Mullard 5-20 high DC Voltages

[Kei-1986]

First post here, so here goes.

Having built myself a single test unit using a wooden chassis. I am in the testing phase and have hit a stumbling block. The amp works and sounds great with no sign of hum or any nasty noises, however the DC voltages I am measuring are approximately 20V high. What makes no sense is the AC voltages out of the secondaries of the mains transformer are measuring low as I'm using the 250V primary tap and my mains voltage seems to sit between 235 and 240V.

The information on the build:
Built as a standard amp following the component values laid out in the mullard book. (The pages are available here http://www.r-type.org/articles/art-003d.htm)

The only tweaks made to the design so far:

1. Raised the value of C15 to 47uF and the other 8uF capacitors are 10uF
2. The HT feeds are both fused individually using 250mA fuses before the rectifier valve
3. There are UF4007's used on the octal socket to reduce the chances of the GZ34/5AR4 from arcing over. (checked and no HT gets through without the valve fitted.
4. The layout, which is why this has been prototyped in wood. (no point getting aluminium machined if the layout doesn't work)

These are the main bits used to build it:
Sowter UP21s output transformers (6.6K + 43% UL taps, 16, 8 & 4 ohm secondary taps)
VVT VTL12158-1440 Chokes (10H 180mA, measured 117.5 ohm resistance)
Primary Windings mains transformers (410V – 0 – 410V @ 180mA, 6.3V @ 4A Centre tapped, 6.3V @ 2.5A Centre tapped, 5V @ 3A)
Tung-Sol EL34B
Sovtek 5AR4
NOS GE 6267
NOS GE 5751

Testing voltages with no input signal sat idle. (using Fluke 28II DMM)

Mains voltage - 237.6Vac
HT voltage - 398.5Vac (directly measured off the 410-0-410 winding)
C15 - 481Vdc
C12 - 465Vdc
C5 - 431Vdc
C4 - 172Vdc

V4 Anode - 457Vdc
V4 G1 - 0.9Vdc
V4 G2 - 454Vdc
V4 G3/Cathode - 32.7Vdc
Heater - 5.74Vac

V3 Anode - 455Vdc
V3 G1 - 0.6Vdc
V3 G2 - 454Vdc
V3 G3/Cathode - 32.2Vdc
Heater - 5.76Vac

V2 Anode 1 - 317Vdc
V2 Anode 2 - 317Vdc
V2 G1 - 83.5Vdc
V2 G2 - 91.6Vdc
V2 Cathode 1&2 - 96Vdc

V1 Anode - 91.7Vdc
V1 G1 - 0 (no input)
V1 G2 - 114.5Vdc
V1 G3/Cathode - 2.18Vdc
Heater - 5.9Vac

Rect pin 8 - 478Vdc
Rect pin 8 - 5.4Vac (to GND)
Rect pin 2 - 7.2Vac (to GND)

Other measurements I made whilst it was running:
14.5Vrms output is about the limit before it seems to begin to oscillate. (1KHz sine wave showing ringing around the edges of the peaks/troughs). The waveform doesn't flatten off top/bottom like I would have expected, it keeps rising with the ringing around the edges. I added 30pF to the feedback capacitance as a test, making C9 360pF. The ringing on appears unchanged.

Output level appears to rise below 9Hz and above 42KHz. (forgot to make a note the output power during that test)

This is the scope trace with 1KHz square wave input.
https://***********************/65535/...250c88f7_h.jpg

This is 10KHz. I've tried added an additional 30pF cap to C8 but it doesn't appear to have made any difference.
https://***********************/65535/...aba07cda_h.jpg

The issue I've noticed is the very slight signs of red plating on the EL34's. Seen here in this picture with a slight glow on the right side of the anodes of both near where they are welded together.
https://***********************/65535/...fea242e4_h.jpg

This is the component layout underneath.
https://***********************/65535/...5f8f253a_h.jpg

Basically, I'm trying to figure out how I can have lower AC voltage going into the valve rectifier than the book figure by ~12V yet end up with 20V higher than the book figure coming out of the other side? Any pointers would be very welcome.

[Electronpusher0]

Not sure if it helps but don't forget that the voltages in the Mullard book are measured with an AVO 8, 20K ohm per volt. If you are using a modern dvm it will not load the point being measured in the same way.

Peter

[ms660]

Increasing the value of the reservoir capacitor will increase the DC voltage across it.

Lawrence.

[PJL]

And the 5-20 was designed to drive a tuner and pre-amp of up to 40mA.

[turretslug]

As well as the points mentioned above, there are a few other things that could cumulatively bump up HT- if the mains transformer HT windings and/or smoothing choke have lower resistance than the originals, if the installed 5AR4 has a lower forward voltage than expected from an original Mullard GZ34 (rectifiers were incrementally improved over time as regards aspects like cathode coatings and electrode clearance tolerance) and if the output valves in particular are drawing less current than original spec.- at least the cathode voltages are close to each other and knowing these and cathode resistance will give current drawn.

[Herald1360]

14.5V rms into 8R is about 26W so no lack of grunt.

The ringing/oscillation on overdrive could be worth worrying about. How does it look fed with a square wave which doesn't overdrive it?

If you're worried about 20V extra HT (only about +5%) add a watty resistor between the rectifier cathode and the reservoir capacitor and tweak to get the voltage you want.

Heater volts all seem to be lowish- what are they back at the transformer? Is the heater wiring a bit on the thin side?

[Richard]

Hi

Nice job.

What value did you use for R26 and R27?
I can not see them in the picture?

"The power supply is conventional and uses a Mullard indirectly-heated, full-wave rectifier, type GZ34, in conjunction with a capacitive input filter. The values of the limiting resistors R26 and R27 will depend on the winding resistances of the mains transformer used. Their purpose, when required, is normally one of voltage control only."

[Kei-1986]

Originally Posted by Electronpusher0

Not sure if it helps but don't forget that the voltages in the Mullard book are measured with an AVO 8, 20K ohm per volt. If you are using a modern dvm it will not load the point being measured in the same way.

That had crossed my mind but I was uncertain as to which parts of the circuit would be affected by the 10meg to 20meg difference in the meter.

Originally Posted by ms660

Increasing the value of the reservoir capacitor will increase the DC voltage across it.

That is something I am in the process of sorting as I have some more of those 10uF capacitors. I assume that with the larger capacitance the effect on the DC side is that the reduction in ripple is where the increase comes from as there is less drop between the pulse peaks.

Originally Posted by Herald1360

14.5V rms into 8R is about 26W so no lack of grunt.
The ringing/oscillation on overdrive could be worth worrying about. How does it look fed with a square wave which doesn't overdrive it?

If you're worried about 20V extra HT (only about +5%) add a watty resistor between the rectifier cathode and the reservoir capacitor and tweak to get the voltage you want.

Heater volts all seem to be lowish- what are they back at the transformer? Is the heater wiring a bit on the thin side?

The square wave scope shots were at 1.5Vrms. (about 0.25W)
https://***********************/65535/...29cc4092_b.jpg
https://***********************/65535/...ca4d7ddf_b.jpg

The heater voltages are low as I'm using the 250V primary instead of the 240V in order to get the lowest voltages out. Thought it the safest choice since I don't have a variac. My mains voltage is usually between 230-240V.

Originally Posted by Richard

What value did you use for R26 and R27?
I can not see them in the picture?

"The power supply is conventional and uses a Mullard indirectly-heated, full-wave rectifier, type GZ34, in conjunction with a capacitive input filter. The values of the limiting resistors R26 and R27 will depend on the winding resistances of the mains transformer used. Their purpose, when required, is normally one of voltage control only."

Currently at least, R26/R27 are not included as I had assumed that buying a mains transformer specifically designed for this amplifier with a wide coverage of primary voltages should be sufficient.

I think the 5AR4 having a much lower voltage drop than expected is the only thing I can think of that would truly explain what I'm seeing

A thought I had considered was to replace the cathode bias resistors with 560 ohms to help reduce the current. I would prefer to avoid roasting the EL34's as they are around £100 for 4 so eking out a longer life would be preferable.

[Richard]

Hi

I would leave the smoothing cap as it is, put transformer to 240V tap, experiment with R26 and R27 until you get the voltages you are after.
This will reduce the work the EL34s are doing a bit, it does look like one is just starting to glow.
The ringing clearly should not be there, if you follow the signal through the amp with the scope you should be able to work out which stage it is coming from.
Make sure you don't zap the scope with the high voltages present, a high voltage cap on the input will protect it from DC.

Richard

[PJL]

I would have thought that any amplifier that has a propagation delay and negative feedback is inevitably going to ring slightly if you hit it with a very fast rising edge.

The ringing is around 80KHz and decays quite quickly so is completely inaudible and the reality is that any audio signal source is not going to deliver transients like this.

Might be interesting to see how other 5-20 builds behave before trying to fix it.

PS: C1/R3 rolls off the HF open loop gain

[ajgriff]

I've always thought that high frequency ringing is almost a standard feature of valve ampifiers due to the inherent resonant characteristics of output transformers. Doesn't necessarily matter much as long as it's inaudible. Happy to be corrected if I'm wrong.

Alan

[Herald1360]

It looks fairly tame, maybe a bit bouncier than ideal, but there does seem to be a slight discontinuity in the 10kHz case as the ring starts. This might suggest that the beast is only conditionally stable at some high frequency which could be the root cause of the oscillation seen when you overdrive it and the loop gain falls off.


It may not matter in practice but it would be interesting to nail it! An open loop bode plot from the feedback injection point to the LS output would be interesting. A 1-2-5 plot (3 frequencies per decade) up to the gain crossover point should be detailed enough. Have fun! At least it's not a 1200Amp DC PSU that's honking......

[Diabolical Artificer]

The red plating on the valve there is odd, considering your DC measurements aren't far off. What do get on the scope, no IP, IP grounded? What AC do you get at C12/ EL34 anodes? Ground the grids of the EL34's, do you still get red plating? Disconnect the NFB, still red plating?

Some pics of the underside of your amp would be nice, could reveal something.

Andy.

[ajgriff]

Originally Posted by Diabolical Artificer

Some pics of the underside of your amp would be nice, could reveal something.

Last link in the first post Andy.

Alan

[Ambientnoise]

Not experienced with valve amps but surely it’s good practice to put an LPF (simple RC) on the input of an amp to limit the input rise times, especially in test scenarios ? Normal programme material will be of limited bandwidth as has been said.

[Kei-1986]

Originally Posted by Diabolical Artificer

The red plating on the valve there is odd, considering your DC measurements aren't far off. What do get on the scope, no IP, IP grounded? What AC do you get at C12/ EL34 anodes? Ground the grids of the EL34's, do you still get red plating? Disconnect the NFB, still red plating?

Some pics of the underside of your amp would be nice, could reveal something.

To be fair, I can't see it with my eyes unless it's run in a very dark room, so it's not severe. The camera sees it a lot better than I do and even then it's still only really visible in the dark.

This album contains all of the photos of the project. (including some of the mistakes along the way)
https://www.***********/photos/399591...57713410461998

Based on my bias sums using this calculator () I am running the EL34's at:
Total Cathode Current = 68.1 DC milliamps = Voltage Across Cathode Resistor / Cathode Resistor.
Total Anode Current = 64.4 DC milliamps = Cathode Current minus approximate screen current of 5.5%.
Anode Current per Valve = 64.4 DC milliamps = Total Anode Current / Number of Valves.
Anode Dissipation per Valve = 27.6 watts = Anode Current per Valve x Anode Voltage.
Anode Dissipation per Valve % = 110.4% Anode Dissipation per Valve % = Anode Dissipation per Valve / Valve Max Rated Dissipation.

So far between suggestions and ideas from others I have a list of potential options:

1. Run with it and increase the cathode resistors to drop the bias. Advantage being no additional heat to deal with. Disadvantage being more likely crossover distortion from class B operation.

2. Change rectifier to a 5U4GB. Looses the soft start and the first filter can't exceed 40uF. Adding a CL90 or CL150 NTC solves the soft start issue.

3. Use the dropping resistors R26/R27 suggested in the mullard book. Disadvantage is the sag they introduce and the heat they generate.

4. Use a zener diode to drop the voltage.

5. Use a mosfet to drop the voltage.

[Chris55000]

Hi!

I am in total agreement with the need for additional series rectifier limiter resistors, begin with 2 X 220 ohm metal–clad about 10W per anode!

Second, I most strongly advise the use of anode, screen–grid and grid "stopping" resistors for the EL34 output valves, they have sufficient mutual–conductance (11mA/V!) to need all three really!

Anode resistors 47 ohms, screen–grid resistors 100 ohm, grid stoppers 1k, all 1W metal power–oxide type!

Note that "stopper" resistors must be connected to the valveholder tags with the shortest possible lead lengths for them to be effective – not more than 5–6mm lead length from the body before wiring to the tags!

I am also of the opinion that modern valves are not as robust as the original Mullard designs, this is another reason why it's best to reduce the h.t. line voltage by about 5% below the figure quoted on Mullard's original plans!

If you try the suggestions quoted here the valves should not be showing any signs of anode glow – if they do then I'd suggest they're sub–standard or defective!

With the years I spent on valve–operated gear, the only "glowing anodes" I had were TV line–output valves being run with low or no drive!

Chris Williams

[G6Tanuki]

Out of interest, I've always wondered why the trend for two series-resistors, one in each rectifier anode lead, came to be a thing.

You can achieve the same result with one resistor on the rectifier-cathode-to-reservoir-capacitor connection or - my favourite - from the transformer HT centre-tap to ground.

(it's my preferred option because it means the resistor is not sitting at high voltage so is incapable of 'biting' you!)

[John Caswell]

Mullard always quoted a minimum resistance per anode for their rectifiers (Rlim min in the handbook), so if your power transformer didn't meet this spec you added small Rs to bring it up to spec.
e.g. EZ80 (Rlim min) = 300R, EZ81 = 230R, GZ34 = 150R

John

[Chris55000]

Hi!

The reason resistors were always added in pairs in the anode side of rectifiers was simply practical – using one resistor in the centre–tap lead of the h.t. secondary of the transformer means it has to be considerably larger in wattage rating for a given value, because the reservoir capacitor's ripple current, as well as the h.t. current, has to flow through it, and also it was usually physically easier (particularly with octal–base rectifier valves where unused pins are omitted from the base in many cases) to suspend smaller wattage components in the under–chassis wiring!

Now that high–wattage metal–clad resistors that can be screwed direct to metalwork are available, it's now only a matter of personal choice!

Chris Williams