diyAudio (
-   Tubes / Valves (
-   -   Distortion from heater sag? (

hihopes 23rd September 2007 05:03 PM

Distortion from heater sag?
My hybrid amp has a slight distortion in certain passages (eg piano with lots of reverb) and I have tried just about everything I can think of to linearize the output mosfets. I have also inserted a resistor between the anode and grid of the EL34ís (previously connected directly). The mods have brought about some good improvements, but the distortion is not completely gone. Now I am starting to suspect it may be a sagging heater supply. The heater voltage sits at 5.5v. Is this maybe low enough to create a little bit of distortion on complex waveforms like I am experiencing?

Merlinb 23rd September 2007 06:21 PM

Provided the heater voltage is constant, that won't be the cause of the distortion. In fact, slightly low heater voltages tend to improve the linearity of valve! (well in triodes and diodes it does, not sure about pentodes).
It doesn't matter what the valve is doing, there should be no reason for the heater voltage to sag.

hihopes 23rd September 2007 08:51 PM

Thanks merlinb. I will have to look elsewhere then.

anatech 23rd September 2007 08:59 PM

Hi hihopes,

The heater voltage sits at 5.5v. Is this maybe low enough to create a little bit of distortion on complex waveforms like I am experiencing?
I think your heater voltage is too low. Why not try to increase it a bit?

Tubes are actually designed to run at the rated heater voltage for the longest tube life. A slight reduction has been noted to reduce noise in signal tubes. Your reduction is more severe than I have seen anywhere before. Try 6 VDC for a change, then 6.2 V. I am assuming you are running DC heaters through a regulator of some kind.


diy4 24th September 2007 04:21 AM

Your amp is push-pull, right?
Your description sounds like crossover distortion; piano usually shows this more than most sounds.
Does it sound like 'fuzziness' surrounding the notes? That's my description, anyway.
Low heater voltage would cause clipping of peaks. 5.5 volts is more than 10 percent low; probably should be fixed anyway.

hihopes 24th September 2007 09:06 PM

Hi diy4, your description is pretty accurate. Most of the time it is crystal clear, but on one particular track with piano and lots of reverb, there is this fuzzy halo that is more pronounced on some notes than on others. For the rest of the CD it can't be heard, but I know it isn't the CD because my other amp, a Hart Linsley Hood kit amp (with the same output mosfets) doesn't reproduce the same effect. I have done all I can think of to clear up the mosfets and I have changed the input transistors, so that is why I started to suspect the valves.
I tried to power the heaters with a regulated supply, but in my zealousness, i damaged a pair of EL34's (must be too much current because the voltage never rose above 6.1v) I still don't know why it happened.

hihopes 24th September 2007 09:24 PM

Hello anatech. Thanks for the input. To answer your questions, no, the supply is AC, from a transformer that probably started out life as a 6v, but providing a few more fractions of a volt than it is doing now. I was concerned about the low voltage and thought i would do the valves a favour and provide them with a regulated supply. I had a 12v 5 amp supply lying around, so I worked out the resistor values required for LM317 to supply 6.25v and implemented it. I figured that the job was going to be too much for a single LM317, and probably even 2 would overheat, so LM317's being cheap, i built it with 4. when I connected it to the valves, it never rose above 6.1v, but soon started to drop fast. I switched it off, but by then the damage was done.
Do you think there is a way to use that 12v PSU, or will I have to start from scratch with another heater supply?

diy4 25th September 2007 12:57 AM

Crossover (or 'notch' ) distortion happens when the output stage bias is incorrect. A 'dead area' results where both output devices are turned off. This could be a design flaw, or incorrect bias. Can you post a schematic?

The EL34 heaters are not the cause; you should raise the heater voltage a bit, but they are probably working ok, almost.

richwalters 25th September 2007 04:56 PM


Originally posted by diy4
they are probably working ok, almost.
Not from this typical output stage; IMO it isnít worth operating power tubes at lower heater voltages. Tests on a 100W p-p AB1 amp using 4x 6550Bís gave markedly worse performance. First figures = heaters.

6.35VAC; Pout =80W; 1Khz; Thd = 0.05%; B+450V; Iq = 85mA each tube, is corect operation.

5.7VAC: Pout = 70W; 1Khz; Thd = 0.15%; B+ 450V; Iq = drops to 73mA each tube.

The clipping level is correspondingly lower with a sharp rise in thd.
At higher frequencies this would be far worse. I can't measure IM thd but the results would be poor.
Generally I find the 6550A,B,C series is heater / emission sensitive, and probably not the only tube in the power class to be so.

Verdict; For optimum power throughput performance use correct heater voltage or a touch higher. I use 6.35V.


anatech 25th September 2007 05:12 PM

Hi rich,
How many people do you think read tube manuals these days? :angel:

Hi hihopes,
If you don't have a tube manual, buy one. The RCA manuals are good as an example.

Understand that EL-34's can draw up to 1.6 amps each. Make sure your filament transformer is rated 1.25 X your total load or higher. Continue using AC for your heaters. You can drop your voltage a couple ways. You can use a series dropping resistor, or you can pre-regulate the AC going into the transformer. This is much more complicated.


I had a 12v 5 amp supply lying around, so I worked out the resistor values required for LM317 to supply 6.25v and implemented it.
The inrush current was probably the final blow to these regulators. You are dropping too much voltage. Even one EL34 is too much for an IC regulator if it's one of the 1.6 amp beasts. Dropping 10 VDC or more across those regulators was most unkind to them. If you are fixed on this course, use a high current switching regulator for each channel. A "soft start" would be an excellent idea. Look into the LM2677 switching regulators for your project.


All times are GMT. The time now is 08:47 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2