|
Vintage Audio (record players, hi-fi etc) Amplifiers, speakers, gramophones and other audio equipment. |
|
Thread Tools |
20th Apr 2019, 3:57 pm | #1 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Mullard 3-3 dc voltages help
Hi
I'm new both to the forum and to vintage audio so I hope I'm posting in the right area and not doing anything wrong. By way of background, I have wanted to build a valve amplifier for a long time and have finally got round to it. I'm too young to have grown up in the valve era but nevertheless have a nostalgia for vintage equipment. I have just completed building a Mullard 3-3 amplifier, trying to follow the Mullard instructions as closely as possible. However, the dc voltage are some way off as follows: Voltage at C9 284 (should be 310) C6 254 (should be 290) C3 183 (210) EL84 anode 270 (290) EL84 screen 239 (290) EL84 cathode 42 (28) EL84 grid 38 (20) EF86 screen 41 (28) EF86 anode 38 (20) I have checked and re-checked the circuit and component values but can't find anything wrong. With the above values the EL84 is running with too high a current, so I have only operated the amplifier for a short period of time but it does appear to be amplifying in a correct, stable manner. In exasperation, I began to experiment with component values one by one. Finally, I tried changing the EF86 control grid leak resistor. A reduction to 5M helped a little. Finally, I have ended up with just 47k. The dc values are now almost spot on: C9 318 (310) C6 292 (290) C3 210 (210) EL84 anode 300 (290) EL84 screen 290 (290) EL84 cathode 29.8 (28) EL84 grid 21.2 (20) EF86 screen 28.8 (28) EF86 anode 21.2 (20) Apart from a loss of bass, the amplifier now functions well and sounds very good (apart from too much hum which I need to tackle). The 3-3 may be low in component count, but with its combination of grid leak biasing, starvation mode, direct stage coupling, and dc feedback between EL84 cathode and EF86 screen, I have to admit I don't fully understand the design. Questions: 1. Please could anyone explain why the amplifier works correctly with the replacement 47k grid leak resistor and not the design value of 10M? 2. Could it be that the EF86 has excessive grid leak? Unfortunately I don't have a spare to check. 3. Is it safe to use with the 47k grid leak? 4. With a modern line source is 47k sufficient loading? Can the bass be restored by increasing the input capacitor? Thank you for any help you may be able to give. Paul. |
20th Apr 2019, 4:27 pm | #2 |
Dekatron
Join Date: Nov 2003
Location: Heckmondwike, West Yorkshire, UK.
Posts: 9,637
|
Re: Mullard 3-3 dc voltages help
For easy reference: http://www.r-type.org/articles/art-003h.htm
|
20th Apr 2019, 4:36 pm | #3 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,583
|
Re: Mullard 3-3 dc voltages help
What make and model of meter did you use for your voltage measurements?
Alan |
20th Apr 2019, 5:32 pm | #4 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Re: Mullard 3-3 dc voltages help
Thank you for the replies. I think the r-type website pages are the same as the Mullard amplifier book which I have tried to follow. The meter is an ISO-TECH IDM93N, which I think has 10M input resistance.
I'm listening to the amp now and it is sounding good (with the 47k input resistor!). Most of the noise has now gone (my wiring from the input socket could have been better). Paul. |
20th Apr 2019, 6:04 pm | #5 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Re: Mullard 3-3 dc voltages help
To follow up on my previous reply:
I built the version with tone controls. Good news about the 47k loading, thank you. For what it's worth I tried experimenting with the original 10M input resistor and changing the input capacitance. Halving or doubling the value made no difference, so as I understand it the grid leakage current creates a small negative bias across the input resistor, and this negative dc voltage is isolated from the input source by the capacitor. The valves were bought new recently, but how new they are is difficult to say. I have a Tungsram EZ80, a Tung Sol EF806SG(EF86), and a Tung Sol EL84. The mains and output transformers are from Primary Windings. There is a centre tap on the heater winding. I struggled a little with the dressing of the heater wiring. I am sure I can do better next time. Your advice on the path of the heater wiring sounds sensible. Looking at the Mullard amplifier book they appear to adopt this approach on the 5-10, although not on the 3-3 for some reason. |
20th Apr 2019, 6:30 pm | #6 |
Dekatron
Join Date: Oct 2003
Location: Croydon, Surrey, UK.
Posts: 7,549
|
Re: Mullard 3-3 dc voltages help
You don't need to change any values. Mullard knew what they were doing and many thousands of these have been built successfully (I've also built two successfully many years ago). Lets see some photos and as asked above, what meter are you using? Just wondering if your amplifier might be running into supersonic oscillation.....you won't hear it but it will greatly upset the operation of the amp and affect the voltage readings.
__________________
There are lots of brilliant keyboard players and then there is Rick Wakeman..... |
20th Apr 2019, 6:40 pm | #7 |
Rest in Peace
Join Date: Mar 2006
Location: N.W. Oxfordshire(Chipping Norton)
Posts: 7,306
|
Re: Mullard 3-3 dc voltages help
Alan makes a good point. If you are using a modern DMM (Digital Multimeter) the voltage readings will be higher than those given it the data, because a DMM typically has an input resistance of around 10Megohsm which doesn't load the circuit in the way that an AVO or other analpgue meter foes. These typically have i/p resistances of between 1000 and 20000 ohms per volt, so, on, say, a250v range, the i/p resistance will be from 250000 (1k/v meter) to 5Meg (20k/v meter). This doesn't, to me at any rate, explain why the EF86 C.G. resistor needed yo be reduced to 47k, when 10 Meg was specified. Obviously the current through the 47k will, theoretically, be 200 times that drawn by a 10Meg resistor.
Sideband has a valid point about supersonic oscillation, though! |
20th Apr 2019, 7:30 pm | #8 |
Heptode
Join Date: Mar 2011
Location: Dorset, UK.
Posts: 519
|
Re: Mullard 3-3 dc voltages help
Hi Paul,
It is just a possibility that the output transformer secondary feedback phase may be incorrect, particularly if "...sounds very good (apart from too much hum which I need to tackle)". Also the feedback is connected to the EF86 cathode via a 6.8k resistor. In terms of bias, if one end of the output transformer secondary is not earthed, not only will EF86 bias be slightly affected but there will be no feedback either. Your experiment proves the EF86 bias is too high since reducing the negative bias generated through the 10M resistor restores sensible operation. What is the EF86 cathode voltage? Rich
__________________
To an optimist a glass is half full; a pessimist half empty; an engineer twice as big as need be! |
20th Apr 2019, 7:38 pm | #9 |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
|
Re: Mullard 3-3 dc voltages help
If a valve audio amp is misbehaving it's often worth disconnecting the negative feedback (assuming it isn't also involved in setting the DC conditions). The best place to do that might be at the junction of R4 and the EF86 cathode.
Are you sure that the feedback components are connected to the correct end of the output transformer secondary (transformer windings have an orientation and it can be very easy to misinterpret what it is) ? If they're not then you'll have positive feedback, not negative, and the amp will become an oscillator, as Sideband has suggested. That might also explain why the hum is louder than you expected - negative feedback will suppress any hum arising within the loop but positive won't. EDIT: I see Rich has raised the same point about the feedback polarity. Cheers, GJ
__________________
http://www.ampregen.com |
21st Apr 2019, 1:12 pm | #10 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Re: Mullard 3-3 dc voltages help
Thank you for the suggestions.
I think I’ll rebuild it and try to improve the wiring. Before I do that is there a simple way to check for the possible ultrasonic resonance? I have tried inputting a signal up to 200khz but didn’t see anything unusual on the oscilloscope. Paul |
21st Apr 2019, 7:02 pm | #11 |
Nonode
Join Date: Aug 2007
Location: Walsall Wood, Aldridge, Walsall, UK.
Posts: 2,853
|
Re: Mullard 3-3 dc voltages help
Hi!
In order to preserve the high–impedance specified in the original Mullard Circuit, try adding a 100k preset across the 560R EL84 cathode–bias resistor, and connect the 10M from the preset slider to the G1 pin of the EF86 – this maintains the negative GB provided by the grid–curremt bias, and allows a small compensatory positive bias to be injected in if, as I suspect to be the case, the negative grid–current of the equivalent EF86 is higher than Mullard designed into their brand of EF86 valve! You can then try adjusting the preset to give the specified voltage of 32V @ the EL84 cathode! Chris Williams PS! Did you include the 1k/0.22uF decoupling filter between the cathode of fhe EL84 and the screen–grid pin of the EF86?
__________________
It's an enigma, that's what it is! This thing's not fixed because it doesn't want to be fixed! Last edited by Chris55000; 21st Apr 2019 at 7:07 pm. |
21st Apr 2019, 8:57 pm | #12 |
Dekatron
Join Date: Oct 2003
Location: Croydon, Surrey, UK.
Posts: 7,549
|
Re: Mullard 3-3 dc voltages help
The best way is to follow the Mullard layout as closely as possible!
__________________
There are lots of brilliant keyboard players and then there is Rick Wakeman..... |
22nd Apr 2019, 8:14 am | #13 |
Octode
Join Date: Dec 2018
Location: Llandeilo, West Wales, UK.
Posts: 1,092
|
Re: Mullard 3-3 dc voltages help
This is one of those odd amplifiers that divide opinion. In more ways than one...
It is very fussy with respect to the measured voltages depending on the individual valves used and how much current they draw in this mode of operation. Google Mullard 3-3 voltages and see you are not alone. If you are completely sure it is not oscillating (layout/ reversed transformer connections etc.) then try different EF86s next. Each will likely give a different result. I suspect Mullard valves might have been tighter in tolerance for starvation mode than today's makes? The BBC tried something very similar with their AM8/2 and AM8/2A amplifiers. The AM8/2 being directly coupled, but the later AM8/2A reverting to capacitor coupling between the stages. I wonder if they too had issues? Enclosed is the info sheet for both. I would put my money into the later version. Alan
__________________
Never Leave Well Enough Alone... |
22nd Apr 2019, 8:49 am | #14 |
Octode
Join Date: May 2009
Location: Stevenage, Herts. UK.
Posts: 1,515
|
Re: Mullard 3-3 dc voltages help
Al, what is the purpose of R9 in the BBC circuits?
|
22nd Apr 2019, 9:03 am | #15 |
Dekatron
Join Date: Sep 2007
Location: Oxfordshire, UK.
Posts: 4,310
|
Re: Mullard 3-3 dc voltages help
The text says that R9 provides some negative feedback, presumably to lower the amplifier's distortion and also its output impedance.
Cheers, GJ
__________________
http://www.ampregen.com |
22nd Apr 2019, 9:32 am | #16 |
Octode
Join Date: May 2009
Location: Stevenage, Herts. UK.
Posts: 1,515
|
Re: Mullard 3-3 dc voltages help
Interesting, not seen that done before. 47k into 150R seems like a very small amount of feedback.
|
22nd Apr 2019, 10:55 am | #17 |
Nonode
Join Date: Aug 2007
Location: Walsall Wood, Aldridge, Walsall, UK.
Posts: 2,853
|
Re: Mullard 3-3 dc voltages help
Hi!
Forgive me for raising awkward questions, but is the complete copy of "BBC Instruction S3" available online anywhere for me to peruse, or can anyone send me a copy by PM to my email or dropbox link, etc? Chris Williams PS! So many times I've found a Member's query is answered by providing a small extract of something, which leads to many more useful resources, either in the full document the extract came from, or the extract refers to a very good textbook I didn't know about myself that I can subsequently buy!
__________________
It's an enigma, that's what it is! This thing's not fixed because it doesn't want to be fixed! |
22nd Apr 2019, 12:50 pm | #18 |
Octode
Join Date: Dec 2018
Location: Llandeilo, West Wales, UK.
Posts: 1,092
|
Re: Mullard 3-3 dc voltages help
Chris,
There is a great amount of BBC Engineering information available on line. Try here first: https://www.bbceng.info/ look at Tech Instructions tab or what you want. You can spend a while in there... Alan
__________________
Never Leave Well Enough Alone... |
23rd Apr 2019, 5:30 pm | #19 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Re: Mullard 3-3 dc voltages help
Thank you for all the useful information. Whilst I am keen to get the original working, the BBC mk2 version is also interesting. Further experimenting will have to wait for a few days but I'll let you know how I get on.
Paul. PS When I said I will 'try to improve the wiring' I meant my own hamfisted assembly, I wasn't suggesting that I know better than the Mullard engineers did! |
27th Apr 2019, 8:46 pm | #20 |
Diode
Join Date: Apr 2019
Location: Cambridge, Cambridgeshire, UK.
Posts: 6
|
Re: Mullard 3-3 dc voltages help
Good news!
I tried a new JJ EF806 valve and everything clicked into place straight away. With the correct 10M grid leak resistor the dc voltages now measure: C9 320 (310 in the Mullard book) C6 295 (290) C3 213 (210) EL84 anode 303 (290) EL84 screen 295 (290) EL84 cathode 29.2 (28) EF86 screen 28.1 (28) EF86 anode 20.3 (20) It will be interesting to see if the original Tungsram EF86 works in a conventional biasing arrangement. Perhaps Mullard valves in period had consistent grid leakage currents, but if anyone out there is building a Mullard 3-3, please be aware that an otherwise functional valve may not work in this design. I'll now do some tests on the audio performance. Paul. Last edited by pbb1234; 27th Apr 2019 at 8:47 pm. Reason: typo |