Baby Huey EBL21

[Richardgr]

Following on from my ECL86 version of the Yves Monmagnon design 'Baby Huey', I decided to try something with Loctal tubes.

I have a big box of 7B6 tubes and a motley crew of EBL21s, and they correspond almost exactly with the triode and pentode parts of the ECL86 respectively.

I also had a pair of Sowter OPTs with 9K primaries. They only have an 8R output and do not have any UL taps, so a good match for the original concept which was designed to work in pentode mode.

It has more or less come together, but I am finding it tricky to tame it. Voltages and the CCS currents are close to the LTSpice simulation, but there are a couple of issues that I am struggling with.

1) The volume control is just a 50K log pot I had in my junk box (this was meant to be a project to use the bits I buy just because they might come in handy). But it is incredibly noisy to make any adjustment with it. I do not detect any DC on the wiper, so I am thinking it is grounding, or could it be the pot? I do not have a input capacitor, but I don't on the other amp I built so don't see that as an issue.

2) It oscillates with a staccato beat when I try to take some measurements, e.g. driver tube grid. What is the normal trigger for this type of oscillation? It can be set off if I turn the volume up, or above a certain voltage threshold on the power supply. Which leads me to ...

3) I have an antique power supply that last saw service in the cold war for the Swedish Defence. I have a variable 0-150V supply which I am using for the CCS, which is floating and linked to the variable 500V supply for the B+. It does take some time to warm up, and the low current draw of the CCS seems to be an issue sometimes with the low voltage supply - it gets stuck. Does anyone have any thoughts on this device? On the whole it seems to do its job OK. It recently relinquished its Mullard EL34s for some JJ ones.

How many valve devices use 'U' (UF80), 'P' (PL82) and 'E' (EL34) type tubes? I suppose some tellies might have done.

[Diabolical Artificer]

1) Try a different pot.
2) Your construction is less than ideal, lots of long leads and wire = stray L & C plus there is no screening. It could be your amp is acting like an antenna/theramin. See if if your hand etc interact with it. Check your grounds.
3) you can't beat a good old tfmr, bridge & a few caps for powering amps. Regulated supplies can do weird things.

Andy.

[Richardgr]

Hi Andy, thanks for the pointers.

1) I can eliminate the pot for testing. I suppose I could try cleaning it, it has probably been gathering dust for a few decades.

2) Yes, I wonder about that too. I have a small enclosure and tried various combinations of placement. I think I can clean it up a bit around the input/driver stage, which is likely the source of sensitivity.

I am worried that I have prioritised form over functionality!

The 7B6 has the triode perched high up inside so is not making it easy.

Plus, there are the two diodes in the bottle too. I have tied everything to the cathode, but perhaps they should be tied to ground and the cathode left floating?

I wanted to use the convenient central tab on the Loctal socket for a star ground, but the valves sit flush on the aluminium plate so the metal bases can be in contact with the top plate, which deterred me from that approach.

3) I shall have to dig into my spares and see what I can do about a temporary supply.

I noticed that I had not grounded the common lead on the speakers, so that is next on the list.

[Diabolical Artificer]

I think I can clean it up a bit around the input/driver stage, which is likely the source of sensitivity.

It's running at very high gain, the triode having hi mu/u. There's a lot of local FB too. As far as I can tell the two triodes have no filtering on the HT. This can cause funny goings on in my experience, whether it is in your case, no idea. Any idea what frequency it's oscillating at? you may be able to cure it with a LPF or step filter.

Form vs functionality is always a compromise in this sort of amp design, it doesn't look like layout should be an issue. Did the circle design on the old plate inspire you?

The cathode has to "float" else you'd be grounding the CCS. If you tie the diode anode's to ground, I would have thought you would be grounding the cathode too. Bit of a tricky one to think about DC current wise but from an AC POV I don't think it's a not a good idea. Tie em to the cathode as you've done.

Re grounding to the chassis I can't where it is but would guess near the RCA IP? Good luck, this is the fun bit : )

Andy.

[Richardgr]

Well some progress.
I had connected negative feedback from the loudspeaker positive, but had not grounded the common side of the speakers. Solving that fixed the stability issue. Also the chassis was not grounded properly.

Next up I gave the volume pot a squirt of CRC Electronic Cleaner and worked the control a few times. I was monitoring it with a resistance test, and it started off all over the place, and then settled down.

Andy pointed out the wiring was a bit of a mess, but I have tried to obey the main precautions for keeping noises at bay, and now it is actually silent on full volume with my test speakers.

I added a load to the negative rail on the valve power supply so that it stabilises better. Now I am starting to appreciate the power supply, even though it is a a complex and weighty beast. I swapped the 3 low usage Mullard EL34's with something more suitable for a guitar amp, so I don't feel quite so profligate when I am using it anymore.

The only issue now is that there is not a lot of bass in the sound, but I am not sure how much of that is the speakers. I am trying to use this amp as a test bed for some quantitative testing, using an oscilloscope and signal generator, and maybe REW and a sound card later, so probably a good job it is not perfect anyway ;-)

[Diabolical Artificer]

Glad your getting somewhere Richard, well done. Re LF response, your using decent OPT's though no idea of their primary inductance & no idea where the LF pole is, but there's no cathode bypass on U1/2 or on the CCS. As an experiment you could try a 47u/100u in parallel with R5/R11 and one in parallel with R19/D6, see if that improves the LF response. This will increase open loop gain though.

BTW, whats your negative PSU? I usually use a bridge rectified supply with a CRC filter using big caps, EG 220u/470u so it's rock solid: you need a good tfmr winding to supply this though capable of 100mA + Most negative bias supply's are pretty weedy capable of supplying only a few 10's mA. Overkill perhaps especially in this case as the OP stage is cathode biased.

Have you tried doing a frequency response test? You can do it with pencil & paper quite easily if you have a sig gen by printing off a graph, see attached for an example. For a quicky start at 10/20hz go up to 100/200hz then jump to 10khz & go up to 30khz. You need to do this really for both open loop & with NFB applied but in this case just the latter.

If you really want to put the amp through it's paces to test for unconditional stability do a no load test and a test with a capacitive load, EG 100n + 1u/2u. These are done at a very low OP power. Look at Patrick Turners old site, he has an article on there covering this in more detail. See - https://webarchive.nla.gov.au/awa/20....au/index.html

Andy.

[Richardgr]

Hi Andy, thanks, lots of grist to the mill there!

Hopefully have some time over the weekend for some tinkering. I have built the same circuit before, using more basic OPTs, but with UL taps, and that has been my daily listener. I know it is not as precise as my Tubelab SPP, but it just is pleasant to listen to, so hoping for the same result with this one. Finding a use for my stash of 7B6 valves is a bonus!

By the way, do you know of a source of good quality NOS Loctal sockets? The Chinese ones are not good for these valves, with their tufinol/cunifer non-flexible pins, the glass on the valve base by the pin has been known to chip a bit (threatening the vacuum), at least until the socket ages a bit and no longer has a death grip. I use wafer bases for the adaptors I make up for 5B/255M and they are quite good and reasonably cheap from Billingtons (because no one likes wafer bases ... but then I look at all those 1930's radios with paxolin wafer bases that soldier on without issues ...).

Cheers, Richard

[Diabolical Artificer]

Testing valve amps in depth can take ages but to be honest with low gain, low power amps or SE design's it's not worth it, unless you have an issue. It's gets interesting with 50w + amps though.

Re loctal sockets, with you there, I hate the new Chinese ceramic bases so try to buy NOS. There's loads on Ebay & Billington as you say. If you need help buying in the UK, drop me a PM. There's a few suppliers of ex Soviet stuff in Latvia, not bought stuff from there before, Ukraine yes, Latvia no. Why not ask on the forum?

Re the Tubelab SPP, is George still selling the PCB's? Andy.

[Richardgr]

George has seen a flurry of interest since announcing his intentions to wind up the firm. I bought a few boards recently, one of each type and some of his 'Unset' powerhead boards. Unlike a lot of my other future project ideas, they don't take up much space! I think he mentioned he had ordered a new batch of SPP boards to keep up with the rejuvenated interest.

Having them delivered to Sweden was a bit painful. It is not so much the MOMS (VAT) they add, but the 'administration' fee of £7.50, that is a single cost regardless of expense. I wish now I had bought a few more and sold them on here.

These are the bases I am after mainly, but the McMurdo ones are OK. The pip on the valve base is meant to indicate which way to lever the valve to break the grip on the spigot, so the bases have to be able to accommodate the sideways movement of the rigid pins. I have enough for my 'big project', Brimar 25P1 Amplifier, but it would be nice to be able to utilise the rest of my Loktal valves (eg 7C5 = 6V6). In my eyes they are the peak of Valve production, and from that point on it was all about cost reduction.

[low_imp]

I just did an upgrade of a DB110 Bogen amp with the EBL21 - which had 6V6GTA.
They are basically a half sized EL60, but with 11W Pa.
Gain of the valve is at least 2.5x higher than the ancient 6V6.

They don't red plate at 13W static and have no worry at well over rated voltage.
A matched pair therefore have Pa of 22W.....

EL60 is basically an early Loctal based EL34 but with 9 pins instead of 8 as per EF50.
Pa is 25W.

No info at all exists for fixed bias operation of the E/U BL21/71 valve but I went ahead and did it anyway.
The result is good because the valve was designed specifically to work at low anode voltages/high swing and the cathode is much larger than the 6V6, so needs 0.8A instead of 0.45. Large cathode = high perveance/high sensitivity.

FYI a PPP version of such an amplifier give SAME result as a pair of EL60 without the reliability problems of the EL60/34, as it appears the grids are better aligned ie, lower g2 emission/current.

In PPP 1.6A per valve pair would be needed, meaning a total of 3.2A, and reduction of A-A load to roughly half. (4-4.5K ohm a common enough transformer).
At 300-350V Va this would give roughly 30W at very low distortion.

The EL60 Vf 6.3 Volts / If 1.3 Ampere would need only heaters of 2.6A.

[low_imp]

Here are some tech details of the conversion.
The 7C5 was an option of course, but with the large changes to bias, decisions had to be taken.

As you can see the 6V6GTA Westinghouse large anode version was the ultimate 6V6, looking almost identical to the much more powerful 7591.
It certainly is better than the Sylvania made 7C5.


Sadly the pair of (LOCTAL) EBL output valves I have for this are not matched.
+
After faffing about with the original design of phase invertor which had had the grid ground line taken down to the cathode bias of the original POV kit, then converting the original 6V6 amp to fixed bias, some inbalance occured in the pi.

In the end after correcting the screen grid supply (pentodes take a lot more current) the inital bias of -6V had to be taken up to -8.16V. This meant one pentode runs 34.6m/a - the other 24.
That's a lot of difference considering the hot one is sitting at 11.5W and the weak one at 8. Official max Pa is 11W compared with the 6V6GTA of 14W.

In the end proper testing confirmed, in fact a rather disappointing outcome after levels had all been equalised.
The IMD figures are identical, and the power output the same. errk!
This was not supposed to happen.

Square wave reponse is very good however.

Some improvements seem to be had in FR and linearity, but more needs to be done here.
I suspect PSU is not up to keeping linearity at anything over 7-7.5V out which is only about 6W.
Driving the 15 ohm tap into 10 ohms, doesn't work so the ideal A-A is still 8-9K ohms.