• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Yet Another Output Tube BIAS Servo Schematic from DIYAudio.ru

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi !

Here is a schematic of output tube BIAS servo published by Hopchik (Oleg from Voronezh), on DIYAudio.ru forum. According to feedback from forum members it works as expected. I didn't tried to assemble this schematic, posting here in hope someone could find it useful.

Link to original post on DIYAudio.ru
C???? ??????????????? ???????? ???????? ????

R19 and R20 connected to the grids of output tubes (CETKA in Russian means Grid).

I send a message to the author yet till now he didn't responded. Copyright notice on DIYAudio.ru states that copying and citation permitted if provides link to forum and nick of the author.

PS. What really bothers me with this and all similar servo schematics that they will modulate BIAS voltage during low-frequency high-level transient, possibly need low-pass RC filter with very high capacitance. Have to do some simulation in LTSpice.

Any feedback appreciated.
 

Attachments

  • auto_bias-diyaudio_ru-Hopchik-1.JPG
    auto_bias-diyaudio_ru-Hopchik-1.JPG
    84.3 KB · Views: 918
  • auto_bias-diyaudio_ru-Hopchik-2.JPG
    auto_bias-diyaudio_ru-Hopchik-2.JPG
    81.1 KB · Views: 973
There are several reasons that cause some to want to use servo bias. Some of those reasons have other methods to accomplish the same thing.

Always think about what reason(s) you have to use such a circuit, versus the complexity, reliability, and how well it will accomplish what you want, before you decide to use it.

The 1970s or 1980s Telarc recording of the 1812 Overture, with real cannons that have major large signal at 6Hz will bring about 99.47% of all tube amps to their knees . . . not to mention your poor ruined loudspeakers.

The woofer cones jumped out of their baskets, when RCA had the monitors turned to high in the recording studio, during RCA's recording sessions of the 1812 Overture in 1959, because they also used real cannons.
 

PRR

Member
Joined 2003
Paid Member
Not just "low frequency".

But it looks like a 6 second time constant.

I have done work with cold-bias class B with extended loud parts which could cut-off this scheme, only slightly before the grid-network cut-off.

Cannons don't BOOM for 6 seconds, and unclipped speech/music is rarely high-power that long.
 
PRR,

Yes, I picked very unusual and demanding music recordings.

I think it is generally true that when difficult recordings are played way too loud for a given amplifier, the bias will shift. That happens whether the output tubes are RC cathode self biased, adjustable fixed bias, RC coupling to the output tube control grids, use a constant current with cap bypassed cathode bias, or some servo bias technique.

If it sounds bad when played too loud, then turn it down, I say.
 

45

Member
Joined 2008
You need something like a clipper in the circuit so that the servo monitors the current in quiescent or pure class A operation but ignores the AC current conduction when the tube doesn't operate in class A anymore. Such a kind of servo bias is in the Morgan Jones book (p. 420 of the 3rd edition) and has also been revisited by Bronskie with some measurements proving it's really doing its job properly. So also search in Tubecad.
 
Sodacose,

As I said in post #2, think about why you would want to use such a circuit. What is your goal?

The circuit you posted has an Integrator. I believe that this Integrator circuit gives you an Average current (a rolling average). i

Is that your goal, Average current that is according to the combination of: 1. Quiescent current during extremely quiet/no sound sections of the music. integrated with 2. Current during the medium and very loud sections of the music.

Just to re-quote DF96, "Like many simple bias servos, this stabilises the wrong current: average current instead of quiescent current. This means that it will be worse than a bypassed resistor. Like a resistor, it will only work for Class A output stages".
 
I guess the design question I'd have would be (kind of directed to DF96):

Is the bias servo supposed to be auto-adjusting or instead referenced to a fixed value?​

For instance, say the goal ideally is "get 25 ma to each output valve". Where “25 ma” is something one (ideally) can "dial in" to the bias servo circuit. It isn't clear how one decouples the current-sense construct from being a long-time-constant averaging topology in the end, while maintaining its idependence from all the signal-imposed variations on the output finals when in operation.

Dunno.
I remain challenged by this problem.

Probably not enough coffee.
Harumph…

GoatGuy
 
I can't see the point of a bias servo, but they are popular so a lot of people think they need one. About the only people who might actually need one are those with toroidal OPTs, as they can saturate easily if unbalanced. However, few people use toroidal OPTs.

Some people want to exactly set output stage bias current. Why? Valves are not exact devices, and their datasheet graphs are not very exact either, so who cares if the bias is 40mA or 50mA? 40mA might be correct for one set of valves; 50mA for the next set. In either case you won't get much difference in power or distortion if you get it wrong, just a slightly different compromise.

Others want 'fixed' voltage bias, but without the effort of adjusting it for new sets of valves or as valves age. They need a servo which only looks at or around signal zero-crossing; such servos do exist, but they are more complex.

Some go for CCS bias, which achieves much the same thing as a simple servo: setting the wrong current (average instead of quiescent). Worse than a bypassed resistor!

In my view the best option is what people used to do: bypassed resistor for Class A; fixed bias for Class AB and B. In some cases a mixture of the two can be used.

For those who are puzzled about the difference between quiescent current and average current, be aware that valves generate significant second-order distortion. This has two components: second harmonic (and IM for multitone inputs) and DC. Yes, an equal amount of DC shift is produced. The second harmonic may cancel out in a push-pull OPT so you don't hear it, but the extra current still remains. For fixed bias it does nothing, except heat the valve a little. For resistor bias it causes extra voltage drop, which has the result of biasing the valve a little 'cooler' during loud music; the usual solution is to set the bias a little 'hotter' for silence and quiet music. CCS bias (and simple servos) are worse, as they are in effect infinite bypassed resistors. Roughly speaking, 10mA of second-order distortion DC causes:
fixed bias: no change
resistor bias: 5mA bias shift during loud music
CCS bias: 10mA bias shift during loud music

I find it slightly amusing that the people who want to exactly set their output stage bias with CCS or simple servos are guaranteeing that their bias will shift by the maximum amount whenever music is present; it is only what they set it to during silence!
 
I can't see the point of a bias servo, but they are popular so a lot of people think they need one. About the only people who might actually need one are those with toroidal OPTs, as they can saturate easily if unbalanced. However, few people use toroidal OPTs. <snip>
Spot on!

In addition: if using new good tubes bias won't change except for mains voltages fluctuation. If one uses tubes from various sources ( to save money?) you might want to invest in a bias servo. But return of investment ??
 
Sodacose,
Like a resistor, it will only work for Class A output stages".

I'm not gingertube to defend this schematic drawing, but I believe it is similar to the Tent Labs bias servo. The diode clamps sample a window about the idle point, allowing it to work just fine in Class AB according to several users who have used it. I have not personally experimented with this circuit but I have taken a passing interest if I ever build something that I want to add a servo to.

Not quibbling with the average vs quiescent distinction though. It's a fair point with the time constants involved.
 
Hi !

Here is a schematic of output tube BIAS servo published by Hopchik (Oleg from Voronezh), on DIYAudio.ru forum. According to feedback from forum members it works as expected. I didn't tried to assemble this schematic, posting here in hope someone could find it useful.

Link to original post on DIYAudio.ru
C???? ??????????????? ???????? ???????? ????

R19 and R20 connected to the grids of output tubes (CETKA in Russian means Grid).

I send a message to the author yet till now he didn't responded. Copyright notice on DIYAudio.ru states that copying and citation permitted if provides link to forum and nick of the author.

PS. What really bothers me with this and all similar servo schematics that they will modulate BIAS voltage during low-frequency high-level transient, possibly need low-pass RC filter with very high capacitance. Have to do some simulation in LTSpice.

Any feedback appreciated.






Hello from Greece! Can i use this circuit,in 6c33c SE? Looks like interesting circuit.These tubes is very unstable.I find too many schematics in net,but all has simple fixed bias circuit,or cathode bias,or mixed bias.Thank you!
 
I work from the premise that anything you do to balance BOTH AC and DC conditions in the output stage is of benefit.
DC balance means no hum and no "stray" signal to intermodulation with. Harmonic distortion does not occur by itself, there is always intermodulation distortion which accompanies it. AC balance (in a push pull output) means the same as it does for any differential gain stage. 2nd harmonic is cancelled by balance and what remains at perfect balance is 3rd (and higher odd harmonic) due to non-linearities of the amplifying device themselves.
If you do the math (PITA math) it clearly shows that Even harmonics are associated with Intermodulation products which are close to the original frequencies while odd harmonics have a set of products, half of which are close to theoriginal frequencied and half of which are remote from the original frequencies. The "take away" (for euphonic sound) from this is to limit as much as possible odd harmonic distortion.

So triodes for the amplifying devices which have some 2nd harmonic but little 3rd (and other odd harmonics).
This is what lead me to the Baby Huey design with forced AC balance via the feedback scheme used. Adding forced DC balance in the output pair can only help.
Aside: Forcing DC balance also forces gm balance (at idle) and as the BH scheme trades output tube gm for reduced rp (which better drives primary inductance, low frequency response, and better drives leakage inductance and winding capacitance better high frequency response, although we are working against 2 time constants at high frequencies) then gm balance helps. I am a retired electronic design eng and one of the things which "presses my buttons" is the use of a microcontroller where "2 bucks" worth of analog will do the same thing.
I have 2 off the Menno/Tentlabs bias servo boards on my shelf and the schematic I posted is part ONLY reverse engineering of those boards. Where its design varies from a many other bias servo. Its design is that it has higher intrinsic gain and the comparator/integrator works in a narrower window about the idle point. That means that the comparator output is rail to rail ( pulse width modulated if you like) and that has the possibility of coupling into the audio if not careful.

John Broskie's site has a heap of info on alternate bias servos if interested.

For Torroidal Output Trannies DC balance is important to make sure that the output trannies do not saturate in one direction or the other.

Hope there is something in this rave which is useful. To be "up front", the world seems to have gone to pot and I decided (for this afternoon) to follow suit.

Cheers
Ian
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.