JCM800 2303 EL34 red plating issue

[TheKT88KilledJFK]

Hi guys,  sorry for lurking so much but I have a question only el34world can answer!

I have constructed a JCM800 2203 amp (all new components) with 2 x el34 power tubes (I'm just trying to get a 50 watt configuration working), 5.6k 2W grid stoppers, 1k 5W screen resistors.  The plate voltage is right at 450VDC.  When I first plugged the amp in, the bias sat at -25VDC and the tubes began to red plate at this setting.  I set the voltage to the spec'd -40VDC setting and the tubes persisted to red plate which is not surprising.  My first thought is that I have the pins connected incorrectly for the tubes, but I've checked at least six or seven times.  This is how I have the power tubes connected:

Pin 1:  not connected
Pin 2: Filament
Pin 3:  OT
Pin 4: To 1k Screen resistor
Pin 5: to 5k6 Grid resistor
Pin 6: not connected
Pin 7: filament
Pin 8: ground

What can I do about this?  The tubes test good in your typical tester (I've got a bk667).  Are they toast anyways?

[drew]

Aren't Pins 1 and 8 supposed to be tied together with EL34s?


Don't try this until one of the smart people weigh in or you verify it yourself.

[SILVERGUN]

Aren't Pins 1 and 8 supposed to be tied together with EL34s?

Yes, you are correct Drew
Good catch

[tubeswell]

Bias is related to current. So just trying to guess the appropriate bias going by the grid voltage is insufficient to determine adequacy. You really need to measure the tube current at the same time (e.g. by measuring the voltage drop across a 1% tolerance 1R or 10R resistor connected in series between the cathode pin of each output tube and the ground return). Therefore, if using 1R measurement resistors, you'd multiply the measured voltage drop (in mV, which represents the tube current in mA*) by the plate voltage to get a rough order estimate of plate current. e.g. 32mA x 450V = 14.4W, which would be 58% of maximum, which is okay for EL34s.

*Ohms Law 1V=1R x 1A, or in this case 1mV = 1R x 1mA

Furthermore, for modern production EL34s, you really don't want to bias them at 70% of 25W, because most modern production** EL34s are mostly unfit to deserve a 25W rating. Lord Valve has opined (for example) that they should only be regarded at 20W tubes at best. This means keeping the bias on the cold side - say by adopting 70% of 20W, or adopting 53 or 54% of 25W as your marker.

** Although I've heard good reports about ST EL34s handling 500 plate volts okay, but I'd still cautiously bias them a lot closer to Class B than Class A/B. As for the New Sensor's EL34 range, I won't run them in any amp over 420.

But when designing voltages for bias supplies, a good rule of thumb is that the bias voltage should cover a range +/- 10% of the plate voltage. So seeing as how you report 450V on the plates, I'd aim to have a bias voltage adjustment that covers -45V to -50V. This may mean decreasing the value of the AC resistor in the bias supply by a 'notch' or two if you are unable to get this range on dialing the bias pot.

[92Volts]

Besides tubeswell's (very good) advice of measuring tube current instead of just setting a "typical" bias voltage, you should check whether that voltage of -25 or -40 is actually making it onto the grids of the tubes. An accidental connection to ground or missing connection of any sort between the bias supply, 5.6k grid resistor, and the tubes will cause that to fail.

A missing or faulty (shorted) coupling cap between the previous stage (usually the phase inverter) and the grids of your power tubes will also cause this, since the plate (output) of the phase inverter is at +100s of volts DC before you block it with a cap.

[TheKT88KilledJFK]

Aren't Pins 1 and 8 supposed to be tied together with EL34s?


Don't try this until one of the smart people weigh in or you verify it yourself.

They're floating on the inside pair of tubes for the schematic I found.  They shouldn't be though, so I soldered a connection there on the socket.

Bias is related to current. So just trying to guess the appropriate bias going by the grid voltage is insufficient to determine adequacy. You really need to measure the tube current at the same time (e.g. by measuring the voltage drop across a 1% tolerance 1R or 10R resistor connected in series between the cathode pin of each output tube and the ground return). Therefore, if using 1R measurement resistors, you'd multiply the measured voltage drop (in mV, which represents the tube current in mA*) by the plate voltage to get a rough order estimate of plate current. e.g. 32mA x 450V = 14.4W, which would be 58% of maximum, which is okay for EL34s.

*Ohms Law 1V=1R x 1A, or in this case 1mV = 1R x 1mA

Furthermore, for modern production EL34s, you really don't want to bias them at 70% of 25W, because most modern production** EL34s are mostly unfit to deserve a 25W rating. Lord Valve has opined (for example) that they should only be regarded at 20W tubes at best. This means keeping the bias on the cold side - say by adopting 70% of 20W, or adopting 53 or 54% of 25W as your marker.

** Although I've heard good reports about ST EL34s handling 500 plate volts okay, but I'd still cautiously bias them a lot closer to Class B than Class A/B. As for the New Sensor's EL34 range, I won't run them in any amp over 420.

But when designing voltages for bias supplies, a good rule of thumb is that the bias voltage should cover a range +/- 10% of the plate voltage. So seeing as how you report 450V on the plates, I'd aim to have a bias voltage adjustment that covers -45V to -50V. This may mean decreasing the value of the AC resistor in the bias supply by a 'notch' or two if you are unable to get this range on dialing the bias pot.

I've got a bias probe that measures the tube hitting 45 mA fast then swelling up to 60+ mA before I just turn the amp off.  This occurs at any voltage.  I've never had tubes do this before, so I assume it's a fault with my circuit.  The only thing is, I designed the PCB for this and had it fabricated.  The PCB has no faults and complies with the schematic. 

Besides tubeswell's (very good) advice of measuring tube current instead of just setting a "typical" bias voltage, you should check whether that voltage of -25 or -40 is actually making it onto the grids of the tubes. An accidental connection to ground or missing connection of any sort between the bias supply, 5.6k grid resistor, and the tubes will cause that to fail.

A missing or faulty (shorted) coupling cap between the previous stage (usually the phase inverter) and the grids of your power tubes will also cause this, since the plate (output) of the phase inverter is at +100s of volts DC before you block it with a cap.

There's no DC on the input aside from the bias voltage.  The caps are brand new, not that doesn't mean they couldn't fail.  I checked the grid input, alas no voltage that you wouldn't expect.  Maybe I will just pull the caps to check. 

So, in summary, I connected pins one and eight.  I checked the cathode current which was at 60mA and rising.  The bias voltage is pretty much anywhere I place it.  I've placed it at -70VDC and 0 VDC and spots between like normal operating areas like 40 VDC like the schem specs (obviously the current is the important parameter though). 

Thanks for the help so far guys. 
 

[DummyLoad]

pull the tubes and measure pin 5: you should have -35V or thereabouts on both EL34 pin 5. if you don't, fix that first. as already stated: pins 1 & 8 need to be tied together for EL34. your tubes are probably ok for shop beaters - if you shut them down right away.

if the screen supply is near or at the 450V plate supply -25V is too hot for a El34 bias.


--pete

[TheKT88KilledJFK]

pull the tubes and measure pin 5: you should have -35V or thereabouts on both EL34 pin 5. if you don't, fix that first. as already stated: pins 1 & 8 need to be tied together for EL34. your tubes are probably ok for shop beaters - if you shut them down right away.

if the screen supply is near or at the 450V plate supply -25V is too hot for a El34 bias.


--pete

I attached a probe to the underside of pin 5 so I could read it during operation even if a tube is there.  I kept the bias probe in place.  When setting the bias to -35vdc, the current just keeps climbing no matter what.  I turned it off again at 60mA. 

I'm not sure how this is possible to be honest.   

Edit:  I just swapped in some known good el34s, and the same thing happens.  I've screwed up something here....

Edit #2:  I disconnected the coupling caps feeding the power tube section, no change in dc bias was observed in the section.  The current still rises with pin 5 bias voltage set at -35vdc. 

This makes absolutely no sense. 

[tubeswell]

pull the tubes and measure pin 5: you should have -35V or thereabouts on both EL34 pin 5. if you don't, fix that first. as already stated: pins 1 & 8 need to be tied together for EL34. your tubes are probably ok for shop beaters - if you shut them down right away.

if the screen supply is near or at the 450V plate supply -25V is too hot for a El34 bias.


--pete

I attached a probe to the underside of pin 5 so I could read it during operation even if a tube is there.  I kept the bias probe in place.  When setting the bias to -35vdc, the current just keeps climbing no matter what.  I turned it off again at 60mA. 

I'm not sure how this is possible to be honest.   

Edit:  I just swapped in some known good el34s, and the same thing happens.  I've screwed up something here....

Whether to not it complies with a Marshall schematic, -35V is not enough. Read my post earlier. You want at least -45V to -50V with a plate voltage of 450ish. You need to make adjustments to the bias supply circuit.

[DummyLoad]

pictures of the power tube sockets wiring would be helpful.


-pete

[DummyLoad]

pull the tubes and measure pin 5: you should have -35V or thereabouts on both EL34 pin 5. if you don't, fix that first. as already stated: pins 1 & 8 need to be tied together for EL34. your tubes are probably ok for shop beaters - if you shut them down right away.

if the screen supply is near or at the 450V plate supply -25V is too hot for a El34 bias.


--pete

I attached a probe to the underside of pin 5 so I could read it during operation even if a tube is there.  I kept the bias probe in place.  When setting the bias to -35vdc, the current just keeps climbing no matter what.  I turned it off again at 60mA. 

I'm not sure how this is possible to be honest.   

Edit:  I just swapped in some known good el34s, and the same thing happens.  I've screwed up something here....

Whether to not it complies with a Marshall schematic, -35V is not enough. Read my post earlier. You want at least -45V to -50V with a plate voltage of 450ish. You need to make adjustments to the bias supply circuit.

oops! yes follow tubeswell's guidance. you need -45V or more. vg2/-mu and mu is about 10.


--pete

[TheKT88KilledJFK]

pictures of the power tube sockets wiring would be helpful.


-pete

No doubt!  Here is the socket I've been running these test on.

[92Volts]

I checked the grid input, alas no voltage that you wouldn't expect.

To be clear, are the grids themselves receiving -25v or -40v or whatever your bias setting is? From what you say, it sounds like they aren't receiving 200v from the previous stage-- which is good-- but you don't mention what they do read.

The only voltage we "expect" at the grids is exactly equal to the bias voltage.

If that's all correct, it sounds like you just need more negative bias voltage.

[tubeswell]

What gets me these days is even Marshall in their reissue 'handwired' 1959 SL100 contains values that were on the schematics that were first drawn in the mid-1960s, in the heyday of tube manufacture, where the quality of tubes was generally superior to what's available today, and they show components values that just don't work with today's tubes. I've had 2 of these RI amps in for servicing in the last couple of months. They run the plate voltage at about 500 and the screens about 8 or 9V below that, and the stock bias circuit only delivers about -40V, and the tubes are idling at 60mA (if they last). To get them to idle at 30mA, you have to decrease the AC (dropping) resistor in the bias supply from the stock 27k to 22k (or better still 18k). Doesn't matter what values are written on the schematic, you need to have the circuit delivering sufficient negative voltage.


And what's critical with EL34s more than any other tubes, is if you're going to run a quad, make sure they are closely matched to within 3 or 4mA of each other. Otherwise the ones that run hotter than that will kick the bucket in no time, and you risk burning up the OT. (I had to replace the OT in one of those 1959s).


The thing about EL34s is the signal grid is wound in relatively close spatial proximity to the cathode, to squeeze every ounce of gain out of them. This requires close manufacturing tolerances to achieve consistency. But modern manufacturing tolerances being lax, you can't guarantee that you receive a batch of tubes that behave under those idle conditions. Even some NOS tubes made in the 1980s, like Soviet EL34G had huge divergence in operating conditions. You can find a quad that will run okay in those conditions, but you have to match them yourself by buying about 2 dozen tubes and measuring them individually on a setup with a screen voltage of about 320*, to find which ones are going to naturally idle the coolest. Same goes for today's New Sensor tubes. I just don't trust them to behave in those circuits.


* The amount of screen voltage you have affects how much negative bias voltage you need to get the tube to turn on. With ~490 plate volts, and -48 grid volts, you need ~480 g2 volts to get about 34mA. But you can get the same tube current with Vg1 = -45, Vg2 = 370 and Va = 520. Conversely, if you lower the Vg2 to 350 with Va = 510, the tube current will drop off to about 10 to 12mA. And riding the B+ ladder (by changing the voltage divider at the screen supply node in a 1959 to deliver 350 at the junction of the two screen supply caps) is one way of rigging the amp to ensure you are going to get a good matched set.

[TheKT88KilledJFK]

What gets me these days is even Marshall in their reissue 'handwired' 1959 SL100 contains values that were on the schematics that were first drawn in the mid-1960s, in the heyday of tube manufacture, where the quality of tubes was generally superior to what's available today, and they show components values that just don't work with today's tubes. I've had 2 of these RI amps in for servicing in the last couple of months. They run the plate voltage at about 500 and the screens about 8 or 9V below that, and the stock bias circuit only delivers about -40V, and the tubes are idling at 60mA (if they last). To get them to idle at 30mA, you have to decrease the AC (dropping) resistor in the bias supply from the stock 27k to 22k (or better still 18k). Doesn't matter what values are written on the schematic, you need to have the circuit delivering sufficient negative voltage.


And what's critical with EL34s more than any other tubes, is if you're going to run a quad, make sure they are closely matched to within 3 or 4mA of each other. Otherwise the ones that run hotter than that will kick the bucket in no time, and you risk burning up the OT. (I had to replace the OT in one of those 1959s).


The thing about EL34s is the signal grid is wound in relatively close spatial proximity to the cathode, to squeeze every ounce of gain out of them. This requires close manufacturing tolerances to achieve consistency. But modern manufacturing tolerances being lax, you can't guarantee that you receive a batch of tubes that behave under those idle conditions. Even some NOS tubes made in the 1980s, like Soviet EL34G had huge divergence in operating conditions. You can find a quad that will run okay in those conditions, but you have to match them yourself by buying about 2 dozen tubes and measuring them individually on a setup with a screen voltage of about 320, to find which ones are going to naturally idle the coolest. Same goes for today's New Sensor tubes. I just don't trust them to behave in those circuits.

I hear you man.  I set the bias down to -60VDC by adding a 680k resistor in series with the bias pot.  After all, it's just a voltage divider setup.  At this point the tubes weren't red plating anymore.  I'm going to have to have to put a large value pot in to find the sweet spot, because this took me too long to do by switching resistors all day.  I might lower the other resistor values instead.  My concern is there is some threshold of parasitic current necessary for voltage to "develop" at the grid for bias.  Honestly this is the first time I've had to confront this problem.  With this current bias setup, the circuit's output impedance is approaching a 750k and that worries me. 

I was very hesitant to modify a design that should just work, but your post kind of confirmed what I was thinking. 

[sluckey]

I set the bias down to -60VDC by adding a 680k resistor in series with the bias pot.

There's something bad wrong with that simple bias circuit if you need a 680K resistor to get -60v. I'd like to see a hi rez pic that clearly shows every component such that values can be read in your bias supply. Also, do you have a separate bias winding on the PT like the original? How much ac voltage is feeding the bias circuit?

[92Volts]

I found this schematic for the 2203: http://drtube.com/schematics/marshall/2203u.gif

And if you're going off that, it completely relies on the value of that transformer winding, which appears to be dedicated to bias. So if that's wrong you'll never have enough voltage.

Reducing the 15k resistor would be another way to get more negative voltage, but if you're heading up towards 680k and increasing the series resistance is still having an effect, it sounds like something is totally wrong or that isn't the circuit you built. Maybe you have a 150k in place of that 15k?

You do want less overall circuit resistance than that. The resistors hooking the EL34s to the bias are 150k, not 1M or something arbitrarily large, because you can get current flow from the grids with overheated/faulty tubes and you want to be able to handle that.

[TheKT88KilledJFK]

I set the bias down to -60VDC by adding a 680k resistor in series with the bias pot.

There's something bad wrong with that simple bias circuit if you need a 680K resistor to get -60v. I'd like to see a hi rez pic that clearly shows every component such that values can be read in your bias supply. Also, do you have a separate bias winding on the PT like the original? How much ac voltage is feeding the bias circuit?

I just checked it and it's about 58 VAC with the line open.  I've attached the picture.  The schematic I've followed is here:

 http://www.thetubestore.com/lib/thetubestore/schematics/Marshall/Marshall-JCM800-Lead-Mstvol-100W-2203-Schematic.pdf

You can see that R1 goes to the 25k bias pot as well as to the 220k network then the grid resistors.  The input from the transformer goes into R3.  The metal boxes are 10uf caps.  I attached my interpretation of the circuit for what it's worth. 

[DummyLoad]

change R3 to 2.2k, R2 to 10K and R1 to 27k. SIm shows -42V to -54V range with all pot value at same and 58VAC supply. change 1/8W resistor to 1/2W or better.


--pete

[TheKT88KilledJFK]

change R3 to 2.2k, R2 to 10K and R1 to 27k. SIm shows -42V to -54V range with all pot value at same and 58VAC supply. change 1/8W resistor to 1/2W or better.


--pete

Wow thanks for the taking the time to do that.  That is definitely not what I was intending to happen, but I will change those resistors tonight. 

So, my assumption that the tubes should not red-plate at a factory spec'd bias bias voltage was simply wrong.  What tubeswell and others mentioned is that it is never appropriate to rely on that reading.  By relying on that reading I lead to the conclusion that there were other issues in the circuit when simply the bias was too hot (even though the lower magnitude bias voltage tap is also still a concern that needs to be sorted). 

[DummyLoad]

please check bias supply after changes before installing output tubes.


--pete

[PRR]

> When I first plugged the amp in, the bias sat at -25VDC and the tubes began to red plate at this setting.  I set the voltage to the spec'd -40VDC setting

ALWAYS turn the bias to MAX negative before putting power tubes in. Start at dead-cold idle, not with the go-pedal mashed.

I agree with Tubeswell's notion that for most audio tubes, the bias should be at-least 1/10th of the *screen* voltage. This comes from most audio tubes having Mu(g2) around 10, so you need about 1/10th of Vg2 to counter-act the suck G2 puts on the electrons and put the tube near cut-off. That won't be the final value, but for most audio power tubes at rational voltages it won't be a melt-down.

Here I assume that Vg2 is very-near your 450V on plates. Most big G-amps run this way. Say 440V?

I looked in the old data. We can estimate G1 vs G2 bias from the "triode curves", because they cover a full range of G2 voltages, and because Plate voltage has little to do with the current you will get.

Funny thing is, the 1964 data says -40V (I had to extrapolate that) IS a good point for 440V on G2. Should give 35mA and 16 Watts Pdiss, well below any likely red-plate.

Tubes HAVE varied over the 50 years and a dozen factories. I found Svetlana 1997 data, which appears to be a new plot, not a re-copy of old data. 400V G2 and -40V G1 looks like 20mA. Corrected (1.5 power law) for 440V G2 this is 23mA, also an OK and safer value. This is WAY down in the knee-curve of the plot, so there will be a lot of variation with microscopic production tolerances. Still, it does seem like 440V and -40V "should be" a fine bias.

I wonder why yours isn't.

[TheKT88KilledJFK]

> When I first plugged the amp in, the bias sat at -25VDC and the tubes began to red plate at this setting.  I set the voltage to the spec'd -40VDC setting

ALWAYS turn the bias to MAX negative before putting power tubes in. Start at dead-cold idle, not with the go-pedal mashed.

I agree with Tubeswell's notion that for most audio tubes, the bias should be at-least 1/10th of the *screen* voltage. This comes from most audio tubes having Mu(g2) around 10, so you need about 1/10th of Vg2 to counter-act the suck G2 puts on the electrons and put the tube near cut-off. That won't be the final value, but for most audio power tubes at rational voltages it won't be a melt-down.

Here I assume that Vg2 is very-near your 450V on plates. Most big G-amps run this way. Say 440V?

I looked in the old data. We can estimate G1 vs G2 bias from the "triode curves", because they cover a full range of G2 voltages, and because Plate voltage has little to do with the current you will get.

Funny thing is, the 1964 data says -40V (I had to extrapolate that) IS a good point for 440V on G2. Should give 35mA and 16 Watts Pdiss, well below any likely red-plate.

Tubes HAVE varied over the 50 years and a dozen factories. I found Svetlana 1997 data, which appears to be a new plot, not a re-copy of old data. 400V G2 and -40V G1 looks like 20mA. Corrected (1.5 power law) for 440V G2 this is 23mA, also an OK and safer value. This is WAY down in the knee-curve of the plot, so there will be a lot of variation with microscopic production tolerances. Still, it does seem like 440V and -40V "should be" a fine bias.

I wonder why yours isn't.

Holy cow you found a chart relating bias voltage to dissipation.  I've been looking for that while keeping in mind I may not be able to rely on it.  As you said, it gives us an idea of range of values that would be potentially acceptable which is exactly what a bias circuit should cover. 

I still have to get home to check values. 

[bnwitt]

You just need to find a data sheet for your tubes.  Every manufacturer has different characteristics.

[bnwitt]

Attached is an EL34 Bias Chart based on 25 watts maximum plate dissipation.

[bnwitt]

And here is an Excel spreadsheet with the most common power tubes.  You just input the manufacturer's max plate dissipation and the rest calculates out.  I don't use this any more as I have an android app called FTBC Lite that does it for me

[bnwitt]

Make sure you verify the max plate dissipation of the tubes you have with the manufacturer's data sheet.

[PRR]

> You just need to find a data sheet

Your JJ sheet only shows 250V screen voltage.

Your general bias chart does not show any G1 or G2 voltage, only current versus dissipation for various plate voltages (which is simple multiplication).

Use the "triode" curves to get a guide for G1 and G2 voltages. Plot the dissipation by hand, as I did to get a "20W" line in accord with what some gurus say is a safe limit for plate alone. (The triode curve shows a 30W line but that is for both P and G2 tied together, which is not quite right for our use.)

[TheKT88KilledJFK]

Sorry this took so long to reply, but today is my off day! 

The sole diode is measuring 127k ohms of resistance in forward bias when removed from the circuit.  It's very intermittent, but became more consistent in it's failure to where it always has that reading (perhaps operator error?).  It's a brand new 1000v 1n4007, but I must have missed something on the spec sheet for a new diode to exhibit this behavior, or it has failed in an odd way.   I mean, it's pretty obvious diodes don't usually work this way.  Looking at the current/power ratings, I have a hard time believing I exceeded any spec even if the grids shorted to ground (or B+) during arcing in the power tubes but perhaps it did and it lead to a run-away condition. 


Edit: upon further reading it looks like this is typical operation for a diode.  I always just looked at forward voltage drop, so maybe the multimeter is giving a linear calculation of a non linear circuit thus inaccurate.   

[sluckey]

So you are using the old forward/reverse resistance reading to check the diode? Maybe even using an analog meter? That's still my preferred method to check most diodes. A good 1N4007 will read very low forward resistance and very high (probably infinity) reverse resistance. The actual readings depend on the resistance range you are using. The actual resistance reading is not important. The front to back ratio is what's important.

***IF*** you are touching the meter leads or diode leads the reading will be erroneous. Your body resist will be in parallel with the diode and will change the high resistance reading. Not saying you are doing this but be aware and never hold a component (resistor, diode, cap, etc.) with your hands when measuring resistance.

If the diode really shows 127K forward resistance then check a new diode to determine if your procedure is faulty of if the diode is indeed bad.

[TheKT88KilledJFK]

So you are using the old forward/reverse resistance reading to check the diode? Maybe even using an analog meter? That's still my preferred method to check most diodes. A good 1N4007 will read very low forward resistance and very high (probably infinity) reverse resistance. The actual readings depend on the resistance range you are using. The actual resistance reading is not important. The front to back ratio is what's important.

***IF*** you are touching the meter leads or diode leads the reading will be erroneous. Your body resist will be in parallel with the diode and will change the high resistance reading. Not saying you are doing this but be aware and never hold a component (resistor, diode, cap, etc.) with your hands when measuring resistance.

If the diode really shows 127K forward resistance then check a new diode to determine if your procedure is faulty of if the diode is indeed bad.

I checked every resistor I have on hand from germanium 1n4148s to 1n4001s and they all read in the 100k+ region.  I'm pretty sure it's just the way my multimeter is making the determination.  In this case, the diode appeared to have high resistance to me because the cap on that node was installed backwards (it hurts to type this.  The part I had wasn't supposed to have a polarity, I swear.  There was no indication of polarity on the package or a lead that was longer.) 

I figured that out when it also exploded.  It was like a fog machine that smelled like bad ramen noodles!



Anyways, I put bigger caps in and it appears all is well.  I think there is an issue with a lack of volume, even though I running the current*plate voltage at 70% of the rated 25w.   

That being said, I ended up swapping in some 30uF and going to a 33k resistor in series with the 25k pot.  The plates on this set of tubes are happy now, and the bias has an adjustment range of -45vDC to ~53vDC.  There is a hum on the input of the plates, probably coming from the bias section.  My o-scope sucks too much even dial this noise in on the screen, but it's there.

The phase inverters have about 12v of swing on each side, I don't know if that's appropriate or not.  I know the setup is biased at -45VDC, so it would seem that I am way under that on the PI side of things.  That just the next thing to check I guess. 

[sluckey]

they all read in the 100k+ region.  I'm pretty sure it's just the way my multimeter is making the determination.

That's OK as long as the reverse reading is near infinity. IT'S THE RATIO THAT COUNTS.

In this case, the diode appeared to have high resistance to me because the cap on that node was installed backwards

But you just said you were checking the diode removed from the circuit.

If you are using a digital meter, just use the diode check function and forget about front to back resistance ratios.

[TheKT88KilledJFK]

they all read in the 100k+ region.  I'm pretty sure it's just the way my multimeter is making the determination.

That's OK as long as the reverse reading is near infinity. IT'S THE RATIO THAT COUNTS.

In this case, the diode appeared to have high resistance to me because the cap on that node was installed backwards

But you just said you were checking the diode removed from the circuit.

If you are using a digital meter, just use the diode check function and forget about front to back resistance ratios.

Of course I measured it in circuit as well, that's what lead me to remove it and test :D .  That node was dropping three quarters of the rectified voltage before it even hit the voltage divider net.  This all boils down to my mistake but it always does as you probably realize lol.   

They all have more resistance reverse biased than what my digital multimeter can register. 

[shooter]

They all have more resistance reverse biased than what my digital multimeter can register.

infinity can be hard to measure

[tubeswell]

The phase inverters have about 12v of swing on each side, I don't know if that's appropriate or not.  I know the setup is biased at -45VDC, so it would seem that I am way under that on the PI side of things.  That just the next thing to check I guess.

Did you say earlier that you were building and testing the amp incrementally from back to front? (Sorry I'm too lazy to read the whole thread again from the start). The PI output should come up as you add more gain stages to the preamp.