• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Tubelab SE Coupling Capacitors

I am collecting my parts to build a Tubelab SE amp with 300B tubes.

The schematic shows the coupling capacitors to be 0.22uF, in the parts list it is a 0.47uF auricap.

How critical is this part value? What effect does a change in it value have.

Anyone try any russian teflon caps in their Tubelab SE?


The value of the coupling cap is not critical. If you go below 0.1 uF you may lose some bass response. Values above 1.0 uF will cause the bias adjustments to react slowly. 0.22 uF to 0.47 uF are ideal and I have used both depending on availability. The voltage rating should be 450 volts or higher.

I have not tried any Teflon caps, but some builders report good results.
The value of the coupling cap is not critical. If you go below 0.1 uF you may lose some bass response. Values above 1.0 uF will cause the bias adjustments to react slowly. 0.22 uF to 0.47 uF are ideal and I have used both depending on availability.

Hello tubelab and All,
I am a boiler and piping kind of mechanical engineer, I do not understand everything I know about this electrical stuff.
Is it a bad thing for the bias to drift slowly? I thought the bias was a constant kind of thing except when there is no cathode resistor bypass capacitor (local cathode feedback).
All just for Fun!
Is it a bad thing for the bias to drift slowly?

It depends. It is normal for the bias in the Tubelab SE to change slightly with line voltage. The change can be abrupt or slowly over an hour or two. If the total tube current doesn't change by more than 5 mA or so, and doesn't need to be readjusted often then things are OK.

If your tube current slowly increases with heat and doesn't settle down, there could be a problem. The usual cause is either a gassy output tube or a leaky coupling cap. It is possible for resistors to drift in value but this should not be an issue with modern components.

As mentioned old capacitors (even teflon) can develop leakage currents. The leakage usually increases with time and temperature. This will cause the bias to drift, and can cause eventual tube runaway.

A tube with some residual air inside (gassy) can cause a similar situation. The air molecules will ionize causing stray currents to flow inside the tube. The problem arises when current flows from the grid to the plate causing the grid voltage to shift in a positive direction. This is why the tube manufacturers specify a maximum grid circuit resistance. The Tubelab SE has a very low grid circuit resistance so mildly gassy tubes should not runaway, but will cause distortion.
If you are talking about 300B tubes my experience is rather limited. My budget doesn't even think about looking at the pricey stuff. I got some early Sovteks nearly 10 years ago for my P-P amp. They all died within a few months after the warantee expired. New Sensor said tough s***. I decided to go back to $5 or less TV tubes and the 300Beast went in the closet.

After a couple of years I got 4 lightly used Sovtek 300B's from ESRC for $25 each. That was about 6 or 7 years ago and they are still in the amp.

When Shuguang started making 300B's they were cheap. I got a few for $28 each. One pair went into my Tubelab SE and they are still in there. That is the extent of my 300B collection.

My amp has been to visit a few nice stereo systems, and in doing so it has been tested with some high dollar tubes including WE's. Maybe they sounded better than the cheap stuff, or maybe we just wanted them to sound better. I don't know, but I am not spending hundreds of dollars for expendables.

A few years ago I was digging through the junk at an outdoor flea market in the middle of nowhere when I found an old Sparton radio chassis. It had a tag stating that all of the tubes were replaced in 1929. I got it for $20. The output tubes were National Union NX-483's. I plugged them into my amp and they have been my favorite tubes ever since. No one knows what they are and replacements aren't going to be found, but they still work after 80+ years.

On the other hand I bought a sealed box of 100 Sylvania 6V6GTA's at a military surplus auction. Over half were gassy and many of the ones that worked have since failed. These were so bad that they lit up like blue neon lights.

Sometimes it is all about the care and cleanliness during manufacture. Clearly the batch of Sylvanias that I got had some kind of contamination causing a slow leak, since they are still failing even without use. Other Sylvanias are regarded as good tubes.
Let me ask a little different question. Will the value of the coupling capacitor have any effect the onset of blocking distortion and overload?

In the Tubelab SE, no. The mosfet source follower eliminates blocking distortion. Overload is possible in any amp, just drive it into clipping. In the Tubelab SE overload RECOVERY is nearly instantaneous. The limiting factor is the Miller capacitance of the output tube and the 20K grid resistor.

In any amplifier where the coupling capacitor is connected directly to the grid of the output tube (including the Simple SE), blocking distortion and overload recovery are valid concerns. The value of the coupling cap, the grid stopper, and the grid return resistor all affect this.

Under low signal conditions the coupling cap is charged to an "equilibrium" value (driver plate voltage on one end, output tube grid voltage on the other end) which doesn't change when signal is applied. The output tube's grid doesn't draw any current so the cap is working into the grid return resistor (100K+). The time constant of the cap (.1uF+) and this resistor is much longer that the period of the audio signal.

When the output tube gets slammed with a large enough signal to drive its grid positive its grid will draw current and become a low impedance. It can be a few hundred ohms on a DHT like the 300B. Now we have the driver tube trying to yank one end of the cap in the positive direction, but the other end is clamped by the output tubes grid. Now the time constant is different too. The grid impedanc is no longer 100K+ it is now a few hundred ohms, so the capacitor's charge equilibrium begins to change. The equilibrium is lost relatively quickly.

Now the overload goes away. The capacitor needs to recover its equilibrium but the original time constant has returned since the grid is no longer drawing current. In fact under extreme conditions the capacitors charge can be changed enough to cause the output tube to be cutoff.

If the amp is going to see clipping regularly (a guitar amp) the component values need to be carefully optimized to avoid bias shift and total cutoff.

How do the component values interplay here?

The coupling cap it the keeper of the charge equilibrium. A big fat cap will be harder to upset, but recovery will be long. Sometimes several seconds will be required for complete recovery from a severe overload. Too small of a cap will limit bass response. A guitar amp may use .05 to .1 uF since the lowest note on a guitar is 80Hz, while a HiFi amp may use .1 to .22 uF. Big caps (1uF) should be avoided especially if GNFB is used.

The grid return resistor value sets the time constant under normal conditions. The upper limit is stated in the tube manual for a given output tube and the lower limit is determined by what the driver can handle. 100K to 470K are common.

The grid stopper is normally used to prevent oscillation. It has a second function here. The grid stopper resistor determines the time constant under overload conditions since it is in series with the grid. It also forms a low pass filter pole with the tubes Miller capacitance so it affects the high frequency response of the amp. A big resistor (10K) will eliminate blocking distortion in most cases, but it will kill your treble response in triode mode. It works in pentode mode though.

So what do I do? I connect a scope to the output tubes grid, connect my guitar preamp up to the amps input, dial up a Marshall stack cranked to 11 preset and cut loose. I try to find component values that allow some blocking distortion in this manner, but try to avoid total cutoff or "farting out" distortion. The amp should not see blocking in normal HiFi use. If you do want to use a Simple SE for a guitar amp I recommend smaller caps, pentode mode and even wiring the Simple SE for mono P-P operation.....but that is a different thread.

How do the component values interplay here?

The coupling cap it the keeper of the charge equilibrium.

Hello George,
I see said the blind man.
That is an in depth explanation. Thank you.
I have copied your post into a word document and printer it out in 14 pitch Arial type. I will reread it with a pencil and yellow highlighter.
Next to digest is PowerDrive.
All just for Fun!
Next to digest is PowerDrive.

PowerDrive simplified:

Part 1:

The CCS provides the driver triode with the ideal load. This lowers its distortion. Use a resistor load if the driver is a pentode.

Part 2:

The mosfet seperates the output tube from that troublesome coupling cap. The mosfet has a very high input impedance. The impedance is mostly the resistor connecting to the bias pot. Adjust the resistor value to make the driver happy. The driver and the coupling cap now see a constant impedance regardless of signal level. They are happy.

The mosfet feeds the grid of the tube with a very low impedance drive voltage. It can source the grid of the output tube with all of the current it can eat if driven hard, and it can deal with Mr. Miller in triode mode.
Hello tubelab and All,
I sat down with your printed posts and PowerDrive from your web site. I made some observations that were interesting to me. TubelabSE has grid bias and Simple SE has cathode bias.
A SET with cathode bias can also have its’ own version of bias drift. If the output goes into cutout the cathode voltage falls towards earth. I can imagine non-musical shoot and overshoot during the readjustment of the equilibrium of the interstage coupling and cathode bias capacitors. None of this shows up on theoretical load lines or in the ideal world of spice.
Thanks for tickling my brain George.
All just for Fun!
A SET with cathode bias can also have its’ own version of bias drift.

So can a cathode biased P-P amp. The cathode bypass cap can be made excessively large which will effectively remove bias shift due to cathode voltage change though.

I have seen small changes in the cathode voltage in the Simple P-P when driven to extreme overload.

Another "bias" issue is unregulated screen grid voltages in pentode amps. The screen current can vary a bunch in some tubes as the drive is increased. If the usual resistor - capacitor screen supply is used, another time constant is created along with the overload and recovery issues.

Any and all of these describe possible modes for overload and recovery issues. More than one of these can be at work in any given amp.