Armstrong Jubilee Mark 2 tuner/amp - general advice

[cathoderay57]

Well, good news. It works really well, especially on FM with just a 1m length of wire stuck into the aerial socket. R4 LW is a bit hissy so I might try to peak up the RF coil a wee bit. I carried out voltage checks per the manual. All are pretty close to spec apart from anode voltage of V6a, the AF pre-amp, which measured initially 71v (supposed to be 130v). I changed R35 which was originally measuring 101k (should have been 82k) but that only brought the anode voltage up to 73.2v. R40 is very close to spec (7.1k v. 6.8k) That means the ECC85 is drawing 0.82 mA. If the anode voltage were correct at 130v it would only be drawing 0.68mA. I mentioned earlier that I had substituted a 1N4148 silicon diode in place of the o/c germanium OA81 for D2 as shown in the attachment. Therefore I think this might have made the negative bias supply slightly less negative. I checked the grid bias voltages using the Avo on 10v dc range, as per manual. V6b (ECC85) grid measures -0.6v (or -1.2v using DVM) which is correct according to the manual. The grid 1 measurements on the ECL82 pentodes measure -2.0v using the Avo (should be -3.2v) or -14v using the DVM. I have found a serviceable 0A85 in the spares box so might give that a try. As far as V6b is concerned, both anode currents tested 8.0 and 9.5mA on the VCM163 valve tester. Final point is that the cicruit shows R42 as 27k whereas in my set a 15k resistor is fitted. Thoughts? Jerry

[cathoderay57]

The feed to R45 (disappearing off the right hand side of the clip) comes directly from one side of the mains transformer HT secondary winding at about 205v AC. Jerry

[Gabe001]

"All are pretty close to spec apart from anode voltage of V6a, the AF pre-amp, which measured initially 71v"

I think you mean v6b (v6a is not shown in the attached snippet)

Assuming no leaky caps in that area and that the valve is good (it should be, you've tested it) I agree with you regarding increasing the negative bias. Looking at the ecc85 curves probably a further -0.2v or so may be sufficient - well within the margin of error of a multimeter.

Possibly remedied by changing r42 to its original value - I find negative voltage rails a bit confusing but I think larger value r42 would make that rail more negative. Happy to be corrected

[cathoderay57]

Hi Gabriel, thanks for the response. You were spot on, and yes, the ECC85 I should have referred to is V6b. Here were the measurements I took with the 1N4148 providing neg bias. I stopped bothering with the Avo and looked up the spec values for grid bias in the databook, and just used the DVM:

ECL82 grid1: -14.2v (databook spec -16.0v)
ECC85 grid: -1.23v (databook spec -2.0v)
ECC85 anode: 74.7v (manual spec 130v)

The ECC85 anode was 73.2v at the previous post and I hadn't changed anything but the mains voltage here fluctuates rather a lot so maybe it was that.

I then substituted an 0A85 for the 1N4148:

ECL82 grid1: -14.2v (databook spec -16.0v)
ECC85 grid: -1.24v (databook spec -2.0v)
ECC85 anode: 75.6v (manual spec 130v)

Clearly negligible difference. I then checked the two potential divider resistors on the ac input side of D2. R45 measured 122k (should be 100k) and R44 measured 7.6k (should be 6.8k). I just replaced R45 then powered up again:

ECL82 grid1: -15.6v (databook spec -16.0v)
ECC85 anode: 80.0v (manual spec 130v)

Getting there. Next I adopted Gabriel's suggestion and put in a 27k resistor for R42 (per circuit) in place of the 15k (as-found):

ECL82 grid1: -15.6v (databook spec -16.0v)
ECC85 grid: -2.1v (databook spec -2.0v)
ECC85 anode: 112.0v (manual spec 130v)

That's close enough so I'm quitting at this point. Next I'll clean up the chassis and have a think about what to do cabinet-wise...... Jerry

[Gabe001]

Glad I could help.
Attached is a suggestion for a cabinet (no affiliation with seller)

[cathoderay57]

Thanks for the tip. However, I've opted for an Ikea eket cabinet in white stained oak finish. I will just need to make a front panel for it. I'm going to use a couple of car speakers that I've had for a while which should handle the power, possibly plus a piezo tweeter. It won't have the vintage look but better than the box it came in, and more headroom for ventilation. I will post a pic when done. Cheers, Jerry

[cathoderay57]

Not quite there yet. A small but annoying fault is still present. After the set has been running on FM for about 5 minutes the frequency suddenly jumps up by about 300 kHz requiring a retune. I think I need to isolate whether the frequency shift is occurring in the mixer/oscillator (V1b) or the AFC valve (V6a). I could do with some advice as to the best way to do this. I cannot just pull V6 because the other half provides audio. Referring to the circuit clip in Post #12, I'm thinking of disconnecting one end of C14 and thereby running without AFC to see if the fault still occurs. Any better ideas welcomed. Jerry

[G6Tanuki]

In the past I have had issues with ceramic trimmers - the type that use silver plated faces as the capacity elements. These can become intermittent and cause some strange effects.

They sometimes reveal their problems if you give them a bit of a tap with a non conductive poking tool.

[cathoderay57]

Thanks. Well, there's C8 which would do it, but it's a beehive trimmer. I have known of these giving trouble so will have a careful prod tomorrow. The only thing is C8 is below decks and, since the problem seems fairly repeatable, it might be thermally related so that would tend to rule out the trimmer. I'll report back tomorrow. Cheers, Jerry

[cathoderay57]

As a precaution I cleaned the valve pins and sockets for both ECC85s but the fault was still there. Had a poke at C8 beehive trimmer and bingo, crackling and frequency shift. After a couple of turns back and forth the screw thread cleaned up and so far has been fault-free. C8 is described in the FM alignment procedure as a neutralising capacitor. Since the sensitivity and sound quality are excellent I won't be fiddling with the alignment. Many thanks for the tip G6Tanuki! Cheers, Jerry

[cathoderay57]

There was some audio distortion on a strong local FM station. On reflection, like Gabriel, I looked at the ECC85 characteristic and this time I realised that for a grid bias of -2.1v and a sub-100v anode voltage, the valve characteristic was non-linear and too close to cut-off so I changed R42 back to 15k (this could have been an official mod) and reduced the anode load resistor R35 from 82k to 47k. Seems better. Jerry

[cathoderay57]

Sorry to keep beating the drum about this set but there is still something I can't get my head around. The sound quality on FM is generally excellent. I have today realigned the FM section and the AFC now works much better - previously it had a tendency to drift. The ongoing issue is that for my local FM station the signal tends to be louder than the others to the extent that it leads to what sounds like audio distortion. I've attached a scan of the area where I think the problem lies which is the use of half an ECC85 as an audio pre-amp. Try as I might I cannot get the anode voltage up to the datasheet predicted value of 130v. It is around 70-80v when using the specified value 82k for R35. I have tested the ECC85 on the VCM163 and it yields the following results: Test settings Va=250, Vg=-2.3.
Triode #1: Ia=8mA (10mA new) gm=4.5mA/v (5.9 new)
Triode #2: Ia=9.5mA (10mA new) gm=5.5mA/v (5.9 new)
I have replaced the 2 electrolytics across the anode/cathode pins of the EB91 ratio detector. Please see attached sketch of the pre-amp circuit. As I have put in the notes in the attachment, if we accept the anode voltage of V6b is indeed supposed to be 130v (per service sheet table) that means the anode current is only 0.6mA. Looking at the transfer characteristic graph in the ECC85 datasheet on the Valve Museum website, it's apparent that under those biasing conditions, the valve is virtually cut-off and certainly at a non-linear operating point right at the bottom of the curve. Positive-going grid excursions are fine but negative-going moves into cut-off which seems to me is a recipe for distortion. I'm actually surprised that other FM stations are undistorted. I can't equate the bias condition (actually -2.1v measured at the grid using a DVM) being consistent with 130v on the anode. Edward commented in Post #17 that the biasing would need some attention. He wasn't kidding! I suppose the question boils down to what should I do next to improve things? Making the grid bias less negative thereby increasing the anode current and decreasing R35 to hold up the anode voltage seem the only options I can think of. Advice requested! Cheers, Jerry

[buggies]

An earlier sketch mentioned a difference in a resistor (82k vs 1M) and a 47 vs 150pf capacitor. Could it just be that there is too high an audio level fed to the valve?

[Gabe001]

Looking at the recommended operating parameters Ra of 82k (R32) seems to be oddly high. The datasheet quotes a value of 12k with a bias of -2v. How low have you gone?

The problem with improving the performance of the amplifier is that it may overdrive the cathodyne and actually make things worse on stronger signals. Maybe it's been deliberately crippled. How about a simpler solution like a100k grid stopper after c51?

[cathoderay57]

Both good suggestions, thanks. I've currently got a 47k in for Ra (R82).I'll also experiment with a counterintuitive approach by altering the fixed bias resistors so as to increase reverse bias until the anode voltage reaches 130v and see what happens then. Jerry

[Edward Huggins]

When I first heard these I was so impressed, even as a "budget" design (i.e. it wasn't a Leak or a Quad), it seemed to me to perfom as a genuine Hi Fi Receiver. I always pondered why they would use an ECC85 (in part) as an AF valve and for it to have consistently stable bias operating conditions......Do persist with this, as with an appropriate high-sensitivity speaker, you will be rewarded in the end.

[cathoderay57]

Thanks. First a correction. In post #35, R82 should read R35. I thought it was about time I put the scope on. Unfortunately I don't have an FM sig gen so I can't input a clean sinewave and will have to make do with a broadcast signal. I used my local FM station that has been causing the distortion. The first trace is at the grid coupling capacitor to V6b, the ECC85 pre-amp. The setting of the scope is 0.5v/div showing a signal output from the ratio detector/input to the ECC85 grid of just over 1v pk-pk. The waveform is symmetrical above and below the zero axis indicating that the output from the ratio detector is undistorted. The second trace is from the low voltage side of C51 the anode coupling capacitor. The scope scale is now 10v per div. The signal is no longer symmetrical and although negative peaks are going down to -30v the positive peaks struggle to exceed +20v. Since the amplifier inverts the signal, this appears to be consistent with the valve going into cut-off for high negative-going input peaks. Off now to play some more games with the biasing. Cheers, Jerry

[cathoderay57]

I've spent the afternoon trying various combinations. I found the best option was to fit a 27k for R42 and an 82k for R35. That gives a grid bias of -2.1v and an anode voltage of 111v. No other combinations produce less distortion. I then tried Gabriel's suggestion of a 100k grid stopper which would also reduce the audio level slightly as suggested by Buggies (note that the 82k vs 1M and a 47 vs 150pf capacitor only affect AM which doesn't distort so I left them alone). The grid stopper had minimal effect so I replaced the 100k with a 1M to effectively halve the signal at the grid. That did work, but of course reduced the volume from other FM stations with lower sig strength and also seemed to attenuate treble response. So in the end I think this is a live-with issue. I can totally eliminate the distortion by simply moving the aerial wire to reduce RF signal from the local station which is a workaround that works best for me. When I put the chassis into a cabinet I'll fit an internal dipole aerial which probably won't pick up as strong a signal as the current vertical wire aerial but, if it does, I can adopt a similar solution by rotating the cabinet for less signal. Cheers, Jerry

[cathoderay57]

I admit I have been a bit out of my depth with this issue. I have read up posts on this Forum to get a better idea of how an FM ratio detector works but I am no closer to understanding if this one is working as well as can be expected, or if there are further fixes or improvements that I can try. I should add that the distortion only occurs on FM, and then only on a very strong local station, and is most noticeable on bass notes. I also wonder whether the AGC is performing as it should. Therefore, to sum up, I have attached 2 images - the first is a clip of the ratio detector from the manufacturer's data, and secondly drawn as-found by me with differences highlighted in red. All of the differences appear to be original since the set does not appear to have ever had any components changed before - all wire wrapped through tag holes and no new solder. I have also added on the second jpeg notes of voltage measurements for AGC, AFC and audio. Any help welcomed. Cheers, Jerry

[Radio Wrangler]

Some comments on the ratio detector:

1) reduced bass.... all of the tuned circuits will have faster rise/fall times than needed for good bass, there are two longer time-constants in play, though.

2) Ratio detectors have a floating electrolytic capacitor which serves as a voltage reference for amplitude scaling their self-limiting action. If dodgy, this causes amplitude problems, poor limiting and distortion weighted more to the low freq end of audio. Worth changing for a known good one if you're doing anything in this area.

3) The other slow time constant is in the AFC network. If you think about it, the discriminator has to discriminate right down to DC frequency shifts in order to correct mistuning. Someone mistunes the receiver and leaves it alone. AFC is expected to not only correct this, but to keep on correcting it indefinitely. So there is a DC coupled discriminator output routed to the AFC - in this case a reactance valve. A reactance valve is a variable gain amplifier with capacitive feedback. As its gain is changed, the effective capacitance.... Miller capacitance varies, so it can pull the tuning of the receiver LO to make the receiver lock onto the signal. It's an example of a frequency locked loop (note, not phase locked)

IF the AFC path to the reactance valve isn't filtered to let only slow changes through, the AFC loop will try to track the audio FM and cancel it! This will happen most at low audio frequencies and the AFC will strip bass off.

It looks like the schematics circulating of this set either have a typo, where the AFC path is shown shorted to ground, or Armstrong got the AFC wrong and decided no AFC was better than what they had, so they really did add the short.


So my general advice would be to replace any drifted resistors, especially the output loads. This circuit relies on balance to try to cancel some effects, so give it the best chance. Replace the ratio detector electrolytic with a good low leakage part. and then look at the route to the reactance valve from the ratio detector and eyeball the capacitor which should be slowing the response, and check things carefully in this area.

For a bit of experimentation, you could put a DC voltmeter on the ratio detector DC output and move a sig gen across the tuned frequency. You should get it move away from zero as you get closer, peak and then steadily ramp across to a peak of the opposite polarity before fading back to zero.

If you stop moving the sig gen, the DC voltage should stay where it was, plus or minus a bit of drifting.

Hope this helps

David