want to couple the two signals together before you twist together. The cathode and plate has signals in opposite polarity, if you feed the plate voltage back to the cathode, you create a negative feedback and reduce the gain but reduce distortion. If you twist the grid and cathode, you are like putting a cap between them. This is the first pass thinking.

BUT, the last thing you want to do is to create parasitic component coupling, it is unreliable, unpredictable and is bad practice. Just because the author wrote it don't mean he/she is right. Hate to be offensive and blunt, there are so many tinkerers in this guitar amp design thinking they know electronics and can even write a book. A lot of them just follow the old myth handing down. ENZO here introduce a book RDH4 copy that is very useful, so far it gives me more useful info than the two new books I bought. Download chapter two.

Electronics theories has gone through a quantum leap in the pass 30 years, there are a lot of things even the so called experts in the older days don't know. What you are talking about is in the field of signal integrity engineer which become very important today as speed of signal is getting faster and faster. Everything is about electromagnetics, coupling and radiation which part of it is grounding. With the guitar amp, even though the speed is crawling, the impedance of the circuit is very high, into hundreds of kilo ohms. These high impedance circuit become very susceptible to radiation and coupling which is all electromagnetics and grounding. Why do you think there are still so many so called designers fighting oscillation, motor boating and all sort of positive feedback? Because they really don't know this kind of signal integrity and grounding stuff. And they end up doing all sort of strange wiring that goes around and around. What you try to do is playing with fire. Make it simple, make it clean. Don't introduce anything into the circuit that is not there. Change the sound by the component and design, not by this kind of dangerous coupling.

The only time I twist two wires are the power supply and return. Like the 6.3V AC wire. This make it become a transmission line that closely couple together. The tighter the coupling, the less EM wave radiate out, the quieter the circuit become. You can twist the plate wires of a push pull stage as they are differential signal and twisting them together reduce radiation of the large swing of the wires to the rest of the sensitive circuit. A little capacitive coupling between the two wires are not that important as there are amps that actually has a few hundred pF cap across the two wire to create a dominant pole to stabilize the closed loop feedback power amp stage. If you look at some vintage Fender amp, you'll see a lot of them has 2200pF cap from each of the lead to ground which is the same as putting the cap between the two lead of the OT. You can twist the speaker wire from the OT to the speaker as that carry a lot of current and generate a lot of magnetic field that can feed back into the sensitive circuit. Other than that. Don't!!!!

If you really think doing some twisting with certain wire can improve the sound, immediately go to the theory, find out why the coupling improve the sound, then put a real component in and separate the wires and have a clean and reproducible circuit.

[Steve Conner]

If I remember right, it is the power supply and ground reference wires to each stage that Merlin likes to twist together, not the actual signal wires leading to the tube socket. His stuff is all correct in theory, but in practice I don't think it makes much of a difference.

My own preference is for a preamp ground bus made out of a piece of house wiring, connected to the chassis at the input jack. And for any long signal connections within the chassis (say a grid wire that goes to a volume control down the other end of the panel) I'll use screened cable.

[Alan0354]

Now it makes a lot of sense to twist the power and ground as the ground is the power return so any noise one the power will have an opposite signal in the ground return and they couple to each other and less radiation.

[stormbringer]

If I remember right, it is the power supply and ground reference wires to each stage that Merlin likes to twist together, not the actual signal wires leading to the tube socket. His stuff is all correct in theory, but in practice I don't think it makes much of a difference.

My own preference is for a preamp ground bus made out of a piece of house wiring, connected to the chassis at the input jack. And for any long signal connections within the chassis (say a grid wire that goes to a volume control down the other end of the panel) I'll use screened cable.

Fig 12.16 of Merlin's (aka the Valve Wizard's) preamp book shows the signal wires twisted. And he says in the text that the low-impedance cathode wire will help to shield the others from external interference.

[exclamationmark]

I always twist my respective pre-amp wires together for the neatness factor, and I haven't really found any negatives in doing so. When you consider the two triodes inside a 12ax7 are only separated by mere millimetres with a potential gain of 10,000 between them, a little wire twisting ought not to hurt (perhaps we should switch to octals to eliminate inter-triode capacitances! I'm joking of course).

Realistically, the capacitance between twisted plate and grid wires is so tiny only frequencies far, far away into the RF region would be subject to negative feedback (which might help to prevent oscillation). The grid to cathode capacitance also has the effect of shunting high frequencies to ground, which again may prevent oscillation. So essentially it acts like shielded wire for free (though I imagine the capacitance would only be in the range of 10-50pf - it turns out maxwells equation's are quite difficult to solve... ). The point of twisting high current wires is to cancel their magnetic field (and thus interference), and really has little to do with capacitive coupling.

[Alan0354]

I always twist my respective pre-amp wires together for the neatness factor, and I haven't really found any negatives in doing so. When you consider the two triodes inside a 12ax7 are only separated by mere millimetres with a potential gain of 10,000 between them, a little wire twisting ought not to hurt (perhaps we should switch to octals to eliminate inter-triode capacitances! I'm joking of course).

Realistically, the capacitance between twisted plate and grid wires is so tiny only frequencies far, far away into the RF region would be subject to negative feedback (which might help to prevent oscillation). The grid to cathode capacitance also has the effect of shunting high frequencies to ground, which again may prevent oscillation. So essentially it acts like shielded wire for free (though I imagine the capacitance would only be in the range of 10-50pf - it turns out maxwells equation's are quite difficult to solve... ). The point of twisting high current wires is to cancel their magnetic field (and thus interference), and really has little to do with capacitive coupling.

Yes, that's what I explained before about the negative feedback and the grid cathode cap. But the point is using twisting is unpredictable and particularly not doable in pcb. It is a bad practice to make things more complicated that it is necessary particular for novice starting out with bad habit.

If you twist grid and plate wires, you form the miller cap. I am not sure you can say it's not significant. I don't think I need to explain that in detail.

If you twist wires from different stages together, you are really asking for it. In cascaded gain stages, gain is so high. You have to deal with guitar talking directly to the tube already, you don't need any more potentially more paths of oscillation.

From my years of circuit design, less is more. Make it simple, make it sweet. In fact, I don't even believe in hand wiring beyond the first prototype. pcb is so much more predictable and reliable. I don't buy for a moment that the old school hand wiring has any advantage. Now a days, fabbing a run of 2 layers pcb is so cheap it's not worth hand wiring beyond the first prototype. Layout is a big part of the design, you put in the effort, do it right on pcb, then you can have a reliable and reproducible product.

[exclamationmark]

Actually, that's a good point about the miller capacitance. In reality though, I haven't heard a difference. The main reason I do it is because it's easier to keep 4 bundles of wires away from each other than 12 when prototyping! It's also surprising to note that a fair few PCB amps take advantage of these 'dirty tricks' to save on component costs. For example I've seen one peavey amplifier actually had a serpentine pattern over a ground plane to artificially increase the grid-cathode capacitance to either save on the cost of a capacitor or stop oscillations! Similar things have also been done to increase the miller capacitance. Ribbon cables are also another commonly used thing in PCB amps for off board mounted triodes (which is essentially the same as twisting triode wires together). I guess at some point bad practices could be considered as 'smart' practices!

I totally agree on the PCB vs handwiring thing though... Unless I'm in a hurry to get a prototype done, I'll spend a day and etch out a board myself.

[Gingertube]

I know you said you wanted high gain. TUT Fig 5-53 gives you that in a single channel VERY Marshall'ish tapped gain arrangement.

I would seriously recommend that you instead look at the two channel version of Fig 5-36 instead. I have built one of those with the fancy footswitching option (covered in TUT5 "Standard" project).

This gives you a "Fender Black face" style clean channel and a lead channel with ample gain. There is a "fancier" version which allows the 2 channels to work at the same time as Fig 5-54, I probably would'nt bother with that.

Cheers,
Ian

[stormbringer]

Thanks Ian. I contemplated making a two channel 'Standard', but I thought I better tackle a single-channel amp first. I do want to make the Standard from TUT5 in the future however.