EL34 Push Pull weirdness

[waner]

I need help!

I have bulit a 30W EL34 Push Pull amp (http://www.lundahl.se/pdfs/claus_byrith/amplifier_30wpp.pdf)that I can`t get to work as it should.

1.I`ve got a huge low freq noise in the amplifier.
2.The amplifier gets distorted very early.
3.There is a pulse sounding in the speaker (similar to the sound when you tapping on a microphone). The pulse is linear not random (oscilation?). If I measure with a multimeter on the anod of the EF86 the pulse disapear. If I look at the amplifier in the dark the EL34 blinks blue in time with the pulse. One EL34 at the time. The bias voltage is also pulsing (not stable).

I have tried to switch the NFB on the secondary, then the amp really started to oscilate.
I have tried drive the amp without NFB that increased the LF noise.
All the dc voltage seems to be okay

I have an external PS to the amp. All the leads to the amp runs in the same cable (both the high voltage and the AC for the filament) could that have something to do with the LF noise?
The Choke in the PS is reversed 180 deg to the maintransformer. Could that cause the LF noise?

Please please help me

Regards
Mattias

 

[cogsncogs]

3.There is a pulse sounding in the speaker (similar to the sound when you tapping on a microphone). The pulse is linear not random (oscilation?). If I measure with a multimeter on the anod of the EF86 the pulse disapear. If I look at the amplifier in the dark the EL34 blinks blue in time with the pulse. One EL34 at the time. The bias voltage is also pulsing (not stable).

What you are describing is called motor boating. This is unwanted coupling of signal either from the power supply or through grounding, resulting in Positive feedback at some low frequency, usually at 1 or 2Hz. Remember all wire has some resistance! This can be caused by:

1. Bad or miss-wired decoupling capacitor, i.e. C4, C3, C12 on the 5-20 schematic.
2. Miss-wired ground wire(s).
3. Bad layout. i.e. using a common ground return path (wire) for all stages of the amplifier to the power supply.


The input/driver circuitry should have their own ground wire separate from the output stage ground wiring, connected only at (inside) the power supply ground.
The heater supply should have it's own wiring, i.e. not inside the same cable. 16-18ga twisted pair depending on the distance between the amp and PS.
Try changing where the spkr ground/secondary connects to ground.
Go back and re-read the manual and make sure everything is connected and goes to where it's supposed to go.

I have tried drive the amp without NFB that increased the LF noise.

This tells me that it definitely has nothing to do with the NFB loop.
Hope this helps and gives you some ideas where to look! If you haven't read this before, it may help.

www.audioxpress.com/resource/audioclass/

Wayne

 

[nilrog]

Have you built this amplifier with the same components as described in the paper? I'm going to build the same amp but i'm still waiting for some parts.

3.There is a pulse sounding in the speaker (similar to the sound when you tapping on a microphone).

The swedish magazine AoE had the same amplifier as a project last year during the summer. We'll, not exactly the same... They used LL1620 transformers and had added back some of the components that was removed by Claus along with some other modifications.

Anyway, they stated that there was an oscillation of ~.5 Hz caused by the 470nF caps before the EL34s so they lowered the value of these caps.

/Roger

 

[Sch3mat1c]

I have bulit a 30W EL34 Push Pull amp (http://www.lundahl.se/pdfs/claus_byrith/amplifier_30wpp.pdf)that I can`t get to work as it should.

Hmm, pretty good circuit. Looks like a cross between two of my schematics actually.

1.I`ve got a huge low freq noise in the amplifier.

TURN IT OFF!

2.The amplifier gets distorted very early.
3.There is a pulse sounding in the speaker (similar to the sound when you tapping on a microphone). The pulse is linear not random (oscilation?).

Motorboating.

If I measure with a multimeter on the anod of the EF86 the pulse disapear. If I look at the amplifier in the dark the EL34 blinks blue in time with the pulse. One EL34 at the time. The bias voltage is also pulsing (not stable).

Because, like any oscillator, it is running between cutoff and saturation. Cutoff is hard enough that the blue glow disappears from the tubes, while saturation is hard enough that grid current appears (from overdriving the output grids), biasing them down. This is class C, 'grid lead biased' (so called from the bias voltage appearing across the grid leak resistor).

I have tried to switch the NFB on the secondary, then the amp really started to oscilate.

Alright, then you have a few too much phase shift somewhere forward. If it screams backwards, then you have NFB connected correctly as-is.

I have tried drive the amp without NFB that increased the LF noise.

Ok, guess not. Must be PSU as others have said.

I have an external PS to the amp. All the leads to the amp runs in the same cable (both the high voltage and the AC for the filament) could that have something to do with the LF noise?

No. Low frequencies like these would never be coupled from wire to wire, they are just too slow to induce any voltage or current in nearby wires.

The Choke in the PS is reversed 180 deg to the maintransformer. Could that cause the LF noise?

Nah. Although it should be 90 degrees...

I tell ya what I would do. I'd trim down the power supply, use only enough resistance to decouple the stages. For the ECC83, instead of a 10k, use 1k. Use 4.7k for the next (supplies the EF86). Or even none for the second - the pentode and dual triode can share the same supply rail. You'll have to rebias the pentode stage (adjust the 390 ohm cathode resistor until plate is around 80V). Oh, and remove the zener diode in the PSU.

Tim

 

[waner]

Thank you so much for your replys!!!

I have grounded all of the components to ne point in the amplifier. This point goes trough the cable to the PS and this is grounded at one point in the PS.
The heater wiring has it one wiring but it goes to the PS in the same cable (it`s a cable with 12 leads) as the DC voltage.
The cable is about 10 feet long.

I will have to take a bite in the sour apple and do some heavy ground rewiring.

Thank you so much Wayne.


All the components are the same as in the schematics exept for the outputtransformer. I have used a LL1682/PP 5:5500. I use 6ohms speakers that will give me a primary of 6500 ohm.
I will check the 470 caps

Thank you Nilrog

 

[waner]

Hi Sch3mat1c

I will try to redesign the PS (it seems easier than to do the rewiring).
I have built a stereopair that is driven by this PS. I have already changed the 10k to 4,7k and the 100k to 47k. Do you think this is enough for two channels?

Thank you so much for your help

Regards
Mattias

ps
This shure is a great forum with many great brains!

 

[Sch3mat1c]

I have built a stereopair that is driven by this PS. I have already changed the 10k to 4,7k and the 100k to 47k. Do you think this is enough for two channels?

Yes (as long as the capacitors have doubled, as well as the power transformer's current ratings). But that may not fix the problem; you need to reduce the number of phase shifts in the amplifier. In this case, it seems to be the output stage bouncing on the first rail, causing the others to bounce along with it. If you reduce the number of stages, by cutting the last one (and rebiasing the first stage as necessary) as I mentioned, that should get it.

Tim

 

[gingertube]

Before you go too far in rebuilds etc. - check the following

Your problem is definitely a power supply decoupling problem causing "motorboating" as described in posts above.

It may well be C4 or C7 electrolytics. Either the electrolytics are stuffed or you have a dry joint on either the high voltage side or the ground side such that the capacitor is not really connected at all.

I had exactly this problem with an EL34 Guitar amp I built. Turned out to be a dry solder joint on the ground side of the electrolytic.

Good Luck with your debugging.
Cheers
Ian

 

[analog_sa]

If I look at the amplifier in the dark the EL34 blinks blue in time with the pulse.

Cmon guys, why spoil the wonderful Christmas decorations? This guy will only need help in January.

 

[gingertube]

One more stab i n the dark

I just re-read the posts and had another look at the schematic.

You mentioned that if you put a multimeter probe onto the anode of the EF86 the problem went away. That is when you added the small capacitance of a multimeter probe and lead.

There is another possibility - rather than your problem being sub-sonic oscillation (motorboating) due to insufficient decoupling of the power supply it may be that you have the opposite problem i.e. bursts of High frequency oscillation (supersonic) and you are only hearing the "envelope" of this.

Check that your 4k7 grid stopper on the EF86 has the body of the resistor as close to the grid pin as the wiring will allow. Also check that you have the EF86 in Triode Mode with the screen properly connected to the anode.

Cheers,
Ian

 

[waner]

Thank you so much for all your replys!

Ian

I haven`t fitted R4, C2, C4 and C3 to the circuit. Could that cause the problem?
In Claus cirucit it isn`t used so I figured that I didn`t needed it.

I will start to check the PS this weekend (it seems easier to start there). I will incresed the capasitors


Regards
Mattias

ps By the way I hate those blinking christmaslights.

 

[cogsncogs]

I haven`t fitted R4, C2, C4 and C3 to the circuit.

In looking at Claus's circuit (4-30) in the pdf file, I don't see any ref to R4, C2, C3, and C4. They are not marked as such. Can you describe what they are and where they go? Are you referring to the 5-20 schematic?

Originally posted by gingertube
There is another possibility - rather than your problem being sub-sonic oscillation (motorboating) due to insufficient decoupling of the power supply it may be that you have the opposite problem i.e. bursts of High frequency oscillation (supersonic) and you are only hearing the "envelope" of this.

Good catch Ian!
Quite possible. Yeah, here's where (again) a scope comes in handy! Without one you're poking around in the dark!

Wayne

 

[Family_Dog]

Hi Waner,

The motor-boating is usually caused by one of the filter electrolytics in the power supply not doing its job - ensure that the negatives are all earthed to the main chassis.

Does the 6v winding of your transformer have a centre-tap, if so, is it earthed? If the winding does not have a centre-tap, fit two 100-ohm resistors across the LT winding and earth them at the centre point.

-Eric

 

[waner]

Hi again

R4, C2, C4 and C3 is in the appendex pdf. R4,C2 is in series with the EF86 anod and the 180V PS.
C4 and C3 is in parallell with the catod of EF86


Hmm. The 4.7k resistor isn`t located near the grid. I`ve totally missed that. Does it make much difrence in stability?
I have access to a scope. How high in freq do you think that the HF osc appear?


Thank you all for your replys. This is the first time I have made an tread in a forum. This is awsome

Cheers
Mattias

 

[Sch3mat1c]

Good catch Ian!
Quite possible. Yeah, here's where (again) a scope comes in handy! Without one you're poking around in the dark!

I've had situations where motorboating was aggravated by this, namely adding capacitance on the output for testing. It starts oscillating at HF, causing saturation of the amplifier, which then leads to oscillation of the stronger (preferred) motorboating mode.
Generally speaking, ANY instability can be found by sweeping the frequency. If there is a tendancy towards motorboating, I guarantee there's a rise in gain at between 0.5 and 30Hz. Likewise for HF, something around 10-100kHz (or higher, if you have a very high quality OPT).

A scope and wide band frequency generator are two indisposable tools here, but only if you know what to look for.

Tim

 

[thoriated]

Hi, Waner -

Why not just solder those 4.7K's right to the output tube socket pins first thing? That's where they should be put, if possible, anyway.

 

[cogsncogs]

@ waner

C2 and R4 form an anti-ringing network (phase advance network). Please connect! Remember touching probes to anode of the EF86?
C3 and C4 = cathode resistor bypass caps, sets AC gain.
R2 (4.7k), R8 (1k), and the 4.7k grid stoppers to the grids of the EL34's. Make sure all the grid stopper resistors are in place, and as physically close to the tubes as possible.
Are you using the PCB's or point to point?

All the components are the same as in the schematics exept for the outputtransformer. I have used a LL1682/PP 5:5500. I use 6ohms speakers that will give me a primary of 6500 ohm.

Using a different OPT than specified may require different compensation techniques...


Wayne

www.lundahl.se/pdfs/claus_byrith/appendix_cb.pdf

 

[ray_moth]

Waner, the 4.7k resistors leading to the control grids of the EF86 and EL34 tubes, the 1k resistor leading to the grid of the ECC83 tube and the 1k resistors leading to the screen grids of the EL34s serve as "stoppers". They help prevent parasitic oscillations and should be soldered as closely as possible to the pins on the respective tube sockets, using resistor leads cut to 3mm or less.

 

[cogsncogs]

As ray_moth pointed out, I'll add to it, the 1k resistor to the screens of the EL34's are very important and must be used in 99% of all cases! Not only do they "stop" parasitic oscillations they also limit the dynamic current "injected-reflected" (for lack of better wording!) in an UL circuit (current limiting in pentode mode), back into the screen grids. Usually the values range from 1k to 100R, with 1k being the most seen/used.
The grid "stoppers" to control grids (grid 1) are used nearly 100% of the time and are an absolute must! The resistor body itself, with as short of lead as possible should be connected to the socket tag. Values here range from 4.7k to 1k, with 1k being the most (again) seen/use (4.7k for an 6L6, 5881 and their variants, etc.).
If you haven't tried connecting the anti-ringing network, please try it! It may solve your problem! Another idea is to change the coupling caps to the grids of the EL34's to a lower value, say 330nF or 220nF. Don't lower the value of grid leak resistors to the EL34 grids because of the high-ish output impedance of the ECC83/12AX7's. This is one reason I don't like using such a hi-mu tube in this position even though you can use more equal plate resistors! And also an EL34 using fixed bias the maximum grid resistance is approx 270k ohms (470k ohms cathode bias), but can be a little higher (390k as in this case) because of the 10 ohm cathode resistors.
Let us know how thing are progressing!

Cheers
Wayne

 

[nilrog]

As I wrote before the Swedish magazine AoE had this amplifier as a project but there are some minor/major changes to it compared to Claus design. I didn't have the schematics at hands then so I couldn't describe what they changed. But now I have it infront of me.

C2 and R4 form an anti-ringing network (phase advance network). Please connect! Remember touching probes to anode of the EF86?

These components was added back by AoE, probably because they used a different OPT (LL1620).

Another idea is to change the coupling caps to the grids of the EL34's to a lower value, say 330nF or 220nF.

These caps was lowered to 100nF due to the fact that their amp had oscillation problems at ~.5 Hz.

Besides this, they had a different power supply...with a trafo that they pushed to the limits (according to it's spec). Overall I don't like their changes, and brief explanantions to why they have changed some things from Claus design. So I will soon try and put together the amp with the same OPT etc. as Claus used and I hope it works fine.

I'll be very interrested in hearing your progress

/Roger