12ax7 plate resistance

[valvetones]

Hello Friends, I was going over the data sheet for a 12AX7 trying to confirm why fender used a 29k primary impedance for the reverb driver transformer.
It appears at 250v it’s Ra is 62.5K per side. The tube is wired in parallel for the reverb circuit cutting the plate resistance in half.
Now at 100v on the data sheet, it’s listed that the Ra is 80k.
So plate resistance goes up when voltage goes down?

The reason I checked in the first place is that I’m doing a bit of an experiment with a Thordarson 20A04 input transformer used in place of a fender reverb transformer.

The fender transformer (as I’m sure you all know) is a 29k primary and 8 ohm secondary.

The Thordarson is a 3-6 ohm primary with a 25-36k secondary.

So I plan on wiring it backwards essentially where the red wire goes to ground, blue to the positive reverb jack, black to B+ and green to the Anodes.

I’m merely guessing that the reflected load impedance phenomenon that applies to speakers and output transformers will also apply to the 8ohm tank and 3-6 ohm primary (wired as the secondary) of the transformer.

Now, assuming this all works, I might assume that I could end up with as high as 50k on the Anodes due to reflective loading.  If the Anodes want 29k and I give them 50K then should I assume the reverb will be weaker?

To compound this line of thinking, if according to the data sheets, lower voltage = higher plate resistance, then I could simply lower the voltage for a better match?

To compound That line of thinking, if it’s true that the plate resistance is 62.5k at 250v but fender ran the anode voltage at 430v for the reverb tube anodes, does that then mean that the actual anode resistance at 430v is actually much lower than 62.5k , thus making the 29k of the reverb driver fender used mismatched in the opposite direction I’m likely to experience with what I’m trying to use?

Lots to think about. Need more great minds to bounce this off of.
Who’s up for careful consideration?

Also, it’s occurred to me that I might should of posted this on the amp board, but it’s really a question more geared toward the effects of voltage, plate resistance, and transformer impedance.

I apologize in advance if I’m in the wrong spot.

Thank you friends, hope you all are doing well.
VT

[sluckey]

The reverb driver is a 12AT7, not a 12AX7.

[PRR]

The 250V vs 100V change probably also means lower current. Which has a lot of effect on plate resistance.

See attached and page 4 of https://frank.pocnet.net/sheets/093/1/12AX7A.pdf

A singe-ended "power" amp wants a plate transformer designed to carry DC current. An "input" transformer will almost certainly NOT be designed for DC current. Bass response falls badly, bass distortion rises. At an extreme, fine Permalloy input transformers may be "ruined" by even a few mA of DC.

And yeah, I think Fender favored 12AT7 for slapping springs. However there was a mini bass-amp with an interstage transformer driven with half a 12AX7.

[HotBluePlates]

... trying to confirm why fender used a 29k primary impedance for the reverb driver transformer. ...

Do you know how to draw a loadline for a single-ended power stage?  Have you learned how to graphically determine a tube's plate resistance from its data sheet & an operating point?

You may have to download & zoom-in on the image at the bottom.  It is a set of 12AT7 curves from the 3rd page of a 12AT7 data sheet.

I snagged a silverface Deluxe Reverb schematic, and noted 6v across a 680Ω cathode resistor and ~400v supply on the other side of the Reverb Transformer.

     The data sheet shows a single-triode, so we get half-current on the data sheet curves.

     Since we have two triodes in parallel, this implies two 58kΩ loads in parallel for a total load impedance of 29kΩ.  The curves will show a single load resistance (58kΩ), and twice the internal plate resistance compared to parallel triodes.
     
     6v / 680Ω = 8.8mA ---> a single triode passes 4.4mA, let's call it "4.5mA" to make graphing easier.

     The schematic says, "410v" but 400v is easier for graphing, so I drop a big Green dot at the operating point: 400v and 4.5mA.

     You note the Reverb Transformer impedance is 29kΩ; our load for a single triode is then 29kΩ x 2 = 58kΩ.

     We need a loadline through the operating point.  400v / 58kΩ = ~6.9mA, let's call it 7mA to make graphing easier.  The Red loadline starts at 0v and (7mA + 4.5mA =) 11.5mA.  The line passes through the operating point at 400v and 4.5mA.  This defines a ~58kΩ loadline because (400v - 0v) / (11.5mA - 4.5mA) = 400v / 7mA = 57kΩ plus some rounding error.

Now we calculate power output:

     A Class A amp has a limit of clean output while swinging up to 2x idle current and down to zero idle current.  So I used a Blue line along the loadline to a current of 2 x 4.5mA = 9mA.  The plate swings down to 140v.  As a check, 58kΩ x 4.5mA = 261v, and 400v - 261v = 139v (close enough to "140v").

     Multiplying the change from the operating point to this peak found with the Blue line gives us Peak Power:
     (400v - 140v) x 4.5mA = 1.17 watts

     RMS Output Power is 1/2 Peak Power, or 585 milliwatts.  But then we have two 12AT7 triodes each contributing the same power for 1.17 watts.

We figure the internal plate resistance of the triode:

     The internal plate resistance is defined by a line tangent to the grid-bias curve at the operating point.

     I estimated a compromise slope between the -6v and -7v curves, and plot it as a Purple line.

     The line has endpoints of voltage at 500v and 290v.  There are endpoints of current at 8.5mA and 0mA.

     Internal Plate Resistance (rp) = (500v - 290v) / 8.5mA = ~24.7kΩ

     Note: 24.7kΩ x 2 = ~50kΩ ----> the load is nearly 2x rp, where theory predicts maximum output power.
     Also, this is the rp of a single triode; 2 triodes in parallel then have an rp of 24.7kΩ / 2 = ~12.4kΩ which is mighty-close to 1/2 the Reverb Transformer load impedance of 29kΩ.

[HotBluePlates]

"But the data sheet says plate resistance of 10-15kΩ"

Yes, but one condition has an idle current of 10mA (much higher, rp will be much lower).

The other condition shows a similar plate current of 3.7mA, but with only 100v on the plate, and 270Ω x 3.7mA = ~1v of bias.

We have much higher plate voltage, but also much higher bias voltage to hold plate current down.  The result is the rp shown using the curves (and demonstrated in-circuit) is much higher than the data sheet conditions would suggest.

[valvetones]

The reverb driver is a 12AT7, not a 12AX7.

It’s for a 6G16 vibroverb, which uses a 12ax7

Sorry for the confusion

[valvetones]

"But the data sheet says plate resistance of 10-15kΩ"

Yes, but one condition has an idle current of 10mA (much higher, rp will be much lower).

The other condition shows a similar plate current of 3.7mA, but with only 100v on the plate, and 270Ω x 3.7mA = ~1v of bias.

We have much higher plate voltage, but also much higher bias voltage to hold plate current down.  The result is the rp shown using the curves (and demonstrated in-circuit) is much higher than the data sheet conditions would suggest.

Starting to make some sense. I’ll have to attempt to figure out what else needs to change then, if it works at all. Thanks for the detailed explanation of drawing load lines. I’ll see what I can come up with.

[valvetones]

“A singe-ended "power" amp wants a plate transformer designed to carry DC current. An "input" transformer will almost certainly NOT be designed for DC current. Bass response falls badly, bass distortion rises.” At an extreme, fine Permalloy input transformers may be "ruined" by even a few mA of DC.

- I’ll report back, I just have to try it.
It is exactly the same size as a fender reverb driver. Not that that means it will work, but it’s allready bolted on and wired up. You guys may be in for a good chuckle at my needing to confirm with my bare hands that the red iron on the stove is hot.

And yeah, I think Fender favored 12AT7 for slapping springs. However there was a mini bass-amp with an interstage transformer driven with half a 12AX7.

-Again, and sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

[HotBluePlates]

... sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

This was Fender's first amp with onboard reverb.  Obviously it worked & they sold amps.  What's fun is that you can get usable sound even with a non-optimum design.

But notice also Fender never used a 12AX7 to drive reverb in any of their other amps.  What's more, they used a 12AT7 in their 63 Vibroverb Reissue.

[valvetones]

... sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

This was Fender's first amps with onboard reverb.  Obviously it worked & they sold amps.  What's fun is that you can get usable sound even with a non-optimum design.

But notice also Fender never used a 12AX7 to drive reverb in any of their other amps.  What's more, they used a 12AT7 in their 63 Vibroverb Reissue.

Yeah I was pondering that last night. The differences are really pretty simple
The cathode resistor changed from 1000 ohms to 2200 ohms, and the reverb transformer changed from 125A20A to 125A20B.

The transformer part is interesting, I was stating the transformer as 29k primary, but those are for the 125A20B.  Last night the thought kicked me right in the face that the 125A20A was likely suited to a 12AX7  and its specs are unavailable anywhere online. Who knows what it’s primary was. For that matter, who knows if it even used the same tank.

I may end up switching the cathode resistor and use a 12AT7, and the $10 repro 125A20B

Thanks for all the replies everybody.
Groovy Cats indeed.
Thank you much.

[HotBluePlates]

... The differences are really pretty simple The cathode resistor changed from 1000 ohms to 2200 ohms ... I may end up switching the cathode resistor and use a 12AT7 ...

I started out plotting a loadline for the 2.2kΩ and saw right away that the design didn't make sense.  Too little current, too much bias voltage, much-reduced output power.

So I switched mid-stream to the later silverface 680Ω setup, and suddenly everything dropped into place as it should.  I wonder about the blackface/early-silverface value of 2.2kΩ being another case of, "... usable sound even with a non-optimum design."

Others may have insights on whether the 680Ω setup has problems not revealed on the curves.

[tubeswell]

... and sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

I'm guesstimating that the 6G16 reverb transformer's reflected load was 'designed' to be about 25k for 2 parallel 12AX7 triodes - for a somewhat steepish load line to coax a bit more plate current. (But I've never actually impedance-tested a 6G16 reverb transformer to know)

[valvetones]

... The differences are really pretty simple The cathode resistor changed from 1000 ohms to 2200 ohms ... I may end up switching the cathode resistor and use a 12AT7 ...

I started out plotting a loadline for the 2.2kΩ and saw right away that the design didn't make sense.  Too little current, too much bias voltage, much-reduced output power.

So I switched mid-stream to the later silverface 680Ω setup, and suddenly everything dropped into place as it should.  I wonder about the blackface/early-silverface value of 2.2kΩ being another case of, "... usable sound even with a non-optimum design."

Others may have insights on whether the 680Ω setup has problems not revealed on the curves.

Very interesting indeed. Yeah I don’t know of anyone ever complaining about blackface reverb response. Lol.
Which schematic were you looking at showing a 680 ohm resistor?

Also, I’ve noticed that the 6G16 is the only amp I see in Ted Weber’s book that calls for a 1watt cathode resistor.
All the other amps with reverb in this book call for a 2200 1/2 watt.
Is 200 ohms really so much lower that I have to use a 1 watt?

[valvetones]

... and sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

I'm guesstimating that the 6G16 reverb transformer's reflected load was 'designed' to be about 25k for 2 parallel 12AX7 triodes - for a somewhat steepish load line to coax a bit more plate current. (But I've never actually impedance-tested a 6G16 reverb transformer to know)

Well your in luck! I just bought a 125A20A off reverb for $140.
A grocery’s bill worth of knowledge can be had for all if you can describe to me how to perform this impedance test.
I also have an original 125A20B to compare it to.
It’ll be fascinating to figure this all out.
Please let me know.

[valvetones]

I started out plotting a loadline for the 2.2kΩ and saw right away that the design didn't make sense.  Too little current, too much bias voltage, much-reduced output power.

So I switched mid-stream to the later silverface 680Ω setup, and suddenly everything dropped into place as it should.  I wonder about the blackface/early-silverface value of 2.2kΩ being another case of, "... usable sound even with a non-optimum design."

Others may have insights on whether the 680Ω setup has problems not revealed on the curves.

This may be in left field, but do you think it may have anything to do with the reverb coupling circuits?

They are different from 6G16 to AB763 but I’m not sure about AB763 to whichever silverface with a 680 ohm you may be talking about.

Which amp specifically were you looking at with that value? Is it also a 1watt or is it 1/2 watt.

This build is completely wired up, and nearly ready to go. I just need to finish reforming these electrolytics and figure out what’s gonna happen with the reverb transformer, bias up some tubes and see how she sounds.
I put the EF86 in the normal channel, and am rather curious to see how that works out when jumpering channels.

Should be cool. I’ll be sure to post some gut shots. It may be my finest work to date.

{edit: fixed lost quote tag- PRR}

[HotBluePlates]

Which schematic were you looking at showing a 680 ohm resistor?

...  I snagged a silverface Deluxe Reverb schematic, and noted 6v across a 680Ω cathode resistor and ~400v supply on the other side of the Reverb Transformer.

I already said the first time. 

... the 6G16 is the only amp I see ... that calls for a 1watt cathode resistor.
All the other amps ... call for a 2200 1/2 watt.

Power = Volts x Current
Current = Volts / Resistance   (Ohm's Law)

Substituting:
Power = Volts x (Volts / Resistance) = Volts² / Resistance

Most of the schematics tell you how much voltage is present at the reverb driver's cathode:
     6G16 Vibroverb    1kΩ 1w    3.3v --->  (3.3v)² / 1000Ω = 11 milliwatts

     Princeton Reverb     2.2kΩ     8v ----->  (8v)² / 2200Ω = 29 milliwatts

     Pull-boost Princeton Reverb    680Ω   5v ----> (5v)² / 680Ω = 37 milliwatts

All of those are well below 1/2 watt much less 1w.

But look around at Fender schematics...  There are a lot of 1.5kΩ cathode resistors, but not too many 1kΩ.  However, there's a 1kΩ 1w resistor in the brown Princeton's power supply.  Fender probably just swiped that part when they needed "1kΩ."  (They could have swiped the 6K6's 1kΩ cathode resistor in the standalone reverb unit, but it was 2w and probably more expensive)

[valvetones]

Which schematic were you looking at showing a 680 ohm resistor?

...  I snagged a silverface Deluxe Reverb schematic, and noted 6v across a 680Ω cathode resistor and ~400v supply on the other side of the Reverb Transformer.

I already said the first time. 

... the 6G16 is the only amp I see ... that calls for a 1watt cathode resistor.
All the other amps ... call for a 2200 1/2 watt.

Power = Volts x Current
Current = Volts / Resistance   (Ohm's Law)

Substituting:
Power = Volts x (Volts / Resistance) = Volts² / Resistance

Most of the schematics tell you how much voltage is present at the reverb driver's cathode:
     6G16 Vibroverb    1kΩ 1w    3.3v --->  (3.3v)² / 1000Ω = 11 milliwatts

     Princeton Reverb     2.2kΩ     8v ----->  (8v)² / 2200Ω = 29 milliwatts

     Pull-boost Princeton Reverb    680Ω   5v ----> (5v)² / 680Ω = 37 milliwatts

All of those are well below 1/2 watt much less 1w.

But look around at Fender schematics...  There are a lot of 1.5kΩ cathode resistors, but not too many 1kΩ.  However, there's a 1kΩ 1w resistor in the brown Princeton's power supply.  Fender probably just swiped that part when they needed "1kΩ."  (They could have swiped the 6K6's 1kΩ cathode resistor in the standalone reverb unit, but it was 2w and probably more expensive)

Looks like I’ll be more than fine with the 1/2 watt I have installed!
Thank you my friend. I was abit concerned.
Really appreciate the math too.
That helps heeeps.

[tubeswell]

... and sorry I didn’t mention it in the beginning, but it’s for a 6G16 vibroverb, which indeed does use a 12ax7.

I'm guesstimating that the 6G16 reverb transformer's reflected load was 'designed' to be about 25k for 2 parallel 12AX7 triodes - for a somewhat steepish load line to coax a bit more plate current. (But I've never actually impedance-tested a 6G16 reverb transformer to know)

Well your in luck! I just bought a 125A20A off reverb for $140.
A grocery’s bill worth of knowledge can be had for all if you can describe to me how to perform this impedance test.
I also have an original 125A20B to compare it to.
It’ll be fascinating to figure this all out.
Please let me know.

Put a lowish voltage 1kHz* 5-6VAC source across the secondary and (carefully - taking all appropriate safety precautions) measure the VAC across the primary, and you have your secondary-primary VAC ratio. Now turn that around and you have your Pri:Sec VAC ratio (LoL), then square the pri:sec vac ratio and you have your Pri:Sec impedance ratio.


Say you used a 5VAC source (from a rectifier winding), and you got 280VAC across the primary, that'd be 5:280 or 1:56. So 56:1 x 56:1 = 3136:1 impedance ratio, and with an 8R input transducer on the secondary, thats a reactive load of 25k.


*no specific need to use 1kHz - you can use 'mains frequency' VAC for this test and it 'works' to give you a ballpark, but if using Mains VAC voltage as the source, then (use it with a 1:1 isolating x-former and) put it across the primary and measure the VAC on the secondary - just because.

[valvetones]

Cool deal, I’ll do that, probably will do it with a variac since I have one handy. So will be at the mains frequency.
I’ll up date you all and congrats, you may all have that chuckle now.
The input transformer did indeed drive the springs and worked, but the hum was unbearable.
So I tossed the old fender unit it and the hum is less, much less really but still getting a little more than expected when the reverb knob is turned up. I need to investigate this.
Part of me thinks it may have been the location of the tank at the time, but it could be related to other anomalies I’m experiencing.
So I did the ac4 style EF86 gainstage on the normal channel, I used 1meg instead of the 5.6meg on the screen as seen in the AC15, and connected the output to the 220k channel mix resistor in the 6G16 circuit.
During the end of the filter cap reforming process, I decided to check my coupling caps for DC leakage as everything in the Amp is nos.
Nothing had leakage except the cap on the EF86 plate, or so I thought, a quick review of the 6G16 layout showed me that the HV was actually coming from the Oscillator tube on the vibrato channel, going through the 1meg mix resistor, through the 220k mix resistor, through the EF86 stage volume pot, and back to the plate coupling cap. So I had DC on the volume pot, I got rid of by putting an equal value capacitor to the EF86 plate cap on the volume pot going to the 220k mix resistor. That keeps the DC off the volume pot, and EF86 plate cap.
Curiously enough, the vibrato functions on the EF86 normal channel but the reverb does not. Although I think that’s awesome, I have no idea why. There is a downside though. The whole point in having a EF86 in the normal channel is to have the channels in phase so I can jumper them. That part works well as intended, however, when trying to use both channels with the vibrato on I notice some strong vibrato pulsating that does not sound musical, and I thing it’s a result of too strong a signal in the vibrato circuit oscillator.
I dunno for sure. Turning the intensity control down passed 5 gets rid of the issue.
Finally, I used .0033 mfd as the EF86 plate cap, the other side of that cap goes to the volume control and a 500pf cap, then the 500pf cap to the tone control. I purposely lowered these values in an attempt to really only roll of the treble frequencies in the tone control. Anyway, I doesnt seem to roll off much at all.
So three issues really
1. Treble / tone control needs to be addressed
2. Need to figure out why I have vibrato on the Normal channel and either figure out how to stop it, or keep it and figure out how to stop the signal from that channel from overloading the oscillation circuit (only when the channels are jumpered).
3. I have some hum in the reverb circuit.

Other than those three things, I do think the amp sounds great.
When I get these things addressed, I’ll start a new thread to showcase the effort and discuss the journey.

Also when the reverb transformer arrives, I’ll test and see what kind of ratio I get.

Thanks everyone for all your patience and thoughtful ideas and considerate education when addressing my posts.
I really like this place.