• 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.

Fixed Bias suggestions

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am looking to use fixed bias for a DHT with a transformer input. I drew up a couple of options and am wondering if anyone has a preference, or any other options.

The upside to A is that the input signal path is a transformer winding, a battery, and the tube which is pretty minimal. The downside is that because the filament supply is biased up from ground, two channels will require separate filament windings which adds a little bit of hassle.

The upside to B is that both filaments can be powered froma single winding. The downside is the addition of a large resistor (~1M or so?) to the input signal path which may or may not be a problem.

The upside of C is that the input signal path eliminates (or at least limits) the battery and the large resistor. The downside is the addition of a capacitor, and being cap-phobic this seems like trouble.

I also thought about using something like tubelab does in his SE amp: http://tubelab.com/TubelabSE.htm my concern is that if you trace the current loop, the bias PS caps are in the signal path, which seems to just trade one electrolytic for another.


  • fixed_bias.jpg
    17.7 KB · Views: 550
Joined 2004
Paid Member
oshifis said:
What is better: a capacitor or a battery (maybe shunted by a capacitor) in the input signal path? I wote for C unless there are "audiophile" batteries.

I have experimented quite a bit with battery bias and have used all three topologies and have generally found "A" and "B" to be preferable. "A" gives very good results with NiCads, (Panasonic/Matsushita) but not such good results with NiMHs, "B" works great with alkaline batteries which may have source impedances of a few ohms or less. "B" in my experience is the most transparent.

"C" adds a lot of additional and IMHO unnecessary components in the signal path. Results can be very good, but the quality of the coupling capacitor is important and hence it will be expensive.
kevinkr said:
Count me in for B as well, incidentally the resistor in series with the battery is not required and not recommended if battery bias is to be used.

Somehow I got it in my head that I needed a large resistor to keep the battery from draining -- not sure where I thought the current was going to go. Sometimes you need someone else to tell you the obvious.

oshifis said:
What is better: a capacitor or a battery (maybe shunted by a capacitor) in the input signal path? I wote for C unless there are "audiophile" batteries.

I don't know. There aren't "audiophile" LED's either, but they still sound better in a signal path than any cap I've tried, and I've tried a few. LEDs are just not practical here as the bias is too high -- I'll be using a pair of 9V batteries if that information is important.

That said, I do think it is worth the experiment to see what works the best. I was just looking for a starting point, and to see if there was anything I forgot.
Yvesm said:
Another vote "B" !

But without any resistor in serie with the bias source that should have the lowest possible internal impedance AND resistance.

The signal must pass thru the bias source and it must be able to accept some current sourced by the grid while still maintening a stable voltage.


Hi Yves

I do not completely understand your reminder, but am I correct assuming that one can not get away by using the standard voltage divider (with a potentiometer) to set the bias, but instead one needs a adjustable voltage regulator*, with the output straight to the IT?

many thanks, Erik

* A regulator as the one proposed by Morgan Jones on p335 of Valve Amplifiers?
Hi Doug, just to throw a wrench into the works. I'm playing with it on the bench with an A2 814 SE biased around -90VDC, ~20 ma through a 4.7K grid resistor. Effectively no caps in the bias grid circuit, I use these handy-dandy IXYS boards I picked up somewhere. ;)


  • fixed_bias_bbbb.jpg
    9.1 KB · Views: 312
Joined 2004
Paid Member
The only issue with this approach is that an audio current does appear across that resistor - it's not a perfect ground, but the 4.7K resistor value shown probably works pretty well.

You can still use batteries with this approach, or a conventional negative bias supply.

Batteries will have a much lower source impedance than that 4.7K resistor, but the impedance will be somewhat frequency dependent based on the battery's internal construction - this can still be dealt with with a small film cap across the battery terminals.

A fresh 9V battery may have an internal impedance of less than 10 - 20 ohms out to several kHz or greater.
The CCS pulls current from the top of the resistor to the negative bias supply. The supply must be sized to acommodate the DC bias + any voltage swing seen by the CCS, the latter determined by the voltage divider ratio formed by the transformer secondary and the resistor to ground. Here's the thing, the latter resistor for most applications is in the 5k range, the CCS dynamic impedance a min 200X greater, making it completely ignorable in parallel with 5k. In normal applications such as in series with a grid choke or, say, 100k grid resistor the swing across the 'bias' resistor is very small and the CCS sees little voltage swing, reducing the potential sonic impact of non-linear capacitances, etc., even further. More than any other grid bias scheme this one appears to be effectively not in the circuit, a kind of magic voltage source. The 814 goes A2 at ~3 watts - cathode follower driver into a 150H/3.4K dcr hammond choke in series with a 4.7k bias resistor - and still pounds out almost 12 watts into a 10k:8 opt with bias stability. I haven't had much opportunity to listen yet but it meters and works well.

Hi kevin. It's a valid point, I agree 100% a battery would be better for a transformer secondary unless you need a pot to make it adjustable. I think the CCS is better than most realizable, low ripple traditional rectified/smoothed bias supplies though. Consider a 100k grid resistor instead of Doug's transformer. Now split it 95K/5K and CCS the bottom 5K. The driver still sees 100K, it's effectively a zero Zout bias supply. The CCS swings ~1/20 the driver swing.
FWIW, I've considered battery bias on a FET, like the IRFBC20. Yes, you must cap. couple the I/P, but only a modest cap. value is needed working into a 2.2 MOhm or possibly larger gate resistor. As the enhancement mode FET requires forward bias, connect the gate resistor to the + side of a 9 V. Lithium battery. Connect the - side of the battery to the FET's source electrode.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.