BBC AM8/2 and Mullard 3-3 circuits

Advertise with us
shop deals on amazon
Shop deals on ebay
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
G

Goodizzy

Active member
Joined
Sep 21, 2015
Messages
42
Location
Virginia, USA
Good afternoon,

I hope everyone is having a pleasant weekend.  I have been looking over this pair of circuits as I have a pair of Klangfilm 5K primary OTs for SE use. 

I intend to create a stereo version of the Mullard 3-3, but with a shared HT supply from a single PT and rectifier.  (I know this will introduce crosstalk but the purpose of the amp isn't for recording and I have all these parts onhand). This is to be used at home, in the library, to power bookshelf speakers at low-medium volume.

I understand the BBC used two copycat circuits of the Mullard 3-3 as shown in the asscociated TI for loudspeakers from the BBC Engineering site.  (http://www.bbceng.info/ti/eqpt/AM8_2.pdf)

They have the two circuits labeled Figs. 57 and 57A.  Both appear different than the original 3-3 circuit without tone controls.

The instruction differentiates between 57 and 57A as DC-coupled and AC-coupled, respectively.  Both circuits also employ RF-rejection networks, as outlined in the TI.  The 57 comprising the earlier serials and the 57A, the latter.

As I understand it from the TI regarding loudspeakers, these amps were used in the programme chain as reproduction units in offices, listening rooms and lesser positions in the studio not critical to recording /reproduction fidelity. According to the TI the programme chain is as follows:

Programme input -> Equaliser LS1A/1 -> LS volume control -> AM8/2 amp -> LS

The equaliser in the chain is designed to filter and better match the Goodmans 8" full-ranger drivers employed as loudspeakers.  The EQ employs a 1:3.16 transformer prior to the filter. 

I guess what I'm failing to understand is why the difference in versions as far as coupling to the output stage is concerned.  Is there an advantage between the two versions?  Why the switch to AC-coupling in the later serials?

Does the circuit, (at least on paper) with the RF rejection networks, perform better than the original 3-3 circuit?  Any other comments on the circuits and their implementation?
 
The original 3-3 uses a clever dc NFB system to set the EF86 screen grid potential from the EL84 cathode voltage. The advantage of this is the NFB operates down to dc sother are lo LF stability problems. the disadvantage is that the cathode is at quite a high voltage which limits output swing. The BBC ac coupled version requires a much lower cathode voltage giving the EL84 a greater output swing. I guess the BBC desired for non critical listening  applications a little more power and a little more distortion was OK.

By the way, I have some PCBs of the original 3-3  that I made a while ago but never got round to building. You are welcome to a couple for the price of postage.

Cheers

ian
 
Ian,

Thanks for the reply.  I see what you are saying regarding the output swing in the AC-coupled schematic.  It seems odd, considering the BBC's strict parameters for quality control on electronics that they would sacrifice fidelity for volume.  If it was in use in non-critical applications, however, I suppose this would be acceptable.  What type of distortion levels are we talking about here?

In regard to the DC-coupled version, (fig 57) does it compare favorably to the original 3-3?  I am tempted to build this over the original 3-3 due to its RF-rejection networks.  But considering the size and similarity between the two I suppose I could do a bit of comparative testing between circuits.

I appreciate the offer on the PCBs.  I will be building this pair P2P, as I have two Masco 2-valve intercom chassis from the 60s that the missess is keen about.  I just finished busting all the rivets and old components off them and so they will work nicely.  I will be providing the filament and HT from a separate PSU as to increase distance from the valves as well as practise Building a separate PSU, P2P, before starting on a few REDD.47 mic preamps later this month.
 
I think they end up essentially the same power. A 560 Ohm resistor moves from cathode to raw B+, giving nearly the same tube drop either way.

The 0-185 version seems over-clever to me. Maybe it was, or just someone got frumpy against cleverness. The 185-504 version must have been equally good (in BBC terms) since they made twice as many.

I also doubt any large change of distortion. Though I do wonder if the first version's driver may have strained to make even the modest drive a '84 needs; or was vulnerable to driver tube production tolerances.

> What type of distortion levels are we talking about here?

That's in the paper, as a limit-spec which covers both variants. The 3 Watt test level is conservative; as a guitar amp it would certainly claim 5 Watts, but not at low THD.
 
Thanks for the reply, PRR.  I must have misread the paper regarding the distortion levels as I thought it applied only to the DC-coupled variant.  Interesting point regarding the change from DC to AC coupling though.

I appreciate all the analysis gents.  I think I will actually tryout all three schematics with this circuit as the parts quantity is small and swapping components should be relatively painless.
 
PRR said:
I think they end up essentially the same power. A 560 Ohm resistor moves from cathode to raw B+, giving nearly the same tube drop either way.
Click to expand...
NIce catch = I missed that completely. The other thing is the original 3-3 design operated the EF86 at very low current/plate volts which gives it very high gain. To avoid loading the EF86 plate, it is direct couple to the EL84 which in turn means the EL84 cathode volts must be raised. Theyb did all this so they could get enough open loop gain to apply 20dB of NFB yet have an input sensitivity of 100mV for use with crystal cartridges. Presumably the BBC would have been using line level inputs so they could afford to lose some open loop gain.

Cheer

Ian
 
I'll be using the amps with a Garrard AT6 which I've refurbished.  I'm using a cheaper medium-quality stereo ceramic cartridge which outputs at 350mV.  This will be far too low to drive the AM8/2 proper, no?

It also appears it might be too high for the 100mV ceramic input the original 3-3 was designed with in mind.

Is there a simple and cheap solution to this?  I'm wondering if it is a worthwhile endeavour to try and improve the quality of replay with a ceramic cartridge system used to play thrift store beaters and copies.
 
The 8-2 appears to be almost identical to the 3-3. Its overall gain is approximately R13/R2 = 56. For one watt into 8 ohms we need 8V rms at the output so the input needs to be 8/56 V rms = 143mV. For four watts you need twice this so say about 300mV - not too far away from your cartridge sensitivity.

The 8-2A gain is similarly set by R6/R4 = 45. So for 1W at the speaker we need 8/45 =  177mV at the input and for four watts we need 355mV. Looks to me like your cartridge will be OK with the 2A design.

it also looks as though the BBC did not alter the sensitivity as much as I expected they would.

However, all the above is wrong because the BBC used 3 ohm speakers and a different output transformer ratio, the result being that the sensitivity remains 100mV input for 3W output as stated in the BBC docs.

So, it looks like however you build it you will likely have a 100mV sensitivity. The best solution I think is therefore to add a volume control at the input.  One version of the 3-3 had a 500K log pot but I think for a ceramic cartridge a 1M LOG would be OK.

Cheers

Ian
 
Ian,

Excellent, thank you for the explanation.  I will build both BBC types with the 1M pot and compare.

The Klangfilm OTs I have for this project have only 4 ohm secondaries.  I have several full-range 4 ohm speakers hanging around to compare as well.

I'm hoping I can finish the garden before the rain and finally get back to these things this afternoon!
 
ruffrecords said:
For one watt into 8 ohms we need 8V rms at the output.
Click to expand...
This is 8 Watt into 8 Ohms. For one watt into 8 ohms, you need sqrt( 8 )=2.8Vrms  ;)
 
> ceramic cartridge

Emphasizing a point Ian touched: you want HIGH input impedance on a crystal needle, or you won't get bass.

At least potentially high. If the cartridge is over-engineered and you can get the specified loading data, you may find it wants a certain Meg to get the 50Hz corner.

Over the decades, the "worth" of crystal/ceramic cartridges has been endlessly debated. I don't think I would worry about amplifier THD with 99% of the "crystals" I have seen. Or even EQ-- they can't easily give the 500Hz and 2KHz corners inside the cart, and rarely specify them externally. Many cheep needles seem to expect a cheap speaker with a response rise 1KHz-2KHz.

But I've listened to ALL sorts of phonographs, very refined to super-crude, and all can be fun.
 
This might be a bit late............

This article on the Mullard 3-3 gives construction details and performance figures and a useful circuit description.
Very little difference from the BBC version.  http://www.r-type.org/articles/art-003h.htm

Even though it will have a percent or two distortion it should sound sweet like a good quality valve radio, which should be ideal for your application.

Good luck
DaveP
 

Similar threads

A
Interesting resonant low-pass filter
Replies
1
Views
368
amageddian
A

Latest posts

Back
Top