redplating 1959 w larmar

[flemingmras]

Not necessarily! If you have the wrong reflected plate load impedance (too low), there is no bias current you can use, even down to zero mA plate current/full class B operation, that will keep you from exceeding the maximum plate dissipation at some point in the power range of the amp. You have to either change the plate-to-plate impedance of the transformer or lower the plate voltage. By the way, the max dissipation point will not occur at full clip output - it will occur somewhere around midscale power levels. The tubes will actually run cooler at full power than they do somewhere in between.

Randall Aiken

Has anyone notified Randall Smith of this?

Aiken, this brings up some interesting thoughts here. Some math I was playing around with earlier...not sure if it's correct and if it isn't please feel free to correct it -

Assuming a 100% efficient zero loss output transformer and single frequency test signal conditions into a resistive load -

With each 1/2 of the OT primary being 400 ohms (roughly), you would need 500mA draw through that 400 ohm load, which would in turn create a 200 volt drop across the 400 ohm load, to get 100 watts on 1/2 the OT primary, which then would get transferred to the secondary.

However, no transformer is even close to 100% efficient, which means that you'd have to be feeding MORE THAN 100 watts of power in to get 100 watts out, which would require more current and therefore a higher voltage drop across each 1/2 of the OT primary.

And of course, as current flow through the tube & the load increases, so does the voltage drop across the load, which causes the voltage drop across the tube (from cathode to plate) to decrease as current flow increases. Effectively, the PI grid drive is modulating the cathode - plate resistance of the tube at input signal frequency. As plate - cathode resistance decreases, plate/load current increases and hence the decrease in plate voltage, which is what puts your plate voltage/current out of phase. Effectively you have the exact same thing happening in the bias supply right at the bias pot as you adjust it.

Am I correct in thinking this?

[Roe]

I keep abusing this amp (running it a full volume for some time)

running with 65% bias one RFT redplates . but with cold bias there's only a very slight hint of redplating NFB at 100k/4ohms. However, with NFB at 20k/4ohms it clearly redplates.
The former is so slight that it is manageable but the latter is not ok. the kt88s dont redplate at all

edit: corrected info about NFB

[raiken]

Do you have any way of measuring B+ supply current? I'd be curious to know what the idle current and full power current is on your 480V supply. If you could insert an multimeter in line with the OT center tap, or if your meter is fuse protected with a 1A slo-blo, you could just remove the fuse and put the multimeter across the fuse holder terminals.

[Roe]

Do you have any way of measuring B+ supply current? I'd be curious to know what the idle current and full power current is on your 480V supply. If you could insert an multimeter in line with the OT center tap, or if your meter is fuse protected with a 1A slo-blo, you could just remove the fuse and put the multimeter across the fuse holder terminals.

I'll try and see what I get.
Im only running a .5A fast blow fuse on the OT and it hasn't blown yet.

multimeter is fused at 10A 60sec.
I guess I should measure between OT CT and b+ (rectifier or mains filter cap)

[raiken]

Do you have any way of measuring B+ supply current? I'd be curious to know what the idle current and full power current is on your 480V supply. If you could insert an multimeter in line with the OT center tap, or if your meter is fuse protected with a 1A slo-blo, you could just remove the fuse and put the multimeter across the fuse holder terminals.

I'll try and see what I get.
Im only running a .5A fast blow fuse on the OT and it hasn't blown yet.

multimeter is fused at 10A 60sec.
I guess I should measure between OT CT and b+ (rectifier or mains filter cap)

Yes, or temporarily stick a 1A fuse in the meter and take the easy route and pull the HT fuse and measure across the holder so you don't have to desolder anything. Of course, please be careful and don't let the meter leads slip, and watch your fingers!

I'm really surprised that 500mA fast blow hasn't popped. You should be drawing nearer to an amp, I would think...

RA

[Roe]

...

Yes, or temporarily stick a 1A fuse in the meter and take the easy route and pull the HT fuse and measure across the holder so you don't have to desolder anything. Of course, please be careful and don't let the meter leads slip, and watch your fingers!

I'm really surprised that 500mA fast blow hasn't popped. You should be drawing nearer to an amp, I would think...

RA

it seems my multimeter can only measure ampers in AC and 1.5V DC (battery)

I measures 0A AC both at idle and under load.

EDIT: measurements added below

[VelvetGeorge]

Just scanned over all 5 pages of this thread, forgive me if I ask questions that have already been answered. Also, having Randall looking at your issue and giving advice is huge!!! One of the most brilliant guys in the industry! Howdy Randall.

A few things to add to the conversation:

the 82k side of the PI will always drive harder. I generally offset this by putting slightly lower spec'ed tubes in that side (V4 and V5 in 100 watts).
In some cases I've had to lower the 82k to a 68k to get closer to equal output from that side. Specifically in vintage amps that red plate under normal operation.
Note that I find you don't want both sides exactly equal or you lose some of the distinct tone that the offset produces. In Marshalls anyway.

Oscillation, which I think has been eliminated as the issue at this point, can be checked off the list after trying these fixes:

lift negative feedback resistor, to check phase (which you did).

separate OT primary and secondary wires.

add 5.6k grid resistors.


Although the original Marshall circuit is not technically correct and doesn't account for voltage and load variations, in almost every case you can make it operate without red-plating.

Other suggestions:

try 1.5k screen grid resistors.

Ground all output tube cathodes at a single point.

I'll check back to see what develops.


george

[Roe]

Just scanned over all 5 pages of this thread, forgive me if I ask questions that have already been answered. Also, having Randall looking at your issue and giving advice is huge!!! One of the most brilliant guys in the industry! Howdy Randall.

A few things to add to the conversation:

the 82k side of the PI will always drive harder. I generally offset this by putting slightly lower spec'ed tubes in that side (V4 and V5 in 100 watts).
In some cases I've had to lower the 82k to a 68k to get closer to equal output from that side. Specifically in vintage amps that red plate under normal operation.
Note that I find you don't want both sides exactly equal or you lose some of the distinct tone that the offset produces. In Marshalls anyway.

Oscillation, which I think has been eliminated as the issue at this point, can be checked off the list after trying these fixes:

lift negative feedback resistor, to check phase (which you did).

separate OT primary and secondary wires.

add 5.6k grid resistors.


Although the original Marshall circuit is not technically correct and doesn't account for voltage and load variations, in almost every case you can make it operate without red-plating.

Other suggestions:

try 1.5k screen grid resistors.

Ground all output tube cathodes at a single point.

I'll check back to see what develops.


george

george, thanks for replying. yes, I'm very happy for getting Randall's help.
here's what I've done sofar

1) 82k is lowered to 68k but this didn't really help & I'm running tubes with the lowest bias in v4-5, I've tried three different PI tubes without any effect
2) yes, removing NFT doesn't help. BUt: there is less redplating with less NFB (I use a 20k in line with a 80k pot at the 4 ohms secondary)
3) OT primaries and secondaries are not even close. both are twisted and as much apart from other wires as possible (I've tried moving the secondaries but it didn't help)
4) I used 5k6s from the start but have now installed 10Ks in v4-5 but it doesn't help. also, I use a 47k grid resistor at the PI grid

I haven't tried the two last things:
5) I'm using screen resistors which measure 1k (the type used in old marshalls). Last I checked, the screens voltage isn't too high. its about 4 volts lower than plate voltage, something which doesn't seem bad
6) now the cathodes are grounded to each individual socket (with a 1ohm 0.6w resistor in between)

[flemingmras]

Just scanned over all 5 pages of this thread, forgive me if I ask questions that have already been answered. Also, having Randall looking at your issue and giving advice is huge!!! One of the most brilliant guys in the industry! Howdy Randall.

A few things to add to the conversation:

the 82k side of the PI will always drive harder. I generally offset this by putting slightly lower spec'ed tubes in that side (V4 and V5 in 100 watts).
In some cases I've had to lower the 82k to a 68k to get closer to equal output from that side. Specifically in vintage amps that red plate under normal operation.
Note that I find you don't want both sides exactly equal or you lose some of the distinct tone that the offset produces. In Marshalls anyway.

Oscillation, which I think has been eliminated as the issue at this point, can be checked off the list after trying these fixes:

lift negative feedback resistor, to check phase (which you did).

separate OT primary and secondary wires.

add 5.6k grid resistors.


Although the original Marshall circuit is not technically correct and doesn't account for voltage and load variations, in almost every case you can make it operate without red-plating.

Other suggestions:

try 1.5k screen grid resistors.

Ground all output tube cathodes at a single point.

I'll check back to see what develops.


george

george, thanks for replying. yes, I'm very happy for getting Randall's help.
here's what I've done sofar

1) 82k is lowered to 68k but this didn't really help & I'm running tubes with the lowest bias in v4-5, I've tried three different PI tubes without any effect
2) yes, removing NFT doesn't help. BUt: there is less redplating with less NFB (I use a 20k in line with a 80k pot at the 4 ohms secondary)
3) OT primaries and secondaries are not even close. both are twisted and as much apart from other wires as possible (I've tried moving the secondaries but it didn't help)
4) I used 5k6s from the start but have now installed 10Ks in v4-5 but it doesn't help. also, I use a 47k grid resistor at the PI grid

I haven't tried the two last things:
5) I'm using screen resistors which measure 1k (the type used in old marshalls)
6) now the cathodes are grounded to each individual socket (with a 1ohm 0.6w resistor in between)

The simplest way to ground the output tubes to a central point while still keeping your 1 ohm cathode resistors would be to solder your 1 ohm resistors between pin 8 and pin 1. Then run buss wire through pin 1 of all 4 sockets, then tie one of the pin 1's to ground. Absolute best place for the tie to ground would be right at the negative side of your first filter caps.

Randall...did you happen to see my question for you at the top?

[Roe]

here's one thing that is a little strange: when I turn the pot connected between NFB resistor and 4ohms out, this affects redplating a little and also hum. It appears that the highest NFB (20k at 4ohms = 28k at 8 ohms) gives less redplating and less hum. the lowest NFB (100k at 4ohms) gives a little more redplating and more hum. (the way stated first turned out to be correct after all)

The hum appear to be hum from PT which is picked up by OT. its not terribly bad. its just a loud as an original 1971 superbass I had on my bench. (my 66 and 67 100watters are more quiet)

[Roe]

Do you have any way of measuring B+ supply current? I'd be curious to know what the idle current and full power current is on your 480V supply. If you could insert an multimeter in line with the OT center tap, or if your meter is fuse protected with a 1A slo-blo, you could just remove the fuse and put the multimeter across the fuse holder terminals.

I'll try and see what I get.
Im only running a .5A fast blow fuse on the OT and it hasn't blown yet.

multimeter is fused at 10A 60sec.
I guess I should measure between OT CT and b+ (rectifier or mains filter cap)

Yes, or temporarily stick a 1A fuse in the meter and take the easy route and pull the HT fuse and measure across the holder so you don't have to desolder anything. Of course, please be careful and don't let the meter leads slip, and watch your fingers!

I'm really surprised that 500mA fast blow hasn't popped. You should be drawing nearer to an amp, I would think...

RA

Randall,

found out how to measure ampers in DC.

EDIT: it measures .15A at idle and .5A or even .6A when cracked

Sorry, it took me some time to learn to use the multimeter correctly.

[raiken]

Also, having Randall looking at your issue and giving advice is huge!!! One of the most brilliant guys in the industry! Howdy Randall.

Hey, George! Thanks, but you give me far too much credit! I'm just another solder-jockey like all the rest of you guys.

This is such an interesting topic, I can't seem to stay out of it. The loadline alone for a Marshall running at those voltages is truly scary! I suspect that older tubes had more robust plate structures and could handle overloads better.

RA

[raiken]

here's one thing that is a little strange: when I turn the pot connected between NFB resistor and 4ohms out, this affects redplating a little and also hum. It appears that the lowest NFB (100k at 4ohms) gives less redplating and less hum. the highest NFB (20k at 4ohms = 27k at 8 ohms) gives a little more redplating and more hum. (at first I mistakenly stated it was the other way around)

The hum appear to be hum from PT which is picked up by OT. its not terribly bad. its just a loud as an original 1971 superbass I had on my bench. (my early 100watter are more quiet)

I suspect the different amounts of NFB is affecting your PI drive symmetry - remember, it is the asymmetrical drive that is causing only one side to redplate anyway. Anything you can do to shift the drive back to the other tube will lessen the dissipation on the 82K side and increase it on the 100K side. Did you ever try any of the other things I suggested, like raising the 10K tail resistor up to 47K or 100K, or adding a 100K in series with the PI input to see the effect? That should balance the PI a bit more and distribute the dissipation better, although at the expense of the nice second-order harmonics you get when driving hard.

RA

[Roe]

here's one thing that is a little strange: when I turn the pot connected between NFB resistor and 4ohms out, this affects redplating a little and also hum. It appears that the lowest NFB (100k at 4ohms) gives less redplating and less hum. the highest NFB (20k at 4ohms = 27k at 8 ohms) gives a little more redplating and more hum. (at first I mistakenly stated it was the other way around)

The hum appear to be hum from PT which is picked up by OT. its not terribly bad. its just a loud as an original 1971 superbass I had on my bench. (my early 100watter are more quiet)

I suspect the different amounts of NFB is affecting your PI drive symmetry - remember, it is the asymmetrical drive that is causing only one side to redplate anyway. Anything you can do to shift the drive back to the other tube will lessen the dissipation on the 82K side and increase it on the 100K side. Did you ever try any of the other things I suggested, like raising the 10K tail resistor up to 47K or 100K, or adding a 100K in series with the PI input to see the effect? That should balance the PI a bit more and distribute the dissipation better, although at the expense of the nice second-order harmonics you get when driving hard.

RA

I will try upping the tail resistor how do I add the 100k in series with v3a's grid? do I add it parallel with wire from .022uf PI input to v3a's grid

added amper readings above by the way

[raiken]

I will try upping the tail resistor how do I add the 100k in series with v3a's grid? do I add it parallel with wire from .022uf PI input to v3a's grid

Just desolder the wire going to pin 2 of the PI (assuming that is the section you used for the input) and solder a 100K to the socket pin 2 and connect the lifted wire to the other end.

RA