Ec Tremolux design musings

[Justin Thomas]

I'm wondering why such a Binky little amp needs a built-in attenuator?

Sissies...

Justin

[olddawg]

I'm wondering why such a Binky little amp needs a built-in attenuator?

Sissies...

Justin

I use an attenuator on a 20 watt amp nearly every gig. I would rather dime a small amp than run a big amp on 2 or have nothing but preamp overdrive! But yeah the club owners, crowds, and ordinances are all sissies nowadays!

[Dave H]

My notes have a 10uF capacitor in parallel with the diode to ground and you have 1uF. I also have 100uF for the filter capacitor and you have 10uF. My version will charge up much faster and it will rival the ripple of your version.

Were the 0.1u caps also changed to 1u in your version? That would make both versions produce the same bias voltage.

The time to charge up is the number of mains cycles it takes to pump up the large capacitor from the smaller ones. Your version is able to transfer 10x the charge each cycle but the output cap is also 10x the value so the time to fully charge should be be the same.

[66 Kicks]

Were the 0.1u caps also changed to 1u in your version? That would make both versions produce the same bias voltage.

The time to charge up is the number of mains cycles it takes to pump up the large capacitor from the smaller ones. Your version is able to transfer 10x the charge each cycle but the output cap is also 10x the value so the time to fully charge should be be the same.

The one that I had built and analyzed was on a 461V power supply and it put out -79V with a 20mV peak ripple. It used a single 1.0uF input capacitor with a 10uF capacitor parallel to the diode to ground and a 100uF filter capacitor. It charged up to -55V in less than one second.

My version was for a 6L6 amp, but the topology is the same as the EC version and the principles are the same. It could put out around -39V if you ran two 1.0uF capacitors in series. I may have misstated the comparisons because I never actually built the exact EC version and I was just making an educated guess. The charge up time probably is the same for both versions. To me, the only saving grace for this bias is the rapid charge rate.

The upshot from all of this is that tubeswell could get more voltage out of the EC bias by replacing one of the 0.1uf capacitors with a 0.22uf capacitor.

[Dave H]

The one that I had built and analyzed was on a 461V power supply and it put out -79V with a 20mV peak ripple. It used a single 1.0uF input capacitor with a 10uF capacitor parallel to the diode to ground and a 100uF filter capacitor. It charged up to -55V in less than one second.

Here's a sim of your circuit.

[Dave H]

And here's a sim of Tubeswell's bias supply with one of the 0.1u caps changed to 0.2u.

[tubeswell]

And here's a sim of Tubeswell's bias supply with one of the 0.1u caps changed to 0.2u.

[ATTACH=CONFIG]42323[/ATTACH]

I'm honoured Dave, but I think it would need to be called 'Dave H's' bias supply ;-)

[loudthud]

What happens to the bias Voltage when the Standby switch is open? After the rectifier tube warms up, there will be an offset sine wave on the HV secondary.

[Daver]

So they skimped on the bias supply but popped for the expensive and overpriced Sprague Atom filter caps? Weird....

[tubeswell]

What happens to the bias Voltage when the Standby switch is open? After the rectifier tube warms up, there will be an offset sine wave on the HV secondary.

With the standby switch off, the bias voltage virtually disappears altogether - except for a couple of mV bouncing around on the meter.

[Gingertube]

Getting to this late - some ravings:
On the "funny sub-harmonic distortion" when dimed I would suggest:
1k5 grid stops are way too low for 6V6 I like to see 10K minimum for 6V6 (The value required is INVERSELY dependent on tube gm, 6V6 gm is 1/3 of a EL34 so would want to see 3 x the 1k5 as a minimum).
PLUS:
The concertina splitter does "weird s..t" when the anode and cathode loads become unbalanced which happens with a dimed amp when the output tubes go into "forward rectification" grid current on signal peaks.
For a concertina splitter - output tube grid stops need to be much higher again (to limit output tube grid current and hence limit upset of the concertina splitter) - 39K or even 47K, I've seen up to 100K used.

The Wiz said this about it:
The Valve Wizard

On a vintage OZ 60W Goldentone with a cathodyne/concertina splitter and 6DQ6 output tubes I took the original output tube grid stops from 10K up to 47K with good results on the overdriven sound.

The Wiz also advocates a large gridstop on the concertina triode grid (470K).
I also (some years back) had some email conversation with Kevin O'Connor about that. He uses Concertina Splitters in all his commercial amps. He said that he definitely agreed with pushing up the output tube grid stops but had never found it necessary to use such large grid stops on the concertina triode grid, but did use gridstops there. I inferred (was not specifically stated) that he used something like 20 to 30K o the concertina triode grid.

On the bias issue:
No bias when switching out of standby also happens with cathode biased output stages. You get aLARGE pulse of current thru' teh output tubes until the cathode bypass cap(s) charge up. I had a own design and build cathode biased EL34 amp (a Trainwreck Clone) which regularly (every couple of weeks) would take out the B+ fuse. I eventually eliminated the standby switch to stop that happening.

Cheers,
Ian

[Leo_Gnardo]

On the "funny sub-harmonic distortion" when dimed I would suggest:
1k5 grid stops are way too low for 6V6 I like to see 10K minimum for 6V6 (The value required is INVERSELY dependent on tube gm, 6V6 gm is 1/3 of a EL34 so would want to see 3 x the 1k5 as a minimum).
PLUS:
The concertina splitter does "weird s..t" when the anode and cathode loads become unbalanced which happens with a dimed amp when the output tubes go into "forward rectification" grid current on signal peaks.
For a concertina splitter - output tube grid stops need to be much higher again (to limit output tube grid current and hence limit upset of the concertina splitter) - 39K or even 47K, I've seen up to 100K used.

I take it the usual situation found in old Fender Princetons could be improved by this advice, no grid stoppers at all on the 6V6's. Often I'll add 1K5 carbon comp, looks like a higher value is in order.

[tubeswell]

Getting to this late - some ravings:
On the "funny sub-harmonic distortion" when dimed I would suggest:
1k5 grid stops are way too low for 6V6 I like to see 10K minimum for 6V6 (The value required is INVERSELY dependent on tube gm, 6V6 gm is 1/3 of a EL34 so would want to see 3 x the 1k5 as a minimum).
PLUS:
The concertina splitter does "weird s..t" when the anode and cathode loads become unbalanced which happens with a dimed amp when the output tubes go into "forward rectification" grid current on signal peaks.
For a concertina splitter - output tube grid stops need to be much higher again (to limit output tube grid current and hence limit upset of the concertina splitter) - 39K or even 47K, I've seen up to 100K used.

The Wiz said this about it:
The Valve Wizard

On a vintage OZ 60W Goldentone with a cathodyne/concertina splitter and 6DQ6 output tubes I took the original output tube grid stops from 10K up to 47K with good results on the overdriven sound.

The Wiz also advocates a large gridstop on the concertina triode grid (470K).
...
Cheers,
Ian

Yes I have all of Merlin's books and have read (and still regularly consult) them numerous times, and am familiar with his account of different types of cathodyne distortion when overdriven and what this does to the output stage bias etc and how to remedy it.

It had already occurred to me that this may well have been what contributed to the uneven plate voltage and tube current readings I got on V4 and V5 - before I re-tubed and re-biased the amp. V5 was idling way too hot. I suspect the screen in that tube had previously taken one or more big hits (along with red-plating) when the gain-spiked cathodyne side clipped the too-hotly-biased 6V6 grid. The other tube (V4) was idling quite high (for fixed bias) as well. See pics attached. Even when I dialled in maximum (-38V) bias voltage, I could only get V5 to idle at 418V@27mA (i.e. after deducting Ig2~2mA, = 10.5W).

Putting JJ6V6S in allowed me to get both new tubes idling close to 76%. I think the B+ voltage is probably too high for the available (~38V) bias range in the stock amp to be able to run fixed-bias 12W tubes practically.

[g1]

Is this perhaps one of those issues where different brands of 5Y3 are leading to problematic B+ levels?
I'm a bit surprised it's running that much higher B+ than the schematic shows considering it's a modern unit.

[tubeswell]

Don't think so, I tried NOS RCA 5Y3GT (that I know is bonafide) and the B+ voltage was 10V higher.