BMK2 used in cathode-biased amp

[K O'Connor]

Hi Guys

A member of the forum brought up a question about adjusting idle in a cathode-biased amp and bought the BMK2 Bias Mod Kit for 2-tubes from us. The BMKs are intended for use with fixed-biased amps and the notes reflect that application. However, you can use BMK2 to hum-balance a cathode-biased amp. Each pot supplied will control one side of the push-pull circuit.

Where is the bias voltage coming from? It is the voltage across Rk. Each tube sees this voltage and adjusts its own current in a bit of a tug and war with the opposite tube(s). The two sides are unlikely to have equal currents, let alone end up at the more important hum-balance point for the OT. Being able to tap this voltage with the pots allows the effective bias voltage to one side to be REDUCED. This will make that side run hotter.

Since cathode-biased amps are notoriously poorly designed, lacking grid-stops and screen-stops, sharing one bias resistor for the entire output stage, and aiming to run each tube at its full dissipation, having a bias adjustment that makes things even hotter is not really a good idea, but it is what we can do. The fix is to increase the value of Rk to a safer operating point that gives leeway for making one side (or both) a little warmer. To do this, we can either add another resistor in series with the stock Rk, or change Rk outright with a higher value. The new net value should be about 20% higher or more. So, a 250R would change to 300R, maybe go to 330R to use a standard value. The 100R in an AC30 would change to 120R minimum. Higher is better as we can make things warmer at will using BMK2.

To limit the adjustment range of the pots, we use the 6k81s in series with the X end of the pot, just as in fixed-bias.

The drawing shows how BMK2 is incorporated into a 2-tube cathode-biased amp. The RED-1, RED-2 and BLK are the meter jacks, along with the 1R current-sense resistors. We can measure 1mV per 1mA through the tube and know how much heat it is dissipating.

If the amp is a type with independent cathode-bias resistors and bypass caps, the BMK2 can still be used with half the kit applied to each tube. You would need a second BLK meter jack at the top of the second Rk,Ck.

[Champ81]

Hi Kevin,

You mentioned MV as well. Would I be correct in saying the R1 goes after it sees R2 for an MV. The current MV setup I have now the wires go from the coupling caps to the first lug then the wires go out the middle lug of MV to the grid stop resistor.

I also wanted to mention I inserted the attachment and did not press "insert into post" instead posted it afte attaching but it seems like it adds the attachment inline.

[K O'Connor]

Hi Guys

The note indicates that R2+4 would be replaced by a dual-pot; the wipers would tie to the grid-stops (R1,3)) In this scheme, pot-X ties to the coupling cap, wiper ties to the grid-stop, por-0 ties to the wiper of the bias pot.

If there is bias-modulated tremolo, things get a little more complicated.

Regarding your attachments: When you cursor into the dashed box to add a link to an image, etc, you will then browse through your files and select one. Maybe you are dragging the file into the box? I do not know what that would do, but it may place the image there instead of the link to the image for upload.

Have fun

[Champ81]

Thank you.

I realized I jotted down the x and o position of the MV switched. It makes sense now. 

One question I have is essentially with the MV in the scheme the grid leaks are variable? . I believe what I have is what they call a "type 2" mv or "post phase". I believe the 220k resistor from the wiper to "o" of the bias adjustment pot is put in to account for this and make it more stable?

For the Rk value I am going to swap it out with a higher than 20percent resistor as you had mentioned. What would be the maximum resistance you would limit at before things would become unstable.

For attachments it could be because I an using my phone to post. I will try again on a PC. Hopefully that should work.

[K O'Connor]

Hi Guys

Champ81: Does your amp have a feedback loop around the power amp? Likely not if it is the typical 18W or 20W with EL-84s.

In any case, we are looking at the cathode-biased output stage in isolation. The note regarding how an MV might be positioned here is for completeness, as many of these amps do have an MV right at the input of the output tubes.

With the stock ground-referenced MV the only caps between the splitter and the output tube grids are the plate coupling caps. Were we to want to retain the ground-referenced MV while adding the bias control system, we have to add a second set of caps to isolate the MV from the bias voltage on the grids which is no longer simply zero volts.

Note that the bias pots are 16mm panel pots BUT they are mounted on the tube plane of the chassis. They are then as accessible and as protected as the tubes. Panel pots are far more reliable than trimpots, the latter which should never be used for bias controls in tube amps.

Yes, safety resistors would be good to add. In the case of the bias pots themselves, the resistor is tied from the pot wiper to the raw bias voltage, which in this case is ground. In a fixed-bias amp, the tie is from wiper to maximum raw bias voltage (which might be pot-0). AND when adding the MV as suggested, a safety resistor would tie to the MV pot between wiper and pot-0.

Safety resistors across bias/MV pots are 10x the pot value.

Increasing Rk will not change the stability of the circuit, it simply reduces idle current. Therefore, there is no limit to how high the Rk value is except for practical considerations.

Suppose the stock Rk=250R. Measure the voltage across it to calculate the idle current. Note this down as the maximum value NOT to exceed.

We might go to 330R to add the bias adjust circuit. You could make this change first simply to read Vk and calculate the current before adding the rest of the circuitry. Say you decide to go higher, to 470R. Again, there will be decreased current. The higher Rk drops more voltage per mA, so Vk may not change a lot, but now there is leeway to adjust the individual and net tube currents back towards stock while hum-balancing the output stage.

As Ik moves towards the original value, the voltage across the new Rk will head higher than the voltage drop across the stock Rk, so you may want to calculate that maximum voltage based on the original calculation of stock idle current. Then make sure the new Rk can dissipate the related power.

Make sure Ck has a voltage rating to accommodate whatever the maximum Rk voltage may be. For example, going from 250R to 500R generates twice the Vk value. Will the stock Ck handle this? If not, change it for a higher-voltage rated cap.

[K O'Connor]

Hi Guys

There is a variation of the scheme presented above that is simpler in two regards, but first I will describe the difference.

The circuit for the most part is like the above schematic. The difference is that the stock Rk + Ck remain as they are and a second R-add is inserted between them and ground. The bias pots are wired in parallel with R-add, oriented as shown with pot-0 to ground and pot-X to the junction of R-add, Rk and Ck.

The first advantage of this version over the first is that the maximum current is still the same as stock. In the first circuit the range resistors in series with the bias pots could allow for higher than stock currents if not chosen correctly based on the pot value.

The second advantage is simply the saving of the range resistors, one R for SE, two Rs for PP, one R per individually biased tube.

Ck could be changed out with a higher-voltage unit per the previous posts, but this scheme allows it to remain stock.

On the general scheme of cathode bias and the standard range of output tubes used in guitar amps, you could view Rk/tube = 500R. So, SE would use Rk=500R, PP would use Rk=250R and four tubes using a shared Rk=125R. The latter is highly NOT recommended despite every AC-30 and clone doing so. These values seem to work reasonably well for the large bottle tubes and EL-84, each adapting to the Rk value based on the tube's own transconductance.

Because the tubes idle at full dissipation, it is extremely important to have proper value screen-stops, as TUTs indicate.

[Champ81]

I finished installing the bias circuit. I did the procedure as indicated in TUT 2. The issue I am having is with the power tubes pulled I am measuring pin 2 of the el84 tubes, with the bias all the way to "x". In TUT it says to measure pin 5 which I believe would be the control grid of an el34 octal tube. 

The voltage from chassis to pin 2 of el84 there is no DC voltage present with bias all the way to x. 


I put in the tubes and I notice the bias range goes from around 42mV and all the way up to 100mV. The other tube there is no change in current when I turn its respective bias pot.  The stock resistor of 150R has not been changed yet but at least it should be working?

I drew up a layout (I apologize for the lack of neatness) going off the schematic.

Edit: I didn't read your previous message carefully regarding the push pull circuit. Essentially it is dialing the currents per tube but pushing and pulling as you mentioned. Previously I had installed sv1 raw bias supply and this seems to work differently. The previous build I can dial in the current without affecting the range of the other tube.


I didn't answer your previous question. The amp doesn't seem to have a feedback loop.
I did manage to balance the hum. The only issue now is the MV is always open now as the grounding reference is at the wiper of the bias pot. I must have misunderstood the wiring and not correctly. But you mentioned that it now has to be referenced and isolated back to zero volts via another set of coupling caps. I'm not sure how to do this visually. Do I just wire in coupling caps from the wiper back to the ground?

[K O'Connor]

Hi Guys

Yes, TUT descriptions incorporating tube pin numbers assume an octal base of the standard pin-out. You can attribute this to my own preference for octal power tubes and a dislike of EL-84s. Never paid much attention to them until I designed an EL-84 version of the STUDIO amp called the KC-25.

Note that cathode-bias is a bias METHOD, that is also known as "self bias". There is no current through Rk unless a tube is plugged in, and therefore no Vk.

NEVER operate less than the full tube complement for shared-Rk output stages.

It looks like your drawing corresponds to my drawing. If one bias pot functions then so too should the other. It may be that for the one pot to show its range the other has to be set for less than maximum current. The tubes already adjusts themselves to a specific maximum Vk (for the specific voltage environment) and once that is achieved there is no more room for adjustment. Try setting both tubes back a bit from maximum, then target the hum-balance.

Note: For hum-balancing an output stage there should be no signal entering the amp. Turn any Volume, Level or MV close to the PA input to zero.

[Champ81]

Thanks Kevin

The bias controls are working as it should I believe. As you mentioned one pot was set too far in the range and so the other had no room. But the adjustment is there. Push pull. Also to note is this circuit seems to work a bit differently than the previous build where I have the SV1.

  The only issue now is the MV is still not referenced and so it is not functioning at the moment.  I have been trying to figure it out as you mentioned you need to put in a new set of coupling caps to account for the voltage now present at the grid.  I can't figure out how to reference it to and where to tie them in. Currently the MV is wide open and the normal volume now acts like an MV.   

In the process of trying to figure this all out and understand this bias circuit essentially this bias system is adjusting the voltage potential between the Rk and the control grid 2 of the el84 am I correct? In that case shouldn't there be a variable voltage when I measure between the Rk and the grid pin 2?

[K O'Connor]

Hi Guys

The DC isolated MV is simple to wire: pot-0 goes to ground, pot-X goes to the splitter cap and pot-wiper goes to a new cap tying into the output stage with its bias controls. Within that circuit, R2,4 remain as fixed values. This is the same as you do for a fixed-bias amp post-PI-MV.

In any tube circuit, you can see the grid voltage if you have one meter lead at the cathode of the tube and the other on the grid. In many circuits, measuring grid voltage from ground is not accurate as the meter resistance changes the biasing. You have to remember that in the bias circuit above for cathode-bias, we are creating a little extra voltage range so there will be adjustment. You should find that taking one tube or the other down from maximum by just a tiny amount will balance the hum. The amount of control voltage available will change as you change the idle current in either tube.

In fixed-bias, the raw bias supply is itself considered to be constant, and it is not effected by the pot settings or the tube currents. That makes the bias pots less interactive than in this situation.

[Champ81]

The last post was confusing so I deleted it and currently I have it hooked up with the investment of a breadboard.
Keeping it simple is the best way for everyone and the current state of the circuit to start. As I think it's somewhat doing what it should.

MV is currently hooked up like this:
- Pot "o" to wiper of bias pot wiper AND grounded from this lug.
- wiper to 8.2k grid resistor
- pot "x" to the coupling cap

Bias pots are wired like this:
- Pot "o" to ground
- wiper to pot "o" of MV
- Pot "x" to Rk via 6.81k resistor

Rk was doubled from 150R to 300R by putting another 150R in series.

This is what I am observing:
Adjusting either bias pot does adjust a  by mere 0.1mV increments but increasing and decreasing. The current draw at test points measure around 25mV and increases by turning to "o" direction and decreases turning to "x" direction by 0.1mV. 

I added an 0.1uf cap inline directly from MV wiper straight to the grid pin 2 in series with the grid resistor.
In this scenario one tube draws in all the current to something too large and so I turn the amp off as it reaches this state. The first tube draws hardly any current. Adjusting does not do anything.

When I remove the cap it goes to the initial state I mentioned at 25mVish. 

Can anyone explain what is going on? I thought I had it right but the adjustment is so small a total range of about 0.8mV from "x" to "o" and it doesn't make sense that the additional of a cap would cause one tube to keep going up to over 100mV.

[K O'Connor]

Hi Guys

First thing: The cap connecting the grid to anywhere other than the signal input source (splitter) is absolutely undesirable.

You might see caps that tie to the bottom of the grid-leak as post-filters of the bias voltage in a fixed-biased supply, often because the raw bias supply is not well filtered to begin with, or simply because in the hifi ideal this point should look like an AC ground for signals. In the case of your observation with the cathode-bias adjust circuit, the cap is likely causing oscillation which make the tube red-plate.

As I suggested earlier, simply changsing the value of Rk to a higher value should show decreased Ik BUT you might always see the same Vk. When you tried the overall circuit with the stock Rk, you indicated that there was control over the currents. Did you achieve hum balance at that point?

What is the supply voltage in this amp? Ik should never exceed a specific value based on the tubes in use, and specifically, based on the tube's plate power rating. For EL84 this is 12W. If Va=300V, then 12W of heat is achieved with just 40mA. Rk determines how hot the tube will get in the given voltage environment, and for any Rk you can calculate a target maximum Vk.

Lots of hobbyists will argue that Rk carries plate plus screen current, which is true, and that you should use the plate dissipation plus the screen dissipation for calculations. Their goal is run the tube at its maximum heat. Mine never is. For the sake of the 2W of screen dissipation, I prefer to leave that out and err on the side of "slightly cooler"especially since EL84 amps tend to have skimpy screen protection. With any tube you can use a Pd value UP TO its max and anywhere below that value.

[Champ81]

Hi Kevin

I measured 392V from chassis to the first power filter cap. As for the cathode voltage from Rk to chassis I measured around 12V. 

With the original Rk I remember achieving hum balance and current varied with the pots at the test points. For a cathode bias adjust circuit what kind of range am I looking at? Definitely not 0.8mV max difference from "o" to "x" am I right? Or is that how it is designed.

Currently I have grounded the MV ground lug pot "o" to ground while that same lug has another wire going to the wiper of the bias pot.  Otherwise the MV is always open and I cannot control it. In my first setup I had no ground to pot "o" but it went only to the wiper. With this setup the bias seemed to work but the MV was now open so I also tied it to ground.
Is this correct?

One thing also to note is currently I am using test leads to connect to the bias pots and other circuitry to make sure it is working as it should.
Also to note is once that tube red plated is it possible that the tube is damaged and so bias adjust doesn't work? I can still play through the amp and it sounds good but could there be a defect now with the tube?

[K O'Connor]

Hi Guys

Cnamp81: Electronics uses precise terms so that circuit descriptions and essays about circuit function will be clear to all readers.

The term "ground" has a precise meaning in electronics, as it is the reference for signal and DC supply voltages. Real ground is also called "earth". There are also AC grounds which is any DC point tied to earth via a capacitor, as all DC supply nodes are.

In your amp, "making a connection", or "tie 'A' to 'B'" os not the same as "grounding" UNLESS one of the related tie points truly is circuit ground.

Your description of how you connected the MV states that the MV pot-0 (pot-zero) is tied to the bias-pot wiper AND to circuit ground. This means the bias pot cannot do anything.

Look at the original PDF and see that the grid-leaks R2,4 can be replaced by the sections of a dual-pot. Pot-0 is NOT tied to ground directly, although the bias pot wiper can make that connection at the extreme of the pot sweep. The bias pot can do what it is supposed to do.

So, remove the link between ground and the MV pot-0 IF you are trying to follow the original PDF.

You can only ground the MV if it is capacitvely isolated from the entire circuit of the PDF.

Regarding the red-plated tube: It may be toast. You can check it in the circuit if you use a higher value Rk just to control it.

Really, the amp needs higher-value screen resistors if there are any, and add some if there are not.

[Champ81]

I see what you are saying. Ground in these terms is a reference point relative to another point in the circuit but not the actual chassis?

The good news is I got the bias working again meaning it changes. With 2X150R in series = 300R cathode resistor the current draw starts at about 22mV.  The current at the test points go higher as I adjust it. I have it wired as in the pdf now. I took the ground wires off the MV pot "o" now.  The only issue now is as I adjust the bias pot up in current the volume increases so basically its as if the reference to ground is shifting. So the physical MV at zero is not zero. 


The tubes in question seem to be functioning now. It's strange. It wasn't before. But I put in new el84s and it wasn't "running" away. So I put in the previous tubes back in to confirm they are toast but they didn't either. Anyways I will draw up a proper layout that is cleaner than the previous hand written one. I just found out Diylc software.

[K O'Connor]

Hi Champ81

You said previously that you have TUT3. Please read the section about grounds. Chassis is tied to earth ground (safety ground) to act as a shield The chassis is not supposed to be used to tie circuit grounds to each other - it is strictly for shielding.

NO terminal on a device is "ground" unless you tie that terminal to the circuit ground.

A typical Volume control, MV or other level control is often connected to circuit ground at the pot-0 end. Without the bias control circuit, your stock cathode-bias output stage has a grid-leak tied to ground for each side of the circuit. These grid-leaks can be replaced by a dual pot with pot-0 tied to ground and wiper to the grid. Pot-X accepts the signal input.

With the bias circuit the grid-leaks are above ground, so if they are replaced by the MV, the pot-0 ends are above ground.

You can make the bottom of the MV (or grid-leaks) AC ground by adding caps across R12 and R13, which is also from each bias pot-wiper to ground AND MV pot-0 to ground. This eliminates the voltage divider effect of the grid-leak and the bias-pot, and therefore the volume change with bias setting.

[K O'Connor]

Hi Guys

I have updated the PDF in post-1 with the AC grounding caps in place. Please have a look.

[Champ81]

Did you mean 47uf? It says 470uf on the schematic.  I looked those up and they are huge can capacitors. I have a few 50uf 50V caps lying around ad well as some 22uf caps as well.  The voltage readings between the caps in circuit maxes at around 3.8V so I guess 50V rated should be fine.

[K O'Connor]

Hi Guys

You need to look at places like Digikey and Mouser for the non-tube-specific parts - which is pretty much everything in an amp other than the tubes and sockets.

470uF-16 or 25V is 10mm diameter and 10-16mm tall as a radial-lead 10khr part. Tiny.

These caps have to be a high value because the portion of the pot resistance they work against can be very low.

For example, say you increase Rk by two times as you did, going from 150R to 300R. If you want to limit the range of adjustment so that the tube currents can never exceed the stock value, then the range resistors R10,11 should equal the pot value. So, with 25k pots R10,11 should be 25k each. Say you had 10k pots, then R11,12 would be 10k each. In both cases, the portion of the pot between the wiper and ground can sweep to zero and may in practical terms be only 1k or so.

At 1k, to have an AC ground with a lower cutoff frequency than the coupling caps of the amp might mean using at least 10uF. That works out to 16Hz, and flat to 160Hz. That seems "good" but we need it to be better. 100uF reduces the frequency roll-off to 1.6Hz. Anything lower is still an improvement and we want this roll-off to be at least a decade away from the lowest roll-off in the amp. 470uF brings it down to 340mHz. We could use 1,000uf (1mF) and cut that in half.

You can measure the pot resistance to ground with the power off and see what time constant you get with 50uF caps. That is an off value in modern manufacturing, except as large cans, you more typically see 47uF then 56uF in the E24 value range. 50 would be part of the older E12 range.

We are using low-voltage caps here.

It is good to make voltage measurements of the stock circuit first. The bias voltage Vk in a 2x EL84 amp is typically <15V. So doubling Rk to potentially have double the Vk at the stock current BUT with a limit of voltage across the pot to be half, means we can use a 16V cap. A 25V cap gives us some leeway.

Remember: Electrolytic cap voltage ratings are "working voltage" meaning you can operate the cap at its rating. Using a 100V cap at 15V for a few years means that the cap "becomes" a 15V cap as it reforms to the voltage applied. If it regularly sees 100V but mostly 15V it will likely retain its 100V forming.

[Champ81]

I must have filtered the results wrong in digkey. I see the smaller ones now. I measured the current going through the 47uf caps I had on hand to test it. There was about 8 volts. I'll get some 16V o 25V 470uf caps.

For the 220k resistors that ties to the wiper and pot "o" and the 6.81k range resistor would you reccomend larger wattage for those? Are they there as fusible or should they be more robust. I'm assuming the 220k is there to be robust in case the pots fail and those resistors are on place for protection to the rest of the circuit.