How do the diodes protect the OT?

[J M Fahey]

I just finished comparing sound of a Classic Tone Marshall OT to Classic Tone Vibrolux. ....... It is very disappointing that I cannot hear any difference it all different setting I can think of. This together with the sound clip I posted on OT comparison, I really question why people worry so much about different brands components and all the fuzz. Why worry all the minor difference when changing value of some of the resistors and capacitors in the signal chain can drastically change the sound of the amp. It seems like people just use one single amp, leaving the circuit totally unchanged. Then just change one component to a different brand and claim this brand is better or not.

Thatīs the point.

FWIW there is one classic example: in the Marshall Book, Ken Bran, the original Marshall Tech and responsible for the very first ones, says that once they got an EL34 Marshall head with, obviously, a very British transformer, pulled it and put the equivalent Fender one, pulled from an amp, as American as can be.
So not only it was a different maker but even some specs didnīt match, besides nominal power, and his conclusion was : "it still sounded like a Marshall" .
Straight from the horseīs mouth .

[Leo_Gnardo]

Ken Bran, the original Marshall Tech and responsible for the very first ones, says that once they got an EL34 Marshall head with, obviously, a very British transformer, pulled it and put the equivalent Fender one, pulled from an amp, as American as can be.
So not only it was a different maker but even some specs didnīt match, besides nominal power, and his conclusion was : "it still sounded like a Marshall" .
Straight from the horseīs mouth .

To agree with Juan, and Ken Bran:

Alan, think of the output transformer as an analog to a car transmission. If you put an Oldsmobile transmission in your Pontiac, that won't make it "drive like a Olds." Ratios might be a bit different, power curve too, but it's still going to transmit the engine's power (output tubes) to the drive wheels (speakers) and zoom, off you go.

If you insist, you can lab test the amps/transformers and maybe find a bit of difference in low and high frequency rolloff points but it's a guitar amp and most do not challenge the limits of the OT's frequency response. Guitars & amps & our expectation of what they should sound like limited generally from 80 Hz to 6 KHz. As you say, changing a cap or 2, resistor or 2, and definitely different speakers result in much more obvious changes in tone, and that's what makes the horse race between brands.

[trobbins]

PP amp operation with no secondary load may have been explained somewhere, but I don't recall coming across anything, so here goes (happy to be corrected on anything that follows!, or pointed to a good reference).

Start with initial idle condition of PP output stage. Assume idle current is almost zero (for now). Apply an ac signal, and follow loadline of first half-cycle of input signal voltage.

Positively driven valve loadline locus would initially move from B+ voltage at zero current, along zero current line to axis origin, even for low level of input signal. Some plate current would flow during that transition due to stray OT capacitance. Some current would start to ramp up due to increasing voltage across OT winding inductance - the change in current would depend on L, R and transition time of input signal. If locus got to origin in a very short time, then inductive current ramp rate would be say 500V/20H = 25mA/ms (based on B+=500V and 20H for OT half-winding incremental inductance at low current). So plate current would increase along relevant grid-voltage dependant plate curve. Effective plate resistance would increase as knee is reached, and then current would flatten to grid curve level as plate voltage rises. If drive signal was present long enough, then plate voltage would appear to settle at IR drop below B+.

When drive voltage cycles back through zero, then loadline locus would hmmm? If drive signal change was slow, and plate voltage had previously risen to near B+, then loadline would be almost vertical down to idle point, as only parasitic winding resistance would load the plate. But rate of change in plate current back to zero mA would force plate voltage to ellipse out to a higher value.

As for the negatively driven grid tube during that initial half cycle of input signal, its voltage would head towards 2B+, but then fall back to B+. There would be some plate current from capacitive charging, but otherwise the plate current would be effectively zero.

I would guess that the risk for OT damage from over-voltage would be during the high dI/dt portions in the driven half-winding, ie. as plate current initially rises up from zero when grid positive swing starts, and at end of input signal half-cycle when plate current falls back to idle level. That dI/dt risk relates principally to the driven half winding as I see it, not to the other half-winding.

Alan, I added the MOV capacitance comment as a general thread comment - not to indicate anything you had said (sorry for confusion on that).

[Alan0354]

PP amp operation with no secondary load may have been explained somewhere, but I don't recall coming across anything, so here goes (happy to be corrected on anything that follows!, or pointed to a good reference).

Start with initial idle condition of PP output stage. Assume idle current is almost zero (for now). Apply an ac signal, and follow loadline of first half-cycle of input signal voltage.

Positively driven valve loadline locus would initially move from B+ voltage at zero current, along zero current line to axis origin, even for low level of input signal. Some plate current would flow during that transition due to stray OT capacitance. Some current would start to ramp up due to increasing voltage across OT winding inductance - the change in current would depend on L, R and transition time of input signal. If locus got to origin in a very short time, then inductive current ramp rate would be say 500V/20H = 25mA/ms (based on B+=500V and 20H for OT half-winding incremental inductance at low current). So plate current would increase along relevant grid-voltage dependant plate curve. Effective plate resistance would increase as knee is reached, and then current would flatten to grid curve level as plate voltage rises. If drive signal was present long enough, then plate voltage would appear to settle at IR drop below B+. At the time when plate current starts to ramp up from 0mA, the inductance of half-winding would generate an extra dI/dt voltage which would push plate voltage below zero volts, and the inductive voltage would fall back to zero as current rate of change slowed down.

When drive voltage cycles back through zero, then loadline locus would hmmm? If drive signal change was slow, and plate voltage had previously risen to near B+, then loadline would be almost vertical down to idle point, as only parasitic winding resistance would load the plate. But rate of change in plate current back to zero mA would force plate voltage to ellipse out to a higher value.

As for the negatively driven grid tube during that initial half cycle of input signal, its voltage would head towards 2B+, but then fall back to B+. There would be some plate current from capacitive charging, but otherwise the plate current would be effectively zero. Perhaps there would be some coupling of winding voltage undershoot from driven side to force an overshoot above 2B+.

I would guess that the risk for OT damage from over-voltage would be during the high dI/dt portions in the driven half-winding, ie. as plate current initially rises up from zero when grid positive swing starts, and at end of input signal half-cycle when plate current falls back to idle level. That dI/dt risk relates principally to the driven half winding as I see it, not to the other half-winding. If so, then the diode clamping protection part would be getting hit as the principal control of that 'undershoot'.

Alan, I added the MOV capacitance comment as a general thread comment - not to indicate anything you had said (sorry for confusion on that).

I have to spend some time reading this. I'll come back if I have more question.

[Alan0354]

To agree with Juan, and Ken Bran:

Alan, think of the output transformer as an analog to a car transmission. If you put an Oldsmobile transmission in your Pontiac, that won't make it "drive like a Olds." Ratios might be a bit different, power curve too, but it's still going to transmit the engine's power (output tubes) to the drive wheels (speakers) and zoom, off you go.

If you insist, you can lab test the amps/transformers and maybe find a bit of difference in low and high frequency rolloff points but it's a guitar amp and most do not challenge the limits of the OT's frequency response. Guitars & amps & our expectation of what they should sound like limited generally from 80 Hz to 6 KHz. As you say, changing a cap or 2, resistor or 2, and definitely different speakers result in much more obvious changes in tone, and that's what makes the horse race between brands.

Ha ha, been on the players forum for too long. I was so surprised listening to the 4 Marshall OT comparison, yes, there is an ever so slightly difference, but it's so unimportant to me. Let's just say I don't have a good ear, but it got to be so little difference I cannot tell after 20 minutes back and fore at all different setting and different speakers.

I am going to use the Marshall OT in the next build ONLY because I bought and received it already. Also, I am planning to use EL34 which might over power the Vibrolux OT.

Yeh, the resistor/cap values combinations are where the mojo lies. That you don't use much electronics. It's a different world. Then choosing the speaker. I just bought the WGS Veteran 30. I think I might want another speaker. Then I have the All American WGC G12C/S, Celestion G12T-75, WGS Veteran 30 that is like Vintage 30. What should be the next one?

You think I would hear the difference between 6L6 and EL34?

[Alan0354]

To stress the point about a cap or a resistor can make a day and night difference. I started out my second Marshall build into the Bassman with the exact preamp as my first Fender style in the KMD combo. I have speaker and input footswitch that I can switch from one amp to the other and switch speakers so I can use the same speaker and cabinet for true A/B comparison. The two amps sound very much the same at the starting point( of cause slightly different as the power amp is slightly off).

I use 0.22uF to bypass the cathode resistors in two stages in the KMD. I removed just that two in the Bassman test bed. I COMPLETELY change the vibe of the amp. That two caps did whole lot more than getting a little more gain on the high end. It change the sound so much, change the harmonic content, it's just totally different. The difference is as drastic as changing a speaker.

This is where the mojo starts.


I just got 2X2 wood to brace up the little Marshall combo. I hope I can make it sounds better.....crossing my fingers. The box is very flimsy.

[loudthud]

The attached zip file contains some X-Y scope video clips that are somewhat relevant to this discussion, but does not show the effect of clamping diodes. I'll try to set something up over the weekend. Don't try this unless you have a properly rated scope probe.
　
The files MVC-080W and MVC-081W were taken from a 5F6A re-issue with 5881 output tubes. Plate voltage is measured with a 1000X scope probe, cathode current measured across a 1 Ohm resistor in series with the cathode. The vertical deflection factor is 100mA per division. The Horizontal deflection factor is 200V per division. The 0,0 origin is 2 div right and 1 div up from the lower left corner. At the beginning of the file you can see the bias point, about 40mA at 430V. The file MVC-081W is with a 2 ohm resistive load. Plate voltage swings from a little less than 100V to just over 700V. Peak plate current is about 400mA. The file MVC-080W is with the stock 4x10 speakers connected. Plate voltage swings from -400V to about +1000V. A different scope shot not included shows that the negative spikes last about 30uS.
　
The file MVC-531W is the current and voltage in a 16 Ohm Marshall 4x12 cabinet driven by (not absolutely sure) a Roccaforte Custom 80. Vertical is the current at 2 Amps per division, horizontal is Voltage at 20V per division. Inductive discharge easily kicks the trace off screen.

[jazbo8]

　
The file MVC-531W is the current and voltage in a 16 Ohm Marshall 4x12 cabinet driven by (not absolutely sure) a Roccaforte Custom 80. Vertical is the current at 2 Amps per division, horizontal is Voltage at 20V per division. Inductive discharge easily kicks the trace off screen.

Just to double check, the horizontal scale is 20V, not 200V per division as the other two?

[trobbins]

Hmmm, I'm thinking that the attached loadline and plots may be a reasonable illustration of no-load conditions for one valve in a PP output stage when overdriven at grid. From idle point (A), loadline moves to axis origin (B) and then tracks along 0V grid curve for plate, and if sufficient time is available for given primary winding inductance then locus will sit at (C) until grid voltage falls.

Leaving (C) will mean discharging energy in primary winding. Plate voltage will be pushed high to use up some energy in charging up stray capacitance. Other half-primary winding voltage will similarly be pushed low due to coupling, and similarly will use up some energy in discharging its stray capacitance. Both valves won't have infinite resistance, so will both provide some level of energy discharge.

The loadline and plots indicate what would likely happen if a flyback diode was being used to clip the plate voltage of the other valve as it swings under 0V.

[Audiotexan]

I really question why people worry so much about different brands components and all the fuzz.

That's an easy answer. We all have a 'golden tone' that we're predisposed to, for whatever reason. Just the fact that in our mind, when we hear "that certain tone"...it makes us smile uncontrollably, and makes the fur fly (hair follicles literally stand on end)!

The hard part is trying to find 'the sound' that does that. That's why there's thousands of amp designs, and tens of thousands of pedals out there. Everyone's fur does not 'fly' equally nor with the same stimulus. Anything that brings us closer to that sound we hear in our head makes us happier. That's even MORE complicated for those with a good ear. Tin ears are happy with anything remotely close. Perfectionists are never happy, but enjoy the journey even while being frustrated to wit's end sometimes along the way.

It's more 'points of fine tuning' vs changing the overall tone. But NO amount of tube rolling cap/resistor changing is going to alter the fundamentals of the amp's design. You have to have an amp that's very close to what you want to hear to start with before you even think about 'tweaking'

[loudthud]

Just to double check, the horizontal scale is 20V, not 200V per division as the other two?

Yes, scope set at 200mV/div, 1000X probe.

[jazbo8]

Yes, scope set at 200mV/div, 1000X probe.

Thank you.

[Alan0354]

Hmmm, I'm thinking that the attached loadline and plots may be a reasonable illustration of no-load conditions for one valve in a PP output stage when overdriven at grid. From idle point (A), loadline moves to axis origin (B) and then tracks along 0V grid curve for plate, and if sufficient time is available for given primary winding inductance then locus will sit at (C) until grid voltage falls.

Leaving (C) will mean discharging energy in primary winding. Plate voltage will be pushed high to use up some energy in charging up stray capacitance. Other half-primary winding voltage will similarly be pushed low due to coupling, and similarly will use up some energy in discharging its stray capacitance. Both valves won't have infinite resistance, so will both provide some level of energy discharge.

The loadline and plots indicate what would likely happen if a flyback diode was being used to clip the plate voltage of the other valve as it swings under 0V.

[ATTACH=CONFIG]30268[/ATTACH]

First off, we are still talking about no load at the secondary.

I've gone through your drawing:
1) I think you are looking at one side of the pp only. From your output voltage plot, if you consider the other half of the pp, B will go below as the other side shoot up and induce the first side to keep going low.

2) I don't see how you can claim point D and E is clipped at 2B+. Who is stopping it from going higher? This can only happen if you use clamping diode as in my original picture of the Fender Prosonic. I don't think you can count on the capacitance of the winding to keep the voltage to 2B+




This is how I see what happen when any signal driving into the power tubes through the PI stage with no load on the secondary of the OT. Here is my drawing:



At the top, I show the two power tube in PP mode. A is input of first tube, B is input of second tube. C is plate voltage of first tube, D is plate voltage of second tube. I expend the time scale so the input at A and B change very slowly to show the detail of what's going on.

When there is no load, impedance of the primary is very high and inductive:

1) At the start, A goes up slightly, the current of first tube starts to increase slightly, but impedance of the primary is high, inductance is high, you'll see a rapid and drastic voltage drop at the plate of the first tube at point C. It will go until point E where the tube is saturate and current stop increasing.

2) BUT when A start to go up, point B at the input of the second tube goes down. This decrease the plate current of the second tube. As the primary is inductive, it want's to keep the idle current going, therefore point D goes up beyond +B.

3) Because point D goes up, Point C is being dragged down by the effect of the center tap transformer configuration. This is shown at point E on the graph. This mean the plate of the first tube at C start to go low, even below 0V into -ve.

If there is no clamping diodes, there is nothing to stop the voltage swing at point C and D other than winding capacitance and plate capacitance and finite impedance of the primary. You don't need to drive to 0V at the grid, just by slight increase or decrease from idle current will bring this effect.

My drawing just cover the very first part of your graph, not even reaching B of your graph. But I take into consideration of the effect of the other half of the circuit that drag the voltage at the plate of the first tube below 0V. In fact, if you don't have ways to clamp it, this will fly to -HV because the plate of the second tube fly up as plate current of the second tube decreases.



EDIT:

I think we are drifting off the subject. My main concern is the power dissipation on the MOV or TVS. I don't see any of the explanation in your two post is telling me that dissipation is not the problem. Only thing you said relating to dissipation is the winding capacitance and the other parasitic capacitance will limit the energy in the coil......which I don't think I can agree. If winding capacitance can limit the energy and voltage, ignition coil will never work........And more important of it all, people will never burn OT and power tube if they forget to plug in the speaker.

Lastly, if You said the winding and parasitic capacitance come into play, it will form a resonance circuit with Q depending on the winding resistance ( low). that is more disastrous.

[trobbins]

Alan, I did state no secondary load conditions, and I did indicate that flyback diode clamping caused D to E characteristic (without going in to affect of leakage inductance causing some additional overshoot). And yes the locus is for one of the PP valves only (the positive grid driven valve), as that allows someone looking at a valves V-I characteristic curves to pencil in what happens.

Yes the example scenario was driving grid to 0V, for convenience of explanation, but the same characteristic occurs for any grid drive above idle bias voltage - i.e. the locus effectively transitions towards the y axis (plate current) with little increase in current above idle, and then moves along the grid drive voltage characteristic curve (which is a little more complex to appreciate if the grid voltage is increasing as part of the signal cycle).

I suggest you have wrongly interpreted what is happening once the driven tube reaches its saturation position near the graph origin. That drive valve is acting like a switch (but with the plate characteristic curves of effectively a constant resistance region from the origin, up to the knee of the grid voltage characteristic curve, and then close to a constant current source) - and as such becomes the 'loading' on the OT. Eg. the incremental resistance can be down at 100 ohm loading for an EL34 if grid voltage had reached 0V. Hence, the driven valve clamps the windings during that initial stage if the input cycle.

Also note that if no diode clamping was present, then yes the voltage during D to E portion would fly higher (but not to infinity) as stray capacitance soaks up more and more energy, and possibly leakage breakdown starts to contribute (eg. across valve base terminals or in the OT).

[loudthud]

Alan and trobbins, neither of you seem to show the plate voltage swinging negative like my scope clips show.

Absent from this conversation is someone running a simulation. There are always problems with the models used and they might be incomplete, not properly accounting for negative plate voltage, leakage inductance, winding capacitance et al. But I wish someone would make the effort.

I've purchased some vintage amps that had arched the tube sockets. But I don't know if the arch was caused by big overdrive or an open load. Could it have been a bad tube? Who's to say? I really don't want to send a signal to an amp with an open load to find out which will fail first, the socket or the OT.