• 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.

replacing cathode bypass cap. with shunt reference voltage chip

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The dynamic resistance of shunt reference voltage chip is very low(for example TL431 has only 0.22ohm), if it is connected in parallet to the cathode resistor, the cathode dynamic impedance can be almozt zero.

Provide additional power supply for reference voltage chip, let the reference voltage chip provide some current output for the cathode resistance(you have to reduce the resistance of the cathode resistor to take on more current).

Try it, the sound is clear.
 
it's not ordinary fixed bias. it's used reference voltage chip instead cathode bypass capacitor. The function of bypass capacitor is to make cathode grounded dynamically, The internal resistance of shunt reference voltage chip is close to zero. parallel connection to the cathode resistor can also make the cathode grounded dynamically, but the sound is better than the electrolytic capacitance.

For example, 12AX7 cathode resistor 1.67K, cathode current is 1.2mA, cathode voltage is 2V, normally it needs a bypass capacitor greater than or equal to 47uF. Reduce the cathode resistor to 1K, set the reference voltage chip to output 2V, and connect the reference voltage chip to 1K cathode resistor in parallel.

Since the cathode voltage is still 2V, the tube current is still 1.2mA, and the additional 0.8mV current is provided by the reference voltage chip. This is it. Reference voltage chips need external power supply, not cathode power supply directly.
 
it's not ordinary fixed bias. it's used reference voltage chip instead cathode bypass capacitor. .

I agree, it's a different method than I have seen before. Of course, there are many ways to generate a fixed voltage, but when the bias for the tube is set by a fixed voltage we still call it "fixed bias". Still, you may not need the parallel resistor at all.
 
Account Closed
Joined 2018
The "EFB" biasing and regulation system designed by David Gillespie caught my attention, and so I converted my custom EL84 amp with standard cathode bias to the EFB system. It tracks any power supply variations due to any "sag" on heavy bass or volume levels, and instantly compensates the output bias accordingly. And regardless whether the power supply is tube rectified, or SS diodes, they "all" have some fluctuation inder heavy load.

This system keeps the tube bias accurately "fixed" throughout its operating range, thus improving transient response and eliminates distortion of the treble range as well. It really works! - I can drive the amp now without it "crapping out" at higher volumes than before, and actually increases output by a few more watts.
 
As KodaBMX and WiseOldTech both imply, the 'solution' of merit is to use a small voltage reference that can 'go low enough' in milliamps to operate correctly, AND to 'bypass' it with a capacitor so that its output impedance is near-zero at audio frequencies.

The LM337 is one such solution. It suffers a bit from needing higher currents than 1.2 mA … (per your example.)

These days though, the conventional wisdom is to use an equally near-constant voltage drop device (you're looking for about 2 volts), which would be the RED colored LED diode.

As DF96 notes though, as with all “fixed voltage bias” approaches, unlike CCSs and cap-bypassed-resistors, they do not self-adjust as the tube ages, or as different tubes are inserted having somewhat differing operating points. This is definitely not a feature. It is a bug. Not a serious bug, but a bug notwithstanding.

Now, I'm a simple goat when it comes down to it; while there definitely are self-designed-and-implemented JFET solutions that are attractive, I tend to like these things, tho' their a bit pricey: https://www.mouser.com/datasheet/2/68/1n5283-5314-43522.pdf

In a nutshell, for about three bucks or so, you get a 2-terminal series current regulator of remarkable durability; it needs only a bypass capacitor to become a very low audio-frequency auto-bias source; the ability of it to function as a self-adjusting bias, given the various ageing of valves and different valves, is excellent.

So. That'd be my solution. Drop-in, 1.2 mA, 1N5299 CCS regulator.

Dumber than a box of rocks, but who cares?

Just saying,
GoatGuy ✓
 
Fortunately, yes.

You just have to accept that as 2-terminal drop-in constant-current sources, they do that job well, and not much else. For a dial-a-current-regulator, we're basically back to one of hmmm… (3) or so conventional designs. 1 - JFET, 2 - bipolar NPN, 3 - current-sense OpAmp+MOSFET CCS regulator.

The trick is to also cobble together a low dropout design that works reliably at low milliamp flow … AND … regulates tightly. The DIYaudio forum is chock full of various designs.

Again … note that the 1N5283–1N5314 series ranges from 0.22 mA to 4.7 mA from Central Semi. The data sheet(s) are at https://www.mouser.com/datasheet/2/68/1n5283-5314-43522.pdf for more info.

GoatGuy ✓
 
I wonder if you could tame the noise with an RC filter in series?

Well … such sources are either being used as autobias devices, or much more demanding hard-reference constant-current sources for differential pairs of tubes / FETs / BJTs / whatever.

The first case is actually the easiest, perhaps obviously. As a “smart resistor” in auto-bias of amplification stages, to actually amplify, the lil' smart chunks of silicon need to be bypassed — just as resistors do — with substantial capacitors. This renders any intrinsic noisiness moot. They turn into reference-current variable-voltage … capacitor bypassed bias supplies. Perfect.

The second case is actually a complicated one: but KodaBMX's idea is a sound one. Placing an RC (or even better, LC?) filter in series squashes the noise in the Audio Frequency band, whilst still allowing the “constant current” to get thru. Indeed … just sitting here thinking about it, perhaps ONLY a choke might be the most effective component.

At near-DC, all they have is the RCOIL series resistance. Won't change the voltage much at all. However in the AF band, one can count on the impedance following the usual formula
Z = 2πFL …​

So yah… I like it.

GoatGuy ✓
 
You can filter it off, but with a few caveats. The voltage sources have near-zero resistance at 0 Hz, so capacitive decoupling won't help much at low audio frequencies. An RC-filter will be more effective, but also increases the resistance, making it a less accurate constant voltage biasing.

Current sources filtered with inductors are the dual case and have the dual problem: the resistance at DC is very high, so a series inductor won't help much at low frequencies. You could shunt the current source with a resistor to solve this, but then the constant current biasing gets less accurate.

So all in all, you would have to find some practical compromise.

When you want current source behaviour at DC and a low impedance at AC, it is indeed simple, like GoatGuy wrote: just shunt the current source with a sufficiently large capacitor.
 
Last edited:
AC couple the cathode back to Vref of TL431 via 3uF and bypass the TL431 with 0.1 in series with 220R... ? I get 2mV ripple from a shunt regulator which uses exactly this - cannot see why it would change in cathode circuit of voltage gain stage.

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