replacing resistors with trim pots/rheostat

[user 41014]

When replacing fixed bias resistors (or other) with trim pots, is there a standard way to connect them? Thinking of what a resistor sees a rheostat is the same two terminals but I never hear anyone say hey replace your bias R with a rheostat. So does the non source terminal go to ground and the wiper is signal out as you'd expect? As a variable voltage divider the extra voltage has to go somewhere or else to heat if one leg is floating like a rheostat/resistor.

I need to upgrade the PS on a IIB I'm working on (all of them for that matter) to MB's recommendation. While I'm wrapping my head around where to put the extra filter I'm eyeing up the bias R for the ability to use 387's too hot for Marks otherwise.

The second reason I ask is I was wondering about preamp anode resistors. Whether for testing, fishing for a sweet spot or versatility has anyone installed trim pots on the plates of V1-V3? I was thinking how handy getting 32k-100k V1A/V1B values on an expression pedal would be with my 76 boogie to dial in the gain/clean compromise. Again, rheostat floating or non-plate leg grounded if using a trim pot?

Also, to manage heat on a trim pot is it a good idea to use smaller value pots in series following a R that takes some of the load? E.g. For a 3k-6k bias R range, is it better to use a 3k R with a 5k trim pot to manage the additional 0-3k variable rather than using a 10k pot to manage 3k-6k?

 

[bandit2013]

Just a few comments in general. Perhaps I am being Capitan Obvious here. Sorry if any of this is rude or unwelcomed.

Considering the potentiometer the wiper terminal can be tied to the unused terminal since the wiper would be a short across that area anyways. Would be more of a failsafe if the wipers breaks contact with the resistive element. The shorted-out area will not be dissipating heat as there would be no current flowing through it. In either case, a trim pot should be used with a series resistor as this will keep you from shorting out the bias voltage. In most bias voltage circuits, it is not a single resistor that is used to set the voltage. It is typically a parallel circuit to a resistance of smaller value thus the adjustment will not have the same level of current passing through it. The smaller valued resistor would be the minimum resistance to maintain the bias voltage in an open circuit condition at a safe operating condition. The added parallel resistance with a variable pot would be there to reduce the circuit resistance that would result in dropping the bias voltage. Lower the bias voltage the higher the plate current. A multi-turn trim pot would provide better accuracy as the change in value will be in smaller increments vs larger changes by a single turn trim pot. It does not take much change in voltage to have a larger change in mA current in the power tube.

Change in plate resistance is generally not done. Sure, it changes the gain characteristic but also changes the load line of the triode circuit. Having a trim pot in those areas would be in a high voltage potential. It is still low current, but the voltage potential would be the concern since most potentiometers do not meet the creepage requirement for the applied voltage. Still the concept of doing this would be cool. I would probably determine what resistance would be the best sound and then replace with a fixed resistor of the same type used originally. Also, you would need to verify if there is any plate voltage shift if there is a channel switch or pull-pot of sorts. Not too familiar with the IIB.

If you want to bring in more low end, add a cap to the cathode. Does not need to be a large value. This usually increases the bass response. If you feel there is too much low end, you can reduce the bypass capacitor value if there is one, to lower the signal gain response in the lower frequencies.

 

[user 41014]

Just a few comments in general. Perhaps I am being Capitan Obvious here. Sorry if any of this is rude or unwelcomed.

Considering the potentiometer the wiper terminal can be tied to the unused terminal since the wiper would be a short across that area anyways. Would be more of a failsafe if the wipers breaks contact with the resistive element. The shorted-out area will not be dissipating heat as there would be no current flowing through it. In either case, a trim pot should be used with a series resistor as this will keep you from shorting out the bias voltage. In most bias voltage circuits, it is not a single resistor that is used to set the voltage. It is typically a parallel circuit to a resistance of smaller value thus the adjustment will not have the same level of current passing through it. The smaller valued resistor would be the minimum resistance to maintain the bias voltage in an open circuit condition at a safe operating condition. The added parallel resistance with a variable pot would be there to reduce the circuit resistance that would result in dropping the bias voltage. Lower the bias voltage the higher the plate current. A multi-turn trim pot would provide better accuracy as the change in value will be in smaller increments vs larger changes by a single turn trim pot. It does not take much change in voltage to have a larger change in mA current in the power tube.

Change in plate resistance is generally not done. Sure, it changes the gain characteristic but also changes the load line of the triode circuit. Having a trim pot in those areas would be in a high voltage potential. It is still low current, but the voltage potential would be the concern since most potentiometers do not meet the creepage requirement for the applied voltage. Still the concept of doing this would be cool. I would probably determine what resistance would be the best sound and then replace with a fixed resistor of the same type used originally. Also, you would need to verify if there is any plate voltage shift if there is a channel switch or pull-pot of sorts. Not too familiar with the IIB.

If you want to bring in more low end, add a cap to the cathode. Does not need to be a large value. This usually increases the bass response. If you feel there is too much low end, you can reduce the bypass capacitor value if there is one, to lower the signal gain response in the lower frequencies.

I could use a 10 ton wall of Captain Obvious falling on me for this one. Thank you for the response.

I've read about tying the wiper and one lug together as a best practise for the same reasons you've laid out which of course is wiring the pot as a rheostat. Until reading that I was wrongly thinking the outer pot lugs provided total pot ohms in series to ground with the added wiper load. I grabbed a pot and tested the tying lug and wiper with my meter. I've been meaning to use a breadboard, pot, resistor and 9V battery to test this further.

Interesting that this PS board bias R may be in parallel with another circuit piece which could only be on the DR board with the PI tubes. Charlie Daniels had trim pots added to the DR boards of his 2 coli's and I've been I meaning to trace that as it looks like it is being used as a pot (all 3 legs being used). It looks professionally done and they can't be more than 1/2W but all the resistors at the PI are 1/2 watt.

I was thinking parallel fixed R with pot if wired as rheostat to reduce current. I have a 20k pot from Mesa but it is a 3296 series which is 1/2 watt while the current bias R is a 1W. At first I thought using a series resistor ahead of the variable R to reduce the V drop across the variable R. Then I started thinking using a CC in parallel with the variable to drop the current flow across the pot. The problem with that is if I'm allowing for 1/2 to 3x (of my now 5.6K ohm) bias swing I'm flowing the bulk of the current through the pot unless using exceptionally hot tubes. Now you have me considering an open circuit condition as well.

I have a 20k 1W on its way to alleviate dissipation concerns and free up the design decision. I normally like the idea of stepped repeatable values and was thinking 22 or 25 turn but went with single turn for infinite tuning.

OG Boogie plate R's:
Thanks for pointing out the load line and by extension maybe the bias point.
The cathodes are all 1.5K ohm Rk with 22uF bypass on the '76 which has an equivalent 50k for V1A/V1B anodes. This seems pretty standard (22 or 25uF bypass) for gain stages and I appreciate the input about adjusting the low end. The manual states 32k-100k loads can be used depending on the signal. I need to just get an oscilloscope but a double bourns 100k pot seemed like a cool alternative and use ears to dial it instead.

I'll still do the breadboard. Rheostat wiring (tying the wiper to outer lug) is the easiest way to wrap my head around but I still like the idea of tying the non-source pot lug to ground. I can't find either way being a standard but I haven't found much info for fixed to variable bias mods.

 

[bandit2013]

I am not all too familiar with the Mark amp you are working on. To actually find a schematic that is legible is probably more difficult. Just as a reference only as a point of discussion. Here is the bias circuit for the Mark V90W amp.

The actual bias resistor is the R155 (33k ohms). Since it creates a voltage divider with the R99 (3.3k ohms). This circuit is fed with a 60V AC source. Half rectified so only one diode drop and lots of ripple. The series resistor R152 keeps the inrush current to the two 470uF caps at bay. However, it also factors in as part of the bias circuit too. It is not as stable as you would think it is. Since most of the current is passing through the 33k ohm resistor, the parallel resistor of 82k ohms is the tuning resistor to set the bias voltage at that point. Adding in a trim pot in series with the 82k would make this adjustable. or change of the 82k to a slightly lower value. Note the actual part that was used is a 1% resistor so it is 82.5k ohms. Also this amp is a Simul-Class type. I had to change the 82k resistor with a 91K to prevent the power tubes on the Class A portion from red plating. Measured resistance of the 91k ohms was 86k as it was a 10% resistor. I could have made it adjustable had I used the 82.5k and placed a 10k or 20k pot in series (wired as a rheostat). Found the 86k prevented red plating and the total output power was 110W vs the 150W I was seeing before meltdown of the power tubes.

Then again, what is following the Phase inverter where the bias is injected. if there is a component mismatch there could lead to an unbalanced pair. The Mark V90 has the bias injected through two 220k ohm resistors. the control grids of the Class A/B section are 2.2k resistors. I believe they are both 1W. The issue was with the extended class A tubes since it has 2.2M resistors to ground and 150K ohm feeds to create a voltage divider and act as a grid stopper at the same time. If the V90 was a class A/B amp, it probably would not be much of an issue.

Considering the applied voltages of a tube amp. It is always wise to use a higher wattage resistor than what may be required. Overhead is a good thing. The bias current is not the factor but the actual voltage is what matters. Sizing the components to prevent failure is the key.

Also be weary of wire wound pots. They can handle more current than the simple resistive elements but can have a negative effect due to the associated inductance of the coil or spiral wrapped resistor.

Perhaps I need to be educated in the likes of what the bias circuit is supposed to behave like. Is it desired to be stable or does it change with influences of signal? I would assume it is meant to be stable as it is DC and not an AC signal.

I only offered up a few thoughts on the subject to make you think about what you want to do and is the goal achievable. I am sure there is a way to do what you want. How do get there may be a bit over my head. I only chimed in as this peaked my interests a bit.

 

[user 41014]

I am not all too familiar with the Mark amp you are working on. To actually find a schematic that is legible is probably more difficult. Just as a reference only as a point of discussion. Here is the bias circuit for the Mark V90W amp.

View attachment 1374

The actual bias resistor is the R155 (33k ohms). Since it creates a voltage divider with the R99 (3.3k ohms). This circuit is fed with a 60V AC source. Half rectified so only one diode drop and lots of ripple. The series resistor R152 keeps the inrush current to the two 470uF caps at bay. However, it also factors in as part of the bias circuit too. It is not as stable as you would think it is. Since most of the current is passing through the 33k ohm resistor, the parallel resistor of 82k ohms is the tuning resistor to set the bias voltage at that point. Adding in a trim pot in series with the 82k would make this adjustable. or change of the 82k to a slightly lower value. Note the actual part that was used is a 1% resistor so it is 82.5k ohms. Also this amp is a Simul-Class type. I had to change the 82k resistor with a 91K to prevent the power tubes on the Class A portion from red plating. Measured resistance of the 91k ohms was 86k as it was a 10% resistor. I could have made it adjustable had I used the 82.5k and placed a 10k or 20k pot in series (wired as a rheostat). Found the 86k prevented red plating and the total output power was 110W vs the 150W I was seeing before meltdown of the power tubes.

Then again, what is following the Phase inverter where the bias is injected. if there is a component mismatch there could lead to an unbalanced pair. The Mark V90 has the bias injected through two 220k ohm resistors. the control grids of the Class A/B section are 2.2k resistors. I believe they are both 1W. The issue was with the extended class A tubes since it has 2.2M resistors to ground and 150K ohm feeds to create a voltage divider and act as a grid stopper at the same time. If the V90 was a class A/B amp, it probably would not be much of an issue.

Considering the applied voltages of a tube amp. It is always wise to use a higher wattage resistor than what may be required. Overhead is a good thing. The bias current is not the factor but the actual voltage is what matters. Sizing the components to prevent failure is the key.

Also be weary of wire wound pots. They can handle more current than the simple resistive elements but can have a negative effect due to the associated inductance of the coil or spiral wrapped resistor.

Perhaps I need to be educated in the likes of what the bias circuit is supposed to behave like. Is it desired to be stable or does it change with influences of signal? I would assume it is meant to be stable as it is DC and not an AC signal.

I only offered up a few thoughts on the subject to make you think about what you want to do and is the goal achievable. I am sure there is a way to do what you want. How do get there may be a bit over my head. I only chimed in as this peaked my interests a bit.

Talking this through helps (I usually talk to myself or dogs lol) but you've given me a lot of good food for thought. What is the +24V rail for on the V90? Is it the EQ since I don't see it being pulled from the -V like old Marks?

It's a IIB coli I'm 1st considering this bias mod on. All the Marks up to the mid 80's were susceptible to over-voltage at the screens causing the 30uF/500V screen cap to bulge too often. I replaced it with the pair of 47uF filters (each with 150K ohm 1W) as per MB and in doing so had to move the bias caps and R out of the way. It seemed the time to take a stab at a bias mod.

Seeing Charlie's coli's bias mods on the DR board gives me a little pause (all IIB's, same SP-9A preamp boards) but the -52V (or close to it) is established on the PS board.

And this would likely be easier to understand on any Mark but a coli due to the dual tube phase inverter that I can't find any literature regarding. I see resistors that tell me it should be a long tail with complications but one tube on the DR board has heaters not even connected and there's a socket for a top hat looking device that may be a current limiting diode on the ground plane. Still it's a good challenge and I understand the 2nd tube is for buffering but if heaters aren't warming the tube what is really going on?

Here is the actual bias circuit of my IIB on top. R2 is the target. Do you know why C2 and R3 were added to the bias circuit over time? The lower schematic of my '76 used just one bias cap with no R3. I didn't think ripple mattered as much for the grid but I see the V is still using the dual bias caps.

I'm pretty far from being an expert but I would assume you'd want the idle current to be as stable as possible just like you wouldn't want your car's idle to be erratic, but that is without a signal.

I was bouncing around the idea of using a 50k pot || fixed CC and that general concept value-wise looks to be in use on the V90. The 1W 20k trimmer will be here this week tho and while I don't really know I'm guessing I want the ability to go higher than 10K tops. A 2.7k in series with a 20k pot || 20k fixed would give 2.7k - 12.7k which is close to the 2.7k - 15k I have been shooting for blindly. Or just pot for 0-20k ohm.

 

[bandit2013]

Yeah, the 24V is used for the GEQ. That circuit is a polar opposite to the GEQs used in the earlier models. It is also the reason for the midrange spike that is difficult to remedy. The gain or distortion is not symmetrical by any means. Just seeing the signal characteristic on the Send circuit reveals some of the issue when compared to the JP2C it is very similar to the measurements I got from the Mark III while I was in college. I borrowed a few toys from work to see what the send levels were on the V as a means to determine its line level characteristics. So the first image is the preamp output under gain. The colors are hard to determine in the photo. The green waveform (marked by the arrows) is the output of the preamp. The sinusoidal waveform was the input signal at 750Hz AC. The positive transision of the compressed signal has a sharp leading edge that is above the clipped voltage. It normalizes to its limit just before transitioning negative. The negative waveform forms a "w" shape which indicates some harmonic influences. I am not too concerned with the negative region but more so with the positive region. That sharp edge is indicating some major harmonic frequencies and the way the amp sounds, would not doubt it is odd and not even harmonics. Without a frequency analyzer, it is only a guess. The Tera Echo was used as a buffer for the FX loop and a point to measure the signal. The Lehel P-split was used to isolate the frequency generator ground from the input of the amp. This is the full picture.

This is a zoom in on the oscilloscope.

With the same setup on the JP2C, same input frequency of 750Hz and no change in signal level from the function generator, the distorted waveform is more uniform and symmetrical. That explains why it sounds pleasing and not ice pick. The leading edge still has a sharp rise but does not over extend pas the clipped plateau.

I did not take a full image shot of the gain test on the JP2C, I did when I was looking at the clean channel as that had the highest signal level on the send. I was not trying to dig into the distortion or its symmetry, only to see what the signal levels were relative to the line level in dBu. Was curious why the Mark V sucks with most effects in the loop but my other amps do not. It is not the signal level but the actual impedance of the output and return's input impedance that results in poor performance of that amp in question.

 

[bandit2013]

Talking this through helps (I usually talk to myself or dogs lol) but you've given me a lot of good food for thought. What is the +24V rail for on the V90? Is it the EQ since I don't see it being pulled from the -V like old Marks?

It's a IIB coli I'm 1st considering this bias mod on. All the Marks up to the mid 80's were susceptible to over-voltage at the screens causing the 30uF/500V screen cap to bulge too often. I replaced it with the pair of 47uF filters (each with 150K ohm 1W) as per MB and in doing so had to move the bias caps and R out of the way. It seemed the time to take a stab at a bias mod.

Seeing Charlie's coli's bias mods on the DR board gives me a little pause (all IIB's, same SP-9A preamp boards) but the -52V (or close to it) is established on the PS board.

And this would likely be easier to understand on any Mark but a coli due to the dual tube phase inverter that I can't find any literature regarding. I see resistors that tell me it should be a long tail with complications but one tube on the DR board has heaters not even connected and there's a socket for a top hat looking device that may be a current limiting diode on the ground plane. Still it's a good challenge and I understand the 2nd tube is for buffering but if heaters aren't warming the tube what is really going on?

Here is the actual bias circuit of my IIB on top. R2 is the target. Do you know why C2 and R3 were added to the bias circuit over time? The lower schematic of my '76 used just one bias cap with no R3. I didn't think ripple mattered as much for the grid but I see the V is still using the dual bias caps.

View attachment 1393

I'm pretty far from being an expert but I would assume you'd want the idle current to be as stable as possible just like you wouldn't want your car's idle to be erratic, but that is without a signal.

I was bouncing around the idea of using a 50k pot || fixed CC and that general concept value-wise looks to be in use on the V90. The 1W 20k trimmer will be here this week tho and while I don't really know I'm guessing I want the ability to go higher than 10K tops. A 2.7k in series with a 20k pot || 20k fixed would give 2.7k - 12.7k which is close to the 2.7k - 15k I have been shooting for blindly. Or just pot for 0-20k ohm.

Still not familiar with the amp but the single capacitor and no current limiter resistor in the bottom schematic places the burden of the stability of the bias voltage on the one cap C1.

Splitting up the bias circuit with the 15k ohm resistor adds to feeding the two capacitors so they are not directly in parallel. That resistor is a current inrush limiter on the 2nd capacitor. That 15k resistor prevents the capacitance of 100uF at turn on that may be too excessive for the 60V winding. In other words, it slows the charge time on C2 and most of the inrush will be through 470 ohm resistor into C1. The time constant of the C1 is = R1*C1. The time constant for the C2 cap is (R1+R3)*C2. Once steady state is achieved, both circuits will be similar, just the one with the two caps will have better filtering. I may be a bit rusty on the analog circuits. Been a while since I had to think about it.

 

[user 41014]

Still not familiar with the amp but the single capacitor and no current limiter resistor in the bottom schematic places the burden of the stability of the bias voltage on the one cap C1.

Splitting up the bias circuit with the 15k ohm resistor adds to feeding the two capacitors so they are not directly in parallel. That resistor is a current inrush limiter on the 2nd capacitor. That 15k resistor prevents the capacitance of 100uF at turn on that may be too excessive for the 60V winding. In other words, it slows the charge time on C2 and most of the inrush will be through 470 ohm resistor into C1. The time constant of the C1 is = R1*C1. The time constant for the C2 cap is (R1+R3)*C2. Once steady state is achieved, both circuits will be similar, just the one with the two caps will have better filtering. I may be a bit rusty on the analog circuits. Been a while since I had to think about it.

Thanks for explaining the extra bias cap/R as a current limiter. I can relate that loosely to a big can in an AC unit or 1kW ballast that is there for start up reasons.

The interesting part of that is at steady state there shouldn't be current flow to the grids, just establishing the negative suppression to the cathode-plate current with grid leaks for the stray electrons picked up. The PT must supply a surge at start-up or it wouldn't have been added. And there is a 1W CC for the bias R suggesting possible surge too.

Makes me wonder if I should add that C2 and 15k R3 on my '76 and other amps that don't have it. Most of my IIB's are coli's with massive Schumachers so I should check my non-coli IIB's to see if the extra C2 was just a coli thing. For that matter I should look at the III's I have too, most of which are non-coli.

I'm going to go ahead with the bias mod. I was playing with the 1W 20k trimmer and am strongly leaning to inserting the pot as the stock bias R2 is positioned with a 62K ohm CC in parallel. I'll pull the 15K R3 from the bias node to attach to the wiper (if it is still called that on trimmers). I have some 2W 62k's but I ordered some 1W's that are reportedly testing about 67-69k. With trimmers in the 18.4k to just under 21k range I'll have 0-15k bias range and if there's ever an issue with the trimmer which I'm doubting the 62-70k resistance will just clamp the output effectively "switching" the output tubes off.

 

[bandit2013]

Your bias circuit may be an issue. You should have a series resistor with the 20k pot. Worse thing you can have is a bias voltage that is less than the desired -voltage. If it goes higher in magnitude no big deal, it could get up to -62V (depends on the actual output voltage on the transformer).

Going smaller in magnitude say below -50V may be red plate region. It does not take much, a few millivolts change on the bias voltage will result in a larger change in plate current. Poof goes the Magic Dragon. Experiment on expendable power tubes first.

Just for reference, I looked up a schematic for the IIB. Was not hard to find. Not a Coli though. Sort of curious where the -56 voltage goes (point F, could not find it on the schematic except here). I circled the basic bias voltage circuit. Dates to 5/29/1981

If I were to add a variable resistance, it would be on the 10K circuit. As I said before, a few millivolts will have a larger plate current change than you realize. Just trying to help. If I am making matters worse, hopefully someone else will chime in and offer their expertise.

 

[user 41014]

Your bias circuit may be an issue. You should have a series resistor with the 20k pot. Worse thing you can have is a bias voltage that is less than the desired -voltage. If it goes higher in magnitude no big deal, it could get up to -62V (depends on the actual output voltage on the transformer).

Going smaller in magnitude say below -50V may be red plate region. It does not take much, a few millivolts change on the bias voltage will result in a larger change in plate current. Poof goes the Magic Dragon. Experiment on expendable power tubes first.

Just for reference, I looked up a schematic for the IIB. Was not hard to find. Not a Coli though. Sort of curious where the -56 voltage goes (point F, could not find it on the schematic except here). I circled the basic bias voltage circuit. Dates to 5/29/1981

If I were to add a variable resistance, it would be on the 10K circuit. As I said before, a few millivolts will have a larger plate current change than you realize. Just trying to help. If I am making matters worse, hopefully someone else will chime in and offer their expertise.



View attachment 1446

My understanding of bias R value is not deepy rooted but is that the higher ohms = "higher" valued negative bias V = less plate current. No?

My thinking is if I used a series R --> trimmer and the trimmer went south open there would be no path to ground at the bias point which I can't think would be good and if it failed closed would be such a small value resistance that again the power tubes would get flooded with electrons and as you say poof!

We agree on the higher (more negative) Vgrid is the safer place to be in a disaster and your experience appears more deep than mine so I better circle the wagons before attacking. Still it's my understanding atm that higher bias ohms provides more negative Vgrid.

The date on that schematic was originally the IIA, crossed off with a newer date scratched over it so I wouldn't place too much faith in that schematic. The 10k || 6.8k bias resistance shown in real life is a 1W 5.6k ohm single piece in my case vs the 4k ohm that 10k || 6.8k would give you.

The negative -56V at point F should be the take off for the EQ which is why I was asking about your +24 on the V90- something new to me. Below is from a IIC schematic. They don't have the coli PT represented for bias -V but you can see they have 2 factory select bias resistors in parallel. I could check my C but idr seeing 2 bias R's on any of my amps.

 

[user 41014]

Was curious why the Mark V sucks with most effects in the loop but my other amps do not. It is not the signal level but the actual impedance of the output and return's input impedance that results in poor performance of that amp in question.

Is it a 12AX7 driving this? You would think if this was a problem a cathode follower would be ideal but it's not like you're going to add a stage.

If it is AX7 have you tried a 12AT7 for more headroom and higher input Z / lower output Z? Would that higher current and headroom correct that green wave ice pick?

I see you like your RA's. I really like the Atlantic series and the other non-mesa mesa's. It doesn't seem like they ever caught on. Maybe it was the rep of the Stiletto ice pick that kept petaluma purists away. That's good for RA/ED prices unless the crowd decides later they overlooked some gems that are not plentiful.

 

[bandit2013]

The reason for the +24V dc is for the GEQ circuitry. It is different when compared to the previous Mark amps before it that had the GEQ following the FX loop. Those run a different topography that require a negative voltage as it uses PNP transistors vs the NPN of the Mark V GEQ circuit. Since the signal always passes through the front end of the GEQ to the back end of the GEQ (active or in bypass mode) the signal for the send level is derived from the GEQ circuits. Not a preamp tube buffered send level. Things would have been different if Mesa used an NPN voltage follower circuit to derive the FX send level. It is a cathode driven signal using a PNP transistor as the final output stage of the GEQ circuit. Output impedance is too high as many effects manufactures discovered in attempt to find a work around solution to use their gear with this amp. I do not recall the references I had found when I was trying to figure out how to make use of the FX loop. My only solution was to toss out what I had and go with Strymon stuff. Through trial and error, that stuff worked. All of the other gear did not.

If it was as simple as changing a preamp tube, that would have been done. Bummer is the return portion of the FX loop is tube buffered, not the send part.

I believe they went back to the roots of the original GEQ when they designed the Mark VII as that appears to follow the FX loop.

Sorry for the TMI here. I had a full pot of coffee this morning and well. I slipped into amp mode.

Yep, The RA is the pinnacle of Mesa's engineering marvels. What they packed into that amp was what made me interested in it. Then after owning one and becoming accustomed to it, I found it hard to go back to the Mark amps. Sure, the Triple Crown was supposed to be the next generation of Royal Atlantic. Not. It may sound similar but they missed the boat as they catered to the Metal crowd wanting to djent. Just a few preamp tube swaps in the RA and it surpasses the TC in all respects. Also it is more dynamic and usable for many other styles of playing. The only way to appreciate the RA is to compare it to other amps.

ElectraDyne is more so related to the Stiletto in all respects. However, there is common ground on the clean channel that is found in the RA and TC. All three amps use op-amp compensation following the tube preamp circuits. The saving grace for the RA and TC is the use of a cathode follower tube buffered send level using a 12AT7. Same tube is used for the return. The ED does not use a tube buffered FX, it is all op-amps. Now for the Hi/Lo gain circuit of the ED, may have the same names used on the panel but they are totally different designs compared to the RA. ED is a Marshall 2203 circuit Lo gain uses a single triode before the cold clipper then followed by the DC coupled cathode follower tone stack driver. Similar to the Stiletto but does not have a change in cold clipper cathode resistance. It is the basic 10K resistor common with Marshall. Hi gain channel introduces another gain stage in front of the cold clipper circuit. Then top it all off with Simul-Class power. Tempted to get one but undecided. The RA hi/lo gain structure is more or less based on the Marshall 2555 Jubilee but I do not believe it uses any LEDs to sub as the cold clipper. Tone stack is plate driven so no cathode followers except for the FX circuit. The difference between the RA and TC is the TC has borrowed some design from the ED in use of the op-amp compensation to make it much tighter and to simulate the presence function. At least the RA retains its non-linear gain structure using all tube circuits. No op-amps to flatten the tone so the Hi/Lo gain circuit is fully tunable with different preamp tubes.

In terms of tonal qualities. Both the RA and TC are dead ringers for the Silver Jubliee. I have one and have enjoyed using that with either amp.

Not sure what was up with the Stiletto. I had one for a week and could not stand it. Not shrill or ice pick just lacking any character. It was a Stage 1 and not a Stage II which would have been an improvement. That is the reason why I have to Royal Atlantic RA100s. The combo was purchased with the Deuce as trade in since I had a time limit to return it to GC (bought it online). I was lucky they had the RA combo available. I did call them up first as I have seen that amp in the store a few months earlier. (original owner had 6L6 tubes installed, that will keep any prospective buyers away from the RA).

Then along came the Badlander. BAD for what it is. BAD to the bone? Similar to the Rectifier series but more closely related to the ED in some respects. No more low end drone to swamp your tone. That was the only thing keeping me from adapting to that amp as a main rig (MWDR and Roadster).

The only fix it to push it to the point high SPL that would require ear plugs. Decent amps but did not feel like entertaining the neighbors all the time. I do get them out once in a while. My new fix is the Badlander. It is very much on par with the Royal Atlantic in terms of dynamics. Now I have two of them. At the moment, I am running the two BAD100 in stereo and have the RA100 as a filler. It is one of the only amps I can use at the same volume level as the 2 BADS and retain the overall tone and character of the gain structure. Simple as they may be, another gem made by Mesa that does not seem to get much following.

 

[bandit2013]

View attachment 1447

It is odd, that when you look at the various schematics available over the model years they have a different approach used for the bias circuit. Sure the amp you are working on has the minimal components. It worked to a point until awareness of the design flaws came about in long term use.

It makes me wonder if the Simul-Class version used an integrated quad like the Mark III did. (6L6/6CA7). Depends on the class A circuit. I can see the -67 for a full quad of 6L6 tubes.

The basic key in all of the circuits I have seen for the bias is the R98//R99 pair. The higher resistance component is just the bias tune resistor. That is what should have the pot. The other being a smaller resistance value would be at a worse case scenario resistance for a safe operation and not instant red plate death of power tubes. In either case if R98 or R99 opens the end result is fail safe -Vgrid voltage. It is not common for a resistor to fail as a short circuit. If it does, it will be momentary and usually caused by overvoltage and arcing across the breech in the resistive element. Ever seen a blown out carbon comp screen resistor? Nasty. Usually caused by thermal overload and when the component cools down it gets micro-fissures in the element. When it heats up again, it becomes an open circuit at a microscopic point of view, leading to arcs that will eventually get to higher temperatures and finally blowout with a puff of smoke and bad smell of burnt phenolic material that hangs around for days. The parallel resistance design is fail safe such that if both become open circuits, the bias voltage will reach a higher negative voltage. Fail without grace is when a lower negative voltage occurs resulting in higher plate currents. That results in more damage such as blown screen resistors, melted wires and other mess if the fuse fails to react quick enough. I personally would avoid the use of any carbon comp resistors as they can momentarily create a bad condition if any micro-fissures occur to do cool down. Once the arc begins, it is game over as the arcing will bypass the resistive element and appear like a short. I have seen this happen in my career. It does not make sense but in reality it does happen. Not often but when it does it is not a good thing. No, this was not with any amps or audio equipment but with driving circuits used to control relays in high power circuits. It was not of my design, it pre-existed my employment at the company I worked at prior to my current job. It was interesting to see the root cause analysis and the x-ray images of the failed components. Considering the comments I recall, it is not possible, no way this would happen in that circuit as the voltage and current is not high enough to cause this, said the chief engineer who designed it. It was a good learning experience for me so I always think about that in my designs. Thermals, and such. Also, avoidance of carbon comp axial leaded parts. Flame proof metal oxide resistors for better stability.

Here is some basic bias stuff. If you scroll down to the bottom where it says "better bias system" that sort of describes a fail safe bias circuit. The author does not use "Fail Safe" but it is a good term to use when making changes. What would happen if the mod made craps out? I have done enough FMEAs for UL certification it is more or less stuck in my head. Not only writing the **** report, having to prove it. FMEA is Failure Mode and Effect Analysis. It is where you take a live circuit, and cause a failure of a component, or multiple components to prove the design will endure this condition and remain safe if active or fail safe (shut down in a safe condition). I am an Electrical Engineer at a company that produces products for combustion, ventilation, and other stuff I never expected to develop and design. I am not claiming to be an expert on tube amps. I am still learning what I can when I find free time. There may be other places to search for conversion of fixt bias to adjustable bias. At least the Coliseum is all Class A/B so no need to worry about the class A portion of the Simul-Class.

https://el34world.com/charts/Biascircuits.htm

 

[user 41014]

The parallel resistance design is fail safe such that if both become open circuits, the bias voltage will reach a higher negative voltage. Fail without grace is when a lower negative voltage occurs resulting in higher plate currents. That results in more damage such as blown screen resistors, melted wires and other mess if the fuse fails to react quick enough. I personally would avoid the use of any carbon comp resistors as they can momentarily create a bad condition if any micro-fissures occur to do cool down.

To a self educating amp nerd there is no TMI lol.

I was going to edit my post after thinking about an open fail of all bias resistance not being a bad thing. It would force a high -V to the grid. If the 5.6k CC that's in there now is setting the right bias for a particular set of tubes or average of the colors (let's say 28-32ma?) then the damage would be a condition where bias R is lower to some degree with said tubes.

I can't think of any failure scenario on the parallel design of 20k variable || 62k fixed that leads to low bias R. And the turn screw isn't where anyone could fiddle. My plan is to set 5.6k on the parallel trimmer output and if anything it will drift up at soldering. Then I can test the -V grid and the miliamps at the plate to get a base line and proportionally to each other.

I'm a handful of classes short of the EE here in Denver tho I stopped working on it when the lab computers started taking a half hour to log into. My main interest was EM so I took all the drive classes there were along with the other main EM ones. It was one of the most helpful teachers and you have to take advantage of the good ones. But there are no vacuum tube amplification classes unless you turn a lab into one.

I have a 27" ED combo that hollywood had for a price I couldn't refuse and it shipped from petaluma direct. I shouldn't have sold off the heads but I was trying to reduce the wall of mesa. I'd say get one. My extra RA went too in the reduction.

My duece is stage II. I want a trident naturally given my coli obsession. It's never ice picked on me and I assume that's due to the 412 since people with ice picks seemed to use small cabs. The thing is an absolute monster given the headroom of a full stack.

For the rectos I've always wanted a dual, it's a G thing which I've had a couple triples of, and I've had the single and the roadster. I loved them all but they had to make way for others. I need another wall. Only the Maverick and Blue Angel remain.

I don't know if/when I'll be able to wean myself from CC's. Despite all the bad they do handle E surges well. I've seen them cooked, crumbled and split but unless they are in too humid of an environment there are usually circuit reasons for failing. The PS supply according to Mike should be upgraded on all the old Marks with 3 x 30uf/500V filters as it can over-voltage the screens where many CC failures occur.

 

[bandit2013]

I agree, when I was in college, they skipped the most important aspects of learning electronics as this is where most if not all of the theory evolved from. "You will not need to learn that, you will never come across it in your lifetime" was their excuse. Ever hear of small signal analysis, Miller Capacitance effect? etc.... this all was derived from the triode circuitry. Not from solid state stuff. It was never brought forward to where this theory came from. That was until I looked at a book my dad had, it was on high frequency design, oscillators, signal analysis, etc. I never really looked at it in detail. However, it was primarily based on vacuum tubes. There were some sections introducing the bipolar transistor and MOSFETS but the focus was all on the Triode.

If the approach does not work out well, you can always make a change. That is part of the learning experience. Correcting the bias on the Mark V90 is one of my tasks. I just do not have the proper tools to make use of an adjustable bias (do not have a bias probe that plugs into the tube socket). I do have a few amps that have an adjustable bias, I have even corrected said bias but they also have ports I can use a multi-meter with to confirm adjustments. Almost need two as a change in one side lends a drift on the other side. (dual bias design or balanced method). Still have to get a bias probe for use with two tubes. That way I can view the plate current on both sides. May even help with the extended class A sockets.

 

[user 41014]

The in between bias socket probes/meters are great. Mine is a bit more labor since one of my test sockets went south (meter circuitry I'm sure). Even one can work for the basics but I need a 4 socket unit and would like to have 6 socket ability all at once. Then all I have to spot probe is the grid V.

Thanks for the link. I see the series bias resistor is there for dummy-proofing not circuit failure. And now I know the 470ohm is called the "bias range resistor"! I can always add a 2.7k series resistor that shifts the bias sweep up by that value. The extra lead lengths might come in handy.

Dummy proof non-floating bias mod:

In reality the 15k R3 will be fed from the trimmer out, not the top 20k || 62k || C1 node.

 

[bandit2013]

If you wired the 20kpot as a rheostat, you would have a -56V to -49V bias range which would be good (assuming the open voltage for the transformer is 60VAC. For some reason, my brain is not letting me see the other method as I cannot seem to split the pot into two resistors based on your last comment. However, ignoring the nodes and or loops, the parallel equivalent is 15k, so if the wiper is at the top, you would have the -56V. When the wiper is at the bottom, the voltage divider changes dramatically and the end result would be -8.6V since it would be connected directly to the 2.7k ohm resistor. That may be a bit much for Class A/B 6L6 tubes. Are you planning on running KT88? Typical bias voltage would be somewhere down to -15VDC but that would be for Class A operation. Class A/B is -45VDC. Depends on plate voltage and such. The actual bias voltage does not matter as that is not the critical thing here, it would be plate current based on the plate voltage of the tube that is critical. That would be ok. Not sure the ease of adjustment if it is a single turn pot. Wide range for only 260 degrees of rotation (typical pot range of movement). The finer the change in resistance with a multi-turn means fine tuning of the plate current more easier to set. Have to wait for the drift and settling and readjust if off. I notice that with the Marshall Silver Jubilee. It took a while to get both sides balanced in the right plate current range. The Carvin V3MC had a crappy trim pot, just the slightest movement on total current would swing dramatically. It was a dumb *** method for setting bias current as that included the full power supply and not just the EL84 power tubes. Current measured from the standby switch and wire to the PCB.

 

[user 41014]

Here is a better visual

Depending on actual R values this circuit would give 2.7k to roughly 17-18k bias resistance. I don't know what the PT tap is supplying or if -52V grid was actual.

No KT88's - not sure how that would sound. I have a set of mesa label 5881's I was going to use for this amp as they are in the very small camp that can take coli plate V's (or the 105 tranny but not all of them are ~500+V).

This mod allows for using the tubes the amp was designed around, the 387 and the next best tube, the winged =C=. STR 415's are insane priced. We all have read on here that they last forever and fly by night tube sellers have picked up on that false assertion. The 454 resales are just as bad.

So while there's a lot of junk and overpriced tubes abound there are 387's that will not work with mesa's color rainbow and therefore are reasonably priced cause no one cares if they work in a peavey.

I have some super 6's that are way too hot for fixed mesa's. Not sure I have a matching set but this opens the door for finding and sorting 387's that don't break the bank for a circuit originally voiced for them.

The single turn trimmers are multi - turn, not just <360. My testing found them to be very granular.

 

[bandit2013]

The bias winding is usually 60VAC.

I would personally connect a jumper from top of R2 to R3 turning the trim pot into a rehostat. This would give you a usable range of bias voltage. -59VDC down to -49VDC which is suitable for a variety of power tubes in the Class A/B.

The way you have the diagram, You would still get the -59VDC but at the other extreme you will end up with -8.6VDC. No need to have that range of bias adjustment. Also would be more difficult to get the bias where you want it. Mostly because it is small increments in bias voltage that go a long way in plate current. It can be less than 1mV change results in 10mA plate current. That is why I suggested a multi-turn pot as it does not take much of a change in voltage to result in a higher change in plate current.

 

[user 41014]

The bias winding is usually 60VAC.

I would personally connect a jumper from top of R2 to R3 turning the trim pot into a rehostat. This would give you a usable range of bias voltage. -59VDC down to -49VDC which is suitable for a variety of power tubes in the Class A/B.

The way you have the diagram, You would still get the -59VDC but at the other extreme you will end up with -8.6VDC. No need to have that range of bias adjustment. Also would be more difficult to get the bias where you want it. Mostly because it is small increments in bias voltage that go a long way in plate current. It can be less than 1mV change results in 10mA plate current. That is why I suggested a multi-turn pot as it does not take much of a change in voltage to result in a higher change in plate current.

Yep- 60.3 VAC tap. Thank you for pushing me to test this. I only had one leg of C2 lifted to make room for the screens feed e-cap V upgrade so I jumpered it back to test the existing 5.6k ohm bias V.

I'm reading 60.2 VAC coming into the diode and -61.2 VDC coming out. Maybe some reactance effecting the reading? Or maybe just the way the rms average calculates in the meter?

The 470ohm drops it to -56.8 VDC. For ease of testing I probed downstream of the 15k R3 on top instead of the blue wire take off. It measured -56.7 VDC. I thought it would be a bigger drop.

I then jumpered a 5.6k across the existing 5.6k ohm to simulate the lowest voltage magnitude of the circuit I plan to implement. It was settling out between 2.95-3k ohm.

This resulted in -47.4V bias voltage that R4 locks in as the least negative V point. It loaded down the circuit as downstream of the 470 ohm it was now -47.5V. For a second test that magnitude drifted up .1V but the relationship stayed the same.

I could pull a leg of the existing 5.6k to add in series to a 11k ohm I have handy and see what the high voltage is but the upper boundary is set by the PT anyway.

Being able to take the negative voltage "lower" is great for fine tuning but wasn't a goal here. If you need to coax your tubes there's a point when they should be tossed. Idk exactly where that is but testing some old GE/other tubes a couple coming in under 10mA started to make the other tube of its pair heat up and clearly was going to red plate.

This bias mod is meant to add more grid suppression as needed to use hot tubes. In reality iirc the mesa tube rainbow goes from the teens to just over 40mA. Obviously there's a point in there that a specific amp is biased for, not the whole range. The mod allows for fine tuning the drivetrain for the gearing that best suits the terrain.

There is one problem with this bias mod on old marks tho and that is the fact this negative DC feed is also used for the EQ (e.g. -38VDC) which is dropped from the "-52VDC". Many of mine don't have that circuit to feed and doing the mod on the PS board is the beefy option.

My Charlie Daniel coli's, also non-EQ and SP-9A as the amp I'm working on, have bias pots installed on the DR board seemingly at the output of the PI. I'll take pics and post them. I see no way of adding a bias mod on the PS board for power tube grids and keep consistent EQ voltage short of using a 2nd trimmer to compensate. That doesn't seem like the path of least resistance.